mirror of
https://github.com/storytold/pdfcraft.git
synced 2026-10-25 02:29:31 -06:00
M6: treat imported MaxLen zero as unlimited
This commit is contained in:
djmushin
@@ -0,0 +1,117 @@
|
||||
# AcroForm `/MaxLen 0` compatibility
|
||||
|
||||
Status: implemented; proposed as a separate compatibility PR. Last updated: 2026-10-19.
|
||||
|
||||
- `crates/forms/src/lib.rs::walk` reads inherited `/MaxLen`; the UI and value validation both
|
||||
consume `Field::max_len`.
|
||||
|
||||
The parser exposed `/MaxLen 0` as `Some(0)`, so both editor and API blocked input. It now
|
||||
exposes only positive limits; zero clears an inherited limit. The original object entry is
|
||||
preserved through save/reopen. This is an interoperability interpretation for zero, **not
|
||||
a claim that ISO explicitly defines zero as unlimited**. ISO 32000-1 Table 229 defines an
|
||||
optional inheritable maximum length without defining a special zero meaning:
|
||||
[public specification](https://opensource.adobe.com/dc-acrobat-sdk-docs/pdfstandards/PDF32000_2008.pdf).
|
||||
Positive limits remain enforced; no field names or character-sheet rules enter product code.
|
||||
|
||||
Regression: `crates/forms/src/tests.rs::zero_max_len_is_unlimited_and_overrides_inherited_limit`.
|
||||
The forms README and `form.fill-text` parity entry document this behavior.
|
||||
|
||||
Historical validation recorded before this review:
|
||||
|
||||
- Forms unit suite: all 68 tests passed. The new zero-limit regression was first run against
|
||||
the old parser and failed (`Some(0)` instead of `None`), then passed with the compatibility
|
||||
interpretation. Its positive-limit control and preserved raw entry also passed.
|
||||
|
||||
The user reports successful Class text input on the integrated release using their Pathfinder
|
||||
character sheet. This is a manual functional result, without timings; the private document
|
||||
was not used for automated regression tests in this review.
|
||||
|
||||
## Independent branch validation
|
||||
|
||||
Validated on upstream `a12fe8f4116984ebbc870eb514346ea922233316` with an isolated
|
||||
Cargo target directory and no performance changes:
|
||||
|
||||
- The focused regression failed with the original parser (`Some(0)` rather than `None`),
|
||||
then passed after restoring the compatibility fix.
|
||||
- `cargo test -p pdfcraft-forms -j 2`: 81 unit tests passed; doc tests passed.
|
||||
- `cargo clippy -p pdfcraft-forms -j 2 --all-targets -- -D warnings`: passed.
|
||||
- `cargo check --target wasm32-unknown-unknown -p pdfcraft-forms -j 2`: passed.
|
||||
- `cargo fmt --check`, `git diff --check`, and `cargo xtask layers`, `assets`, `parity`: passed.
|
||||
|
||||
These are local checks, not hosted CI. No private PDF was used.
|
||||
|
||||
## Rebase validation
|
||||
|
||||
Rebased onto freshly fetched upstream main
|
||||
`932bf9c6bfdd63e2a21bb831696a3447b9def9cf` on 2026-10-18. The only conflict was
|
||||
in the M6 roadmap row; upstream CSV/XML updates were retained alongside the compatibility
|
||||
note. The implementation and original regression are unchanged from the independent branch.
|
||||
|
||||
Validation used a fresh Cargo target directory, two build jobs, and locked, offline builds:
|
||||
|
||||
- The original regression failed with the upstream parser (`Some(0)` instead of `None`),
|
||||
then passed with the compatibility fix restored.
|
||||
- Forms: 83 unit tests passed. Engine: 282 unit and 9 integration tests passed.
|
||||
JavaScript: 47 unit tests passed. Doc tests passed. The timing probe and two private-file
|
||||
tests remained ignored.
|
||||
- The automation `javascript_through_tools` regression passed.
|
||||
- Scoped Clippy for forms, engine and JavaScript, with `--all-targets -- -D warnings`, passed.
|
||||
- WASM checks for those three crates passed.
|
||||
- Rust formatting, whitespace, layering, assets and parity checks passed.
|
||||
- A freshly built CLI filled synthetic fields with direct and inherited `/MaxLen 0`, rendered
|
||||
their values, saved/reopened them and accepted longer text after reopening. Its positive-limit
|
||||
control rejected six characters for `/MaxLen 5`; the synthetic input file stayed unchanged.
|
||||
|
||||
No performance optimizations, JavaScript snapshot/diff changes or profiling instrumentation
|
||||
are included. Existing executables, backups and private PDFs were not modified.
|
||||
|
||||
## PR #3323 refresh
|
||||
|
||||
Revalidated against freshly fetched upstream main
|
||||
`4db6ac4ae2448df6389d8bb0bc54db6ba175721c` on 2026-10-18. Conflicts in both
|
||||
roadmaps were resolved by keeping upstream's button-label and content-editing notes alongside
|
||||
the `/MaxLen 0` documentation. The original compatibility implementation and regression are
|
||||
preserved; no performance changes were added to the PR.
|
||||
|
||||
Forms: 85 unit tests passed. Engine: 283 unit and 15 integration tests passed. JavaScript:
|
||||
48 unit tests passed. The automation JavaScript regression, scoped Clippy with warnings denied,
|
||||
WASM, formatting, whitespace, layering, assets and parity checks all passed. The synthetic CLI
|
||||
fill/render/save/reopen and positive-limit control passed. The original regression again failed
|
||||
with the old parser and passed with the fix restored. Private-file tests remained ignored.
|
||||
|
||||
## 2026-10-19 revalidation
|
||||
|
||||
Rebased onto freshly fetched upstream main
|
||||
`459379ef9b127648cc30c0b2fe2b20c2b2375f04`. Only `ROADMAP.md` required manual
|
||||
conflict resolution; all upstream entries and the compatibility note were retained.
|
||||
The implementation, original regression and forms README remain byte-for-byte identical
|
||||
to the previously reviewed compatibility head. No performance changes were introduced.
|
||||
|
||||
Validation used a fresh isolated build directory, two jobs, and locked offline builds:
|
||||
|
||||
- The original regression failed with the old parser and passed with the fix restored.
|
||||
- Forms: 93 unit tests passed. Engine: 283 unit and 15 integration tests passed.
|
||||
JavaScript: 48 tests passed. Doc tests passed; the timing probe and two private-file
|
||||
tests remained ignored.
|
||||
- Automation JavaScript, scoped all-targets Clippy with warnings denied, WASM, formatting,
|
||||
whitespace, layering, assets and parity passed.
|
||||
- Synthetic CLI fill/render/save/reopen checks passed for direct zero and zero overriding
|
||||
an inherited positive limit. Longer text after reopening was accepted. The `/MaxLen 5`
|
||||
control rejected six characters, and the synthetic input stayed unchanged.
|
||||
|
||||
These are local checks, not hosted CI. Existing executables, backups and private PDFs
|
||||
were not modified; only the compatibility worktree was changed.
|
||||
|
||||
## Revision history
|
||||
|
||||
- 2026-10-19: refresh PR #3323 onto upstream `459379ef`, retain all upstream roadmap entries
|
||||
and independently repeat regression, native/WASM, repository and CLI validation.
|
||||
|
||||
- 2026-10-18: refresh PR #3323 onto upstream `4db6ac4a`, retaining both roadmap updates and
|
||||
independently revalidating compatibility and repository checks.
|
||||
|
||||
- 2026-10-18: rebase onto fetched upstream `932bf9c6`; preserve upstream roadmap updates and
|
||||
independently revalidate the regression, forms/engine/JavaScript suites and quality gates.
|
||||
|
||||
- 2026-10-18: isolate the compatibility change on current upstream main; verify the
|
||||
regression before and after the fix, the forms suite and scoped quality gates.
|
||||
Reference in New Issue
Block a user