Crate pdf_inspector

Expand description

Smart PDF detection and text extraction using lopdf

§Quick start

// Full processing (detect + extract + markdown) with defaults
let result = pdf_inspector::process_pdf("document.pdf").unwrap();
println!("type: {:?}, pages: {}", result.pdf_type, result.page_count);
if let Some(md) = &result.markdown {
    println!("{md}");
}

// Fast metadata-only detection (no text extraction)
let info = pdf_inspector::detect_pdf("document.pdf").unwrap();
println!("type: {:?}, pages: {}", info.pdf_type, info.page_count);

// Custom options via builder
use pdf_inspector::{PdfOptions, ProcessMode};
let result = pdf_inspector::process_pdf_with_options(
    "document.pdf",
    PdfOptions::new().mode(ProcessMode::Analyze),
).unwrap();

Re-exports§

pub use detector::detect_pdf_type;
pub use detector::detect_pdf_type_mem;
pub use detector::detect_pdf_type_mem_with_config;
pub use detector::detect_pdf_type_with_config;
pub use detector::DetectionConfig;
pub use detector::PdfType;
pub use detector::PdfTypeResult;
pub use detector::ScanStrategy;
pub use extractor::extract_text;
pub use extractor::extract_text_with_positions;
pub use extractor::extract_text_with_positions_and_rotations_mem;
pub use extractor::extract_text_with_positions_and_rotations_mem_in_frame;
pub use extractor::extract_text_with_positions_and_rotations_mem_with_options;
pub use extractor::extract_text_with_positions_mem;
pub use extractor::extract_text_with_positions_mem_in_frame;
pub use extractor::extract_text_with_positions_mem_with_options;
pub use extractor::extract_text_with_positions_pages;
pub use extractor::extract_text_with_positions_pages_with_password;
pub use extractor::PositionFrame;
pub use extractor::PositionOptions;
pub use markdown::to_markdown;
pub use markdown::to_markdown_from_items;
pub use markdown::to_markdown_from_items_with_rects;
pub use markdown::to_markdown_from_items_with_rects_and_page_count;
pub use markdown::MarkdownOptions;
pub use markdown::MarkdownProfile;
pub use process_mode::ProcessMode;
pub use types::BoldSource;
pub use types::LayoutComplexity;
pub use types::PdfLine;
pub use types::PdfRect;
pub use types::TextItem;

Modules§

adobe_korea1: Adobe-Korea1 CID-to-Unicode mapping table.
detector: Smart PDF type detection without full document load
extractor: Text extraction from PDF using lopdf
glyph_names: Glyph name to Unicode mapping
markdown: Markdown conversion with structure detection.
process_mode
structure_tree: Tagged PDF structure tree parser.
tables: Table detection and formatting.
text_utils: Character classification and text utility functions.
tounicode: ToUnicode CMap parsing for PDF text extraction
types: Shared types used across the extraction and markdown pipelines.
vision: Optional native vision primitives used by OCR pipelines.

Structs§

FontCMapGaps: A font whose ToUnicode CMap (or, for a font without one, the embedded program’s own cmap table) had no entry for some of the codes the document shows through it, and what became of those codes.
PageMarkdown: Per-page markdown extraction result.
PageOcrReasons: OCR reasons for a single 1-indexed page.
PageRegionResult: Result for a page’s region extractions.
PagesExtractionResult: Combined per-page markdown extraction and layout classification result.
PdfClassification: Lightweight classification result for routing decisions.
PdfOptions: Configuration for process_pdf_with_options and friends.
PdfProcessResult: High-level PDF processing result.
RegionText: Result for a single region’s text extraction.
StructureElement: One structure-tree element reference from a tagged PDF, resolved to a page and Marked Content ID.
TableExtractionResult: Markdown for one extracted table plus a diagnostic flag describing which path produced it.
TsrTableInput: Input for extract_tables_with_structure_mem: one cropped table region plus the raw structure-recovery output for it.
VectorGridDetection: Region-scoped vector grid detection result for TSR-compatible callers.

Enums§

PageRotation: Which way a page’s coordinate frame was turned so that predominantly rotated text reads along +x.
PdfError

Constants§

OCR_REASON_INVISIBLE_TEXT_LAYER: OCR reason: every text-showing operator on the page leaves nothing to see — text render mode 3 (invisible), or mode 7 (clip only) with nothing painted through the clip — while an image covers at least half of the page: a scan carrying a text layer nobody sees. What that layer says is not what the page shows, so the page is read from its raster.
OCR_REASON_NO_TEXT: OCR reason: the page has no extractable text and no image to OCR — blank, or content the parser cannot reach.
OCR_REASON_SCANNED: OCR reason: the page is a scanned image (a full-page raster / image-only page) with no usable text layer.
OCR_REASON_SUSPECTED_GARBLED_TEXT: OCR reason emitted when the extracted text layer appears garbled due to broken font decoding or mojibake.
OCR_REASON_VECTOR_TEXT: OCR reason: the page’s text is drawn as vector outlines (path operators) rather than real text operators, so it cannot be extracted as characters.

Functions§

classify_pdf_mem: Classify a PDF from a memory buffer without extracting text. Returns the PDF type and which pages need OCR (~10-50ms).
collect_text_in_region: Collect text items that fall within a region bbox (top-left origin, PDF points) and return them as a single string in reading order.
collect_text_in_region_in_frame: collect_text_in_region with the page’s coordinate frame given explicitly instead of inferred: rotation is the turn extract_text_with_positions_and_rotations_mem reported for the items’ page (PageRotation::Upright for pages absent from that map). The region bbox stays in top-left page coordinates and is turned to match.
detect_pdf: Fast metadata-only detection — no text extraction or markdown generation.
detect_pdf_mem: Fast metadata-only detection from a memory buffer.
detect_vector_grid_in_region_mem: Detect a vector ruled-line / rectangle grid inside one page region.
extract_pages_markdown: Path-based wrapper for extract_pages_markdown_mem.
extract_pages_markdown_mem: Extract formatted markdown for pages of a PDF, with layout classification metadata.
extract_structure_elements: Path-based wrapper for extract_structure_elements_mem.
extract_structure_elements_mem: Extract structure-tree element references from a tagged PDF in memory.
extract_tables_in_regions_mem: Extract tables within bounding-box regions from a PDF in memory.
extract_tables_in_regions_mem_in_frame: extract_tables_in_regions_mem with the frame of the region rects given explicitly — see extract_text_in_regions_mem_in_frame.
extract_tables_in_regions_mem_with_options: extract_tables_in_regions_mem_in_frame with every option given as a PositionOptions — see extract_text_in_regions_mem_with_options.
extract_tables_with_structure_auto_mem: Auto-fallback variant of extract_tables_with_structure_mem: runs the TSR-hybrid path, checks the resulting cells for known SLANet detection pathologies (phantom rows, multi-row-in-cell text), and falls back to the heuristic extract_tables_in_regions_mem for any input where the TSR path looks compromised.
extract_tables_with_structure_cells_mem: Extract structured cells using externally-supplied structure recovery.
extract_tables_with_structure_mem: Extract markdown tables using externally-supplied structure recovery.
extract_text_in_regions_mem: Extract text within bounding-box regions from a PDF in memory.
extract_text_in_regions_mem_in_frame: extract_text_in_regions_mem with the frame of the region rects given explicitly. PositionFrame::Sheet reads them as that function does; PositionFrame::Display reads them on the rendered page — the visible page box turned clockwise by the page’s inheritable /Rotate, top-left origin, y down — the frame a layout model working on a page image reports boxes in. Pages whose text is predominantly rotated are handled the same way in both frames.
extract_text_in_regions_mem_with_options: extract_text_in_regions_mem_in_frame with every option given as a PositionOptions: the frame the region rects are read in, and whether bold is also read from the font’s weight class (bold_from_weight: a weight class of bold_weight_threshold or more, 600 by default, is bold, and a run the weight makes bold is then its own item while the region’s lines are assembled). The default options are extract_text_in_regions_mem.
process_pdf: Process a PDF file with full extraction (detect → extract → markdown).
process_pdf_mem: Process a PDF from a memory buffer with full extraction.
process_pdf_mem_with_configDeprecated: Process PDF from memory buffer with custom detection and markdown configuration.
process_pdf_mem_with_options: Process a PDF from a memory buffer with custom options.
process_pdf_with_configDeprecated: Process a PDF file with custom detection and markdown configuration.
process_pdf_with_config_pagesDeprecated: Process a PDF file with custom configuration and optional page filter.
process_pdf_with_options: Process a PDF file with custom options.
widen_degenerate_form_bboxes_mem: The PDF written back out with the /BBox of its Form XObjects repaired — a zero-area box widened, numerals too large for any parser saturated — for callers that render the document with their own renderer. Rust only: the Python, Node.js and WebAssembly bindings do not expose it.

Crate pdf_inspector

Crate pdf_inspector Copy item path

§Quick start

Re-exports§

Modules§

Structs§

Enums§

Constants§

Functions§

Crate pdf_inspector