bump version to 1.0.7

certs, html and acme are shared across nginx-proxy, acme-companion and
the acme-ip service, so they become named volumes (dropping the nginx-
prefix). The vhost.d mount is removed: acme-companion only needs it with
ACME_HTTP_CHALLENGE_LOCATION=true or letsencrypt_user_data, and users
who customize nginx config can mount their own. ./data stays a bind
mount so the SQLite state stays directly backup-able. README updated.

.dockerignore

@@ -28,5 +28,9 @@ LICENSE
 README.md
 data/
 cert/
+nginx-certs/
+nginx-vhost.d/
+nginx-html/
+nginx-acme/
 **/target/
 .idea/

.github/workflows/docker.yml

@@ -12,12 +12,35 @@ env:
   DOCKERHUB_REPO: yoursfunny/telegram-twitter-media-bot
 
 jobs:
+  # A tag push and a branch push to the same commit fire two workflow runs;
+  # build only once. Tag runs always build; master runs build only when the
+  # pushed commit is not already tagged (the tag run covers it).
+  should-build:
+    runs-on: ubuntu-latest
+    outputs:
+      build: ${{ steps.check.outputs.build }}
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          fetch-depth: 0
+      - id: check
+        shell: bash
+        run: |
+          if [ "$GITHUB_REF_TYPE" = "branch" ] && git tag --points-at "$GITHUB_SHA" | grep -q .; then
+            echo "commit already tagged; the tag run builds the image"
+            echo "build=false" >> "$GITHUB_OUTPUT"
+          else
+            echo "build=true" >> "$GITHUB_OUTPUT"
+          fi
+
   docker:
+    needs: should-build
+    if: needs.should-build.outputs.build == 'true'
     runs-on: ubuntu-latest
     steps:
       - name: Docker meta
         id: meta
-        uses: docker/metadata-action@v5
+        uses: docker/metadata-action@v6
         with:
           images: ${{ env.DOCKERHUB_REPO }}
           tags: |
@@ -28,22 +51,30 @@ jobs:
             type=sha
       -
         name: Set up QEMU
-        uses: docker/setup-qemu-action@v3
+        uses: docker/setup-qemu-action@v4
       -
         name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
+        uses: docker/setup-buildx-action@v4
       -
         name: Login to Docker Hub
-        uses: docker/login-action@v3
+        uses: docker/login-action@v4
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
+      # Buildkit cache via the GitHub Actions cache backend (uses the
+      # automatic GITHUB_TOKEN, no extra secrets). mode=max keeps every
+      # stage's layers so the cargo-deps and ffmpeg layers are restored
+      # instead of re-downloaded/recompiled. The scope must be pinned to a
+      # fixed string: the gha backend defaults to the current git ref, which
+      # would give every new tag a cold cache on release builds.
       -
         name: Build and push
-        uses: docker/build-push-action@v5
+        uses: docker/build-push-action@v7
         with:
           push: true
           build-args: |
             APP_NAME=${{ env.APP_NAME }}
           tags: ${{ steps.meta.outputs.tags }}
-          labels: ${{ steps.meta.outputs.labels }}
+          labels: ${{ steps.meta.outputs.labels }}
+          cache-from: type=gha,scope=tgxmb-build
+          cache-to: type=gha,mode=max,scope=tgxmb-build

.gitignore

@@ -2,6 +2,10 @@
 __pycache__/
 cert/
 data/
+nginx-certs/
+nginx-vhost.d/
+nginx-html/
+nginx-acme/
 docker-compose.yml
 
 .env

AGENTS.md

@@ -0,0 +1,99 @@
+# Repository Guidelines
+
+## Project Overview
+
+Telegram bot (teloxide) that turns post links from X/Twitter, Pixiv, and Bluesky into media messages (images, video, GIF) with the post's title, author, and tags. It supports batch media splitting, retry with persistence, inline queries, forward-channel rebinding with caption templates, and Pixiv ugoira→MP4 transcoding. README and user-facing strings are in Chinese. The project is a Rust port of a Python predecessor (see `queue.rs` comments referencing `utils/task_queue.py`).
+
+Two-crate Cargo workspace (both v1.0.3, edition 2024, resolver 3):
+
+- **`crates/x-media`** — library that fetches and normalizes media from the three sites. Pure, no Telegram knowledge.
+- **`crates/xmedia-bot`** — the bot binary: teloxide dispatcher, SQLite-backed chat state, persistent task queue.
+
+## Architecture & Data Flow
+
+```
+Telegram update → Dispatcher (polling or axum webhook) → dptree branches
+  ├─ message    → commands (any chat) / URL links (private chat only)
+  ├─ inline_query → InlineQueryResult Photo/Video/Mpeg4Gif
+  └─ callback_query → "forward" (copy to channel) / "template|<name>" (apply caption template)
+```
+
+Message flow: `message_handler` extracts URLs (from `url`/`text_link` entities, text + caption, deduped) → `x_media::site::fetch(url)` → `Fetched` → builds a `Task` → `send::send_media_sequence` (media groups ≤ 9, caption on first item) or `send::send_animation`. On Telegram URL-fetch failure or size error (`send_batch_via_upload`): download via `x_media::site::download_media` to a temp file (≤ 10 MiB), sniff magic bytes (`sniff_ext`), upload via multipart; oversized items fall back to `fallback_url`. On failure: `enqueue_retry` persists resume-state `Task` into the SQLite queue → single worker leases (120 s lock TTL) → retry with exponential backoff (≤ 30 s, `MAX_RETRIES = 2`) → dead-letter → `notify_failure`. Success → `post_send_actions`: edit-before-forward prompt with inline buttons, or `copy_messages` to the bound forward channel.
+
+The `x-media` library: `site::fetch(url)` dispatches (in order) twitter → bsky → pixiv via per-site regex `PATTERN` and returns `Ok(None)` for unmatched URLs. `Fetched { source_url, caption, title, media: Vec<Media>, sensitive, … }`; `caption_with(format)` substitutes `{url} {author} {author_url} {title} {tags}`.
+
+## Key Directories
+
+| Path | Purpose |
+|---|---|
+| `crates/x-media/src/` | Fetch library. `site/mod.rs` = dispatcher + `Fetched`/`FetchError`/`download_media`/`media_size`; `media.rs` = `Media` enum; `examples/fetch.rs` = end-to-end usage sample |
+| `crates/x-media/src/site/<twitter\|pixiv\|bsky>/` | One directory per site: `mod.rs` (re-exports), `interface.rs` (PATTERN, `enabled()`, `fetch_from_url()`, site struct, `From<SiteStruct> for Fetched`), `model.rs` (serde DTOs). Pixiv adds `api.rs` (auth + transport); twitter adds `auth.rs` (logged-in GraphQL `TweetDetail` fallback for NSFW tweets, gated on `TWITTER_AUTH_TOKEN`) |
+| `crates/xmedia-bot/src/main.rs` | Entry point: env/log init, queue worker start, pixiv validation, 300 s edit-expiry sweep, dptree handler tree, webhook vs polling dispatch |
+| `crates/xmedia-bot/src/config.rs` | Manual env parsing into `Config` |
+| `crates/xmedia-bot/src/handlers.rs` | `Command` enum (teloxide `BotCommands`), message/inline/callback handlers, URL extraction, global statics; per-URL work spawned with a `Semaphore(8)` cap (teloxide's per-chat workers are sequential — batch-forwards need concurrency) |
+| `crates/xmedia-bot/src/state.rs` | `ChatStore`: parking_lot `Mutex<HashMap>` cache + SQLite write-through (`chat_state` table) |
+| `crates/xmedia-bot/src/link_cache.rs` | `LinkCache`: SQLite-backed cache (`link_cache` table) of successfully sent posts — raw caption fields + Telegram `file_id`s; repeat links re-send locally (no fetch/upload), TTL + prune, invalidated on permanent send failure |
+| `crates/xmedia-bot/src/queue.rs` | `PersistentTaskQueue`: SQLite-backed queue (`tasks` table), `QUEUE_WORKERS = 4` concurrent workers (lease via `BEGIN IMMEDIATE` + `locked_until` TTL), retry→dead-letter, `Notify::notify_waiters` wakeup, `busy_timeout` on all connections |
+| `crates/xmedia-bot/src/send.rs` | Media senders, upload fallback, error classification, queue task handlers |
+
+## Development Commands
+
+```bash
+export TELOXIDE_TOKEN=<token>          # required; PIXIV_REFRESH_TOKEN optional (Pixiv disabled without it)
+cargo run -p xmedia-bot                # run the bot (polling by default)
+cargo run -p x-media --example fetch -- <url>   # test a link through the fetch library
+cargo test --workspace                 # full test suite (no CI test step exists — run locally)
+cargo build --release -p xmedia-bot    # release build (Dockerfile does this)
+cargo clippy --workspace --all-targets # lint (Clippy is the configured IDE linter)
+cargo fmt --check                      # formatting
+```
+
+Docker: `docker build -t tgxmb .` then `docker run --rm -d --name tgxmb --env-file .env -v ./data:/app/data tgxmb`. Runtime requires **ffmpeg** (built into the image).
+
+## Code Conventions & Common Patterns
+
+- **No anyhow/thiserror.** Errors are hand-rolled enums with manual `Display`/`source()`/`From` impls: `QueueError` (`Retryable { delay_seconds, payload }` / `Permanent`), `SendError` (Retryable/Permanent), `FetchError` (`Http`/`Json`/`Pixiv`/`NotFound`/`Blocked`), `PixivError`, `Classification`. New errors should follow this pattern.
+- **Global state via `std::sync::LazyLock` statics**, not DI: `CONFIG`, `CHAT_STORE`, `TASK_QUEUE` in `handlers.rs`; shared reqwest `CLIENT` in `x-media/src/site/mod.rs`. `Bot` is passed/cloned into handlers; queue workers rebuild `Bot::from_env()`.
+- **Async**: tokio multi-thread runtime (`#[tokio::main]` default). All rusqlite I/O inside `tokio::task::spawn_blocking`. Long loops use `tokio::select!` with `tokio::sync::{watch, Notify}` stop/wake channels. No streams.
+- **Blocking sync primitives**: `parking_lot::Mutex` for hot caches, `tokio::sync::Mutex` for async-shared state (pixiv token cache), `AtomicBool` for feature gates.
+- **Site adapter convention** (no trait, no enum dispatch — follow the existing convention): each site module exports `PATTERN: LazyLock<Regex>`, `enabled() -> bool`, `fetch_from_url(url) -> Result<Fetched, FetchError>`; `site/mod.rs` re-exports the site struct and `fetch_once` adds one guarded if-branch. Adding a site = new `site/<name>/{mod.rs,interface.rs,model.rs}` + one branch in `fetch_once`.
+- **Serde**: per-site `model.rs` are pure `Deserialize` DTOs mirroring API JSON; site structs in `interface.rs` have private fields, a `caption()` builder, and `impl From<SiteStruct> for Fetched`. Persisted payloads use internally-tagged enums (`#[serde(tag = "kind")]` / `type`).
+- **Naming**: module-per-concern, snake_case files, `CamelCase` types, `snake_case` fns. `//!` module docs and `///` docs on non-obvious logic (syndication token, ugoira encoding, `display_text_range`).
+- **Retries**: only `x-media::site::fetch` retries (3 attempts, `1 << attempt` backoff, HTTP errors only). Queue retries are explicit `QueueError::Retryable` with computed delay (`retry_delay_seconds`).
+- Logging via `log` macros (`pretty_env_logger`, level from `RUST_LOG`).
+
+## Important Files
+
+| File | Why it matters |
+|---|---|
+| `crates/xmedia-bot/src/main.rs` | Startup sequence, webhook vs polling, graceful shutdown (SIGINT via teloxide ctrlc / SIGTERM via `stop_token` for docker, → sweep stop → admin msg → queue stop) |
+| `crates/xmedia-bot/src/handlers.rs` | `CHAT_STORE`/`TASK_QUEUE`/`CONFIG` singletons (open `data/task_queue.db` **relative to CWD**); command dispatch; URL extraction; retry enqueue |
+| `crates/xmedia-bot/src/send.rs` | Constants `MAX_MEDIA_GROUP = 9`; fallback chain; `classify_request_error`; download-and-reupload fallback triggered only by Telegram API errors (`is_media_fetch_failure` / `is_size_error`) |
+| `crates/xmedia-bot/src/photo.rs` | Pure-Rust photo processing (no ffmpeg): `png` (image-png) decode/encode + `zune-jpeg` decode + `fast_image_resize` Lanczos3 downscale + `jpeg-encoder`. Photos over Telegram's limits (width + height > 10000 px → `PHOTO_INVALID_DIMENSIONS`; bytes > 10 MiB) are decoded, downscaled keeping the format, PNG bit depth > 24 (RGBA 32-bit / 16-bit per channel) reduced to 24-bit RGB with alpha flattened white (≤24-bit untouched, never upconverted), and transcoded to JPEG only if still over the cap; memory budget guarded, otherwise the item's smaller fallback URL |
+| `crates/x-media/src/site/mod.rs` | Dispatcher, `Fetched`/`FetchError`, shared `CLIENT`, `download_media` (adds `Referer: https://www.pixiv.net/` for `pximg.net` hotlink protection) |
+| `crates/x-media/src/site/pixiv/api.rs` | OAuth token exchange (hardcoded app client id/secret), access-token cache, ugoira zip→MP4 via ffmpeg in `spawn_blocking` |
+| `Dockerfile` | Multi-stage: cached dep layer via stub sources + `touch *.rs` mtime hack, static ffmpeg from ffmpeg.martin-riedl.de (`FFMPEG_URL` arg, `unzip -t` integrity check), `debian:bookworm-slim` runtime, entrypoint |
+| `docker-entrypoint.sh` | Privilege drop: `useradd` with `LOCAL_USER_ID` (default 9001) + `setpriv` (no gosu on bookworm-slim) |
+| `docker-compose.yml.example` | Deployment env reference (real `docker-compose.yml` is gitignored). Ships nginx-proxy + acme-companion: webhook mode needs TLS termination in front (teloxide's axum listener is HTTP-only; `WEBHOOK_CERT` only feeds `set_webhook`), bot exposes `VIRTUAL_HOST`/`VIRTUAL_PORT` on the shared `proxy` network, no host port; container names `nginx-proxy`/`acme-companion`/`tgxmb`, start order via `depends_on` (proxy → acme → bot) |
+| `.github/workflows/docker.yml` | CI: build+push to Docker Hub on tag `v*`/master; **no test step**; buildx gha cache (`cache-from`/`cache-to`, scope `tgxmb-build`, `mode=max`) so cargo deps + ffmpeg layers are restored across runs |
+| `README.md` | Feature docs + command table (Chinese) |
+
+## Runtime/Tooling Preferences
+
+- **Rust, stable, edition 2024**, workspace resolver 3. No `rust-version`/MSRV pin, no `rust-toolchain.toml` — recent stable is assumed. No nightly features.
+- Package manager: **Cargo** (workspace with path dep `x-media` ← `xmedia-bot`). No `[workspace.package]`/shared deps — each crate lists deps independently.
+- **Two reqwest versions coexist in the lock** (0.12.28 via teloxide, 0.13.3 in x-media) — don't unify casually.
+- Config is **environment-variable driven** (dotenv loads `.env`, gitignored; no `.env.example` exists). Key vars: `TELOXIDE_TOKEN` (required), `PIXIV_REFRESH_TOKEN`, `TWITTER_AUTH_TOKEN` (optional; x.com `auth_token` cookie — enables the logged-in GraphQL fallback that fetches NSFW tweets syndication withholds), `BOT_ADMIN` (comma-separated ids), `EDIT_MESSAGE_TTL_SECONDS` (default 86400), `LINK_CACHE_TTL_SECONDS` (default 604800), `WEBHOOK`/`WEBHOOK_URL`/`WEBHOOK_LISTEN`/`WEBHOOK_PORT`/`WEBHOOK_CERT`/`WEBHOOK_SECRET_TOKEN` (webhook mode requires URL/listen/port, `.expect`ed; `WEBHOOK_CERT` is Telegram-facing self-signed validation only — TLS must be terminated by a reverse proxy), `RUST_LOG`, `TELOXIDE_PROXY`, `LOCAL_USER_ID` (entrypoint only).
+- SQLite via `rusqlite` with `bundled` feature (no system libsqlite needed). DB file `data/task_queue.db` is CWD-relative — run from the workspace root, or `/app` in Docker. Mount `./data` and `./cert` volumes.
+- `.gitattributes` enforces LF for `*.sh` (CRLF breaks shebangs in containers). `.gitignore`: `.env`, `data/`, `cert/`, `docker-compose.yml`, `/target`, `.idea/`.
+- Docs are in Chinese; user-facing bot strings too. Keep that convention when editing captions/templates/docs.
+
+## Testing & QA
+
+- **~51 tests, all inline `#[cfg(test)] mod tests`** — no `tests/` integration directories. Framework: built-in Rust test + `#[tokio::test]` (dev-deps only in `x-media`: tokio macros/rt-multi-thread, dotenv).
+- No mocking framework anywhere (no mockito/wiremock/mockall). Conventions: pure-function units (regex parsing, serde round-trips, chunking, retry math) tested synchronously; async tests use real dependencies — file-backed SQLite via `tempfile` (`queue.rs::new_queue()` helper), live network fetches.
+- Live-network tests exist in `site/twitter/interface.rs` (3), `site/bsky/interface.rs` (2), `site/pixiv/interface.rs`/`api.rs` (env-gated on `PIXIV_REFRESH_TOKEN`/dotenv, skip by early return). Run the full suite with `cargo test --workspace`.
+- Fixtures are inline `serde_json::json!` builder fns (`fixture()`, `thread_json()`, `illust_json()`), not files. The shared `CLIENT` sets `pool_max_idle_per_host(0)` under `#[cfg(test)]` to avoid cross-runtime `DispatchGone`.
+- **CI runs no tests** — `.github/workflows/docker.yml` only builds/pushes the image; verification is a local responsibility.
+- Untested and hard to test without a mock seam: `handlers.rs` (depends directly on teloxide `Bot`); `main.rs`, `config.rs`, `state.rs`; `media.rs`, `lib.rs`, all `model.rs`.
+- No coverage tracking, no lint gate in CI.

Cargo.lock

@@ -505,6 +505,15 @@ dependencies = [
  "syn",
 ]
 
+[[package]]
+name = "document-features"
+version = "0.2.12"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d4b8a88685455ed29a21542a33abd9cb6510b6b129abadabdcef0f4c55bc8f61"
+dependencies = [
+ "litrs",
+]
+
 [[package]]
 name = "dotenv"
 version = "0.15.0"
@@ -599,12 +608,33 @@ version = "0.1.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7360491ce676a36bf9bb3c56c1aa791658183a54d2744120f27285738d90465a"
 
+[[package]]
+name = "fast_image_resize"
+version = "6.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e9c50201dc184ba6553da1695aac20a042efffbe2d84542cee31917c86c3ab1e"
+dependencies = [
+ "cfg-if",
+ "document-features",
+ "num-traits",
+ "thiserror",
+]
+
 [[package]]
 name = "fastrand"
 version = "2.4.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9f1f227452a390804cdb637b74a86990f2a7d7ba4b7d5693aac9b4dd6defd8d6"
 
+[[package]]
+name = "fdeflate"
+version = "0.3.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1e6853b52649d4ac5c0bd02320cddc5ba956bdb407c4b75a2c6b75bf51500f8c"
+dependencies = [
+ "simd-adler32",
+]
+
 [[package]]
 name = "find-msvc-tools"
 version = "0.1.9"
@@ -1305,6 +1335,12 @@ dependencies = [
  "libc",
 ]
 
+[[package]]
+name = "jpeg-encoder"
+version = "0.7.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a0370574b86f7eca156b9f298392b5e69a23f8c86f3f865add60bbc2e79467a6"
+
 [[package]]
 name = "js-sys"
 version = "0.3.98"
@@ -1352,6 +1388,12 @@ version = "0.8.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "92daf443525c4cce67b150400bc2316076100ce0b3686209eb8cf3c31612e6f0"
 
+[[package]]
+name = "litrs"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "11d3d7f243d5c5a8b9bb5d6dd2b1602c0cb0b9db1621bafc7ed66e35ff9fe092"
+
 [[package]]
 name = "lock_api"
 version = "0.4.14"
@@ -1604,6 +1646,19 @@ version = "0.3.33"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "19f132c84eca552bf34cab8ec81f1c1dcc229b811638f9d283dceabe58c5569e"
 
+[[package]]
+name = "png"
+version = "0.18.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "60769b8b31b2a9f263dae2776c37b1b28ae246943cf719eb6946a1db05128a61"
+dependencies = [
+ "bitflags",
+ "crc32fast",
+ "fdeflate",
+ "flate2",
+ "miniz_oxide",
+]
+
 [[package]]
 name = "potential_utf"
 version = "0.1.5"
@@ -3296,12 +3351,13 @@ checksum = "1ffae5123b2d3fc086436f8834ae3ab053a283cfac8fe0a0b8eaae044768a4c4"
 
 [[package]]
 name = "x-media"
-version = "1.0.2"
+version = "1.0.7"
 dependencies = [
  "bytes",
  "dotenv",
  "html-escape",
  "log",
+ "rand 0.8.6",
  "regex",
  "reqwest 0.13.3",
  "serde",
@@ -3314,12 +3370,15 @@ dependencies = [
 
 [[package]]
 name = "xmedia-bot"
-version = "1.0.2"
+version = "1.0.7"
 dependencies = [
  "dotenv",
+ "fast_image_resize",
  "html-escape",
+ "jpeg-encoder",
  "log",
  "parking_lot",
+ "png",
  "pretty_env_logger",
  "rand 0.8.6",
  "regex",
@@ -3331,6 +3390,7 @@ dependencies = [
  "tokio",
  "url",
  "x-media",
+ "zune-jpeg",
 ]
 
 [[package]]
@@ -3534,3 +3594,18 @@ dependencies = [
  "cc",
  "pkg-config",
 ]
+
+[[package]]
+name = "zune-core"
+version = "0.5.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cb8a0807f7c01457d0379ba880ba6322660448ddebc890ce29bb64da71fb40f9"
+
+[[package]]
+name = "zune-jpeg"
+version = "0.5.15"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "27bc9d5b815bc103f142aa054f561d9187d191692ec7c2d1e2b4737f8dbd7296"
+dependencies = [
+ "zune-core",
+]

Dockerfile

@@ -1,12 +1,16 @@
 # ---------- build stage ----------
 # rust:1-bookworm (full, not slim) ships the C toolchain needed by
-# rusqlite's bundled SQLite, plus wget/xz for the ffmpeg download.
+# rusqlite's bundled SQLite, plus wget/unzip for the ffmpeg download.
 FROM rust:1-bookworm AS builder
 
 ARG APP_NAME=telegram-twitter-media-bot
-# Statically compiled ffmpeg (ugoira MP4 encoding). amd64 by default; override
-# for other platforms or pin a different johnvansickle build.
-ARG FFMPEG_URL=https://johnvansickle.com/ffmpeg/releases/ffmpeg-7.0.2-amd64-static.tar.xz
+# Prebuilt static ffmpeg (glibc-linked, includes libx264) for ugoira MP4
+# encoding. Served from https://ffmpeg.martin-riedl.de (Cloudflare CDN,
+# built on Debian 12 — glibc-compatible with the bookworm-slim runtime).
+# johnvansickle.com throttles datacenter IPs and served garbage from GitHub
+# runners. `/redirect/latest/` floats to the newest release build; each build
+# also ships a .sha256. Swap `amd64` for `arm64` when building arm64 images.
+ARG FFMPEG_URL=https://ffmpeg.martin-riedl.de/redirect/latest/linux/amd64/release/ffmpeg.zip
 
 WORKDIR /build
 
@@ -22,11 +26,14 @@ RUN mkdir -p crates/x-media/src crates/xmedia-bot/src \
     && cargo build --release -p xmedia-bot
 
 # 2. Static ffmpeg next (cached unless FFMPEG_URL changes), so source edits
-#    never re-download it. The johnvansickle tarball has a
-#    `{build}/ffmpeg` layout, so strip one path component.
-RUN wget -q -O /tmp/ffmpeg.tar.xz "$FFMPEG_URL" \
-    && tar -xJf /tmp/ffmpeg.tar.xz -C /usr/local/bin --strip-components=1 --wildcards '*/ffmpeg' \
-    && rm /tmp/ffmpeg.tar.xz \
+#    never re-download it. The zip contains a single `ffmpeg` binary at the
+#    root. `unzip -t` verifies the archive before extraction so a bad
+#    download fails loudly here instead of a cryptic later error.
+RUN wget -q -O /tmp/ffmpeg.zip "$FFMPEG_URL" \
+    && unzip -tq /tmp/ffmpeg.zip \
+    && unzip -q /tmp/ffmpeg.zip -d /usr/local/bin \
+    && chmod +x /usr/local/bin/ffmpeg \
+    && rm /tmp/ffmpeg.zip \
     && /usr/local/bin/ffmpeg -version >/dev/null
 
 # 3. Real sources last: only our crates recompile on source changes. The

README.md

@@ -10,6 +10,7 @@ Telegram 机器人，将 X / Twitter、Pixiv、Bluesky 的帖子链接转换为
 - 可绑定转发频道自动转发；支持转发前编辑 caption 与自定义模板
 - 发送失败自动重试并持久化，重试耗尽后通知用户
 - Pixiv ugoira 动图自动转码为 MP4
+- 链接结果本地缓存：成功发送后缓存 Telegram file id 与 caption 等，再次收到相同链接直接本地重发，不再请求源站、不保存媒体文件（`LINK_CACHE_TTL_SECONDS` 控制过期，默认 7 天）
 
 ## 快速开始
 
@@ -28,7 +29,73 @@ docker build -t tgxmb .
 docker run --rm -d --name tgxmb --env-file .env -v ./data:/app/data tgxmb
 ```
 
-环境变量：`TELOXIDE_TOKEN`（必填）、`PIXIV_REFRESH_TOKEN`、`BOT_ADMIN`、`EDIT_MESSAGE_TTL_SECONDS`、`RUST_LOG`、`WEBHOOK*`。
+环境变量：`TELOXIDE_TOKEN`（必填）、`PIXIV_REFRESH_TOKEN`、`BOT_ADMIN`、`EDIT_MESSAGE_TTL_SECONDS`、`LINK_CACHE_TTL_SECONDS`、`RUST_LOG`、`WEBHOOK*`、`TWITTER_AUTH_TOKEN`（可选）。
+
+NSFW 推文：公开的 syndication 接口不返回敏感内容。设置 `TWITTER_AUTH_TOKEN`（登录 x.com 后浏览器 Cookie 里的 `auth_token` 值）后，bot 会仅在遇到 NSFW 推文时以登录态获取媒体；未设置则提示无媒体。
+
+### Webhook 部署（需要反向代理）
+
+`docker-compose.yml.example` 内置了 [nginx-proxy](https://github.com/nginx-proxy/nginx-proxy) + [acme-companion](https://github.com/nginx-proxy/acme-companion) 反向代理编排，按部署环境二选一：
+
+**有域名**
+1. DNS A 记录指向服务器
+2. compose 里设 `VIRTUAL_HOST`、`WEBHOOK_URL` 为域名，并取消注释 `ACME_HOST`（设为域名）
+3. acme-companion 自动签发与续期证书，无需手动处理
+
+**只有 IP**
+Let's Encrypt 支持为公网 IP 签发证书（2026 年起可用，有效期约 7 天，须 `shortlived` profile）。用 [acme.sh](https://github.com/acmesh-official/acme.sh) 自动签发与续期，无需手动证书：
+
+1. compose 里增加 acme-ip 服务（签发 + 每日检查自动续期）：
+   ```yaml
+   acme-ip:
+     image: neilpang/acme.sh
+     container_name: acme-ip
+     command: daemon
+     restart: always
+     volumes:
+       - certs:/acme.sh
+       - html:/usr/share/nginx/html
+       - /var/run/docker.sock:/var/run/docker.sock:ro
+     networks: [proxy]
+   ```
+2. 首次签发（把 `<SERVER_IP>` 换成服务器公网 IP，IPv6 同样支持，多个 `-d` 可并列）：
+   ```bash
+   docker compose exec acme-ip acme.sh --issue --server letsencrypt \
+     -d <SERVER_IP> --cert-profile shortlived --days 3 \
+     --webroot /usr/share/nginx/html \
+     --install-cert --cert-file /acme.sh/<SERVER_IP>.crt \
+     --key-file /acme.sh/<SERVER_IP>.key \
+     --reloadcmd "curl --unix-socket /var/run/docker.sock -X POST http://localhost/containers/nginx-proxy/kill?signal=HUP"
+   ```
+3. compose 里设 `VIRTUAL_HOST: '<SERVER_IP>'`、`WEBHOOK_URL: 'https://<SERVER_IP>/'`，无需 `WEBHOOK_CERT`。续期由 acme.sh daemon 自动完成（`--days 3` = 每 3 天续一次，证书 7 天有效有缓冲），续期成功后自动 HUP 通知 nginx-proxy 加载新证书。
+
+   限制：证书约 7 天有效；验证仅支持 http-01/tls-alpn-01（80 端口必须公网可达）；不支持 DNS-01、私有 IP 与 IP 段；同一 IP 集合每 168 小时限签发 5 张。建议先用 `--server letsencrypt_test` 试签，成功后再切正式服务器。
+
+Telegram 只接受 443/80/88/8443 端口。
+
+<details>
+<summary>环境变量说明</summary>
+
+| 变量 | 说明 |
+|---|---|
+| `TELOXIDE_TOKEN` | Bot token（必填） |
+| `PIXIV_REFRESH_TOKEN` | Pixiv 刷新令牌；未设置则禁用 Pixiv |
+| `BOT_ADMIN` | 管理员聊天 ID，逗号分隔；接收启动/停止通知 |
+| `EDIT_MESSAGE_TTL_SECONDS` | 转发前编辑记录过期秒数，默认 86400 |
+| `LINK_CACHE_TTL_SECONDS` | 链接结果缓存过期秒数，默认 604800（7 天） |
+| `RUST_LOG` | 日志级别 |
+| `LOCAL_USER_ID` | 容器内运行用户 UID，默认 9001 |
+| `VIRTUAL_HOST` | 对外域名或 IP，nginx-proxy 按此路由 |
+| `VIRTUAL_PORT` | bot 容器内监听端口，nginx-proxy 的转发目标 |
+| `ACME_HOST` | 域名部署：设为域名时由 acme-companion 自动签发/续期证书 |
+| `DEFAULT_HOST` | nginx-proxy 将未知 Host 的请求路由到该 vhost（IP 访问时需要） |
+| `DEFAULT_EMAIL` | acme-companion 证书通知邮箱 |
+| `WEBHOOK` | `true` 启用 webhook 模式（默认轮询） |
+| `WEBHOOK_LISTEN` / `WEBHOOK_PORT` | bot 容器内监听地址/端口 |
+| `WEBHOOK_URL` | 对外公网 HTTPS 地址（`https://域名/` 或 `https://IP/`） |
+| `WEBHOOK_SECRET_TOKEN` | 更新校验令牌（`X-Telegram-Bot-Api-Secret-Token`） |
+
+</details>
 
 ## 命令
 
@@ -45,6 +112,6 @@ docker run --rm -d --name tgxmb --env-file .env -v ./data:/app/data tgxmb
 
 ## 备注
 
-- 数据持久化于 `data/task_queue.db`，容器部署需挂载该目录
+- 数据持久化于 `data/task_queue.db`，compose 部署使用 bind mount `./data`（保持目录形式便于备份）
 - 运行环境需安装 ffmpeg（Docker 镜像已内置）
 - 测试：`cargo test --workspace`

crates/x-media/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "x-media"
-version = "1.0.2"
+version = "1.0.7"
 edition = "2024"
 
 [dependencies]
@@ -13,6 +13,7 @@ url = "2.5.2"
 bytes = "1"
 zip = "2"
 tempfile = "3"
+rand = "0.8"
 log = "0.4"
 tokio = { version = "1.40", features = ["time"] }
 

crates/x-media/src/media.rs

@@ -14,6 +14,24 @@ impl Media {
             Media::Animated { thumbnail_url, .. } => Some(thumbnail_url),
         }
     }
+
+    /// A smaller variant of this media's file (used as the fallback when the
+    /// primary URL or upload exceeds Telegram's size limits). None when no
+    /// smaller variant exists (videos, animated gifs).
+    pub fn smaller_url(&self) -> Option<&str> {
+        match self {
+            Media::Illustration {
+                url,
+                fallback_url,
+                thumbnail_url,
+                ..
+            } => fallback_url
+                .as_deref()
+                .or(thumbnail_url.as_deref())
+                .filter(|smaller| *smaller != url),
+            Media::Video { .. } | Media::Animated { .. } => None,
+        }
+    }
 }
 
 #[derive(Debug)]

crates/x-media/src/site/mod.rs

@@ -68,18 +68,73 @@ impl Fetched {
     /// caption.
     pub fn caption_with(&self, format: &str) -> String {
         match (&self.render_data, format.is_empty()) {
-            (Some(data), false) => {
-                let escaped = html_escape::encode_text(format).into_owned();
-                escaped
-                    .replace("{url}", &data.url)
-                    .replace("{author}", &data.author)
-                    .replace("{author_url}", &data.author_url)
-                    .replace("{title}", &data.title)
-                    .replace("{tags}", &data.tags)
-            }
+            (Some(data), false) => caption_from_fields(
+                format,
+                "",
+                &data.url,
+                &data.author,
+                &data.author_url,
+                &data.title,
+                &data.tags,
+            ),
             _ => self.caption.clone(),
         }
     }
+
+    /// The pre-escaped placeholder values (author, author_url, title, tags)
+    /// a caller needs to rebuild a caption later, e.g. for a cached post
+    /// where the [`Fetched`] is no longer available.
+    pub fn render_fields(&self) -> Option<(&str, &str, &str, &str)> {
+        self.render_data.as_ref().map(|d| {
+            (
+                d.author.as_str(),
+                d.author_url.as_str(),
+                d.title.as_str(),
+                d.tags.as_str(),
+            )
+        })
+    }
+}
+
+/// Renders a user-supplied caption format from raw (already-escaped) field
+/// values with the same escaping/substitution rules as
+/// [`Fetched::caption_with`]. An empty format returns `built_in` unchanged.
+pub fn caption_from_fields(
+    format: &str,
+    built_in: &str,
+    url: &str,
+    author: &str,
+    author_url: &str,
+    title: &str,
+    tags: &str,
+) -> String {
+    if format.is_empty() {
+        return built_in.to_string();
+    }
+    let escaped = html_escape::encode_text(format).into_owned();
+    escaped
+        .replace("{url}", url)
+        .replace("{author}", author)
+        .replace("{author_url}", author_url)
+        .replace("{title}", title)
+        .replace("{tags}", tags)
+}
+
+/// Stable per-post cache key derived from any supported URL, so variant
+/// domains (x.com / twitter.com / fxtwitter.com, mobile, `/photo/N`
+/// suffixes) map to the same post. Returns `"twitter:<id>"`,
+/// `"pixiv:<id>"` or `"bsky:<handle>/<rkey>"`.
+pub fn cache_key(url: &str) -> Option<String> {
+    if let Some(caps) = twitter::PATTERN.captures(url) {
+        return Some(format!("twitter:{}", &caps[1]));
+    }
+    if let Some(caps) = pixiv::PATTERN.captures(url) {
+        return Some(format!("pixiv:{}", &caps[1]));
+    }
+    if let Some(caps) = bsky::PATTERN.captures(url) {
+        return Some(format!("bsky:{}/{}", &caps[1], &caps[2]));
+    }
+    None
 }
 
 #[derive(Debug)]
@@ -89,6 +144,9 @@ pub enum FetchError {
     Pixiv(PixivError),
     NotFound,
     Blocked,
+    /// The post exists but its content is withheld (twitter NSFW /
+    /// age-restricted tweets come back as an empty `{}` from syndication).
+    Sensitive,
 }
 
 impl fmt::Display for FetchError {
@@ -99,6 +157,7 @@ impl fmt::Display for FetchError {
             FetchError::Pixiv(e) => write!(f, "pixiv error: {e}"),
             FetchError::NotFound => write!(f, "not found"),
             FetchError::Blocked => write!(f, "blocked"),
+            FetchError::Sensitive => write!(f, "content withheld (sensitive)"),
         }
     }
 }
@@ -109,7 +168,7 @@ impl std::error::Error for FetchError {
             FetchError::Http(e) => Some(e),
             FetchError::Json(e) => Some(e),
             FetchError::Pixiv(e) => Some(e),
-            FetchError::NotFound | FetchError::Blocked => None,
+            FetchError::NotFound | FetchError::Blocked | FetchError::Sensitive => None,
         }
     }
 }
@@ -195,6 +254,19 @@ async fn fetch_once(url: &str) -> Result<Option<Fetched>, FetchError> {
 /// fetch of a media URL is blocked (hotlink protection), the bot downloads
 /// the file itself and uploads it via multipart. Site-appropriate headers:
 /// pixiv image hosts need the `Referer` header.
+/// Returns the Content-Length of a media URL, or `None` when the server does
+/// not report one. Used to check whether a file fits Telegram's size limits
+/// before downloading/uploading it.
+pub async fn media_size(url: &str) -> Result<Option<u64>, FetchError> {
+    let mut request = CLIENT.get(url);
+    let lower = url.to_ascii_lowercase();
+    if lower.contains("pximg.net") {
+        request = request.header("Referer", "https://www.pixiv.net/");
+    }
+    let response = request.send().await?;
+    Ok(response.content_length())
+}
+
 pub async fn download_media(url: &str) -> Result<bytes::Bytes, FetchError> {
     let mut request = CLIENT.get(url);
     let lower = url.to_ascii_lowercase();
@@ -209,6 +281,55 @@ pub async fn download_media(url: &str) -> Result<bytes::Bytes, FetchError> {
 mod tests {
     use super::*;
 
+    #[test]
+    fn cache_key_normalizes_domain_variants() {
+        assert_eq!(
+            cache_key("https://x.com/user/status/1234567890/photo/1"),
+            Some("twitter:1234567890".into())
+        );
+        assert_eq!(
+            cache_key("https://mobile.twitter.com/user/status/1234567890"),
+            Some("twitter:1234567890".into())
+        );
+        assert_eq!(
+            cache_key("https://fxtwitter.com/user/status/1234567890"),
+            Some("twitter:1234567890".into())
+        );
+        assert_eq!(
+            cache_key("https://www.pixiv.net/artworks/123456"),
+            Some("pixiv:123456".into())
+        );
+        assert_eq!(
+            cache_key("https://bsky.app/profile/handle.example/post/3lorem"),
+            Some("bsky:handle.example/3lorem".into())
+        );
+        assert_eq!(cache_key("https://example.com/not-a-post"), None);
+    }
+
+    #[test]
+    fn caption_from_fields_substitutes_and_escapes() {
+        // The format string is escaped, the field values are substituted
+        // verbatim (callers pass the already-escaped render data).
+        let out = caption_from_fields(
+            "see {author} at {url} — {title}",
+            "",
+            "https://x.com/u/status/1",
+            "A &amp; B",
+            "https://x.com/u",
+            "hello <world>",
+            "",
+        );
+        assert_eq!(
+            out,
+            "see A &amp; B at https://x.com/u/status/1 — hello <world>"
+        );
+        // Empty format keeps the built-in caption untouched.
+        assert_eq!(
+            caption_from_fields("", "built-in", "u", "a", "au", "t", "g"),
+            "built-in"
+        );
+    }
+
     #[tokio::test]
     async fn unsupported_url_returns_none() {
         let result = fetch("https://example.com/some/article").await;

crates/x-media/src/site/twitter/auth.rs

@@ -0,0 +1,376 @@
+//! Authenticated fallback for tweets the public syndication endpoint refuses
+//! to serve (NSFW / age-restricted tweets come back as an empty `{}`).
+//!
+//! Mirrors nazurin's web API client ([`web.py`]) and is used *only* when
+//! syndication reports [`FetchError::Sensitive`]: the private GraphQL
+//! `TweetDetail` endpoint, authenticated with a browser session cookie from
+//! `TWITTER_AUTH_TOKEN` (the `auth_token` cookie value of a logged-in x.com
+//! session). A fresh random `ct0` is generated per call; X checks that the
+//! `x-csrf-token` header matches the cookie, not that it issued the value.
+//!
+//! [`web.py`]: https://github.com/y-young/nazurin/blob/master/nazurin/sites/twitter/api/web.py
+//!
+//! # Caveats
+//! - X rotates the GraphQL query id when it rolls the web app; if requests
+//!   start failing, update [`TWEET_DETAIL_QUERY_ID`]. Fresh references from
+//!   the actively maintained FxEmbed/FxEmbed: TweetDetail
+//!   `R9IzzyzQBV87-DOWpcvDmw`, TweetResultByRestId `f2sagi1jweVHFkTUIHzmMQ`
+//!   (the latter is anonymous and surfaces NSFW tweets as
+//!   `reason: NsfwLoggedOut`).
+//! - `x-client-transaction-id` is only required for `SearchTimeline`
+//!   (verified against FxEmbed's `proxy/allowlist.ts`) — TweetDetail works
+//!   without it; no need for the nazurin home-page/JS-bundle derivation.
+
+use std::sync::LazyLock;
+
+use serde_json::{json, Value};
+
+use crate::site::FetchError;
+
+use super::interface::Tweet;
+
+/// `auth_token` cookie of a logged-in x.com session; enables the fallback.
+/// Trimmed: a CRLF `.env` (Windows) leaves a trailing `\r` on the value,
+/// which would make the Cookie header invalid.
+static AUTH_TOKEN: LazyLock<Option<String>> = LazyLock::new(|| {
+    std::env::var("TWITTER_AUTH_TOKEN")
+        .ok()
+        .map(|s| s.trim().to_string())
+        .filter(|s| !s.is_empty())
+});
+
+/// Public "logged in" client token used by the x.com web app.
+const LOGGED_IN_BEARER: &str =
+    "Bearer AAAAAAAAAAAAAAAAAAAAANRILgAAAAAAnNwIzUejRCOuH5E6I8xnZz4puTs%3D1Zv7ttfk8LF81IUq16cHjhLTvJu4FA33AGWWjCpTnA";
+
+/// `TweetDetail` query id (from nazurin; still valid as of 2026-08,
+/// corroborated by the current FxEmbed build — see module caveats).
+const TWEET_DETAIL_QUERY_ID: &str = "_8aYOgEDz35BrBcBal1-_w";
+
+fn variables(id: &str) -> Value {
+    json!({
+        "focalTweetId": id,
+        "with_rux_injections": false,
+        "includePromotedContent": false,
+        "withCommunity": true,
+        "withQuickPromoteEligibilityTweetFields": false,
+        "withBirdwatchNotes": false,
+        "withVoice": true,
+    })
+}
+
+fn features() -> Value {
+    json!({
+        "rweb_video_screen_enabled": false,
+        "profile_label_improvements_pcf_label_in_post_enabled": true,
+        "rweb_tipjar_consumption_enabled": true,
+        "verified_phone_label_enabled": false,
+        "creator_subscriptions_tweet_preview_api_enabled": true,
+        "responsive_web_graphql_timeline_navigation_enabled": true,
+        "responsive_web_graphql_skip_user_profile_image_extensions_enabled": false,
+        "premium_content_api_read_enabled": false,
+        "communities_web_enable_tweet_community_results_fetch": true,
+        "c9s_tweet_anatomy_moderator_badge_enabled": true,
+        "responsive_web_grok_analyze_button_fetch_trends_enabled": false,
+        "responsive_web_grok_analyze_post_followups_enabled": true,
+        "responsive_web_jetfuel_frame": false,
+        "responsive_web_grok_share_attachment_enabled": true,
+        "articles_preview_enabled": true,
+        "responsive_web_edit_tweet_api_enabled": true,
+        "graphql_is_translatable_rweb_tweet_is_translatable_enabled": true,
+        "view_counts_everywhere_api_enabled": true,
+        "longform_notetweets_consumption_enabled": true,
+        "responsive_web_twitter_article_tweet_consumption_enabled": true,
+        "tweet_awards_web_tipping_enabled": false,
+        "responsive_web_grok_show_grok_translated_post": false,
+        "responsive_web_grok_analysis_button_from_backend": true,
+        "creator_subscriptions_quote_tweet_preview_enabled": false,
+        "freedom_of_speech_not_reach_fetch_enabled": true,
+        "standardized_nudges_misinfo": true,
+        "tweet_with_visibility_results_prefer_gql_limited_actions_policy_enabled": true,
+        "longform_notetweets_rich_text_read_enabled": true,
+        "longform_notetweets_inline_media_enabled": true,
+        "responsive_web_grok_image_annotation_enabled": true,
+        "responsive_web_enhance_cards_enabled": false,
+    })
+}
+
+/// Whether the authenticated fallback is available.
+pub fn enabled() -> bool {
+    AUTH_TOKEN.is_some()
+}
+
+/// Fetches a tweet as the logged-in user via the private GraphQL API.
+/// Returns the syndication-shaped [`Tweet`] (media included for NSFW posts).
+pub async fn fetch(id: &str) -> Result<Tweet, FetchError> {
+    let token = AUTH_TOKEN
+        .as_deref()
+        .ok_or(FetchError::Sensitive)?;
+    // 16 random bytes as 32 hex chars: X rejects ct0 values of any other
+    // length with 403 code 353 ("matching csrf cookie and header").
+    let ct0: String = (0..16)
+        .map(|_| format!("{:02x}", rand::random::<u8>()))
+        .collect();
+
+    let response = crate::site::CLIENT
+        .get(format!(
+            "https://x.com/i/api/graphql/{TWEET_DETAIL_QUERY_ID}/TweetDetail"
+        ))
+        .query(&[
+            ("variables", variables(id).to_string()),
+            ("features", features().to_string()),
+        ])
+        .header("authorization", LOGGED_IN_BEARER)
+        .header("x-csrf-token", &ct0)
+        .header("x-twitter-auth-type", "OAuth2Session")
+        .header("cookie", format!("auth_token={token}; ct0={ct0}"))
+        .header("x-twitter-client-language", "en")
+        .header("x-twitter-active-user", "yes")
+        .header("referer", "https://x.com/")
+        .send()
+        .await?;
+    if !response.status().is_success() {
+        log::warn!("twitter auth fetch {id}: HTTP {}", response.status());
+        return Err(FetchError::NotFound);
+    }
+    let text = response.text().await?;
+    let json: Value = serde_json::from_str(&text)?;
+    let result = parse_tweet_result(&json, id)?;
+    let syndication_shape = to_syndication_shape(&result)
+        .ok_or_else(|| FetchError::Json(serde_json::Error::io(std::io::Error::new(
+            std::io::ErrorKind::InvalidData,
+            "missing tweet fields in GraphQL response",
+        ))))?;
+    Tweet::from_syndication_json(&syndication_shape.to_string()).map_err(FetchError::Json)
+}
+
+/// Locates the tweet for `id` in a `TweetDetail` response and unwraps
+/// visibility wrappers / retweets, mirroring nazurin's `_process_response`.
+fn parse_tweet_result(json: &Value, id: &str) -> Result<Value, FetchError> {
+    if let Some(errors) = json.get("errors").and_then(|e| e.as_array()) {
+        let messages: Vec<&str> = errors
+            .iter()
+            .filter_map(|e| e.get("message").and_then(|m| m.as_str()))
+            .collect();
+        log::warn!("twitter auth fetch {id} failed: {}", messages.join("; "));
+        return Err(FetchError::NotFound);
+    }
+
+    let instructions = json
+        .pointer("/data/threaded_conversation_with_injections_v2/instructions")
+        .and_then(|v| v.as_array())
+        .ok_or(FetchError::NotFound)?;
+    for instruction in instructions {
+        if instruction.get("type").and_then(|t| t.as_str()) != Some("TimelineAddEntries") {
+            continue;
+        }
+        let entries = instruction
+            .get("entries")
+            .and_then(|e| e.as_array())
+            .ok_or(FetchError::NotFound)?;
+        let wanted = format!("tweet-{id}");
+        for entry in entries {
+            if entry.get("entryId").and_then(|i| i.as_str()) == Some(wanted.as_str()) {
+                let result = entry
+                    .pointer("/content/itemContent/tweet_results/result")
+                    .ok_or(FetchError::NotFound)?;
+                return normalize_tweet_result(result);
+            }
+        }
+    }
+    Err(FetchError::NotFound)
+}
+
+/// Unwraps TweetTombstone/TweetUnavailable errors, the
+/// TweetWithVisibilityResults wrapper and retweets, returning the
+/// `{core, legacy, ...}` tweet object.
+fn normalize_tweet_result(result: &Value) -> Result<Value, FetchError> {
+    match result.get("__typename").and_then(|t| t.as_str()) {
+        Some("TweetTombstone") => {
+            let text = result
+                .pointer("/tombstone/text/text")
+                .and_then(|t| t.as_str())
+                .unwrap_or("tweet is unavailable");
+            log::warn!("twitter auth fetch: tombstone: {text}");
+            return Err(FetchError::NotFound);
+        }
+        Some("TweetUnavailable") => {
+            let reason = result
+                .get("reason")
+                .and_then(|r| r.as_str())
+                .unwrap_or("unknown");
+            log::warn!("twitter auth fetch: tweet unavailable: {reason}");
+            return Err(FetchError::NotFound);
+        }
+        _ => {}
+    }
+
+    // TweetWithVisibilityResults (e.g. limited replies) nests the real tweet.
+    let tweet = result.get("tweet").unwrap_or(result);
+    // A retweet's media lives on the original tweet.
+    if let Some(original) = tweet.pointer("/legacy/retweeted_status_result/result") {
+        return Ok(original.clone());
+    }
+    Ok(tweet.clone())
+}
+
+/// Maps a GraphQL `{core, legacy, ...}` tweet onto the syndication JSON
+/// shape [`Tweet::from_syndication_json`] parses, so the existing text /
+/// media handling (t.co expansion, `name=orig`, mp4 variant) is reused.
+fn to_syndication_shape(tweet: &Value) -> Option<Value> {
+    let legacy = tweet.get("legacy")?;
+    let user = tweet.pointer("/core/user_results/result/legacy")?;
+    Some(json!({
+        "id_str": legacy.get("id_str"),
+        "text": legacy.get("full_text"),
+        "user": {
+            "name": user.get("name"),
+            "screen_name": user.get("screen_name"),
+        },
+        "possibly_sensitive": legacy.get("possibly_sensitive"),
+        "display_text_range": legacy.get("display_text_range"),
+        "entities": legacy.get("entities"),
+        "mediaDetails": legacy.pointer("/extended_entities/media"),
+    }))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn tweet_result() -> Value {
+        json!({
+            "__typename": "Tweet",
+            "core": {
+                "user_results": {
+                    "result": {
+                        "legacy": { "name": "Display Name", "screen_name": "nsfw_author" }
+                    }
+                }
+            },
+            "legacy": {
+                "id_str": "2083868672721039569",
+                "full_text": "nsfw content https://t.co/abc123",
+                "display_text_range": [0, 12],
+                "possibly_sensitive": true,
+                "entities": {
+                    "urls": [
+                        { "url": "https://t.co/abc123", "expanded_url": "https://example.com/x" }
+                    ]
+                },
+                "extended_entities": {
+                    "media": [
+                        {
+                            "type": "photo",
+                            "media_url_https": "https://pbs.twimg.com/media/nsfw.jpg",
+                            "original_info": { "width": 1200, "height": 800 }
+                        },
+                        {
+                            "type": "video",
+                            "media_url_https": "https://pbs.twimg.com/thumb.jpg",
+                            "video_info": {
+                                "variants": [
+                                    { "content_type": "application/x-mpegURL", "url": "https://x.com/pl.m3u8" },
+                                    { "content_type": "video/mp4", "url": "https://video.twimg.com/nsfw.mp4" }
+                                ]
+                            }
+                        }
+                    ]
+                }
+            }
+        })
+    }
+
+    fn conversation(tweet: Value) -> Value {
+        json!({
+            "data": {
+                "threaded_conversation_with_injections_v2": {
+                    "instructions": [
+                        { "type": "TimelineAddEntries", "entries": [
+                            { "entryId": "tweet-2083868672721039569",
+                              "content": { "itemContent": { "tweet_results": { "result": tweet } } } }
+                        ]}
+                    ]
+                }
+            }
+        })
+    }
+
+    #[test]
+    fn parses_graphql_tweet_into_fetched() {
+        let json = conversation(tweet_result());
+        let result = parse_tweet_result(&json, "2083868672721039569").unwrap();
+        let shape = to_syndication_shape(&result).unwrap();
+        let tweet = Tweet::from_syndication_json(&shape.to_string()).unwrap();
+        let fetched: crate::site::Fetched = tweet.into();
+
+        assert!(fetched.sensitive);
+        assert_eq!(fetched.media.len(), 2);
+        match &fetched.media[0] {
+            crate::media::Media::Illustration { url, .. } => {
+                assert_eq!(url, "https://pbs.twimg.com/media/nsfw.jpg?name=orig");
+            }
+            other => panic!("expected illustration, got {other:?}"),
+        }
+        match &fetched.media[1] {
+            crate::media::Media::Video { url, .. } => {
+                assert_eq!(url, "https://video.twimg.com/nsfw.mp4");
+            }
+            other => panic!("expected video, got {other:?}"),
+        }
+        assert_eq!(fetched.source_url, "https://x.com/nsfw_author/status/2083868672721039569");
+        // display_text_range cuts the trailing t.co link.
+        assert_eq!(fetched.title, "nsfw content");
+    }
+
+    #[test]
+    fn unwraps_retweet_to_original() {
+        let original = tweet_result();
+        let mut rt = tweet_result();
+        rt["legacy"]["retweeted_status_result"] = json!({ "result": original });
+        let json = conversation(rt);
+        let result = parse_tweet_result(&json, "2083868672721039569").unwrap();
+        assert!(result.pointer("/legacy/retweeted_status_result").is_none());
+        assert_eq!(result.pointer("/legacy/id_str").unwrap(), "2083868672721039569");
+    }
+
+    #[test]
+    fn error_response_maps_to_not_found() {
+        let json = json!({ "errors": [{ "message": "NsfwLoggedOut" }] });
+        assert!(matches!(
+            parse_tweet_result(&json, "1"),
+            Err(FetchError::NotFound)
+        ));
+    }
+
+    #[test]
+    fn missing_entry_maps_to_not_found() {
+        let json = conversation(json!({ "__typename": "Tweet" }));
+        assert!(matches!(
+            parse_tweet_result(&json, "999"),
+            Err(FetchError::NotFound)
+        ));
+    }
+
+    #[test]
+    fn tombstone_maps_to_not_found() {
+        let tombstone = json!({
+            "__typename": "TweetTombstone",
+            "tombstone": { "text": { "text": "Age-restricted adult content" } }
+        });
+        let json = conversation(tombstone);
+        assert!(matches!(
+            parse_tweet_result(&json, "2083868672721039569"),
+            Err(FetchError::NotFound)
+        ));
+    }
+
+    #[test]
+    fn visibility_wrapper_unwraps() {
+        let inner = tweet_result();
+        let wrapped = json!({ "__typename": "TweetWithVisibilityResults", "tweet": inner });
+        let json = conversation(wrapped);
+        let result = parse_tweet_result(&json, "2083868672721039569").unwrap();
+        assert_eq!(result.get("__typename").unwrap(), "Tweet");
+    }
+}

crates/x-media/src/site/twitter/interface.rs

@@ -19,7 +19,44 @@ pub async fn fetch_from_url(url: &str) -> Result<Fetched, FetchError> {
         .and_then(|caps| caps.get(1))
         .map(|m| m.as_str())
         .ok_or(FetchError::NotFound)?;
-    Ok(fetch(id).await?.into())
+    match fetch(id).await {
+        Ok(tweet) => Ok(tweet.into()),
+        // Syndication withholds NSFW/age-restricted tweets (empty `{}`).
+        // Retry as the logged-in user when TWITTER_AUTH_TOKEN is set;
+        // otherwise degrade to an empty result (the bot replies
+        // "No media found").
+        Err(FetchError::Sensitive) => {
+            if super::auth::enabled() {
+                match super::auth::fetch(id).await {
+                    Ok(tweet) => Ok(tweet.into()),
+                    Err(e) => {
+                        log::warn!("twitter auth fallback failed for {id}: {e}");
+                        Ok(empty_fetched(url))
+                    }
+                }
+            } else {
+                log::info!(
+                    "tweet {id} is sensitive; set TWITTER_AUTH_TOKEN to fetch NSFW media"
+                );
+                Ok(empty_fetched(url))
+            }
+        }
+        Err(e) => Err(e),
+    }
+}
+
+/// A Fetched with no media for withheld tweets: the bot replies
+/// "No media found" and moves on instead of erroring.
+fn empty_fetched(url: &str) -> Fetched {
+    Fetched {
+        source_url: url.to_string(),
+        caption: url.to_string(),
+        title: String::new(),
+        media: vec![],
+        sensitive: true,
+        render_data: None,
+        _keep_alive: None,
+    }
 }
 
 /// Fetches a tweet from the syndication endpoint. Deleted/blocked tweets
@@ -44,7 +81,16 @@ pub async fn fetch(id: &str) -> Result<Tweet, FetchError> {
     {
         return Err(FetchError::NotFound);
     }
-    Ok(Tweet::from_syndication_json(&text).map_err(FetchError::Json)?)
+    // NSFW / age-restricted tweets exist but are served as an empty `{}` —
+    // they surface as FetchError::Sensitive so the caller can retry as a
+    // logged-in user.
+    if serde_json::from_str::<serde_json::Value>(&text)
+        .map(|v| v.get("id_str").is_none())
+        .unwrap_or(false)
+    {
+        return Err(FetchError::Sensitive);
+    }
+    Tweet::from_syndication_json(&text).map_err(FetchError::Json)
 }
 
 /// The syndication token: JS `((id / 1e15) * PI).toString(36)` (the
@@ -129,7 +175,9 @@ impl Tweet {
                     title: None,
                     url: original_twimg_url(&item.media_url_https),
                     thumbnail_url: None,
-                    fallback_url: None,
+                    // The param-less base URL is a reduced-size variant;
+                    // used as the fallback when the original is too large.
+                    fallback_url: Some(item.media_url_https.clone()),
                 }),
                 "video" => media.push(Media::Video {
                     title: None,
@@ -157,21 +205,17 @@ impl Tweet {
 }
 
 /// The raw syndication `text` ends with the appended media short link
-/// (" https://t.co/wmI8McgXul"). `display_text_range` (UTF-16 indices) marks
-/// the visible text; a regex strips any remaining trailing t.co link when the
-/// range is absent or a tweet ends in a URL short link.
+/// (" https://t.co/wmI8McgXul"). `display_text_range` marks the visible text;
+/// a regex strips any remaining trailing t.co link when the range is absent
+/// or a tweet ends in a URL short link.
+///
+/// X reports these indices in Unicode **code points**, not UTF-16 units
+/// (verified against GraphQL responses containing emoji: cutting an emoji
+/// tweet by UTF-16 units silently drops the character after the emoji).
 fn strip_trailing_short_links(text: &str, display_text_range: Option<[usize; 2]>) -> String {
     let mut out = match display_text_range {
         Some([start, end]) if start < end => {
-            let units: Vec<u16> = text
-                .encode_utf16()
-                .skip(start)
-                .take(end - start)
-                .collect();
-            // Drop the replacement char that a surrogate cut at the boundary
-            // would produce (the range end is a valid UTF-16 boundary in
-            // practice, so this is just a safety net).
-            String::from_utf16_lossy(&units).replace('\u{FFFD}', "")
+            text.chars().skip(start).take(end - start).collect()
         }
         _ => text.to_string(),
     };

crates/x-media/src/site/twitter/mod.rs

@@ -1,3 +1,4 @@
+mod auth;
 mod interface;
 mod model;
 

crates/x-media/src/site/twitter/model.rs

@@ -10,7 +10,7 @@ pub struct SyndicationTweet {
     #[serde(default)]
     pub possibly_sensitive: Option<bool>,
     /// Visible-text span; the raw `text` field has the appended media short
-    /// link after it. Indices are UTF-16 code units.
+    /// link after it. Indices are Unicode code points (not UTF-16 units).
     #[serde(default, rename = "display_text_range")]
     pub display_text_range: Option<[usize; 2]>,
     #[serde(default)]

crates/xmedia-bot/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "xmedia-bot"
-version = "1.0.2"
+version = "1.0.7"
 edition = "2024"
 
 [dependencies]
@@ -18,4 +18,8 @@ rusqlite = { version = "0.32", features = ["bundled"] }
 rand = "0.8"
 tempfile = "3"
 parking_lot = "0.12"
+png = "0.18"
+zune-jpeg = "0.5"
+fast_image_resize = "6"
+jpeg-encoder = "0.7"
 x-media = { path = "../x-media" }

crates/xmedia-bot/src/config.rs

@@ -10,6 +10,8 @@ pub struct Config {
     pub admin_ids: Vec<i64>,
     /// EDIT_MESSAGE_TTL_SECONDS, default 86400 (24h).
     pub edit_message_ttl: Duration,
+    /// LINK_CACHE_TTL_SECONDS, default 604800 (7 days).
+    pub link_cache_ttl: Duration,
     // Webhook settings (moved out of main; names/defaults unchanged).
     pub webhook_enabled: bool,
     pub webhook_url: Option<url::Url>,
@@ -36,17 +38,30 @@ impl Config {
         .map(Duration::from_secs)
         .unwrap_or(Duration::from_secs(86400));
 
+    let link_cache_ttl = env::var("LINK_CACHE_TTL_SECONDS")
+        .ok()
+        .and_then(|s| s.parse::<u64>().ok())
+        .map(Duration::from_secs)
+        .unwrap_or(Duration::from_secs(7 * 24 * 3600));
+
     let webhook_enabled = env::var("WEBHOOK")
         .is_ok_and(|v| matches!(v.to_lowercase().as_str(), "true" | "yes" | "1"));
     let webhook_url = env::var("WEBHOOK_URL").ok().and_then(|s| s.parse().ok());
     let webhook_listen = env::var("WEBHOOK_LISTEN").ok().and_then(|s| s.parse().ok());
     let webhook_port = env::var("WEBHOOK_PORT").ok().and_then(|s| s.parse().ok());
-    let webhook_cert = env::var("WEBHOOK_CERT").ok();
-    let webhook_secret_token = env::var("WEBHOOK_SECRET_TOKEN").ok();
+    // Empty strings count as unset (e.g. `-e WEBHOOK_CERT=` to disable a
+    // value that would otherwise come from `.env`).
+    let webhook_cert = env::var("WEBHOOK_CERT")
+        .ok()
+        .filter(|s| !s.is_empty());
+    let webhook_secret_token = env::var("WEBHOOK_SECRET_TOKEN")
+        .ok()
+        .filter(|s| !s.is_empty());
 
         Config {
             admin_ids,
             edit_message_ttl,
+            link_cache_ttl,
             webhook_enabled,
             webhook_url,
             webhook_listen,

crates/xmedia-bot/src/handlers.rs

@@ -1,10 +1,12 @@
 use crate::config::Config;
+use crate::link_cache::{CachedMediaKind, CachedPost, LinkCache};
 use crate::queue::PersistentTaskQueue;
 use crate::send::{self, MediaItemPayload, Task};
-use crate::state::{ChatStore, unix_now};
+use crate::state::{ChatData, ChatStore, unix_now};
 use std::collections::HashSet;
 use std::sync::LazyLock;
 use teloxide::prelude::*;
+use tokio::sync::Semaphore;
 use teloxide::types::{
     CallbackQuery, ChatAction, ChatId, ChatKind, InlineQuery, InlineQueryResult,
     InlineQueryResultMpeg4Gif, InlineQueryResultPhoto, InlineQueryResultVideo, Message,
@@ -19,8 +21,18 @@ pub static CHAT_STORE: LazyLock<ChatStore> = LazyLock::new(|| {
 });
 pub static TASK_QUEUE: LazyLock<PersistentTaskQueue> =
     LazyLock::new(|| PersistentTaskQueue::new("data/task_queue.db"));
+pub static LINK_CACHE: LazyLock<LinkCache> =
+    LazyLock::new(|| LinkCache::open("data/task_queue.db"));
 pub static CONFIG: LazyLock<Config> = LazyLock::new(Config::load);
 
+/// Cap on concurrent per-URL processing. teloxide dispatches updates to a
+/// per-chat worker that handles them sequentially, so a batch-forward of many
+/// messages would otherwise be processed one at a time (fetch + send each,
+/// roughly a second per message). Moving the work into spawned tasks trades
+/// per-chat reply ordering for throughput; the semaphore bounds how many run
+/// at once so a big burst cannot hammer Telegram's rate limits.
+static URL_TASKS: LazyLock<Semaphore> = LazyLock::new(|| Semaphore::new(8));
+
 #[derive(BotCommands, Clone)]
 #[command(rename_rule = "snake_case", description = "")]
 enum Command {
@@ -322,22 +334,29 @@ fn thumbnail_for(media: &Media) -> Option<String> {
 }
 
 fn media_to_payload(media: &Media, sensitive: bool) -> MediaItemPayload {
+    let fallback_url = media.smaller_url().map(str::to_string);
     match media {
         // A gif inside a group becomes a video item; a lone gif takes the
         // animation path (see url_media).
         Media::Illustration { .. } => MediaItemPayload::Photo {
             media: media.url().to_string(),
             has_spoiler: sensitive,
+            fallback_url,
+            file_id: false,
         },
         Media::Video { .. } => MediaItemPayload::Video {
             media: media.url().to_string(),
             has_spoiler: sensitive,
             thumbnail: thumbnail_for(media),
+            fallback_url,
+            file_id: false,
         },
         Media::Animated { .. } => MediaItemPayload::Video {
             media: media.url().to_string(),
             has_spoiler: sensitive,
             thumbnail: thumbnail_for(media),
+            fallback_url,
+            file_id: false,
         },
     }
 }
@@ -350,11 +369,148 @@ async fn enqueue_retry(task: Task, delay_seconds: f64) {
     }
 }
 
+/// Sends a task and handles the outcome: post-send actions on success, retry
+/// enqueue on retryable failure, reply + link-cache invalidation on
+/// permanent failure (a stale cached file id must not repeat forever).
+async fn dispatch_send(bot: Bot, message: &Message, task: &Task, url: &str) {
+    let result = match task {
+        Task::SendAnimation { .. } => send::send_animation(&bot, task).await,
+        Task::SendMediaSequence { .. } => send::send_media_sequence(&bot, task).await,
+        Task::ForwardMessages { .. } => unreachable!(),
+    };
+    match result {
+        Ok(message_ids) => {
+            log::info!("sent {} message(s) for {url}", message_ids.len());
+            send::post_send_actions(&bot, task, message_ids).await;
+        }
+        Err(send::SendError::Retryable { delay_seconds, task }) => {
+            log::info!("send for {url} failed, queued for retry in {delay_seconds:.1}s");
+            enqueue_retry(task, delay_seconds).await;
+            let _ = reply(bot, message.clone(), "Send failed. Task queued for retry.").await;
+        }
+        Err(send::SendError::Permanent {
+            message: err_message,
+            task,
+        }) => {
+            send::invalidate_cache(&task).await;
+            log::error!("send for {url} failed permanently: {err_message}");
+            let _ = reply(bot, message.clone(), format!("Send failed: {err_message}")).await;
+        }
+    }
+}
+
+/// Builds the send task from ready-made items, sharing the payload shape
+/// between the fresh-fetch and link-cache paths.
+#[allow(clippy::too_many_arguments)]
+fn build_send_task(
+    chat_data: &ChatData,
+    message: &Message,
+    source_url: String,
+    caption: String,
+    items: Vec<MediaItemPayload>,
+    cache_data: Option<CachedPost>,
+) -> Task {
+    let chat_id = message.chat.id.0;
+    if items.len() == 1 && matches!(items[0], MediaItemPayload::Animation { .. }) {
+        Task::SendAnimation {
+            chat_id,
+            reply_to_message_id: message.id.0 as i64,
+            caption,
+            animation: items.into_iter().next().unwrap(),
+            source_url,
+            edit_before_forward: chat_data.edit_before_forward,
+            forward_channel_id: chat_data.forward_channel_id,
+            notify_chat_id: Some(chat_id),
+            notify_message_id: Some(message.id.0 as i64),
+            cache_data,
+        }
+    } else {
+        Task::SendMediaSequence {
+            chat_id,
+            reply_to_message_id: message.id.0 as i64,
+            caption,
+            media_batches: send::chunk_media_items(items),
+            batch_index: 0,
+            sent_message_ids: vec![],
+            source_url,
+            edit_before_forward: chat_data.edit_before_forward,
+            forward_channel_id: chat_data.forward_channel_id,
+            notify_chat_id: Some(chat_id),
+            notify_message_id: Some(message.id.0 as i64),
+            cache_data,
+        }
+    }
+}
+
 async fn url_media(bot: Bot, message: &Message, url: &str) {
     let chat_id = message.chat.id.0;
     if let Err(e) = bot.send_chat_action(ChatId(chat_id), ChatAction::Typing).await {
         log::error!("send_chat_action failed: {e}");
     }
+
+    // Link cache: a post sent before is re-sent from Telegram file ids —
+    // no source-site request, no download, no upload. Keyed by the
+    // normalized post id so x.com / fxtwitter / /photo/N variants collide.
+    if let Some(key) = x_media::site::cache_key(url)
+        && let Some(cached) = LINK_CACHE.get(&key, CONFIG.link_cache_ttl).await
+    {
+        log::info!("link cache hit for {url}");
+        let chat_data = CHAT_STORE.get(chat_id).await;
+        let site = key.split(':').next().unwrap_or("unknown");
+        let format = chat_data
+            .message_format
+            .get(site)
+            .cloned()
+            .unwrap_or_default();
+        let caption = if format.is_empty() {
+            cached.caption.clone()
+        } else {
+            x_media::site::caption_from_fields(
+                &format,
+                "",
+                &cached.url,
+                &cached.author,
+                &cached.author_url,
+                &cached.title,
+                &cached.tags,
+            )
+        };
+        let items: Vec<MediaItemPayload> = cached
+            .media
+            .iter()
+            .map(|m| match m.kind {
+                CachedMediaKind::Photo => MediaItemPayload::Photo {
+                    media: m.file_id.clone(),
+                    has_spoiler: cached.sensitive,
+                    fallback_url: None,
+                    file_id: true,
+                },
+                CachedMediaKind::Video => MediaItemPayload::Video {
+                    media: m.file_id.clone(),
+                    has_spoiler: cached.sensitive,
+                    thumbnail: None,
+                    fallback_url: None,
+                    file_id: true,
+                },
+                CachedMediaKind::Animation => MediaItemPayload::Animation {
+                    media: m.file_id.clone(),
+                    has_spoiler: cached.sensitive,
+                    file_id: true,
+                },
+            })
+            .collect();
+        let task = build_send_task(
+            &chat_data,
+            message,
+            cached.url.clone(),
+            caption,
+            items,
+            Some(cached),
+        );
+        dispatch_send(bot, message, &task, url).await;
+        return;
+    }
+
     log::info!("fetching {url}");
     match x_media::site::fetch(url).await {
         // Unsupported links are ignored silently (Python parity).
@@ -384,66 +540,34 @@ async fn url_media(bot: Bot, message: &Message, url: &str) {
                 .cloned()
                 .unwrap_or_default();
             let caption = fetched.caption_with(&format);
-            let task = if fetched.media.len() == 1
-                && matches!(fetched.media[0], Media::Animated { .. })
-            {
-                Task::SendAnimation {
-                    chat_id,
-                    reply_to_message_id: message.id.0 as i64,
-                    caption: caption.clone(),
-                    animation: MediaItemPayload::Animation {
-                        media: fetched.media[0].url().to_string(),
-                        has_spoiler: fetched.sensitive,
-                    },
-                    source_url: fetched.source_url.clone(),
-                    edit_before_forward: chat_data.edit_before_forward,
-                    forward_channel_id: chat_data.forward_channel_id,
-                    notify_chat_id: Some(chat_id),
-                    notify_message_id: Some(message.id.0 as i64),
+            // Raw render data for the link cache; the send fills in the
+            // Telegram file ids and persists the entry.
+            let cache_data = fetched.render_fields().map(|(author, author_url, title, tags)| {
+                CachedPost {
+                    url: fetched.source_url.clone(),
+                    caption: fetched.caption.clone(),
+                    title: title.to_string(),
+                    author: author.to_string(),
+                    author_url: author_url.to_string(),
+                    tags: tags.to_string(),
+                    sensitive: fetched.sensitive,
+                    media: vec![],
                 }
-            } else {
-                let items: Vec<MediaItemPayload> = fetched
-                    .media
-                    .iter()
-                    .map(|media| media_to_payload(media, fetched.sensitive))
-                    .collect();
-                Task::SendMediaSequence {
-                    chat_id,
-                    reply_to_message_id: message.id.0 as i64,
-                    caption: caption.clone(),
-                    media_batches: send::chunk_media_items(items),
-                    batch_index: 0,
-                    sent_message_ids: vec![],
-                    source_url: fetched.source_url.clone(),
-                    edit_before_forward: chat_data.edit_before_forward,
-                    forward_channel_id: chat_data.forward_channel_id,
-                    notify_chat_id: Some(chat_id),
-                    notify_message_id: Some(message.id.0 as i64),
-                }
-            };
-            let result = match &task {
-                Task::SendAnimation { .. } => send::send_animation(&bot, &task).await,
-                Task::SendMediaSequence { .. } => send::send_media_sequence(&bot, &task).await,
-                Task::ForwardMessages { .. } => unreachable!(),
-            };
-            match result {
-                Ok(message_ids) => {
-                    log::info!("sent {} message(s) for {url}", message_ids.len());
-                    send::post_send_actions(&bot, &task, message_ids).await;
-                }
-                Err(send::SendError::Retryable { delay_seconds, task }) => {
-                    log::info!("send for {url} failed, queued for retry in {delay_seconds:.1}s");
-                    enqueue_retry(task, delay_seconds).await;
-                    let _ = reply(bot, message.clone(), "Send failed. Task queued for retry.").await;
-                }
-                Err(send::SendError::Permanent {
-                    message: err_message,
-                    ..
-                }) => {
-                    log::error!("send for {url} failed permanently: {err_message}");
-                    let _ = reply(bot, message.clone(), format!("Send failed: {err_message}")).await;
-                }
-            }
+            });
+            let items: Vec<MediaItemPayload> = fetched
+                .media
+                .iter()
+                .map(|media| media_to_payload(media, fetched.sensitive))
+                .collect();
+            let task = build_send_task(
+                &chat_data,
+                message,
+                fetched.source_url.clone(),
+                caption,
+                items,
+                cache_data,
+            );
+            dispatch_send(bot, message, &task, url).await;
         }
     }
 }
@@ -477,7 +601,13 @@ pub async fn message_handler(bot: Bot, message: Message) -> Result<(), RequestEr
             log::info!("extracted {} URL(s): {urls:?}", urls.len());
         }
         for url in urls {
-            url_media(bot.clone(), &message, &url).await;
+            let bot = bot.clone();
+            let message = message.clone();
+            tokio::spawn(async move {
+                // Held for the whole task; the semaphore is never closed.
+                let _permit = URL_TASKS.acquire().await.expect("URL semaphore closed");
+                url_media(bot, &message, &url).await;
+            });
         }
     }
     respond(())
@@ -502,11 +632,19 @@ pub async fn inline_query_handler(bot: Bot, query: InlineQuery) -> Result<(), Re
                     .unwrap_or_else(|| url.clone());
                 let caption = fetched.caption.clone();
                 let result = match media {
-                    Media::Illustration { .. } => InlineQueryResult::Photo(
-                        InlineQueryResultPhoto::new(id, url, thumbnail)
-                            .caption(caption)
-                            .parse_mode(ParseMode::Html),
-                    ),
+                    Media::Illustration { .. } => {
+                        // Inline photo results have their own (smaller) size
+                        // cap; use the reduced variant when one exists.
+                        let photo_url = media
+                            .smaller_url()
+                            .and_then(|u| url::Url::parse(u).ok())
+                            .unwrap_or_else(|| url.clone());
+                        InlineQueryResult::Photo(
+                            InlineQueryResultPhoto::new(id, photo_url, thumbnail)
+                                .caption(caption)
+                                .parse_mode(ParseMode::Html),
+                        )
+                    }
                     Media::Video { .. } => InlineQueryResult::Video(
                         InlineQueryResultVideo::new(
                             id,

crates/xmedia-bot/src/link_cache.rs

@@ -0,0 +1,230 @@
+//! Persistent cache of successfully sent posts.
+//!
+//! After a media send succeeds, the raw render data plus the Telegram
+//! `file_id`s of the sent items are stored keyed by [`crate::site` cache
+//! key]. A repeated link is then answered entirely from local state — no
+//! re-fetch of the source site, no re-upload — and no media file is stored
+//! on disk (the file ids point at Telegram's servers). Entries expire after
+//! [`Config::link_cache_ttl`]; a stale entry is dropped lazily on read and
+//! by the periodic prune in `main`.
+
+use rusqlite::{params, Connection};
+use serde::{Deserialize, Serialize};
+use std::time::Duration;
+
+#[derive(Serialize, Deserialize, Clone, Debug, PartialEq)]
+#[serde(rename_all = "snake_case")]
+pub enum CachedMediaKind {
+    Photo,
+    Video,
+    Animation,
+}
+
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct CachedMedia {
+    pub kind: CachedMediaKind,
+    pub file_id: String,
+}
+
+/// Everything needed to re-send a post without touching the source site:
+/// the canonical URL, pre-escaped caption fields, and the file ids produced
+/// by the original successful send.
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct CachedPost {
+    pub url: String,
+    /// The site's built-in caption (used when the chat has no format
+    /// override).
+    pub caption: String,
+    pub title: String,
+    pub author: String,
+    pub author_url: String,
+    pub tags: String,
+    pub sensitive: bool,
+    pub media: Vec<CachedMedia>,
+}
+
+/// SQLite-backed cache sharing `data/task_queue.db` with the queue and chat
+/// state (same `open_db` pattern: busy timeout, `spawn_blocking` I/O).
+pub struct LinkCache {
+    db_path: String,
+}
+
+fn open_db(path: &str) -> rusqlite::Result<Connection> {
+    let conn = Connection::open(path)?;
+    conn.busy_timeout(Duration::from_secs(5))?;
+    Ok(conn)
+}
+
+impl LinkCache {
+    pub fn open(db_path: &str) -> Self {
+        if let Ok(conn) = Connection::open(db_path)
+            && let Err(e) = conn.execute_batch(
+                "CREATE TABLE IF NOT EXISTS link_cache (url TEXT PRIMARY KEY, \
+                 payload TEXT NOT NULL, created_at REAL NOT NULL);",
+            )
+        {
+            log::error!("failed to initialize link cache schema: {e}");
+        }
+        Self {
+            db_path: db_path.to_string(),
+        }
+    }
+
+    /// Returns the cached post if present and not expired; a stale entry is
+    /// removed on the spot.
+    pub async fn get(&self, key: &str, ttl: Duration) -> Option<CachedPost> {
+        let db_path = self.db_path.clone();
+        let key = key.to_string();
+        let ttl = ttl.as_secs_f64();
+        tokio::task::spawn_blocking(move || -> rusqlite::Result<Option<CachedPost>> {
+            let conn = open_db(&db_path)?;
+            let mut stmt =
+                conn.prepare("SELECT payload, created_at FROM link_cache WHERE url = ?1")?;
+            let mut rows = stmt.query(params![key])?;
+            let Some(row) = rows.next()? else {
+                return Ok(None);
+            };
+            let payload: String = row.get(0)?;
+            let created_at: f64 = row.get(1)?;
+            if now_f64() - created_at > ttl {
+                conn.execute("DELETE FROM link_cache WHERE url = ?1", params![key])?;
+                return Ok(None);
+            }
+            serde_json::from_str(&payload).map(Some).map_err(|e| {
+                rusqlite::Error::ToSqlConversionFailure(Box::new(e))
+            })
+        })
+        .await
+        .expect("link cache read worker panicked")
+        .unwrap_or_else(|e| {
+            log::error!("link cache read failed: {e}");
+            None
+        })
+    }
+
+    pub async fn put(&self, key: &str, post: &CachedPost) {
+        let db_path = self.db_path.clone();
+        let key = key.to_string();
+        let payload = serde_json::to_string(post).expect("cached post serializes");
+        tokio::task::spawn_blocking(move || -> rusqlite::Result<()> {
+            let conn = open_db(&db_path)?;
+            conn.execute(
+                "INSERT OR REPLACE INTO link_cache (url, payload, created_at) VALUES (?1, ?2, ?3)",
+                params![key, payload, now_f64()],
+            )?;
+            Ok(())
+        })
+        .await
+        .expect("link cache write worker panicked")
+        .unwrap_or_else(|e| log::error!("link cache write failed: {e}"));
+    }
+
+    /// Drops an entry (e.g. a cached file id that turned out invalid).
+    pub async fn remove(&self, key: &str) {
+        let db_path = self.db_path.clone();
+        let key = key.to_string();
+        tokio::task::spawn_blocking(move || -> rusqlite::Result<()> {
+            let conn = open_db(&db_path)?;
+            conn.execute("DELETE FROM link_cache WHERE url = ?1", params![key])?;
+            Ok(())
+        })
+        .await
+        .expect("link cache delete worker panicked")
+        .unwrap_or_else(|e| log::error!("link cache delete failed: {e}"));
+    }
+
+    /// Removes expired entries; returns how many were deleted.
+    pub async fn prune(&self, ttl: Duration) -> usize {
+        let db_path = self.db_path.clone();
+        let cutoff = now_f64() - ttl.as_secs_f64();
+        tokio::task::spawn_blocking(move || -> rusqlite::Result<usize> {
+            let conn = open_db(&db_path)?;
+            conn.execute(
+                "DELETE FROM link_cache WHERE created_at < ?1",
+                params![cutoff],
+            )
+        })
+        .await
+        .expect("link cache prune worker panicked")
+        .unwrap_or_else(|e| {
+            log::error!("link cache prune failed: {e}");
+            0
+        })
+    }
+}
+
+fn now_f64() -> f64 {
+    std::time::SystemTime::now()
+        .duration_since(std::time::UNIX_EPOCH)
+        .map(|d| d.as_secs_f64())
+        .unwrap_or(0.0)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn entry() -> CachedPost {
+        CachedPost {
+            url: "https://x.com/u/status/1".into(),
+            caption: "cap".into(),
+            title: "t".into(),
+            author: "a".into(),
+            author_url: "au".into(),
+            tags: "".into(),
+            sensitive: true,
+            media: vec![CachedMedia {
+                kind: CachedMediaKind::Photo,
+                file_id: "AgAC...".into(),
+            }],
+        }
+    }
+
+    #[tokio::test]
+    async fn put_get_roundtrip() {
+        let dir = tempfile::tempdir().unwrap();
+        let cache = LinkCache::open(dir.path().join("c.db").to_str().unwrap());
+        cache.put("twitter:1", &entry()).await;
+        let got = cache.get("twitter:1", Duration::from_secs(3600)).await;
+        assert!(got.is_some());
+        let got = got.unwrap();
+        assert_eq!(got.url, "https://x.com/u/status/1");
+        assert_eq!(got.media[0].file_id, "AgAC...");
+    }
+
+    #[tokio::test]
+    async fn expired_entry_removed_on_read() {
+        let dir = tempfile::tempdir().unwrap();
+        let cache = LinkCache::open(dir.path().join("c.db").to_str().unwrap());
+        cache.put("twitter:1", &entry()).await;
+        // Force the row into the past so a 1s TTL expires it.
+        {
+            let conn = Connection::open(dir.path().join("c.db")).unwrap();
+            conn.execute(
+                "UPDATE link_cache SET created_at = created_at - 100",
+                [],
+            )
+            .unwrap();
+        }
+        assert!(cache.get("twitter:1", Duration::from_secs(1)).await.is_none());
+        assert!(cache.get("twitter:1", Duration::from_secs(3600)).await.is_none());
+    }
+
+    #[tokio::test]
+    async fn remove_and_prune() {
+        let dir = tempfile::tempdir().unwrap();
+        let cache = LinkCache::open(dir.path().join("c.db").to_str().unwrap());
+        cache.put("twitter:1", &entry()).await;
+        cache.put("pixiv:2", &entry()).await;
+        cache.remove("twitter:1").await;
+        assert!(cache.get("twitter:1", Duration::from_secs(3600)).await.is_none());
+        assert!(cache.get("pixiv:2", Duration::from_secs(3600)).await.is_some());
+        {
+            let conn = Connection::open(dir.path().join("c.db")).unwrap();
+            conn.execute("UPDATE link_cache SET created_at = created_at - 100", [])
+                .unwrap();
+        }
+        assert_eq!(cache.prune(Duration::from_secs(1)).await, 1);
+        assert!(cache.get("pixiv:2", Duration::from_secs(3600)).await.is_none());
+    }
+}

crates/xmedia-bot/src/main.rs

@@ -1,18 +1,39 @@
 use dotenv::dotenv;
 use teloxide::dptree::endpoint;
+use teloxide::stop::StopToken;
 use teloxide::types::{ChatId, InputFile, MessageId};
-use teloxide::update_listeners::webhooks;
+use teloxide::update_listeners::{self, webhooks, UpdateListener};
 use teloxide::prelude::*;
 use tokio::sync::watch;
 use x_media::site;
 
 mod config;
 mod handlers;
+mod link_cache;
+mod photo;
 mod queue;
 mod send;
 mod state;
 
-use handlers::{CHAT_STORE, CONFIG, TASK_QUEUE};
+use handlers::{CHAT_STORE, CONFIG, LINK_CACHE, TASK_QUEUE};
+
+/// Docker `stop` / `compose down` delivers SIGTERM, which teloxide's ctrlc
+/// handler (SIGINT only) never sees — without this the process would die
+/// before the graceful shutdown below (admin notice, queue drain). Stopping
+/// the token unwinds the dispatcher exactly like Ctrl+C does.
+#[cfg(unix)]
+fn spawn_sigterm_handler(stop_token: StopToken) {
+    tokio::spawn(async move {
+        let mut sigterm = tokio::signal::unix::signal(tokio::signal::unix::SignalKind::terminate())
+            .expect("failed to install SIGTERM handler");
+        sigterm.recv().await;
+        log::info!("SIGTERM received, stopping the dispatcher");
+        stop_token.stop();
+    });
+}
+
+#[cfg(not(unix))]
+fn spawn_sigterm_handler(_stop_token: StopToken) {}
 
 #[tokio::main]
 async fn main() {
@@ -66,6 +87,10 @@ async fn main() {
                 }
                 let ttl = CONFIG.edit_message_ttl;
                 let removed = CHAT_STORE.prune_expired(ttl).await;
+                let pruned = LINK_CACHE.prune(CONFIG.link_cache_ttl).await;
+                if pruned > 0 {
+                    log::info!("link cache: pruned {pruned} expired entr(ies)");
+                }
                 for (chat_id, prompt_message_id) in removed {
                     // If the prompt was already deleted, this fails with a
                     // 400 "message to edit not found" — log and ignore.
@@ -96,7 +121,8 @@ async fn main() {
             .webhook_url
             .clone()
             .expect("WEBHOOK_URL is not set");
-        bot.set_webhook(url.clone()).await.unwrap();
+        // `webhooks::axum` calls set_webhook itself (with the full options,
+        // secret token included) — no explicit registration here.
         let listen = CONFIG.webhook_listen.expect("WEBHOOK_LISTEN is not set");
         let port = CONFIG.webhook_port.expect("WEBHOOK_PORT is not set");
         let mut options = webhooks::Options::new((listen, port).into(), url);
@@ -107,20 +133,36 @@ async fn main() {
             options = options.secret_token(secret.clone());
         }
 
+        let mut listener = webhooks::axum(bot.clone(), options)
+            .await
+            .expect("Failed to create webhook listener");
+        let stop_token = listener.stop_token();
+        spawn_sigterm_handler(stop_token);
+
         dispatcher
             .dispatch_with_listener(
-                webhooks::axum(bot.clone(), options)
-                    .await
-                    .expect("Failed to create webhook listener"),
+                listener,
                 LoggingErrorHandler::with_custom_text("Error from update listener"),
             )
             .await;
     } else {
         log::info!("running in polling mode");
-        dispatcher.dispatch().await;
+        // Same listener `dispatch()` builds internally — using
+        // `dispatch_with_listener` just exposes its stop token so SIGTERM can
+        // unwind the dispatcher before the graceful shutdown below.
+        let mut listener = update_listeners::polling_default(bot.clone()).await;
+        let stop_token = listener.stop_token();
+        spawn_sigterm_handler(stop_token);
+
+        dispatcher
+            .dispatch_with_listener(
+                listener,
+                LoggingErrorHandler::with_custom_text("Error from update listener"),
+            )
+            .await;
     }
 
-    // Graceful stop (Ctrl+C): stop the sweep, notify the admin, drain the queue.
+    // Graceful stop (Ctrl+C / SIGTERM): stop the sweep, notify the admin, drain the queue.
     log::info!("Stopping bot");
     let _ = stop_tx.send(true);
     if let Some(admin) = CONFIG.admin_ids.first() {

crates/xmedia-bot/src/photo.rs

@@ -0,0 +1,515 @@
+//! Pure-Rust photo processing: brings a downloaded photo within Telegram's
+//! limits (width + height ≤ 10000 px, bytes ≤ 10 MiB) without ffmpeg.
+//!
+//! Stack: `png` (image-png) for PNG decode/encode, `zune-jpeg` for JPEG
+//! decode, `fast_image_resize` (Lanczos3) for downsampling, `jpeg-encoder`
+//! for JPEG output.
+//!
+//! Bit-depth rule: a PNG above 24 bits (32-bit RGBA or 16-bit per channel)
+//! is reduced to 24-bit RGB; 24-bit and lower depths are left untouched —
+//! gray stays gray, never upconverted. The only upconversion is palette
+//! expansion, which resampling requires. Alpha is flattened onto white (JPEG
+//! and 24-bit RGB have no alpha channel).
+
+use std::io::Write;
+
+use fast_image_resize as fir;
+use tempfile::NamedTempFile;
+
+/// Telegram rejects photos whose width + height exceed this limit
+/// (PHOTO_INVALID_DIMENSIONS). Verified empirically: 6300x3730 (sum 10030)
+/// fails, 6100x3900 (sum 10000) passes.
+pub const PHOTO_MAX_DIMENSION_SUM: u32 = 10000;
+/// Resize target with a safety margin so rounding cannot cross the cap.
+pub const PHOTO_TARGET_DIMENSION_SUM: u32 = 9900;
+/// Upload cap (bytes): files above this are not uploaded; the bot falls back
+/// to a smaller media URL instead.
+pub const MAX_UPLOAD_BYTES: u64 = 10 * 1024 * 1024;
+/// Decode budget (bytes): a larger intermediate buffer is not worth the peak
+/// memory; the photo degrades to the smaller URL instead.
+const MAX_DECODE_BYTES: u64 = 512 * 1024 * 1024;
+/// JPEG output quality (1-100).
+const JPEG_QUALITY: u8 = 90;
+
+/// What to upload for a downloaded photo.
+pub enum PhotoPrep {
+    /// Upload this file (the original when within limits, else the processed
+    /// copy).
+    Upload(NamedTempFile),
+    /// The photo cannot be brought within Telegram's limits — the caller
+    /// falls back to the item's smaller URL.
+    UseFallback,
+}
+
+/// A decoded image buffer tagged with its channel layout.
+#[derive(Debug)]
+enum PixBuf {
+    Gray(Vec<u8>),
+    GrayAlpha(Vec<u8>),
+    Rgb(Vec<u8>),
+}
+
+impl PixBuf {
+    fn pixel_type(&self) -> fir::PixelType {
+        match self {
+            PixBuf::Gray(_) => fir::PixelType::U8,
+            PixBuf::GrayAlpha(_) => fir::PixelType::U8x2,
+            PixBuf::Rgb(_) => fir::PixelType::U8x3,
+        }
+    }
+
+    fn into_vec(self) -> Vec<u8> {
+        match self {
+            PixBuf::Gray(v) | PixBuf::GrayAlpha(v) | PixBuf::Rgb(v) => v,
+        }
+    }
+}
+
+/// Entry point: detects the format and processes the photo if needed.
+pub fn prepare_photo(file: NamedTempFile) -> Result<PhotoPrep, String> {
+    let bytes = std::fs::read(file.path()).map_err(|e| format!("prepare read failed: {e}"))?;
+    if bytes.starts_with(b"\x89PNG\r\n\x1a\n") {
+        prepare_png(file, bytes)
+    } else if bytes.starts_with(&[0xFF, 0xD8, 0xFF]) {
+        prepare_jpeg(file, bytes)
+    } else {
+        log::warn!("photo in unsupported format; falling back to smaller media");
+        Ok(PhotoPrep::UseFallback)
+    }
+}
+
+/// Parses the PNG IHDR (bytes 8..26: signature + length + "IHDR" + width +
+/// height + bit depth + color type).
+fn parse_png_header(bytes: &[u8]) -> Option<(u32, u32, png::BitDepth, png::ColorType)> {
+    if !bytes.starts_with(b"\x89PNG\r\n\x1a\n") || bytes.len() < 26 {
+        return None;
+    }
+    let w = u32::from_be_bytes(bytes.get(16..20)?.try_into().ok()?);
+    let h = u32::from_be_bytes(bytes.get(20..24)?.try_into().ok()?);
+    let depth = match *bytes.get(24)? {
+        1 => png::BitDepth::One,
+        2 => png::BitDepth::Two,
+        4 => png::BitDepth::Four,
+        8 => png::BitDepth::Eight,
+        16 => png::BitDepth::Sixteen,
+        _ => return None,
+    };
+    let color = match *bytes.get(25)? {
+        0 => png::ColorType::Grayscale,
+        2 => png::ColorType::Rgb,
+        3 => png::ColorType::Indexed,
+        4 => png::ColorType::GrayscaleAlpha,
+        6 => png::ColorType::Rgba,
+        _ => return None,
+    };
+    Some((w, h, depth, color))
+}
+
+/// Output channels of a decoded frame for the given color type (post
+/// STRIP_16; palette expands to RGB).
+fn output_channels(color: png::ColorType) -> usize {
+    match color {
+        png::ColorType::Grayscale => 1,
+        png::ColorType::GrayscaleAlpha => 2,
+        png::ColorType::Rgb | png::ColorType::Indexed => 3,
+        png::ColorType::Rgba => 4,
+    }
+}
+
+/// The 32→24 rule: RGBA (32-bit) becomes RGB with alpha composited onto
+/// white; 16-bit per channel was already stripped to 8-bit at decode.
+fn flatten_rgba_to_rgb(rgba: &[u8]) -> Vec<u8> {
+    let mut rgb = Vec::with_capacity(rgba.len() / 4 * 3);
+    for px in rgba.chunks_exact(4) {
+        let a = px[3] as u32;
+        for v in &px[..3] {
+            // Over white: C = C*a/255 + 255*(1 - a/255).
+            let v = (*v as u32 * a + 255 * (255 - a)) / 255;
+            rgb.push(v.min(255) as u8);
+        }
+    }
+    rgb
+}
+
+/// Lanczos3 downsampling via fast_image_resize.
+fn resize_pix(pix: PixBuf, w: u32, h: u32, nw: u32, nh: u32) -> Result<PixBuf, String> {
+    let pixel_type = pix.pixel_type();
+    let src = fir::images::Image::from_vec_u8(w, h, pix.into_vec(), pixel_type)
+        .map_err(|e| format!("resize input: {e}"))?;
+    let mut dst = fir::images::Image::new(nw, nh, pixel_type);
+    let mut resizer = fir::Resizer::new();
+    let options = fir::ResizeOptions::default()
+        .resize_alg(fir::ResizeAlg::Convolution(fir::FilterType::Lanczos3));
+    resizer
+        .resize(&src, &mut dst, &options)
+        .map_err(|e| format!("resize: {e}"))?;
+    let buf = dst.into_vec();
+    Ok(match pixel_type {
+        fir::PixelType::U8 => PixBuf::Gray(buf),
+        fir::PixelType::U8x2 => PixBuf::GrayAlpha(buf),
+        _ => PixBuf::Rgb(buf),
+    })
+}
+
+fn encode_png(out: &mut Vec<u8>, pix: &PixBuf, w: u32, h: u32) -> Result<(), png::EncodingError> {
+    let (color, buf) = match pix {
+        PixBuf::Gray(v) => (png::ColorType::Grayscale, v.as_slice()),
+        PixBuf::GrayAlpha(v) => (png::ColorType::GrayscaleAlpha, v.as_slice()),
+        PixBuf::Rgb(v) => (png::ColorType::Rgb, v.as_slice()),
+    };
+    let mut encoder = png::Encoder::new(out, w, h);
+    encoder.set_color(color);
+    encoder.set_depth(png::BitDepth::Eight);
+    let mut writer = encoder.write_header()?;
+    writer.write_image_data(buf)?;
+    Ok(())
+}
+
+fn encode_jpeg(pix: &PixBuf, w: u32, h: u32) -> Result<Vec<u8>, String> {
+    use jpeg_encoder::{ColorType, Encoder};
+    let mut out = Vec::new();
+    let encoder = Encoder::new(&mut out, JPEG_QUALITY);
+    match pix {
+        PixBuf::Gray(v) => encoder
+            .encode(v, w as u16, h as u16, ColorType::Luma)
+            .map_err(|e| format!("jpeg encode: {e}"))?,
+        PixBuf::GrayAlpha(v) => {
+            // JPEG has no alpha: composite onto white, output as gray.
+            let gray: Vec<u8> = v
+                .chunks_exact(2)
+                .map(|px| {
+                    let (g, a) = (px[0] as u32, px[1] as u32);
+                    ((g * a + 255 * (255 - a)) / 255).min(255) as u8
+                })
+                .collect();
+            encoder
+                .encode(&gray, w as u16, h as u16, ColorType::Luma)
+                .map_err(|e| format!("jpeg encode: {e}"))?;
+        }
+        PixBuf::Rgb(v) => encoder
+            .encode(v, w as u16, h as u16, ColorType::Rgb)
+            .map_err(|e| format!("jpeg encode: {e}"))?,
+    }
+    Ok(out)
+}
+
+fn write_temp(bytes: &[u8], ext: &str) -> Result<NamedTempFile, String> {
+    let mut file = tempfile::Builder::new()
+        .suffix(&format!(".{ext}"))
+        .tempfile()
+        .map_err(|e| format!("temp file failed: {e}"))?;
+    file.as_file_mut()
+        .write_all(bytes)
+        .map_err(|e| format!("temp file write failed: {e}"))?;
+    Ok(file)
+}
+
+fn target_dims(w: u32, h: u32) -> (u32, u32) {
+    let scale = PHOTO_TARGET_DIMENSION_SUM as f64 / (w + h) as f64;
+    (
+        ((w as f64 * scale).round() as u32).max(1),
+        ((h as f64 * scale).round() as u32).max(1),
+    )
+}
+
+/// PNG branch: decode (16→8, palette→RGB; gray/GA stay), flatten RGBA to
+/// RGB, Lanczos-downscale beyond the dimension cap, encode PNG — a PNG still
+/// over the upload cap afterwards becomes JPEG.
+fn prepare_png(file: NamedTempFile, bytes: Vec<u8>) -> Result<PhotoPrep, String> {
+    let (w, h, _bit_depth, color_type) =
+        parse_png_header(&bytes).ok_or("invalid PNG header")?;
+    let size_over = bytes.len() as u64 > MAX_UPLOAD_BYTES;
+    if w + h <= PHOTO_MAX_DIMENSION_SUM && !size_over {
+        return Ok(PhotoPrep::Upload(file));
+    }
+    log::info!("photo {w}x{h} ({_bit_depth:?} {color_type:?}, {} bytes) needs processing", bytes.len());
+
+    let channels = output_channels(color_type);
+    if (w as u64) * (h as u64) * channels as u64 > MAX_DECODE_BYTES {
+        log::warn!("photo decode buffer exceeds the memory budget; falling back to smaller media");
+        return Ok(PhotoPrep::UseFallback);
+    }
+
+    // STRIP_16 drops 16-bit to 8-bit (the depth-reduction step); palette
+    // expands to RGB (resampling requires it). Gray and gray-alpha are kept.
+    let transforms = match color_type {
+        png::ColorType::Indexed => png::Transformations::EXPAND,
+        _ => png::Transformations::STRIP_16,
+    };
+    let mut decoder = png::Decoder::new(std::io::Cursor::new(&bytes));
+    decoder.set_transformations(transforms);
+    let mut reader = decoder.read_info().map_err(|e| format!("png decode: {e}"))?;
+    let out_w = reader.info().width;
+    let out_h = reader.info().height;
+    let mut buf = vec![
+        0u8;
+        reader
+            .output_buffer_size()
+            .ok_or("png output buffer size")?
+    ];
+    reader
+        .next_frame(&mut buf)
+        .map_err(|e| format!("png frame: {e}"))?;
+
+    let mut pix = match color_type {
+        png::ColorType::Rgba => PixBuf::Rgb(flatten_rgba_to_rgb(&buf)),
+        png::ColorType::Grayscale => PixBuf::Gray(buf),
+        png::ColorType::GrayscaleAlpha => PixBuf::GrayAlpha(buf),
+        png::ColorType::Rgb | png::ColorType::Indexed => PixBuf::Rgb(buf),
+    };
+
+    let (mut w, mut h) = (out_w, out_h);
+    if w + h > PHOTO_MAX_DIMENSION_SUM {
+        let (nw, nh) = target_dims(w, h);
+        pix = resize_pix(pix, w, h, nw, nh)?;
+        (w, h) = (nw, nh);
+        log::info!("downscaled photo to {w}x{h} (Lanczos3)");
+    }
+
+    let mut png_bytes = Vec::new();
+    encode_png(&mut png_bytes, &pix, w, h).map_err(|e| format!("png encode: {e}"))?;
+    if png_bytes.len() as u64 <= MAX_UPLOAD_BYTES {
+        return Ok(PhotoPrep::Upload(write_temp(&png_bytes, "png")?));
+    }
+    log::info!("PNG still over the upload cap after processing; transcoding to JPEG");
+    let jpeg_bytes = encode_jpeg(&pix, w, h)?;
+    if jpeg_bytes.len() as u64 <= MAX_UPLOAD_BYTES {
+        return Ok(PhotoPrep::Upload(write_temp(&jpeg_bytes, "jpg")?));
+    }
+    log::warn!("processed photo still exceeds the upload cap; falling back to smaller media");
+    Ok(PhotoPrep::UseFallback)
+}
+
+/// JPEG branch: zune-jpeg decode → Lanczos downscale → jpeg-encoder output.
+fn prepare_jpeg(file: NamedTempFile, bytes: Vec<u8>) -> Result<PhotoPrep, String> {
+    let mut decoder = zune_jpeg::JpegDecoder::new(std::io::Cursor::new(&bytes));
+    // Decodes to RGB by default. Headers first so dimensions are known before
+    // the (potentially huge) pixel decode.
+    decoder
+        .decode_headers()
+        .map_err(|e| format!("jpeg headers: {e}"))?;
+    let info = decoder.info().ok_or("jpeg info unavailable")?;
+    let (w, h) = (info.width as u32, info.height as u32);
+    let size_over = bytes.len() as u64 > MAX_UPLOAD_BYTES;
+    if w + h <= PHOTO_MAX_DIMENSION_SUM && !size_over {
+        return Ok(PhotoPrep::Upload(file));
+    }
+    if (w as u64) * (h as u64) * 3 > MAX_DECODE_BYTES {
+        log::warn!("photo decode buffer exceeds the memory budget; falling back to smaller media");
+        return Ok(PhotoPrep::UseFallback);
+    }
+    let pixels = decoder.decode().map_err(|e| format!("jpeg decode: {e}"))?;
+    let mut pix = PixBuf::Rgb(pixels);
+    let (mut w, mut h) = (w, h);
+    if w + h > PHOTO_MAX_DIMENSION_SUM {
+        let (nw, nh) = target_dims(w, h);
+        pix = resize_pix(pix, w, h, nw, nh)?;
+        (w, h) = (nw, nh);
+        log::info!("downscaled jpeg to {w}x{h} (Lanczos3)");
+    }
+    let jpeg_bytes = encode_jpeg(&pix, w, h)?;
+    if jpeg_bytes.len() as u64 <= MAX_UPLOAD_BYTES {
+        return Ok(PhotoPrep::Upload(write_temp(&jpeg_bytes, "jpg")?));
+    }
+    log::warn!("processed photo still exceeds the upload cap; falling back to smaller media");
+    Ok(PhotoPrep::UseFallback)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn png_header(w: u32, h: u32, depth: u8, color: u8) -> Vec<u8> {
+        let mut bytes = b"\x89PNG\r\n\x1a\n\x00\x00\x00\rIHDR".to_vec();
+        bytes.extend(w.to_be_bytes());
+        bytes.extend(h.to_be_bytes());
+        bytes.extend([depth, color, 0, 0, 0]);
+        bytes
+    }
+
+    #[test]
+    fn parses_png_header() {
+        let bytes = png_header(8979, 5316, 16, 6); // 16-bit RGBA
+        let (w, h, depth, color) = parse_png_header(&bytes).unwrap();
+        assert_eq!((w, h), (8979, 5316));
+        assert_eq!(depth, png::BitDepth::Sixteen);
+        assert_eq!(color, png::ColorType::Rgba);
+
+        let (_, _, depth, color) = parse_png_header(&png_header(10, 10, 8, 0)).unwrap();
+        assert_eq!(depth, png::BitDepth::Eight);
+        assert_eq!(color, png::ColorType::Grayscale);
+
+        assert!(parse_png_header(b"not a png").is_none());
+    }
+
+    #[test]
+    fn flatten_rgba_to_rgb_composites_over_white() {
+        // opaque red stays red
+        assert_eq!(flatten_rgba_to_rgb(&[255, 0, 0, 255]), vec![255, 0, 0]);
+        // fully transparent → white
+        assert_eq!(flatten_rgba_to_rgb(&[0, 0, 0, 0]), vec![255, 255, 255]);
+        // half alpha red → (255+255)/2 = 255, (0*128 + 255*127)/255 = 127
+        let out = flatten_rgba_to_rgb(&[255, 0, 0, 128]);
+        assert_eq!(out[0], 255);
+        assert_eq!(out[1], 127);
+        assert_eq!(out[2], 127);
+    }
+
+    #[test]
+    fn target_dims_stay_under_the_cap() {
+        for (w, h) in [(12000u32, 7000u32), (10000, 10000), (8979, 5316)] {
+            let (nw, nh) = target_dims(w, h);
+            assert!(nw + nh <= PHOTO_MAX_DIMENSION_SUM, "{w}x{h} -> {nw}x{nh}");
+            assert!(nw >= 1 && nh >= 1);
+        }
+        // already within limits: no change expected from the caller, but the
+        // helper must not produce zero dimensions.
+        let (nw, nh) = target_dims(500, 400);
+        assert!(nw >= 1 && nh >= 1);
+    }
+
+    #[test]
+    fn resize_pix_changes_dimensions() {
+        // 300x200 RGB → 100x66
+        let buf: Vec<u8> = (0..300 * 200 * 3).map(|i| (i % 251) as u8).collect();
+        let resized = resize_pix(PixBuf::Rgb(buf), 300, 200, 100, 66).unwrap();
+        match resized {
+            PixBuf::Rgb(v) => assert_eq!(v.len(), 100 * 66 * 3),
+            other => panic!("expected rgb, got {other:?}"),
+        }
+    }
+
+    #[test]
+    fn png_encode_roundtrip_keeps_gray() {
+        let gray = vec![128u8; 4 * 4];
+        let mut out = Vec::new();
+        encode_png(&mut out, &PixBuf::Gray(gray), 4, 4).unwrap();
+        assert!(!out.is_empty());
+        let (_, _, depth, color) = parse_png_header(&out).unwrap();
+        assert_eq!(depth, png::BitDepth::Eight);
+        assert_eq!(color, png::ColorType::Grayscale);
+    }
+
+    #[test]
+    fn jpeg_encode_produces_bytes() {
+        let rgb = vec![128u8; 8 * 8 * 3];
+        let out = encode_jpeg(&PixBuf::Rgb(rgb), 8, 8).unwrap();
+        assert!(out.len() > 100);
+        assert!(out.starts_with(&[0xFF, 0xD8]));
+    }
+
+    /// Writes a small dimension-oversized PNG (9999x2 → sum 10001) to a temp
+    /// file and runs the full pipeline.
+    fn run_pipeline(w: u32, h: u32, color: png::ColorType, fill: u8) -> Result<PhotoPrep, String> {
+        let (channels, data): (usize, Vec<u8>) = match color {
+            png::ColorType::Grayscale => (1, vec![fill; (w * h) as usize]),
+            png::ColorType::Rgb => (3, vec![fill; (w * h * 3) as usize]),
+            _ => unreachable!(),
+        };
+        let mut bytes = Vec::new();
+        {
+            let mut encoder = png::Encoder::new(&mut bytes, w, h);
+            encoder.set_color(color);
+            encoder.set_depth(png::BitDepth::Eight);
+            let mut writer = encoder.write_header().unwrap();
+            writer.write_image_data(&data).unwrap();
+        }
+        assert_eq!(data.len(), channels * (w * h) as usize);
+
+        let mut file = tempfile::Builder::new().suffix(".png").tempfile().unwrap();
+        std::io::Write::write_all(file.as_file_mut(), &bytes).unwrap();
+        prepare_photo(file)
+    }
+
+    #[test]
+    fn pipeline_downscales_oversized_png_keeping_format() {
+        let prep = run_pipeline(9999, 2, png::ColorType::Rgb, 128).unwrap();
+        match prep {
+            PhotoPrep::Upload(file) => {
+                let out = std::fs::read(file.path()).unwrap();
+                let (w, h, depth, color) = parse_png_header(&out).unwrap();
+                assert!(w + h <= PHOTO_MAX_DIMENSION_SUM, "{w}x{h}");
+                assert_eq!(depth, png::BitDepth::Eight);
+                assert_eq!(color, png::ColorType::Rgb);
+            }
+            PhotoPrep::UseFallback => panic!("over-dimension PNG should have been resized"),
+        }
+    }
+
+    #[test]
+    fn pipeline_keeps_gray_png_gray() {
+        let prep = run_pipeline(9999, 2, png::ColorType::Grayscale, 200).unwrap();
+        match prep {
+            PhotoPrep::Upload(file) => {
+                let out = std::fs::read(file.path()).unwrap();
+                let (_, _, _, color) = parse_png_header(&out).unwrap();
+                assert_eq!(color, png::ColorType::Grayscale, "gray must not upconvert");
+            }
+            PhotoPrep::UseFallback => panic!("over-dimension gray PNG should have been resized"),
+        }
+    }
+
+    #[test]
+    fn pipeline_resizes_oversized_jpeg() {
+        // Build a small over-dimension JPEG with jpeg-encoder.
+        let (w, h) = (9999u16, 2u16);
+        let rgb = vec![90u8; (w as usize) * (h as usize) * 3];
+        let mut bytes = Vec::new();
+        {
+            let encoder = jpeg_encoder::Encoder::new(&mut bytes, 90);
+            encoder.encode(&rgb, w, h, jpeg_encoder::ColorType::Rgb).unwrap();
+        }
+        let mut file = tempfile::Builder::new().suffix(".jpg").tempfile().unwrap();
+        std::io::Write::write_all(file.as_file_mut(), &bytes).unwrap();
+        match prepare_photo(file).unwrap() {
+            PhotoPrep::Upload(file) => {
+                let out = std::fs::read(file.path()).unwrap();
+                assert!(out.starts_with(&[0xFF, 0xD8]), "output must stay jpeg");
+                // 9999x2 downscaled: the buffer length tells the new dims.
+                assert!(out.len() > 100);
+            }
+            PhotoPrep::UseFallback => panic!("over-dimension JPEG should have been resized"),
+        }
+    }
+
+    #[test]
+    #[ignore = "heavy: generates a >10 MiB PNG (run explicitly)"]
+    fn pipeline_transcodes_oversized_png_to_jpeg() {
+        // 6000x4000 (sum 10000 — under the dimension cap) smooth gradient with
+        // small per-pixel noise: PNG-incompressible (delta filters defeated)
+        // but JPEG-friendly (DCT smooths the small noise). Verified with
+        // ffmpeg: 8000x6000 amp-5 variant is a 59 MB PNG / 3.3 MB JPEG.
+        let (w, h) = (6000u32, 4000u32);
+        let mut rng = 0x1234_5678_9abc_def0u64;
+        let mut data = Vec::with_capacity((w * h * 3) as usize);
+        for y in 0..h {
+            for x in 0..w {
+                let base = (x + y) * 255 / (w + h);
+                rng = rng.wrapping_mul(6364136223846793005).wrapping_add(1442695040888963407);
+                let n = ((rng >> 33) % 11) as i32 - 5; // noise in [-5, 5]
+                let v = (base as i32 + n).clamp(0, 255) as u8;
+                data.extend_from_slice(&[v, v, v]);
+            }
+        }
+        let mut bytes = Vec::new();
+        {
+            let mut encoder = png::Encoder::new(&mut bytes, w, h);
+            encoder.set_color(png::ColorType::Rgb);
+            encoder.set_depth(png::BitDepth::Eight);
+            let mut writer = encoder.write_header().unwrap();
+            writer.write_image_data(&data).unwrap();
+        }
+        assert!(bytes.len() as u64 > MAX_UPLOAD_BYTES, "test needs a >10MiB PNG, got {}", bytes.len());
+
+        let mut file = tempfile::Builder::new().suffix(".png").tempfile().unwrap();
+        std::io::Write::write_all(file.as_file_mut(), &bytes).unwrap();
+        match prepare_photo(file).unwrap() {
+            PhotoPrep::Upload(file) => {
+                let out = std::fs::read(file.path()).unwrap();
+                assert!(out.starts_with(&[0xFF, 0xD8]), "must transcode to JPEG");
+                assert!(out.len() as u64 <= MAX_UPLOAD_BYTES);
+            }
+            PhotoPrep::UseFallback => panic!("PNG over the byte cap must transcode to JPEG"),
+        }
+    }
+}

crates/xmedia-bot/src/queue.rs

@@ -6,7 +6,7 @@
 //! replaced by dedicated columns.
 
 use parking_lot::Mutex;
-use rusqlite::{params, Connection};
+use rusqlite::{params, Connection, TransactionBehavior};
 use serde_json::Value;
 use std::pin::Pin;
 use std::sync::atomic::{AtomicBool, AtomicU64, Ordering};
@@ -18,6 +18,12 @@ use tokio::task::JoinHandle;
 pub const MAX_RETRIES: u32 = 2;
 pub const LOCK_TTL_SECONDS: f64 = 120.0;
 
+/// Number of concurrent worker loops. Tasks are independent (retries and
+/// forward resumes); leases serialize row claims via SQLite transactions, so
+/// extra workers drain backlogs faster. Each worker can be mid-send to
+/// Telegram at the same time as handler tasks, so keep this modest.
+const QUEUE_WORKERS: usize = 4;
+
 /// What a handler returns instead of throwing. The payload it carries is the
 /// (possibly updated) task state to persist for the next attempt.
 pub enum QueueError {
@@ -42,7 +48,7 @@ pub struct PersistentTaskQueue {
     db_path: String,
     notify: Arc<Notify>,
     stop: Arc<AtomicBool>,
-    worker: Mutex<Option<JoinHandle<()>>>,
+    worker: Mutex<Vec<JoinHandle<()>>>,
     counter: AtomicU64,
 }
 
@@ -68,6 +74,15 @@ fn now_f64() -> f64 {
         .unwrap_or(0.0)
 }
 
+/// Opens the queue DB with a busy timeout. Handler tasks enqueue while
+/// workers lease/update rows concurrently; without the timeout a concurrent
+/// write fails immediately with SQLITE_BUSY and the operation is lost.
+fn open_db(path: &str) -> rusqlite::Result<Connection> {
+    let conn = Connection::open(path)?;
+    conn.busy_timeout(Duration::from_secs(5))?;
+    Ok(conn)
+}
+
 fn ensure_schema(conn: &Connection) -> rusqlite::Result<()> {
     conn.execute_batch(
         "CREATE TABLE IF NOT EXISTS tasks (id TEXT PRIMARY KEY, payload TEXT NOT NULL, \
@@ -96,12 +111,12 @@ impl PersistentTaskQueue {
             db_path: db_path.to_string(),
             notify: Arc::new(Notify::new()),
             stop: Arc::new(AtomicBool::new(false)),
-            worker: Mutex::new(None),
+            worker: Mutex::new(Vec::new()),
             counter: AtomicU64::new(0),
         }
     }
 
-    /// Starts the worker loop. Also recovers rows left `in_progress` by a
+    /// Starts the worker loops. Also recovers rows left `in_progress` by a
     /// previous process (lease expired).
     pub async fn start<H, F, D, G>(&self, handler: H, dead_letter: D)
     where
@@ -114,21 +129,25 @@ impl PersistentTaskQueue {
         let dead_letter: Arc<DeadLetter> =
             Arc::new(move |payload, message| Box::pin(dead_letter(payload, message)));
         self.recover_stale().await;
-        let worker = QueueWorker {
-            db_path: self.db_path.clone(),
-            notify: Arc::clone(&self.notify),
-            stop: Arc::clone(&self.stop),
-            handler,
-            dead_letter,
-        };
-        let worker = tokio::spawn(worker.run_loop());
-        *self.worker.lock() = Some(worker);
+        let mut handles = Vec::with_capacity(QUEUE_WORKERS);
+        for _ in 0..QUEUE_WORKERS {
+            let worker = QueueWorker {
+                db_path: self.db_path.clone(),
+                notify: Arc::clone(&self.notify),
+                stop: Arc::clone(&self.stop),
+                handler: Arc::clone(&handler),
+                dead_letter: Arc::clone(&dead_letter),
+            };
+            handles.push(tokio::spawn(worker.run_loop()));
+        }
+        *self.worker.lock() = handles;
     }
 
     pub async fn stop(&self) {
         self.stop.store(true, Ordering::Relaxed);
-        self.notify.notify_one();
-        if let Some(handle) = self.worker.lock().take() {
+        self.notify.notify_waiters();
+        let handles = std::mem::take(&mut *self.worker.lock());
+        for handle in handles {
             let _ = handle.await;
         }
     }
@@ -145,8 +164,8 @@ impl PersistentTaskQueue {
         let payload = payload.to_string();
         let db_path = self.db_path.clone();
         log::info!("enqueued {id} (run_after {run_after:.1})");
-        let result = tokio::task::spawn_blocking(move || -> rusqlite::Result<()> {
-            let conn = Connection::open(&db_path)?;
+        tokio::task::spawn_blocking(move || -> rusqlite::Result<()> {
+            let conn = open_db(&db_path)?;
             conn.execute(
                 "INSERT OR REPLACE INTO tasks (id, payload, run_after, attempts, status, locked_until, created_at) \
                  VALUES (?1, ?2, ?3, 0, 'pending', 0, ?4)",
@@ -156,14 +175,16 @@ impl PersistentTaskQueue {
         })
         .await
         .expect("queue insert worker panicked")?;
-        self.notify.notify_one();
-        Ok(result)
+        // Wake every sleeping worker: with several workers the one that finds
+        // nothing due must not starve the newly inserted row.
+        self.notify.notify_waiters();
+        Ok(())
     }
 
     async fn recover_stale(&self) {
         let db_path = self.db_path.clone();
         tokio::task::spawn_blocking(move || -> rusqlite::Result<()> {
-            let conn = Connection::open(&db_path)?;
+            let conn = open_db(&db_path)?;
             conn.execute(
                 "UPDATE tasks SET status='pending', locked_until=0 WHERE status='in_progress' AND locked_until < ?1",
                 params![now_f64()],
@@ -206,11 +227,15 @@ impl QueueWorker {
     async fn lease_next(&self) -> Option<LeasedRow> {
         let db_path = self.db_path.clone();
         tokio::task::spawn_blocking(move || -> rusqlite::Result<Option<LeasedRow>> {
-            let mut conn = Connection::open(&db_path)?;
-            let tx = conn.transaction()?;
+            let mut conn = open_db(&db_path)?;
+            // BEGIN IMMEDIATE: with several workers, a deferred transaction
+            // that read before another worker's lease commit would fail with
+            // SQLITE_BUSY_SNAPSHOT. Taking the write lock up front serializes
+            // leases and re-reads the freshest committed state.
+            let tx = conn.transaction_with_behavior(TransactionBehavior::Immediate)?;
             let now = now_f64();
             let row = tx.query_row(
-                "SELECT id, payload, attempts FROM tasks WHERE status='pending' AND run_after <= ?1 \
+                "SELECT id, payload, attempts FROM tasks WHERE status='pending' AND run_after <= ?1 AND locked_until <= ?1 \
                  ORDER BY run_after LIMIT 1",
                 params![now],
                 |r| {
@@ -251,7 +276,7 @@ impl QueueWorker {
     async fn earliest_run_after(&self) -> Option<f64> {
         let db_path = self.db_path.clone();
         tokio::task::spawn_blocking(move || -> rusqlite::Result<Option<f64>> {
-            let conn = Connection::open(&db_path)?;
+            let conn = open_db(&db_path)?;
             let mut stmt = conn.prepare("SELECT MIN(run_after) FROM tasks WHERE status='pending'")?;
             let mut rows = stmt.query([])?;
             match rows.next()? {
@@ -314,7 +339,7 @@ impl QueueWorker {
         let db_path = self.db_path.clone();
         let id = id.to_string();
         tokio::task::spawn_blocking(move || -> rusqlite::Result<()> {
-            let conn = Connection::open(&db_path)?;
+            let conn = open_db(&db_path)?;
             conn.execute("DELETE FROM tasks WHERE id = ?1", params![id])?;
             Ok(())
         })
@@ -328,7 +353,7 @@ impl QueueWorker {
         let id = id.to_string();
         let payload = payload.to_string();
         tokio::task::spawn_blocking(move || -> rusqlite::Result<()> {
-            let conn = Connection::open(&db_path)?;
+            let conn = open_db(&db_path)?;
             conn.execute(
                 "UPDATE tasks SET payload=?1, run_after=?2, attempts=?3, status='pending', locked_until=0 WHERE id=?4",
                 params![payload, now_f64() + delay_seconds, attempts, id],
@@ -338,7 +363,7 @@ impl QueueWorker {
         .await
         .expect("queue reschedule worker panicked")
         .unwrap_or_else(|e| log::error!("queue reschedule failed: {e}"));
-        self.notify.notify_one();
+        self.notify.notify_waiters();
     }
 }
 

crates/xmedia-bot/src/send.rs

@@ -3,7 +3,9 @@
 //! URL is blocked by hotlink protection; the bot downloads the file itself
 //! and uploads it via multipart).
 
-use crate::handlers::{CHAT_STORE, TASK_QUEUE};
+use crate::handlers::{CHAT_STORE, LINK_CACHE, TASK_QUEUE};
+use crate::link_cache::{CachedMedia, CachedMediaKind, CachedPost};
+use crate::photo::{self, PhotoPrep, MAX_UPLOAD_BYTES};
 use crate::queue::QueueError;
 use crate::state::{EditMessage, unix_now};
 use rand::Rng;
@@ -25,18 +27,43 @@ pub enum MediaItemPayload {
     Photo {
         media: String,
         has_spoiler: bool,
+        /// Smaller variant used when the primary media exceeds Telegram's
+        /// size limits.
+        #[serde(default)]
+        fallback_url: Option<String>,
+        /// `media` is a Telegram file id (link-cache hit), not a URL.
+        #[serde(default)]
+        file_id: bool,
     },
     Video {
         media: String,
         has_spoiler: bool,
         thumbnail: Option<String>,
+        #[serde(default)]
+        fallback_url: Option<String>,
+        /// `media` is a Telegram file id (link-cache hit), not a URL.
+        #[serde(default)]
+        file_id: bool,
     },
     Animation {
         media: String,
         has_spoiler: bool,
+        /// `media` is a Telegram file id (link-cache hit), not a URL.
+        #[serde(default)]
+        file_id: bool,
     },
 }
 
+impl MediaItemPayload {
+    fn fallback_url(&self) -> Option<&str> {
+        match self {
+            MediaItemPayload::Photo { fallback_url, .. }
+            | MediaItemPayload::Video { fallback_url, .. } => fallback_url.as_deref(),
+            MediaItemPayload::Animation { .. } => None,
+        }
+    }
+}
+
 #[derive(Serialize, Deserialize, Clone, Debug)]
 #[serde(tag = "type", rename_all = "snake_case")]
 pub enum Task {
@@ -52,6 +79,10 @@ pub enum Task {
         forward_channel_id: Option<i64>,
         notify_chat_id: Option<i64>,
         notify_message_id: Option<i64>,
+        /// Raw render data captured on a cache miss; the send fills in the
+        /// Telegram file ids and persists the entry (see `link_cache`).
+        #[serde(default)]
+        cache_data: Option<CachedPost>,
     },
     SendAnimation {
         chat_id: i64,
@@ -63,6 +94,10 @@ pub enum Task {
         forward_channel_id: Option<i64>,
         notify_chat_id: Option<i64>,
         notify_message_id: Option<i64>,
+        /// Raw render data captured on a cache miss; the send fills in the
+        /// Telegram file id and persists the entry (see `link_cache`).
+        #[serde(default)]
+        cache_data: Option<CachedPost>,
     },
     ForwardMessages {
         from_chat_id: i64,
@@ -73,8 +108,109 @@ pub enum Task {
     },
 }
 
+impl Task {
+    fn cache_data(&self) -> Option<&CachedPost> {
+        match self {
+            Task::SendMediaSequence { cache_data, .. }
+            | Task::SendAnimation { cache_data, .. } => cache_data.as_ref(),
+            Task::ForwardMessages { .. } => None,
+        }
+    }
+
+    fn source_url(&self) -> Option<&str> {
+        match self {
+            Task::SendMediaSequence { source_url, .. }
+            | Task::SendAnimation { source_url, .. } => Some(source_url),
+            Task::ForwardMessages { .. } => None,
+        }
+    }
+
+    /// True when the media payloads are Telegram file ids from the link cache
+    /// (a cached file id that goes permanently bad should be dropped so the
+    /// next request re-fetches).
+    fn is_cached_send(&self) -> bool {
+        self.cache_data().is_some_and(|c| !c.media.is_empty())
+    }
+}
+
+/// Telegram file id of the message's media, matched to the payload kind.
+fn file_id_of_message(message: &Message, item: &MediaItemPayload) -> Option<String> {
+    match item {
+        // `photo()` returns all sizes, smallest first — the largest carries
+        // the file id of the sent media.
+        MediaItemPayload::Photo { .. } => {
+            message.photo().and_then(|sizes| sizes.last()).map(|p| p.file.id.to_string())
+        }
+        MediaItemPayload::Video { .. } => message.video().map(|v| v.file.id.to_string()),
+        MediaItemPayload::Animation { .. } => message.animation().map(|a| a.file.id.to_string()),
+    }
+}
+
+fn kind_of_item(item: &MediaItemPayload) -> CachedMediaKind {
+    match item {
+        MediaItemPayload::Photo { .. } => CachedMediaKind::Photo,
+        MediaItemPayload::Video { .. } => CachedMediaKind::Video,
+        MediaItemPayload::Animation { .. } => CachedMediaKind::Animation,
+    }
+}
+
+/// Collects the Telegram file ids of a sent media group, aligned to the
+/// batch's items.
+fn collect_file_ids(messages: &[Message], batch: &[MediaItemPayload], out: &mut Vec<CachedMedia>) {
+    for (message, item) in messages.iter().zip(batch.iter()) {
+        if let Some(file_id) = file_id_of_message(message, item) {
+            out.push(CachedMedia {
+                kind: kind_of_item(item),
+                file_id,
+            });
+        }
+    }
+}
+
+/// Persists a successful send under the post's cache key. Only runs for a
+/// fresh (non-resumed) task that carried raw cache data with no file ids yet.
+async fn cache_sent_task(task: &Task, media: Vec<CachedMedia>) {
+    let Some(cache_data) = task.cache_data() else {
+        return;
+    };
+    if !cache_data.media.is_empty() || media.is_empty() {
+        return;
+    }
+    let mut post = cache_data.clone();
+    post.media = media;
+    if let Some(key) = x_media::site::cache_key(&post.url) {
+        LINK_CACHE.put(&key, &post).await;
+        log::info!("cached send for {}", post.url);
+    }
+}
+
+/// Persists a lone animation send under the post's cache key.
+async fn cache_animation_send(task: &Task, message: &Message) {
+    if let Some(file_id) = message.animation().map(|a| a.file.id.to_string()) {
+        cache_sent_task(
+            task,
+            vec![CachedMedia {
+                kind: CachedMediaKind::Animation,
+                file_id,
+            }],
+        )
+        .await;
+    }
+}
+
+/// A cached Telegram file id failed permanently (stale/expired); drop the
+/// cache entry so the next request re-fetches instead of repeating it.
+pub async fn invalidate_cache(task: &Task) {
+    if task.is_cached_send()
+        && let Some(url) = task.source_url()
+        && let Some(key) = x_media::site::cache_key(url)
+    {
+        log::info!("removing stale link cache entry for {url}");
+        LINK_CACHE.remove(&key).await;
+    }
+}
+
 pub const MAX_MEDIA_GROUP: usize = 9;
-pub const MAX_UPLOAD_BYTES: u64 = 50 * 1024 * 1024; // Telegram Bot API upload cap
 
 /// Splits media into batches of at most [`MAX_MEDIA_GROUP`] items.
 pub fn chunk_media_items<T: Clone>(items: Vec<T>) -> Vec<Vec<T>> {
@@ -91,17 +227,32 @@ pub fn retry_delay_seconds(attempts: u32) -> f64 {
 /// these errors are handled by the download-and-reupload fallback, NOT by a
 /// queue retry (resending the URL cannot succeed).
 pub fn is_media_fetch_failure(e: &ApiError) -> bool {
-    const MARKERS: [&str; 5] = [
+    const MARKERS: [&str; 6] = [
         "webpage_media_empty",
         "media_empty",
         "empty_web_media",
         "webpage_curl_failed",
         "timeout",
+        // Oversized photos (width + height > 10000 px) are rejected on URL
+        // sends too; route them to the download-and-resize fallback.
+        "PHOTO_INVALID_DIMENSIONS",
     ];
     let description = e.to_string().to_lowercase();
     MARKERS.iter().any(|marker| description.contains(marker))
 }
 
+/// Telegram reported the media file as too large (HTTP 413 on multipart
+/// upload, or a "too large" message for URL-fetched media). These errors are
+/// handled by the size-check fallback (use a smaller media URL), NOT by a
+/// queue retry.
+pub fn is_size_error(e: &ApiError) -> bool {
+    if matches!(e, ApiError::RequestEntityTooLarge) {
+        return true;
+    }
+    let description = e.to_string().to_lowercase();
+    ["too large", "too big"].iter().any(|marker| description.contains(marker))
+}
+
 /// Task-free classification of a Telegram request error. The callers attach
 /// the (updated) task when building a [`SendError`].
 pub enum Classification {
@@ -168,6 +319,32 @@ fn input_file_for(media: &str) -> Result<InputFile, String> {
     }
 }
 
+impl MediaItemPayload {
+    /// The input for a send: a cached file id goes out as `InputFile::file_id`
+    /// (no fetch, no upload), URLs go to Telegram, anything else is a local
+    /// path (transient upload fallback).
+    fn input_file(&self) -> Result<InputFile, String> {
+        match self {
+            MediaItemPayload::Photo {
+                media,
+                file_id: true,
+                ..
+            }
+            | MediaItemPayload::Video {
+                media,
+                file_id: true,
+                ..
+            }
+            | MediaItemPayload::Animation {
+                media,
+                file_id: true,
+                ..
+            } => Ok(InputFile::file_id(media.clone().into())),
+            _ => input_file_for(item_url(self)),
+        }
+    }
+}
+
 fn photo_media(file: InputFile, caption: Option<&str>, spoiler: bool) -> InputMedia {
     let mut photo = InputMediaPhoto::new(file).parse_mode(ParseMode::Html);
     if let Some(caption) = caption {
@@ -214,24 +391,22 @@ fn build_media_group(
             let item_caption = if i == 0 { caption } else { None };
             Ok(match item {
                 MediaItemPayload::Photo {
-                    media,
-                    has_spoiler,
-                } => photo_media(input_file_for(media)?, item_caption, *has_spoiler),
+                    has_spoiler, ..
+                } => photo_media(item.input_file()?, item_caption, *has_spoiler),
                 MediaItemPayload::Video {
-                    media,
                     has_spoiler,
                     thumbnail,
+                    ..
                 } => {
-                    let mut video = video_media(input_file_for(media)?, item_caption, *has_spoiler);
+                    let mut video = video_media(item.input_file()?, item_caption, *has_spoiler);
                     if let (Some(thumb), InputMedia::Video(v)) = (thumbnail, &mut video) {
                         *v = v.clone().thumbnail(input_file_for(thumb)?);
                     }
                     video
                 }
                 MediaItemPayload::Animation {
-                    media,
-                    has_spoiler,
-                } => animation_media(input_file_for(media)?, item_caption, *has_spoiler),
+                    has_spoiler, ..
+                } => animation_media(item.input_file()?, item_caption, *has_spoiler),
             })
         })
         .collect()
@@ -258,8 +433,18 @@ fn sniff_ext(bytes: &[u8]) -> &'static str {
 enum FallbackError {
     Retryable { delay_seconds: f64 },
     Permanent { message: String },
+    /// The downloaded file exceeds the upload cap; the caller falls back to
+    /// the item's smaller URL.
+    MediaTooLarge,
 }
 
+/// Brings a downloaded photo within Telegram's limits via the pure-Rust
+/// chain in [`crate::photo`] (no ffmpeg): dimension cap / upload cap
+/// exceeded photos are decoded, downscaled with Lanczos3, PNG bit depth
+/// reduced (>24-bit → 24-bit RGB, ≤24-bit untouched) and transcoded to JPEG
+/// only if still too big. Anything that cannot be fixed falls back to the
+/// item's smaller URL.
+///
 /// Downloads one media item to a temp file (deleted on drop). Network errors
 /// are retryable; size over the upload cap and other download errors are not.
 async fn download_to_temp(item: &MediaItemPayload) -> Result<NamedTempFile, FallbackError> {
@@ -281,10 +466,12 @@ async fn download_to_temp(item: &MediaItemPayload) -> Result<NamedTempFile, Fall
             });
         }
     };
-    if bytes.len() as u64 > MAX_UPLOAD_BYTES {
-        return Err(FallbackError::Permanent {
-            message: "media too large".into(),
-        });
+    // Photos are downloaded even over the cap so `prepare_photo` can
+    // downscale / transcode them; only videos/animations short-circuit.
+    if !matches!(item, MediaItemPayload::Photo { .. })
+        && bytes.len() as u64 > MAX_UPLOAD_BYTES
+    {
+        return Err(FallbackError::MediaTooLarge);
     }
     let ext = sniff_ext(&bytes);
     let mut file = tempfile::Builder::new()
@@ -302,7 +489,48 @@ async fn download_to_temp(item: &MediaItemPayload) -> Result<NamedTempFile, Fall
     Ok(file)
 }
 
-/// Download-and-reupload fallback for one media batch.
+/// Builds the media group item from an uploaded file.
+fn media_from_file(
+    item: &MediaItemPayload,
+    path: std::path::PathBuf,
+    caption: Option<&str>,
+) -> InputMedia {
+    match item {
+        MediaItemPayload::Photo { has_spoiler, .. } => {
+            photo_media(InputFile::file(path), caption, *has_spoiler)
+        }
+        MediaItemPayload::Video { has_spoiler, .. } => {
+            video_media(InputFile::file(path), caption, *has_spoiler)
+        }
+        MediaItemPayload::Animation { has_spoiler, .. } => {
+            animation_media(InputFile::file(path), caption, *has_spoiler)
+        }
+    }
+}
+
+/// Builds the media group item from a (smaller) URL.
+fn media_from_url(
+    item: &MediaItemPayload,
+    url: &str,
+    caption: Option<&str>,
+) -> Result<InputMedia, String> {
+    Ok(match item {
+        MediaItemPayload::Photo { has_spoiler, .. } => {
+            photo_media(input_file_for(url)?, caption, *has_spoiler)
+        }
+        MediaItemPayload::Video { has_spoiler, .. } => {
+            video_media(input_file_for(url)?, caption, *has_spoiler)
+        }
+        MediaItemPayload::Animation { has_spoiler, .. } => {
+            animation_media(input_file_for(url)?, caption, *has_spoiler)
+        }
+    })
+}
+
+/// Download-and-reupload fallback for one media batch. Files over the upload
+/// cap are not downloaded/uploaded; the item falls back to its smaller URL
+/// (which Telegram fetches itself). Returns the fallback-error without the
+/// task attached; callers wrap it with the updated task state.
 async fn send_batch_via_upload(
     bot: &Bot,
     chat_id: i64,
@@ -313,22 +541,90 @@ async fn send_batch_via_upload(
     let mut files = Vec::new();
     let mut items = Vec::new();
     for (i, item) in batch.iter().enumerate() {
-        let file = download_to_temp(item).await?;
-        let path = file.path().to_path_buf();
         let item_caption = if i == 0 { caption } else { None };
-        let media = match item {
-            MediaItemPayload::Photo { has_spoiler, .. } => {
-                photo_media(InputFile::file(path), item_caption, *has_spoiler)
+        // Size check before downloading/uploading: over the cap, use the
+        // smaller URL instead of the file. Photos are exempt — they are
+        // downloaded and processed (downscale / PNG→JPEG) before uploading.
+        let too_large = match x_media::site::media_size(item_url(item)).await {
+            Ok(Some(size)) => size > MAX_UPLOAD_BYTES,
+            _ => false,
+        };
+        let too_large = too_large && !matches!(item, MediaItemPayload::Photo { .. });
+        let media = if too_large {
+            match item.fallback_url() {
+                Some(url) => match media_from_url(item, url, item_caption) {
+                    Ok(media) => media,
+                    Err(message) => {
+                        return Err(FallbackError::Permanent { message });
+                    }
+                },
+                None => {
+                    return Err(FallbackError::Permanent {
+                        message: "media too large".into(),
+                    });
+                }
             }
-            MediaItemPayload::Video { has_spoiler, .. } => {
-                video_media(InputFile::file(path), item_caption, *has_spoiler)
-            }
-            MediaItemPayload::Animation { has_spoiler, .. } => {
-                animation_media(InputFile::file(path), item_caption, *has_spoiler)
+        } else {
+            match download_to_temp(item).await {
+                Ok(file) => {
+                    // Telegram rejects photos wider+taller than 10000 px
+                    // combined (PHOTO_INVALID_DIMENSIONS): downscale the
+                    // downloaded file before uploading; photos that cannot be
+                    // brought within the limits degrade to the smaller URL.
+                    if matches!(item, MediaItemPayload::Photo { .. }) {
+                        // CPU-heavy (decode/resize/encode): run off the async
+                        // executor thread.
+                        let prep = tokio::task::spawn_blocking(move || photo::prepare_photo(file))
+                            .await
+                            .map_err(|e| FallbackError::Permanent {
+                                message: format!("photo worker panicked: {e}"),
+                            })?
+                            .map_err(|message| FallbackError::Permanent { message })?;
+                        match prep {
+                            PhotoPrep::Upload(upload) => {
+                                let path = upload.path().to_path_buf();
+                                files.push(upload);
+                                media_from_file(item, path, item_caption)
+                            }
+                            PhotoPrep::UseFallback => match item.fallback_url() {
+                                Some(url) => match media_from_url(item, url, item_caption) {
+                                    Ok(media) => media,
+                                    Err(message) => {
+                                        return Err(FallbackError::Permanent { message });
+                                    }
+                                },
+                                None => {
+                                    return Err(FallbackError::Permanent {
+                                        message:
+                                            "photo dimensions exceed Telegram limits and no smaller variant is available"
+                                                .into(),
+                                    });
+                                }
+                            },
+                        }
+                    } else {
+                        let path = file.path().to_path_buf();
+                        files.push(file);
+                        media_from_file(item, path, item_caption)
+                    }
+                }
+                Err(FallbackError::MediaTooLarge) => match item.fallback_url() {
+                    Some(url) => match media_from_url(item, url, item_caption) {
+                        Ok(media) => media,
+                        Err(message) => {
+                            return Err(FallbackError::Permanent { message });
+                        }
+                    },
+                    None => {
+                        return Err(FallbackError::Permanent {
+                            message: "media too large".into(),
+                        });
+                    }
+                },
+                Err(e) => return Err(e),
             }
         };
         items.push(media);
-        files.push(file);
     }
     let result = bot
         .send_media_group(ChatId(chat_id), items)
@@ -355,12 +651,14 @@ fn updated_sequence_task(task: &Task, batch_index: usize, sent_message_ids: Vec<
             reply_to_message_id,
             caption,
             media_batches,
+            batch_index: _,
+            sent_message_ids: _,
             source_url,
             edit_before_forward,
             forward_channel_id,
             notify_chat_id,
             notify_message_id,
-            ..
+            cache_data,
         } => Task::SendMediaSequence {
             chat_id: *chat_id,
             reply_to_message_id: *reply_to_message_id,
@@ -373,6 +671,7 @@ fn updated_sequence_task(task: &Task, batch_index: usize, sent_message_ids: Vec<
             forward_channel_id: *forward_channel_id,
             notify_chat_id: *notify_chat_id,
             notify_message_id: *notify_message_id,
+            cache_data: cache_data.clone(),
         },
         _ => unreachable!("updated_sequence_task requires a SendMediaSequence task"),
     }
@@ -397,6 +696,10 @@ pub async fn send_media_sequence(bot: &Bot, task: &Task) -> Result<Vec<i64>, Sen
     let chat_id = *chat_id;
     let reply_to = *reply_to_message_id;
     let mut sent = sent_message_ids.clone();
+    // File ids accumulated across batches for the link cache. Only a fresh
+    // (non-resumed) full send populates the cache.
+    let mut cached_media: Vec<CachedMedia> = Vec::new();
+    let fresh_send = *batch_index == 0 && sent.is_empty();
     for idx in *batch_index..media_batches.len() {
         let batch = &media_batches[idx];
         let caption = if idx == 0 { Some(caption.as_str()) } else { None };
@@ -420,18 +723,21 @@ pub async fn send_media_sequence(bot: &Bot, task: &Task) -> Result<Vec<i64>, Sen
                     media_batches.len(),
                     batch.len()
                 );
+                collect_file_ids(&messages, batch, &mut cached_media);
                 sent.extend(messages.into_iter().map(|m| m.id.0 as i64));
             }
-            Err(RequestError::Api(api)) if is_media_fetch_failure(&api) => {
+            Err(RequestError::Api(api))
+                if is_media_fetch_failure(&api) || is_size_error(&api) =>
+            {
                 log::info!(
                     "Telegram could not fetch media for batch {idx} ({}), downloading and reuploading",
-                    batch
-                        .first()
-                        .map(|item| item_url(item))
-                        .unwrap_or("?")
+                    batch.first().map(item_url).unwrap_or("?")
                 );
                 match send_batch_via_upload(bot, chat_id, reply_to, batch, caption).await {
-                    Ok(messages) => sent.extend(messages.into_iter().map(|m| m.id.0 as i64)),
+                    Ok(messages) => {
+                        collect_file_ids(&messages, batch, &mut cached_media);
+                        sent.extend(messages.into_iter().map(|m| m.id.0 as i64));
+                    }
                     Err(FallbackError::Retryable { delay_seconds }) => {
                         return Err(SendError::Retryable {
                             delay_seconds,
@@ -444,6 +750,7 @@ pub async fn send_media_sequence(bot: &Bot, task: &Task) -> Result<Vec<i64>, Sen
                             task: updated_sequence_task(task, idx, sent),
                         });
                     }
+                    Err(FallbackError::MediaTooLarge) => unreachable!("handled inside upload"),
                 }
             }
             Err(e) => {
@@ -454,6 +761,9 @@ pub async fn send_media_sequence(bot: &Bot, task: &Task) -> Result<Vec<i64>, Sen
             }
         }
     }
+    if fresh_send {
+        cache_sent_task(task, cached_media).await;
+    }
     Ok(sent)
 }
 
@@ -494,6 +804,7 @@ pub async fn send_animation(bot: &Bot, task: &Task) -> Result<Vec<i64>, SendErro
         MediaItemPayload::Animation {
             media,
             has_spoiler,
+            ..
         } => (media, *has_spoiler),
         MediaItemPayload::Photo { .. } | MediaItemPayload::Video { .. } => {
             unreachable!("SendAnimation carries an Animation payload")
@@ -506,34 +817,76 @@ pub async fn send_animation(bot: &Bot, task: &Task) -> Result<Vec<i64>, SendErro
     match send_animation_inner(bot, chat_id, reply_to, caption, has_spoiler, url_file)
         .await
     {
-        Ok(message) => Ok(vec![message.id.0 as i64]),
-        Err(RequestError::Api(api)) if is_media_fetch_failure(&api) => {
+        Ok(message) => {
+            let id = message.id.0 as i64;
+            cache_animation_send(task, &message).await;
+            Ok(vec![id])
+        }
+        Err(RequestError::Api(api))
+            if is_media_fetch_failure(&api) || is_size_error(&api) =>
+        {
             log::info!(
                 "Telegram could not fetch animation URL, downloading and reuploading: {}",
                 media_url
             );
-            let file = match download_to_temp(animation).await {
-                Ok(file) => file,
+            match download_to_temp(animation).await {
+                Ok(file) => {
+                    let path = file.path().to_path_buf();
+                    match send_animation_inner(
+                        bot,
+                        chat_id,
+                        reply_to,
+                        caption,
+                        has_spoiler,
+                        InputFile::file(path),
+                    )
+                    .await
+                    {
+                        Ok(message) => {
+                            let id = message.id.0 as i64;
+                            cache_animation_send(task, &message).await;
+                            Ok(vec![id])
+                        }
+                        Err(e) => Err(classify_to_send_error(&e, task.clone())),
+                    }
+                }
+                // Over the upload cap: fall back to the smaller URL.
+                Err(FallbackError::MediaTooLarge) => match animation.fallback_url() {
+                    Some(url) => match input_file_for(url) {
+                        Ok(file) => {
+                            match send_animation_inner(
+                                bot,
+                                chat_id,
+                                reply_to,
+                                caption,
+                                has_spoiler,
+                                file,
+                            )
+                            .await
+                            {
+                                Ok(message) => {
+                                    let id = message.id.0 as i64;
+                                    cache_animation_send(task, &message).await;
+                                    Ok(vec![id])
+                                }
+                                Err(e) => Err(classify_to_send_error(&e, task.clone())),
+                            }
+                        }
+                        Err(message) => {
+                            Err(SendError::Permanent { message, task: task.clone() })
+                        }
+                    },
+                    None => Err(SendError::Permanent {
+                        message: "media too large".into(),
+                        task: task.clone(),
+                    }),
+                },
                 Err(FallbackError::Retryable { delay_seconds }) => {
-                    return Err(SendError::Retryable { delay_seconds, task: task.clone() });
+                    Err(SendError::Retryable { delay_seconds, task: task.clone() })
                 }
                 Err(FallbackError::Permanent { message }) => {
-                    return Err(SendError::Permanent { message, task: task.clone() });
+                    Err(SendError::Permanent { message, task: task.clone() })
                 }
-            };
-            let path = file.path().to_path_buf();
-            match send_animation_inner(
-                bot,
-                chat_id,
-                reply_to,
-                caption,
-                has_spoiler,
-                InputFile::file(path),
-            )
-            .await
-            {
-                Ok(message) => Ok(vec![message.id.0 as i64]),
-                Err(e) => Err(classify_to_send_error(&e, task.clone())),
             }
         }
         Err(e) => Err(classify_to_send_error(&e, task.clone())),
@@ -731,6 +1084,7 @@ pub async fn handle_task(payload: serde_json::Value) -> Result<(), QueueError> {
                     });
                 }
                 Err(SendError::Permanent { message, task }) => {
+                    invalidate_cache(&task).await;
                     return Err(QueueError::Permanent {
                         message,
                         payload: serde_json::to_value(task).expect("task serializes"),
@@ -782,6 +1136,14 @@ pub async fn dead_letter_notify(payload: serde_json::Value, message: String) {
 mod tests {
     use super::*;
 
+    #[test]
+    fn oversized_photo_boundary() {
+        // The empirical Telegram limit: sum 10000 passes, 10001 fails.
+        assert!(crate::photo::PHOTO_MAX_DIMENSION_SUM == 10000);
+        assert!(6100 + 3900 <= crate::photo::PHOTO_MAX_DIMENSION_SUM);
+        assert!(6300 + 3730 > crate::photo::PHOTO_MAX_DIMENSION_SUM);
+    }
+
     #[test]
     fn chunk_media_items_sizes() {
         assert_eq!(chunk_media_items::<i32>(vec![]), Vec::<Vec<i32>>::new());
@@ -819,6 +1181,36 @@ mod tests {
         }
     }
 
+    #[test]
+    fn is_size_error_matches_known_errors() {
+        // 413 upload cap.
+        let e = ApiError::RequestEntityTooLarge;
+        assert!(is_size_error(&e), "{e:?}");
+        // Unknown descriptions with size wording.
+        for description in [
+            "Bad Request: file is too large",
+            "Bad Request: media is too big",
+            "Bad Request: url file size is too big",
+        ] {
+            let api = ApiError::Unknown(description.to_string());
+            assert!(is_size_error(&api), "{description}");
+        }
+        // Unrelated errors must not match.
+        for description in ["Bad Request: WEBPAGE_MEDIA_EMPTY", "Bad Request: message is not modified"] {
+            let api = ApiError::Unknown(description.to_string());
+            assert!(!is_size_error(&api), "{description}");
+        }
+    }
+
+    #[test]
+    fn media_item_payload_fallback_url_serde_default() {
+        // Old queued payloads without the field deserialize with None.
+        let json = serde_json::json!({"kind": "photo", "media": "https://a/b.jpg", "has_spoiler": false});
+        let photo: MediaItemPayload = serde_json::from_value(json).unwrap();
+        assert!(matches!(photo, MediaItemPayload::Photo { fallback_url: None, .. }));
+        assert_eq!(photo.fallback_url(), None);
+    }
+
     #[test]
     fn classification_mapping() {
         use teloxide::types::Seconds;
@@ -858,11 +1250,15 @@ mod tests {
                 vec![MediaItemPayload::Photo {
                     media: "https://a/b.jpg".into(),
                     has_spoiler: true,
+                    fallback_url: Some("https://a/b_small.jpg".into()),
+                    file_id: false,
                 }],
                 vec![MediaItemPayload::Video {
                     media: "https://a/v.mp4".into(),
                     has_spoiler: false,
                     thumbnail: Some("https://a/t.jpg".into()),
+                    fallback_url: None,
+                    file_id: false,
                 }],
             ],
             batch_index: 1,
@@ -872,6 +1268,7 @@ mod tests {
             forward_channel_id: Some(333),
             notify_chat_id: Some(111),
             notify_message_id: Some(222),
+            cache_data: None,
         };
         let json = serde_json::to_value(&task).unwrap();
         assert_eq!(json["type"], "send_media_sequence");
@@ -900,6 +1297,8 @@ mod tests {
         let photo = MediaItemPayload::Photo {
             media: "https://a/b.jpg".into(),
             has_spoiler: false,
+            fallback_url: None,
+            file_id: false,
         };
         let json = serde_json::to_value(&photo).unwrap();
         assert_eq!(json["kind"], "photo");

crates/xmedia-bot/src/state.rs

@@ -74,6 +74,10 @@ impl ChatStore {
         let db_path = self.db_path.clone();
         let payload = tokio::task::spawn_blocking(move || -> rusqlite::Result<Option<String>> {
             let conn = Connection::open(&db_path)?;
+            // Concurrent handler tasks (batch-forwards) may write chat_state
+            // while this read runs; without a busy timeout a write lock
+            // collision fails the query immediately.
+            conn.busy_timeout(std::time::Duration::from_secs(5))?;
             let mut stmt = conn.prepare("SELECT payload FROM chat_state WHERE chat_id = ?1")?;
             let mut rows = stmt.query(params![chat_id.to_string()])?;
             match rows.next()? {
@@ -100,6 +104,7 @@ impl ChatStore {
         let db_path = self.db_path.clone();
         tokio::task::spawn_blocking(move || -> rusqlite::Result<()> {
             let conn = Connection::open(&db_path)?;
+            conn.busy_timeout(std::time::Duration::from_secs(5))?;
             conn.execute(
                 "INSERT OR REPLACE INTO chat_state (chat_id, payload) VALUES (?1, ?2)",
                 params![chat_id.to_string(), payload],

docker-compose.yml.example

@@ -1,30 +1,66 @@
 services:
+  nginx-proxy:
+    image: nginxproxy/nginx-proxy:1.11.6-alpine
+    restart: always
+    ports:
+      - '80:80'
+      - '443:443'
+    volumes:
+      - /var/run/docker.sock:/tmp/docker.sock:ro
+      - certs:/etc/nginx/certs:ro
+      - html:/usr/share/nginx/html:ro
+    networks: [proxy]
+    labels:
+      - 'com.github.nginx-proxy.nginx=true'
+    container_name: nginx-proxy
+
+  acme-companion:
+    image: nginxproxy/acme-companion
+    restart: always
+    environment:
+      DEFAULT_EMAIL: 'admin@yoursfunny.top'
+    volumes:
+      - /var/run/docker.sock:/var/run/docker.sock:ro
+      - certs:/etc/nginx/certs:rw
+      - html:/usr/share/nginx/html:rw
+      - acme:/etc/acme.sh
+    networks: [proxy]
+    container_name: acme-companion
+    depends_on:
+      - nginx-proxy
+
   tgxmb:
     image: yoursfunny/telegram-twitter-media-bot:latest
     restart: always
-#    ports:
-#      - "8443:8443"
     environment:
-      # docker-entrypoint.sh drops privileges to this uid.
       LOCAL_USER_ID: '1000'
-      # Bot token (BotFather). Required.
       TELOXIDE_TOKEN: ''
-      # Comma-separated admin chat ids; receives startup/shutdown notices.
       BOT_ADMIN: ''
-      # Required for pixiv support; pixiv is disabled when unset.
       PIXIV_REFRESH_TOKEN: ''
-      # Edit-before-forward records expire after this many seconds (default 86400 = 24h).
+      TWITTER_AUTH_TOKEN: ''
       EDIT_MESSAGE_TTL_SECONDS: '86400'
+      LINK_CACHE_TTL_SECONDS: '604800'
       RUST_LOG: 'info'
-      # Webhook mode is off by default (polling). The listener binds inside the
-      # container, so use 0.0.0.0 and publish the port if you enable it.
-      WEBHOOK: 'false'
+      VIRTUAL_HOST: '<YOUR_DOMAIN>'
+      VIRTUAL_PORT: '8443'
+      # ACME_HOST: 'your.domain.com'
+      WEBHOOK: 'true'
       WEBHOOK_LISTEN: '0.0.0.0'
       WEBHOOK_PORT: '8443'
-      WEBHOOK_URL: 'https://example.com'
-      WEBHOOK_CERT: './cert/cert.pem'
-      WEBHOOK_SECRET_TOKEN: 'secret-token'
+      WEBHOOK_URL: 'https://<YOUR_DOMAIN>/'
+      WEBHOOK_SECRET_TOKEN: ''
     volumes:
       - ./data:/app/data
-#      - ./cert:/app/cert
+    networks: [proxy]
+    depends_on:
+      - nginx-proxy
     container_name: tgxmb
+
+volumes:
+  certs:
+  html:
+  acme:
+
+networks:
+  proxy:
+    name: proxy

docker-entrypoint.sh

@@ -5,9 +5,21 @@ if [ "$(id -u)" -eq '0' ]
 then
    USER_ID=${LOCAL_USER_ID:-9001}
 
-   useradd --shell /bin/bash -u ${USER_ID} -o -c "" -m user > /dev/null 2>&1
-   usermod -a -G root user > /dev/null 2>&1
-   chown -R `id -u user`:`id -u user` /app > /dev/null 2>&1
+   # `docker compose restart` / `docker restart` reuse the same container, so
+   # the overlay fs keeps the user created on first boot. A second `useradd`
+   # then fails with exit code 9, which would trip `set -e` and kill the
+   # container on every restart. Create only if missing; align the UID
+   # otherwise so LOCAL_USER_ID changes still apply.
+   if ! id user > /dev/null 2>&1
+   then
+      useradd --shell /bin/bash -u ${USER_ID} -o -c "" -m user > /dev/null 2>&1 || true
+   else
+      usermod -u ${USER_ID} -o user > /dev/null 2>&1 || true
+   fi
+   usermod -a -G root user > /dev/null 2>&1 || true
+   # Bind-mounted volumes may not support chown; a failure here must not kill
+   # the container either.
+   chown -R `id -u user`:`id -u user` /app > /dev/null 2>&1 || true
 
    export HOME=/home/user
    # setpriv (util-linux, present in bookworm-slim) replaces gosu: drop to the