fix: strip Scroll to Text Fragment anchors in link checks#86
fix: strip Scroll to Text Fragment anchors in link checks#86zeitlinger merged 5 commits intomainfrom
Conversation
There was a problem hiding this comment.
Pull request overview
Updates the lychee link-check remap rules so GitHub “Scroll to Text Fragment” anchors (#:~:text=...) are stripped from /blob/ URLs before verification, avoiding false failures caused by remap precedence and fragment validation limitations in static HTML.
Changes:
- Add a repo-specific
/blob/remap rule that strips#:~:text=...fragments and remaps to the PR head branch. - Add a global GitHub
/blob/remap rule that strips#:~:text=...fragments for all repositories. - Update inline documentation/comments to reflect the new remap ordering.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| # 1. Line-number anchors (#L123): strip fragment, remap to head branch | ||
| echo "--remap" | ||
| echo "^${base_url}/blob/${base_ref}/(.*?)#L[0-9]+\$ ${head_url}/blob/${head_ref}/\$1" | ||
|
|
||
| # 2. Other fragment URLs (#section): remap to raw.githubusercontent.com | ||
| # 2. Scroll to Text Fragment anchors (#:~:text=...): browser-only, | ||
| # strip fragment, remap to head branch | ||
| echo "--remap" | ||
| echo "^${base_url}/blob/${base_ref}/(.*?)#:~:text=.*\$ ${head_url}/blob/${head_ref}/\$1" |
There was a problem hiding this comment.
The repo-specific line-anchor remap only matches #L<digits> at end of URL (#L[0-9]+$). GitHub also generates range fragments like #L10-L20; those won’t match rule (1) and will fall through to the “other fragments → raw.githubusercontent.com” rule, where lychee still can’t verify the fragment. Update this rule (and the analogous /tree/ rule) to also match range anchors (e.g., #L[0-9]+(-L[0-9]+)?$ or similar) so they get stripped instead of remapped to raw.
There was a problem hiding this comment.
Fixed — changed the repo-specific blob and tree rules from #L[0-9]+$ to #L[0-9]+.*$ to match range fragments like #L10-L20, consistent with the global remap. Added a same-repo range test case (links.sh#L6-L10) to cover this.
Scroll to Text Fragment anchors (#:~:text=...) are a browser-only feature that lychee cannot verify in static HTML. Without this fix, the "other fragments" remap rule catches these URLs first and redirects them to raw.githubusercontent.com, where the fragment validation fails. This was observed in open-telemetry/opentelemetry-java-instrumentation where a link to build.gradle#:~:text=extendedAgent fails in CI but works locally (because build_remap_args skips when on the base branch). Fixes the issue by adding explicit rules (in both repo-specific and global remaps) to strip the fragment before the catch-all fragment rule matches. Signed-off-by: Gregor Zeitlinger <gregor.zeitlinger@grafana.com>
Signed-off-by: Gregor Zeitlinger <gregor.zeitlinger@grafana.com>
The repo-specific remap used #L[0-9]+$ which didn't match range fragments like #L10-L20. These would fall through to the "other fragments → raw.githubusercontent.com" rule where they also can't be verified. Changed to #L[0-9]+.*$ to match ranges, consistent with the global remap rule. Signed-off-by: Gregor Zeitlinger <gregor.zeitlinger@grafana.com>
Signed-off-by: Gregor Zeitlinger <gregor.zeitlinger@grafana.com>
Signed-off-by: Gregor Zeitlinger <gregor.zeitlinger@grafana.com>
c989255 to
1156abc
Compare
🤖 I have created a release *beep* *boop* --- ## [0.7.1](v0.7.0...v0.7.1) (2026-03-02) ### Bug Fixes * strip Scroll to Text Fragment anchors in link checks ([#86](#86)) ([b630cdf](b630cdf)) --- > [!IMPORTANT] > Close and reopen this PR to trigger CI checks. Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
## flint v0.20.0 — Rust rewrite flint has been rewritten from scratch in Rust. The bash + Docker stack is replaced by a fast, cross-platform native binary with no container dependency. ### Highlights - **Native binary** — no Docker, no bash wrappers; runs anywhere mise can install a Rust crate - **Parallel linting** — all checks run concurrently - **Diff-aware** — only checks files changed since the merge base by default (`--full` for everything) - **`flint update`** — migrates obsolete `mise.toml` tool keys automatically (e.g. `ubi:` → `github:`, `npm:markdownlint-cli` → `npm:markdownlint-cli2`) - **`flint hook install`** — installs a pre-commit hook without any mise task knowledge - **New linters**: `gofmt`, `google-java-format`, `ktlint`, `dotnet-format`, `markdownlint-cli2`, `xmllint` (via `cargo:xmloxide`), `license-header` (pure-Rust, no binary) - **Windows support** — ktlint self-executing JAR handled via explicit `java -jar` invocation ### Migration See [AGENTS-V2.md](https://github.com/grafana/flint/blob/main/AGENTS-V2.md) and [README.md](https://github.com/grafana/flint/blob/main/README.md) for full setup instructions. Quick reference for existing consumers: ```toml [tools] # While installing from source (pre-crates.io release): "cargo:https://github.com/grafana/flint" = "branch:main" [env] FLINT_CONFIG_DIR = ".github/config" [tasks.lint] run = "flint run" [tasks."lint:fix"] run = "flint run --fix" ``` Remove: `lint:super-linter`, `lint:links`, `lint:renovate-deps`, `setup:native-lint-tools`, `setup:pre-commit-hook` tasks. Run `flint update` to auto-migrate any obsolete tool keys. > [!NOTE] > The changelog below includes entries from the legacy v1 bash era — see [README-V1.md](https://github.com/grafana/flint/blob/main/README-V1.md) for v1 docs. --- ## [0.20.0](flint-v0.19.0...flint-v0.20.0) (2026-04-13) ### Features * add flint v2 Rust binary ([#139](#139)) ([19f2b25](19f2b25)) * add native linting mode and version mapping infrastructure ([#93](#93)) ([24b06da](24b06da)) * add Renovate shareable preset for consuming repos ([#17](#17)) ([8a06590](8a06590)) * consolidate link checking and add autofix flags ([#7](#7)) ([086a5e9](086a5e9)) * flint update command, explicit JAR flag, v0.20.0 ([#146](#146)) ([b43bf52](b43bf52)) * handle line-number anchors and issue comments globally ([#56](#56)) ([cf751df](cf751df)) * **links:** add GitHub URL remaps for line-number and fragment anchors ([#28](#28)) ([5b59065](5b59065)) * **links:** auto-remap base-branch GitHub URLs to PR branch ([#18](#18)) ([dd6cc61](dd6cc61)) * **renovate:** support SHA-pinned URLs in Renovate preset ([#21](#21)) ([4fd1f28](4fd1f28)) * **super-linter:** default to slim image ([#24](#24)) ([c8eeab8](c8eeab8)) * support NATIVE env var for container-free linting ([#107](#107)) ([0a8193d](0a8193d)) ### Bug Fixes * activate mise environment in native lint mode ([#123](#123)) ([d0fec45](d0fec45)) * add 'mise run fix' hint to lint failure output ([#90](#90)) ([5b4ad5d](5b4ad5d)) * decouple version mapping generation from pinned super-linter version ([#112](#112)) ([5370e77](5370e77)) * **deps:** update rust crate crossterm to 0.29 ([#156](#156)) ([c59ae3e](c59ae3e)) * **deps:** update rust crate similar to v3 ([#160](#160)) ([684be4e](684be4e)) * **deps:** update rust crate toml to v1 ([#161](#161)) ([3aae614](3aae614)) * **deps:** update rust crate toml_edit to 0.25 ([#158](#158)) ([42d9efd](42d9efd)) * exclude GitHub compare links from lychee checks ([#10](#10)) ([e714608](e714608)) * fail native lint when enabled tools are missing ([#111](#111)) ([163bb6b](163bb6b)) * improve link checker reliability against GitHub rate limiting ([#95](#95)) ([7a5282d](7a5282d)) * include staged files in native lint file list ([#135](#135)) ([34412d6](34412d6)) * **links:** add regex anchors to remap patterns ([#19](#19)) ([2e17348](2e17348)) * native lint in worktrees, trust toml, use ec binary, drop isort ([#134](#134)) ([8594bba](8594bba)) * **release-please:** fix footer not appearing on release PRs ([#40](#40)) ([d7a55e4](d7a55e4)) * remap same-repo GitHub URLs to local file paths ([#100](#100)) ([b4feadd](b4feadd)) * **renovate-deps:** forward GITHUB_TOKEN as GITHUB_COM_TOKEN ([#132](#132)) ([4d6510b](4d6510b)) * replace broken release-please PR comment with docs ([#12](#12)) ([817b37d](817b37d)) * run shellcheck on .bats files in native mode ([#137](#137)) ([a4fd3f8](a4fd3f8)) * strip Scroll to Text Fragment anchors in link checks ([#86](#86)) ([b630cdf](b630cdf)) * tighten markdownlint config for native mode ([#106](#106)) ([6ef25b2](6ef25b2)) * use remap instead of exclude for issue comment anchors ([#58](#58)) ([656f355](656f355)) --- > [!IMPORTANT] > Close and reopen this PR to trigger CI checks. Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Summary
#:~:text=...) from GitHub blob URLs before lychee tries to verify thembuild_remap_args) and global (build_global_github_args) remapsContext
Scroll to Text Fragment anchors are a browser-only feature — they highlight text on the page but don't correspond to any HTML element ID. Lychee can't verify them in static HTML.
Without this fix, the "other fragments → raw.githubusercontent.com" remap rule catches these URLs first and redirects them to raw content, where fragment validation fails because
#:~:text=extendedAgentisn't a valid anchor.The consuming repo's
lychee.tomlalready has a remap to strip this specific fragment, but it never fires because CLI--remapargs from flint match first (first-match-wins).Observed in open-telemetry/opentelemetry-java-instrumentation CI — the link works locally because
build_remap_argsskips when on the base branch.Test plan
lychee.toml)