Repository navigation
Report a media href that reaches its file only under case folding - #47
Merged
Merged
Conversation
A media href is now resolved against the folder the way a companion .lift-ranges already is, so a Windows-authored folder read on a case-sensitive filesystem no longer reports as missing-media files that are right there. Unlike a companion, whose spelling only sil-lift reads, a media href is read by whatever serves the folder afterwards -- a web export, an APK build -- so the fold is reported rather than resolved silently: a new media-href-mismatch warning says the file is here but a case-sensitive host will not find it, and missing-media narrows to no file under any spelling. Resolution reads the directory rather than probing the href's spelling, so the verdict is the same on a case-folding filesystem as on a case-sensitive one; an exact probe answers for a name it can never report, since SDD.PNG stats true against an on-disk sdd.png. Every component folds, not just the last. The conventional audio/pictures subfolder is sil-lift's own guess, so a folder spelling it another way resolves without being reported, while a folder the href itself misspells is one rename however many references cross it, and is reported once with no entry. A href reaching outside the folder is probed as written, since which directory to fold in would be a guess. check-media reports the same mismatches and exits 1 for them, and no longer calls a misspelled file orphaned as well as missing: it compares folded relative paths rather than resolved ones, which is what the library now means by a href naming a file. --no-check-media suppresses both media codes, since the media it exists for lives outside the folder, where no spelling can be verified at all. missing_media() is replaced by check_media(), which classifies each reference rather than answering one boolean, and returns only the references that did not resolve cleanly. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
check-media built its referenced set from the paths the hrefs spell rather than the files they reach, so a href crossing a dot segment -- audio/sub/../one.wav against an on-disk audio/one.wav -- matched nothing the folder walk found and printed as orphaned though it resolves. The set now holds the files media resolution returns, compared by identity, which also spares the command from re-deriving rules the model owns. A folder mismatch was keyed by the one component that differs, so A/Foo and B/Foo collapsed into one finding although each is its own rename. The key is the whole path the href reached, which the message now names too. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A folder mismatch was keyed by the whole path an href spells, so one folder reached through two spellings of its parent -- A/Foo and a/Foo over a single a/foo -- reported twice although both name the one rename. The key is now the on-disk path the href reaches, together with the component it spells that folder with: enough to tell a/foo from b/foo, and no longer split by an ancestor that has nothing to do with either. The message names that component and the path it reaches rather than the whole written path, so it identifies a real folder instead of one href's way of writing its way down to it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A folder finding was keyed by a path aligned against the href, which describes a/foo and pictures/a/foo identically: the conventional subfolder that separates them is the lookup's, and the document never wrote it. Two such folders merged into one finding, and the survivor named the wrong path. The key is now the directory the resolved file sits in, together with the component the href spells it with, so one folder written two ways still reports twice. Folder paths are reported relative to the folder holding the .lift, which is what lets the message tell pictures/a/foo from a/foo. Every path that was already unambiguous reads exactly as before. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
jasonleenaylor
requested changes
Sep 25, 2026
jasonleenaylor
left a comment
jasonleenaylor
left a comment
There was a problem hiding this comment.
Looks good overall. Two things to fix before merge, one nit, one question, all inline.
This review was assisted by Claude Fable 5.1.
Brings in validate --allow. The semantic-check count in validate.md stays at thirteen with media-href-mismatch, and the --allow description joins the bulleted validate options list in cli.md, in synopsis order. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An href of "" or "." has no components, so appending it to the conventional subfolder named the subfolder itself; a regular file called pictures or audio then matched, and comparing the href's spelling against it raised ValueError. check_media() now reports such an href missing before any probe, and check-media's orphan scan skips it. The folder walk in _media_matches stats entries through helpers that treat any OSError as "not there": before Python 3.14, pathlib lets a PermissionError escape is_file()/is_dir(). Folder listings are sorted, so which of several folding matches is reported, and the folder a mismatch is grouped under, is the same on every filesystem. CONTRIBUTING now asks for a PROVENANCE.md entry only for fetched fixtures, matching the rule that file states for hand-authored negative/ fixtures. The CLI guide's exit codes count misspelled media as a finding. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
imnasnainaec
commented
Oct 8, 2026
imnasnainaec
left a comment
imnasnainaec
left a comment
Collaborator
Author
There was a problem hiding this comment.
@imnasnainaec reviewed all commit messages and made 4 comments.
Reviewable status: 0 of 19 files reviewed, all discussions resolved (waiting on imnasnainaec).
jasonleenaylor
approved these changes
Oct 8, 2026
jasonleenaylor
left a comment
jasonleenaylor
left a comment
There was a problem hiding this comment.
Thanks, all four addressed. LGTM.
This review was assisted by Claude Opus 5.5.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #34.
media-href-mismatchwarning: a media href that reaches its file only under case folding or NFC.missing-medianarrows to "no file under any spelling", and now includes an empty href.audio//pictures/subfolder folds silently, since sil-lift guesses it rather than the document writing it.missing_media()→check_media(), returning aMediaResolutionper reference that did not resolve cleanly.check-mediareports mismatches and exits 1 on them, and no longer calls a misspelled file orphaned as well as missing.--no-check-mediacovers both media codes.negative/fixtures instead of listing them; CONTRIBUTING follows.Decisions recorded in #34.
🤖 Generated with Claude Code
This change is