kreuzberg 4.8.2

High-performance document intelligence library for Rust. Extract text, metadata, and structured data from PDFs, Office documents, images, and 91+ formats and 248 programming languages via tree-sitter code intelligence with async/sync APIs.
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
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72

//! Unified internal DTO for the PDF markdown pipeline.
//!
//! All extraction backends (pdfium structure tree, OCR)
//! produce `PageContent` which the shared pipeline converts to markdown.

use super::geometry::Rect;

/// Semantic role hint from the extraction source.
///
/// Set from PDF structure tree tags (`ContentRole`) or layout model predictions.
/// `None` when the source doesn't provide semantic information.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub(crate) enum SemanticRole {
    Heading { level: u8 },
    Paragraph,
    ListItem,
    Code,
    Formula,
    Caption,
    TableCell,
    Figure,
    BlockQuote,
    Other,
}

/// Granularity level of a content element.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub(crate) enum ElementLevel {
    /// Single word (OCR word-level, or whitespace-split from structure tree).
    Word,
    /// A full line of text (pdfium segment = one baseline run).
    Line,
    /// A block/paragraph of text (structure tree block).
    Block,
}

/// A single content element with normalized spatial and style data.
///
/// This is the unified representation that all extraction backends produce.
/// Fields that a backend cannot provide are set to defaults (`None`/`false`).
#[derive(Debug, Clone)]
pub(crate) struct ContentElement {
    /// Text content.
    pub text: String,
    /// Bounding box in PDF coordinate space (points, y=0 at page bottom).
    /// `None` when positional data is unavailable (e.g. some structure tree blocks).
    pub bbox: Option<Rect>,
    /// Font size in points. `None` for OCR (no font info available).
    pub font_size: Option<f32>,
    /// Whether the text is bold.
    pub is_bold: bool,
    /// Whether the text is italic.
    pub is_italic: bool,
    /// Whether the font is monospace/fixed-pitch.
    pub is_monospace: bool,
    /// Semantic role from the extraction source (structure tree or layout model).
    pub semantic_role: Option<SemanticRole>,
    /// Granularity of this element.
    pub level: ElementLevel,
    /// List item label (e.g. "1.", "a)", "•") when `semantic_role == ListItem`.
    pub list_label: Option<String>,
    /// Layout class hint from a layout detection model (e.g. OCR block type).
    /// `None` when layout detection is unavailable or not applicable.
    pub layout_class: Option<super::types::LayoutHintClass>,
}

/// All content extracted from a single page.
#[derive(Debug, Clone)]
pub(crate) struct PageContent {
    /// Extracted content elements in reading order.
    pub elements: Vec<ContentElement>,
}