1.14.2 — Hardened parsing for pathological PDFs

Packages

All packages were built from source checkpoint 4bee4f9.

• Rust crate pdf-inspector 1.14.2 — cargo add pdf-inspector@1.14.2
• Python pdf-inspector 1.14.2 — pip install pdf-inspector==1.14.2
• Node @firecrawl/pdf-inspector 1.14.2 — npm install @firecrawl/pdf-inspector@1.14.2
• WebAssembly @firecrawl/pdf-inspector-wasm 1.14.2 — npm install @firecrawl/pdf-inspector-wasm@1.14.2

Node platform packages

• Linux x64 glibc
• Linux x64 musl
• Linux ARM64 glibc
• Linux ARM64 musl
• macOS ARM64
• Windows x64 MSVC

Highlights

• Caps Form XObject expansion, CID /W ranges, Encoding/ToUnicode CMaps, content-stream decode, detector Tj/TJ lookback, and disjoint-rect clustering so crafted PDFs cannot force unbounded CPU or memory (#370, #372, #373, #375, #379, #380, #381).
• Form XObjects now track the text line matrix and honor T* / TL / ' / " / Tc / Tw, so nested form text keeps the right spacing (#369).
• Small-caps runs merge instead of being read as extra table columns (#371).

Changes since 1.14.1

Source: f4aab3b...4bee4f9

Resource bounds

• Bound Form XObject expansion per page (#370)
• Bound CID /W range expansion (#372)
• Cap content-stream decode before allocating operators (#373)
• Bound Encoding CMap cidrange expansion (#375)
• Bound ToUnicode bfrange expansion during subset remap (#379)
• Bound detector Tj/TJ operand lookback to the previous operator (#380)
• Bound disjoint-rect table clustering so overlap tests stay subquadratic (#381)

Extraction quality

• Track the text line matrix and handle T* / TL / ' / " / Tc / Tw in Form XObjects (#369)
• Merge small-caps runs so they stop reading as table columns (#371)