fix(ocr): detect Homebrew tessdata on Apple Silicon and custom prefixes - #146
Merged
Conversation
_find_tessdata only probed a tessdata folder adjacent to the binary and, as a fixed fallback, /usr/local/share/tessdata (Intel Homebrew). On an Apple Silicon Mac the binary is /opt/homebrew/bin/tesseract but the data lives in /opt/homebrew/share/tessdata, so the function returned None and the app never set TESSDATA_PREFIX. Extra language packs such as Hebrew were therefore invisible on a MacBook Air M5. Fix (additive, no existing behaviour changed): - Derive <prefix>/share/tessdata relative to the binary (<prefix>/bin/tesseract -> <prefix>/share/tessdata), checked right after the adjacent bin/tessdata probe. Covers Intel, Apple Silicon and non-standard install prefixes generically. - Add /opt/homebrew/share/tessdata to the fixed linux/darwin fallback list for when the binary is resolved via PATH and the derived prefix does not match. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The prefix-relative derivation <prefix>/share/tessdata ran before the versioned Linux lookup /usr/share/tesseract-ocr/<version>/tessdata. For /usr/bin/tesseract the derivation yields the flat /usr/share/tessdata, so on a system where both that stale dir and a newer versioned dir exist, the stale one shadowed the versioned one - inverting the issue #27 invariant (Ubuntu 24.04 ships v5 but the baked-in default points at 4.00). Reorder _find_tessdata precedence: (1) adjacent bin/tessdata, (2) reverse-sorted versioned /usr/share/tesseract-ocr/*/tessdata, (3) relative <prefix>/share/tessdata derivation (still resolves Apple Silicon /opt/homebrew, Intel /usr/local and custom prefixes since the versioned glob finds nothing there), (4) fixed fallbacks. The relative derivation keeps its `if tess_exe:` guard. Add a discriminative Linux conflict test that fails against the old ordering and passes after the reorder. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Deploying pdfapps with
|
| Latest commit: |
cc3311a
|
| Status: | ✅ Deploy successful! |
| Preview URL: | https://0e58539f.pdfapps.pages.dev |
| Branch Preview URL: | https://fix-tessdata-apple-silicon.pdfapps.pages.dev |
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
1 participant
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.
Problem
On a macOS Apple Silicon Mac (a user's MacBook Air M5), extra OCR language
packs such as Hebrew never appeared. Root cause:
_find_tessdatainapp/tools/ocr.pycould not locate thetessdatadirectory on HomebrewApple Silicon.
Homebrew keeps the binary under
<prefix>/binbut the language data under<prefix>/share/tessdata:/usr/local/bin/tesseract+/usr/local/share/tessdata/opt/homebrew/bin/tesseract+/opt/homebrew/share/tessdataThe old function only probed:
tessdatafolder adjacent to the binary (<dir>/tessdata) — whichdoes not exist on Homebrew, and
/usr/local/share/tessdata(Intel only),plus the Linux
/usr/share/...paths and snap./opt/homebrew/share/tessdatawas in neither, so on Apple Silicon thefunction returned
None, the app never setTESSDATA_PREFIX, and anynon-baked-in language pack was invisible.
Fix (additive, no existing behaviour changed)
bin/tessdataprobe, derive
<prefix>/share/tessdatafrom the binary(
os.path.dirname(os.path.dirname(tess_exe))+share/tessdata). Thiscovers Intel, Apple Silicon and non-standard install prefixes in one
generic step.
/opt/homebrew/share/tessdatato the fixedlinux/darwinfallback list (next to the existing/usr/local/share/tessdata) for the case where the binary is resolved viaPATHand the derived prefix does not match.Windows-adjacent lookup and the Linux
/usr/share/tesseract-ocr/<version>/tessdatareverse-sorted preference areuntouched.
Tests
New
tests/test_ocr_tessdata.py(10 tests). All monkeypatchos.path.isdir/glob.glob/sys.platformwith a fictitioustess_exe,so they run identically on the Windows and Linux CI runners (no macOS
runner, no dependency on real system paths):
/opt/homebrew/share/tessdata(wasNone)/usr/local/share/tessdata/opt/custom/share/tessdata(proves generic derivation)/opt/homebrew/share/tessdatafallback when derived prefix missesbin/tessdatatakes precedence (Windows layout)None;tess_exe=None->NoneThe four Apple-Silicon / Intel / custom-prefix / fallback tests fail against
the pre-fix code (verified), confirming they are discriminative.
Full suite: 508 passed, 3 skipped, 0 failed.
ruff check --select F,E9 app/tools/ocr.py: clean.🤖 Generated with Claude Code