Skip to content
PDF Oxide

Page API Reference

Since v0.3.34 every binding exposes a Page object so you can iterate a document and call extraction methods on the page directly, instead of threading page_index through every extraction call. The type is named Page consistently in Python, Node.js, C#, and Go; Rust exposes the same shape through PdfPage.

Quick Example

Python

from pdf_oxide import PdfDocument

with PdfDocument("paper.pdf") as doc:
    for page in doc:                       # len(doc), doc[i], doc[-1] also work
        print(page.text[:80])
        md = page.markdown(detect_headings=True)

Rust

use pdf_oxide::api::Pdf;

let mut doc = Pdf::open("paper.pdf")?;
for i in 0..doc.page_count()? {
    let page = doc.page(i)?;
    println!("{}", &page.text()?[..80]);
}

JavaScript / TypeScript (Node)

const { PdfDocument } = require("pdf-oxide");

const doc = new PdfDocument("paper.pdf");
for (const page of doc) {
  console.log(page.extractText().slice(0, 80));
}
doc.close();

Go

package main

import (
    "fmt"
    "log"
    pdfoxide "github.com/yfedoseev/pdf_oxide/go"
)

func main() {
    doc, err := pdfoxide.Open("paper.pdf")
    if err != nil { log.Fatal(err) }
    defer doc.Close()

    pages, _ := doc.Pages()
    for _, page := range pages {
        text, _ := page.ExtractText()
        fmt.Println(text[:80])
    }
}

C#

using PdfOxide;

using var doc = PdfDocument.Open("paper.pdf");
foreach (var page in doc.Pages)
{
    Console.WriteLine(page.ExtractText()[..Math.Min(80, page.ExtractText().Length)]);
}

Python — Page

Lazy property surface — content is parsed on first access and cached on the Page.

Member Returns Description
page.text str Extracted text (column-aware)
page.chars list[Char] Character-level records with bbox, font
page.words list[Word] Word-level records with bbox
page.lines list[TextLine] Text lines with bbox
page.spans list[Span] Styled spans (font, size, weight)
page.tables list[Table] Structured table rows + cell bboxes
page.images list[Image] Image metadata
page.paths list[Path] Vector path records
page.annotations list[Annotation] Annotations on this page
page.markdown(detect_headings=True) str Markdown conversion
page.plain_text() str Plain text (no layout hints)
page.html() str HTML conversion
page.render(format="png") bytes Render page as PNG / JPEG
page.search(term, case_sensitive=False) list[SearchResult] Find text on this page
page.region(rect) PageRegion Scoped extraction inside a rect 
with PdfDocument("paper.pdf") as doc:
    page = doc[0]                 # or doc.page(0)
    for word in page.words:       # first access parses; subsequent calls cached
        print(word.text, word.bbox)

    # Scoped extraction
    header = page.region((0, 700, 612, 92)).extract_text()

The pre-existing editor PdfPage class (for writing) is unchanged; the new Page is strictly read-only.

Rust — PdfPage

use pdf_oxide::api::Pdf;

let mut doc = Pdf::open("paper.pdf")?;
let page = doc.page(0)?;

let text = page.text()?;
let words = page.extract_words()?;
let tables = page.extract_tables()?;
let md = page.to_markdown(true)?;

Methods available on PdfPage:

  • text(), plain_text(), to_markdown(detect_headings), to_html()
  • extract_chars(), extract_words(), extract_lines(), extract_spans()
  • extract_tables(), extract_paths(), extract_images()
  • annotations(), render(format)
  • search(term) — scoped search
  • find_text_containing(substring) — DOM-level hit list with IDs

Node.js — Page

const { PdfDocument } = require("pdf-oxide");

const doc = new PdfDocument("paper.pdf");
const page = doc.page(0);

console.log(page.width, page.height, page.rotation);  // cached
console.log(page.extractText());
const words = page.extractWords();
const tables = page.extractTables();
const md = page.toMarkdown();

PdfDocument supports for..of via Symbol.iterator, plus doc.page(i) and doc.pageCount().

Six previously native-only methods are now available on both Page and PdfDocument via the TS layer:

  • extractWords
  • extractTextLines
  • extractTables
  • extractPaths
  • getEmbeddedImages
  • ocrExtractText

Each method has an async sibling — extractTextAsync, toMarkdownAsync, etc.

Go — Page

doc, _ := pdfoxide.Open("paper.pdf")
defer doc.Close()

page, _ := doc.Page(0)
text, _ := page.ExtractText()
md, _   := page.ToMarkdown()
tables, _ := page.ExtractTables()

// Iterate every page
all, _ := doc.Pages()
for i, p := range all {
    t, _ := p.ExtractText()
    fmt.Printf("page %d: %d chars\n", i, len(t))
}

Go’s Page struct has the full method surface: ExtractText, ToMarkdown, ToHtml, ToPlainText, ExtractWords, ExtractTextLines, ExtractTables, ExtractChars, ExtractPaths, Annotations, Images, Fonts, RenderPage, Search.

C# — Page

using PdfOxide;

using var doc = PdfDocument.Open("paper.pdf");

Page page = doc[0];                            // or doc.Pages[0] or doc.Page(0)
string text = page.ExtractText();
string md   = page.ToMarkdown();
Table[] tables = page.ExtractTables();

// Async variants
string textAsync = await page.ExtractTextAsync();
string mdAsync   = await page.ToMarkdownAsync();

doc.Pages is IReadOnlyList<Page>. Every sync method has an async Task<T> counterpart with CancellationToken support.

Structured Table Shape

extract_tables() (available on both PdfDocument and Page) returns a consistent Table type across languages:

Language Type Cell access
Rust Table iterate rows[i].cells[j]
Python dict row["cells"][i]["text"]
Go Table table.CellText(row, col)
C# Table table.CellText(row, col)
Node.js Table interface table.cells[row][col]

Each cell carries text plus a bounding box so you can correlate the extraction back to coordinates on the page.

Migration from doc.extract_*(page_index)

Old (still supported):

doc = PdfDocument("paper.pdf")
for i in range(doc.page_count()):
    print(doc.extract_text(i))
    print(doc.to_markdown(i, detect_headings=True))
    print(doc.extract_tables(i))

New (v0.3.34+):

with PdfDocument("paper.pdf") as doc:
    for page in doc:
        print(page.text)
        print(page.markdown(detect_headings=True))
        print(page.tables)

Both styles stay supported; the Page style reads better for per-page pipelines and avoids repeated index bookkeeping.