anydoc 0.2.4

Convert documents (doc, docx, odt, rtf, epub, pdf, presentations, spreadsheets, csv) to GitHub-Flavored Markdown
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52

//! PDF via [pdf-inspector]: classification plus direct Markdown extraction.
//!
//! Unlike the other frontends, pdf-inspector emits Markdown itself, so PDFs
//! bypass the document model and the shared GFM writer. OCR is out of scope
//! here: a document with scanned or image-only pages errors naming them,
//! whether that is every page or one of a hundred, because output missing
//! those pages would read as complete.
//!
//! [pdf-inspector]: https://github.com/firecrawl/pdf-inspector

use crate::error::ConvertError;
use pdf_inspector::PdfError;

pub fn to_markdown(bytes: &[u8]) -> Result<String, ConvertError> {
    let result = pdf_inspector::process_pdf_mem(bytes).map_err(map_error)?;
    if !result.pages_needing_ocr.is_empty() {
        // Detection samples content streams and over-reports short or
        // image-heavy text pages; extraction knows which of them yielded none.
        let flagged: Vec<u32> = result.pages_needing_ocr.iter().map(|page| page - 1).collect();
        let pages = pdf_inspector::extract_pages_markdown_mem(bytes, Some(&flagged))
            .map_err(map_error)?
            .pages_needing_ocr;
        if !pages.is_empty() {
            return Err(ConvertError::NeedsOcr { pages, page_count: result.page_count });
        }
    }
    if result.has_encoding_issues {
        log::warn!("broken font encodings detected; extracted text may be garbled");
    }
    match result.markdown {
        Some(mut markdown) if !markdown.trim().is_empty() => {
            if !markdown.ends_with('\n') {
                markdown.push('\n');
            }
            Ok(markdown)
        }
        _ => Err(ConvertError::Unsupported(format!(
            "PDF has no extractable text ({:?}, {} pages)",
            result.pdf_type, result.page_count
        ))),
    }
}

fn map_error(e: PdfError) -> ConvertError {
    match e {
        PdfError::Encrypted => ConvertError::Encrypted,
        PdfError::Io(e) => ConvertError::Io(e),
        PdfError::NotAPdf(detail) => ConvertError::malformed(format!("not a PDF: {detail}")),
        PdfError::InvalidStructure => ConvertError::malformed("invalid PDF structure"),
        PdfError::Parse(detail) => ConvertError::malformed(detail),
    }
}