refactor: make Flash PDF semantic assembly stages explicit - #43
Merged
Conversation
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 subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Motivation
Document orchestration, title statistics/classification and text grouping/merging have broad responsibilities, while vector-table construction combines all rejection and materialization stages in one function.
Modification
Make document text classification and inline materialization explicit stages. Split title body statistics, prototypes and classifiers, and separate text-row grouping from block assembly. Extract typed vector-track, topology and materialization stages, preserving rejection order and diagnostic fields.
This is stage 6/6 of the Flash PDF equivalence refactor, based on
codex/flash-pdf-05-inline. Review and integrate stages in order.Validation
Final validation: 1,269 tests pass, with one previously failing demo-table manifest test explicitly deselected after reproducing it on the original implementation. pdftext 0.6.3 compatibility: 229 tests pass. All 31 documents / 298 pages retain identical complete model-list and Middle JSON output. Sum of five-run document medians: 45.02s → 31.49s (30.1% reduction); no sample exceeds the 5% time/RSS regression threshold. Six reviewed overlay pages are pixel-identical. Wheel build includes all 62 PDF modules; isolated wheel installation produces identical two-page financial-table output.
The original
test_demo_sparse_table_confidence_manifestbbox expectation failure is documented separately and its expected data is not modified.Compatibility
The existing
PdfModel.predict(), PDFDocument methods, ModelJson/MiddleJson output and renderer contracts remain unchanged. Recognition thresholds, candidate priority and fallback behavior are preserved.Checklist
git diff --checkpass.Remote check status
license/clasucceeds. The separate CLAAssistant workflow fails because the configured signature branch is missing (Branch cla not found), then requests contributor signatures. Workflow/signature settings are unchanged by this refactor. See https://github.com/myhloli/Magic-PDF/actions/runs/33952244239/job/101269092002 .