Abimael MartellandClaude Opus 4.7
9cce4dd161
Closes the residual run-on-cell pattern observed on FNBO branch list
after 1.6.2 deployed:
|Shawnee Blue Valley Parkway|Kansas|6301 Pflumm, ...|0523.04|
|Sonoma Plaza|Kansas Kansas|<addr1> <addr2>|0531.05|
|Mitchell Woonsocket|South Dakota|<addr1>|9628.01|
Cause: when col-N cells across multiple consecutive rows are y-shifted
the same way (a local SLANet drift), the stage 2 orphan pass sees
multiple orphans qualifying for the same nearest empty cell. The cell
gets all of them appended in order, producing "RowA-text RowB-text".
Fix: track the y-coordinate of the first orphan that lands in each
cell. Subsequent orphans only join that cell if their y is within
half-a-row-height of the first orphan's y (same line). Cross-line
orphans skip that cell and look for the next-nearest empty cell on
their own line.
Same-line slack preserves multi-token branch names like
"Blue Valley Parkway" (3 PDF text items at the same y) — all three
stack into the same cell. Cross-row stacking is what gets rejected.
Two new tests:
- stage2_rejects_cross_line_stacking_into_same_cell
Two orphans on different rows, both equidistant to the same empty
cell. First wins; second routes to its own row's cell.
- stage2_allows_same_line_orphans_to_stack_into_one_cell
Three same-line orphans (multi-token branch name) all land in the
same empty cell, joined by spaces.
The existing four stage-2 tests + the cell-bleed regression test from
PR #62 + #63 all pass: 416 lib + 115 integration + 2 doctests.
clippy + fmt clean.
Bump @firecrawl/pdf-inspector to 1.6.3 (patch — refines 1.6.2; no API
changes).
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
PDF Inspector
Fast PDF classification and region-based text extraction for Node.js/Bun. Native Rust performance via napi-rs.
Built by Firecrawl for hybrid OCR pipelines — extract text from PDF structure where possible, fall back to OCR only when needed.
Install
npm install @firecrawl/pdf-inspector
# or
bun add @firecrawl/pdf-inspector
Prebuilt binaries included for linux-x64 and macOS ARM64. No Rust toolchain needed.
API
classifyPdf(buffer: Buffer): PdfClassification
Classify a PDF as TextBased, Scanned, Mixed, or ImageBased (~10-50ms). Returns which pages need OCR.
import { classifyPdf } from '@firecrawl/pdf-inspector'
import { readFileSync } from 'fs'
const pdf = readFileSync('document.pdf')
const result = classifyPdf(pdf)
console.log(result.pdfType) // "TextBased" | "Scanned" | "Mixed" | "ImageBased"
console.log(result.pageCount) // 42
console.log(result.pagesNeedingOcr) // [5, 12, 15] (0-indexed)
console.log(result.confidence) // 0.875
extractTextInRegions(buffer: Buffer, pageRegions: PageRegions[]): PageRegionTexts[]
Extract text within bounding-box regions from a PDF. Designed for hybrid OCR pipelines where a layout model detects regions in rendered page images, and this function extracts text from the PDF structure for text-based pages — skipping GPU OCR.
Each region result includes a needsOcr flag that signals unreliable extraction (empty text, GID-encoded fonts, garbage text, encoding issues).
import { extractTextInRegions } from '@firecrawl/pdf-inspector'
const result = extractTextInRegions(pdf, [
{
page: 0, // 0-indexed
regions: [
[0, 0, 300, 400], // [x1, y1, x2, y2] in PDF points, top-left origin
[300, 0, 612, 400],
]
}
])
for (const region of result[0].regions) {
if (region.needsOcr) {
// Unreliable text — send this region to OCR instead
} else {
console.log(region.text) // Extracted text in reading order
}
}
Types
interface PdfClassification {
pdfType: string // "TextBased" | "Scanned" | "Mixed" | "ImageBased"
pageCount: number
pagesNeedingOcr: number[] // 0-indexed page numbers
confidence: number // 0.0 - 1.0
}
interface PageRegions {
page: number // 0-indexed
regions: number[][] // [[x1, y1, x2, y2], ...] in PDF points, top-left origin
}
interface PageRegionTexts {
page: number
regions: RegionText[]
}
interface RegionText {
text: string
needsOcr: boolean // true when text is unreliable
}
Platforms
| Platform | Architecture | Supported |
|---|---|---|
| Linux | x64 | Yes |
| macOS | ARM64 | Yes |
License
MIT