Update Bluesky component versions and add text encoding functionality #18443
Conversation
- Bump version of @pipedream/bluesky to 0.1.2 - Update action versions for create-post (0.1.0), like-post (0.0.2), retrieve-thread (0.0.2), new-follower-on-account (0.0.2), new-posts-by-author (0.0.2), and new-timeline-posts (0.0.2) - Introduce textEncoding module for detecting facets in post content - Remove deprecated methods from create-post action
WalkthroughIntroduces hashtag/link/mention facet detection via a new common textEncoding module and integrates it into the Bluesky Create Post action, replacing prior internal helpers. Adds a TLD list module. Bumps versions for multiple Bluesky actions, sources, and the package. No other logic changes outside create-post facet construction. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
actor User
participant Action as CreatePost Action
participant TE as textEncoding.detectFacets
participant App as app (resolveHandle)
participant API as Bluesky API
User->>Action: invoke({ text, ... })
Action->>TE: detectFacets(text, app)
alt mention facets
TE->>App: resolveHandle(@handle)
App-->>TE: did
end
TE-->>Action: facets[]
Action->>API: createRecord({ text, facets })
API-->>Action: ack/uri
Action-->>User: result
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Poem
Pre-merge checks and finishing touches❌ Failed checks (2 warnings)
✅ Passed checks (3 passed)
✨ Finishing touches
🧪 Generate unit tests
Tip 👮 Agentic pre-merge checks are now available in preview!Pro plan users can now enable pre-merge checks in their settings to enforce checklists before merging PRs.
Please see the documentation for more information. Example: reviews:
pre_merge_checks:
custom_checks:
- name: "Undocumented Breaking Changes"
mode: "warning"
instructions: |
Pass/fail criteria: All breaking changes to public APIs, CLI flags, environment variables, configuration keys, database schemas, or HTTP/GraphQL endpoints must be documented in the "Breaking Change" section of the PR description and in CHANGELOG.md. Exclude purely internal or private changes (e.g., code not exported from package entry points or explicitly marked as internal).Please share your feedback with us on this Discord post. Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (11)
components/bluesky/common/tlds.mjs (1)
1-2: Export a Set for O(1) lookups and document source/update process.
Membership checks on a large list will be faster with a Set. Also, add a brief header comment noting the source and update procedure for maintainability.Apply this diff:
- export default [ + // Source: IANA TLD list and known IDN TLDs. + // Consider maintaining via a small script that refreshes from: + // https://data.iana.org/TLD/tlds-alpha-by-domain.txt + export const TLD_LIST = [ @@ - ]; + ]; + export const TLD_SET = new Set(TLD_LIST); + export default TLD_LIST;Also applies to: 1441-1441
components/bluesky/actions/create-post/create-post.mjs (3)
32-33: Validate 300-char limit and only include facets when present.
Preempt server-side errors and avoid sending empty arrays.["$type"]: constants.RESOURCE_TYPE.POST, text, - facets: await textEncoding.detectFacets(text, app), + // Bluesky posts are limited to 300 characters (Unicode code points) + // Early validate to provide a clearer error + ...(Array.from(text).length > 300 ? (() => { throw new Error("Text exceeds Bluesky 300 character limit"); })() : {}), + // Compute facets and include only when present + ...(await (async () => { + const facets = await textEncoding.detectFacets(text, app); + return facets && facets.length ? { facets } : {}; + })()), createdAt: new Date().toISOString(),
13-17: Clarify prop description to reflect auto-faceting.
Make it clear hashtags/mentions/links are auto-linked via rich-text facets.- description: "The text content of the post.", + description: "The text content of the post. @mentions, #hashtags, and URLs are automatically linked via rich-text facets.",
3-3: detectFacets: DID present and UTF‑8 offsets correct — pass $ through to resolveHandle
- detectFacets sets features[0].did = handle.did and computes byteStart/byteEnd via utf16IndexToUtf8Index (TextEncoder) — this produces UTF‑8 byte offsets and handles emoji/surrogates.
- Issue: detectFacets calls app.resolveHandle(...) without the action
$context. Forward$into detectFacets and call resolveHandle({$, params: { handle } }) (also update caller in components/bluesky/actions/create-post/create-post.mjs to await textEncoding.detectFacets(text, app, $ )).components/bluesky/common/textEncoding.mjs (7)
3-11: Domain validation: make case-insensitive, allow hyphens, and support multi-label TLDs efficientlyCurrent logic is case-sensitive and O(|TLDs|) per call. It also doesn’t validate label hyphens correctly. Recommend lowercasing, validating labels, and checking multi-label TLDs via a precomputed Set.
Apply this diff to replace isValidDomain:
-function isValidDomain(str) { - return !!TLDs.find((tld) => { - const i = str.lastIndexOf(tld); - if (i === -1) { - return false; - } - return str.charAt(i - 1) === "." && i === str.length - tld.length; - }); -} +function isValidDomain(str) { + const s = String(str || "").toLowerCase(); + // Validate labels: alnum + hyphen, no leading/trailing hyphen in any label + if (!/^[a-z0-9](?:[a-z0-9-]*[a-z0-9])?(?:\.[a-z0-9](?:[a-z0-9-]*[a-z0-9])?)+$/.test(s)) { + return false; + } + // Support multi-label TLDs (e.g., .co.uk) + const parts = s.split("."); + for (let i = 0; i < parts.length; i++) { + const suffix = parts.slice(i).join("."); + if (TLD_SET.has(suffix)) return true; + } + return false; +}Add this near the import (module scope):
// after import TLDs const TLD_SET = new Set(TLDs.map((t) => t.toLowerCase()));
14-16: Avoid re‑creating TextEncoder; reuse a single instanceThis runs for every index conversion. Reuse one encoder to reduce allocations.
Apply this diff:
-function utf16IndexToUtf8Index(i, utf16) { - const encoder = new TextEncoder(); - return encoder.encode(utf16.slice(0, i)).byteLength; -} +function utf16IndexToUtf8Index(i, utf16) { + return encoder.encode(utf16.slice(0, i)).byteLength; +}Add this at module scope:
const encoder = new TextEncoder();
36-36: Facet consistency: drop top-level $type (keep it only in features)Only mention facets include the top-level $type. Recommend omitting it for consistency (features keep $type).
Apply this diff:
- $type: "app.bsky.richtext.facet",
51-52: Links: allow hyphens in domain labelsCurrent regex excludes valid domains with hyphens. Update to accept label hyphens (not leading/trailing).
Apply this diff:
- const re2 = - /(^|\s|\()((https?:\/\/[\S]+)|((?<domain>[a-z][a-z0-9]*(\.[a-z0-9]+)+)[\S]*))/gim; + const re2 = + /(^|\s|\()((https?:\/\/\S+)|((?<domain>[a-z0-9](?:[a-z0-9-]*[a-z0-9])?(?:\.[a-z0-9](?:[a-z0-9-]*[a-z0-9])?)+)\S*))/gim;
62-76: Links: compute start from match.index; fix comment typoUse match.index plus leading group length for deterministic start. Also fix “puncuation” typo.
Apply this diff:
- const start = text.indexOf(match[2], match.index); + const start = match.index + (match[1]?.length || 0); @@ - // strip ending puncuation + // strip ending punctuation
108-110: Byte end is exclusive; fix misleading commentThe code uses exclusive byteEnd, which is correct. The comment says “inclusive”.
Apply this diff:
- byteEnd: utf16IndexToUtf8Index(index + tag.length, text), // inclusive of last char + byteEnd: utf16IndexToUtf8Index(index + tag.length, text), // exclusive
18-48: Optional: parallelize handle resolution and dedupe mentionsMultiple mentions resolve sequentially. For posts with several mentions, gather unique handles and resolve with Promise.all, then map DIDs back to positions.
If you’d like, I can provide a refactor that collects all mention matches first, resolves unique handles concurrently, and then builds facets.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (9)
-
components/bluesky/actions/create-post/create-post.mjs(2 hunks) -
components/bluesky/actions/like-post/like-post.mjs(1 hunks) -
components/bluesky/actions/retrieve-thread/retrieve-thread.mjs(1 hunks) -
components/bluesky/common/textEncoding.mjs(1 hunks) -
components/bluesky/common/tlds.mjs(1 hunks) -
components/bluesky/package.json(1 hunks) -
components/bluesky/sources/new-follower-on-account/new-follower-on-account.mjs(1 hunks) -
components/bluesky/sources/new-posts-by-author/new-posts-by-author.mjs(1 hunks) -
components/bluesky/sources/new-timeline-posts/new-timeline-posts.mjs(1 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
components/bluesky/common/textEncoding.mjs (1)
components/bluesky/actions/retrieve-thread/retrieve-thread.mjs (2)
app(49-52)app(54-59)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (4)
- GitHub Check: Lint Code Base
- GitHub Check: pnpm publish
- GitHub Check: Publish TypeScript components
- GitHub Check: Verify TypeScript components
🔇 Additional comments (10)
components/bluesky/package.json (1)
3-3: Version bump LGTM.
Consistent with action/source bumps in this PR.components/bluesky/sources/new-follower-on-account/new-follower-on-account.mjs (1)
9-9: Version bump LGTM.
No functional changes introduced here.components/bluesky/sources/new-timeline-posts/new-timeline-posts.mjs (1)
9-9: Version bump LGTM.
No behavior changes in this module.components/bluesky/actions/retrieve-thread/retrieve-thread.mjs (1)
7-7: Version bump LGTM.
No logic changes detected.components/bluesky/actions/like-post/like-post.mjs (1)
8-8: Version bump LGTM.
No behavioral changes in this action.components/bluesky/sources/new-posts-by-author/new-posts-by-author.mjs (1)
9-9: Version bump LGTM.
Aligned with the broader release.components/bluesky/actions/create-post/create-post.mjs (1)
9-9: Version bump LGTM.
Reflects public API changes (facet detection integration).components/bluesky/common/tlds.mjs (1)
1-1441: components/bluesky/common/tlds.mjs — UTF‑8 valid; no duplicates found.
Verified by reading the file as UTF‑8 and counting entries: 1439 total, 1439 unique, 0 duplicates.components/bluesky/common/textEncoding.mjs (2)
120-123: Return value shape: confirm downstream handling of undefined vs []You return undefined when there are no facets. Ensure callers (e.g., create-post) handle both cases correctly.
If create-post expects an array, consider returning [] to simplify call sites and avoid optional chaining checks.
29-35: Compute mention slice deterministically; destructure DID from resolveHandleUse match.index + (match[1]?.length || 0) to get the '@' position (avoids indexOf pitfalls) and destructure did from resolveHandle. resolveHandle accepts an args object (resolveHandle(args = {})) and can receive an optional
$param — other callers pass$; include it here only if detectFacets needs request context.Apply this diff:
- const start = text.indexOf(match[3], match.index) - 1; - const handle = await app.resolveHandle({ - params: { - handle: match[3], - }, - }); + const start = match.index + (match[1]?.length || 0); // position of '@' + const { did } = await app.resolveHandle({ + params: { handle: match[3] }, + }); @@ - { - $type: "app.bsky.richtext.facet#mention", - did: handle.did, // must be resolved afterwards - }, + { + $type: "app.bsky.richtext.facet#mention", + did, // resolved DID + },Also applies to: 41-46
|
The latest updates on your projects. Learn more about Vercel for GitHub. 2 Skipped Deployments
|
vunguyenhung
deleted the
15446-action-bluesky-hash-tags-rich-text-facets
branch
Resolves #15446
Summary by CodeRabbit