Struct ParallelExtractor 

pub struct ParallelExtractor;

Available on crate feature parallel only.

Parallel extractor that processes PDF pages across multiple threads.

Uses a batched strategy: divides pages into chunks and each rayon worker opens a single PdfDocument instance to process its chunk sequentially. This amortizes the cost of document opening, xref parsing, page tree walks, and font loading across many pages instead of paying it per-page.

All results are returned in page order regardless of which thread processed each page.

Implementations

impl ParallelExtractor

pub fn extract_all_text(path: &Path) -> Result<Vec<String>>

Extract plain text from every page of a PDF in parallel.

Opens the document once on the calling thread to determine the page count, then divides pages into batches distributed across rayon worker threads. Each worker opens a single PdfDocument and extracts all pages in its batch.

Returns a Vec<String> with one entry per page, in page order.

Errors

Returns the first error encountered by any worker. If multiple workers fail, only one error is propagated (rayon semantics).

Example

use std::path::Path;
use pdf_oxide::parallel::ParallelExtractor;

let pages = ParallelExtractor::extract_all_text(Path::new("report.pdf"))?;
assert_eq!(pages.len(), 42);

pub fn extract_all_markdown( path: &Path, options: &ConversionOptions, ) -> Result<Vec<String>>

Extract Markdown from every page of a PDF in parallel.

Behaves like extract_all_text but converts each page to Markdown using the supplied ConversionOptions.

Returns a Vec<String> with one entry per page, in page order.

Errors

Returns the first error encountered by any worker.

Example

use std::path::Path;
use pdf_oxide::parallel::ParallelExtractor;
use pdf_oxide::converters::ConversionOptions;

let opts = ConversionOptions::default();
let pages = ParallelExtractor::extract_all_markdown(
    Path::new("report.pdf"),
    &opts,
)?;