AnyDocument

class paperjam.AnyDocument

A format-agnostic document for non-PDF files: DOCX, XLSX, PPTX, HTML, and EPUB. Provides text extraction, table extraction, structured content parsing, image extraction, Markdown conversion, and format conversion.

AnyDocument is returned automatically by paperjam.open() when the input is not a PDF. You can also instantiate it directly.

Use as a context manager for automatic resource cleanup:

import paperjam

# Via paperjam.open() (recommended)
with paperjam.open("report.docx") as doc:
    print(doc.extract_text())

# Direct instantiation
from paperjam import AnyDocument

doc = AnyDocument("report.docx")
doc = AnyDocument(raw_bytes, format="pptx")

---

Constructor

AnyDocument(
    path_or_bytes: str | os.PathLike | bytes,
    *,
    format: str | None = None,
)

Parameters

Name	Type	Description
path_or_bytes	str, PathLike, or bytes	File path or raw document bytes
format	str | None	Format hint: "docx", "xlsx", "pptx", "html", "epub". Required when passing bytes without an extension to infer from.

Raises ParseError if the file cannot be parsed. ValueError if format is required but not provided.

---

Properties

format

doc.format: str

The detected document format: "docx", "xlsx", "pptx", "html", or "epub".

page_count

doc.page_count: int

Number of logical pages. For DOCX this is the estimated page count. For XLSX it is the number of sheets. For PPTX it is the number of slides. For HTML and EPUB it is based on section breaks.

metadata

doc.metadata: Metadata

Document metadata as a frozen Metadata dataclass. Available fields depend on the format. See types reference.

bookmarks

doc.bookmarks: list[Bookmark]

Table of contents or bookmark tree extracted from the document. Returns an empty list if the format has no bookmark equivalent.

---

Text extraction

extract_text

doc.extract_text(
    *,
    pages: list[int] | None = None,
) -> str

Extract plain text from the document. Pass pages (1-indexed) to limit extraction to specific pages/sheets/slides. None extracts from all pages.

extract_text_lines

doc.extract_text_lines(
    *,
    pages: list[int] | None = None,
) -> list[TextLine]

Extract text as a list of TextLine objects. Bounding box information is available when the format supports it (DOCX, PPTX) and None otherwise.

---

Table extraction

extract_tables

doc.extract_tables(
    *,
    pages: list[int] | None = None,
    min_rows: int = 2,
    min_cols: int = 2,
) -> list[Table]

Extract tables. For XLSX, every sheet range with data is returned as a table. For DOCX and PPTX, embedded tables are extracted. For HTML, <table> elements are parsed.

for table in doc.extract_tables():
    print(f"Headers: {table.headers}")
    print(f"Rows: {len(table.rows)}")
    print(table.to_csv())

---

Structure extraction

extract_structure

doc.extract_structure(
    *,
    heading_size_ratio: float = 1.2,
    detect_lists: bool = True,
    include_tables: bool = True,
) -> list[ContentBlock]

Extract structured content blocks: headings, paragraphs, list items, and tables. Returns a list of ContentBlock objects.

---

Image extraction

extract_images

doc.extract_images(
    *,
    pages: list[int] | None = None,
) -> list[Image]

Extract embedded images from the document.

---

Conversion

to_markdown

doc.to_markdown(
    *,
    heading_offset: int = 0,
    include_page_numbers: bool = False,
    html_tables: bool = False,
) -> str

Convert the entire document to Markdown.

md = doc.to_markdown(heading_offset=1, include_page_numbers=True)

convert_to

doc.convert_to(
    format: str,
    *,
    output: str | os.PathLike | None = None,
) -> bytes | None

Convert the document to another format. If output is provided, the result is written to disk and None is returned. Otherwise, the converted bytes are returned.

# Get bytes
pdf_bytes = doc.convert_to("pdf")

# Write to file
doc.convert_to("pdf", output="report.pdf")

Supported target formats: "pdf", "docx", "html", "epub", "markdown". Not all conversions are supported for all source formats. See the Format Conversion guide for the full matrix.

---

Saving

save

doc.save(path: str | os.PathLike) -> None

Save the document to disk in its original format.

save_bytes

doc.save_bytes() -> bytes

Serialize the document to bytes in its original format.

close

doc.close() -> None

Release resources. Called automatically when used as a context manager.

---

Comparison with Document

AnyDocument provides the extraction and conversion surface that works across all formats. PDF-specific features are only available on Document:

Feature	AnyDocument	Document (PDF)
extract_text	yes	yes
extract_text_lines	yes	yes
extract_tables	yes	yes
extract_structure	yes	yes
extract_images	yes	yes
to_markdown	yes	yes
convert_to	yes	yes
metadata / bookmarks	yes	yes
search	no	yes
pages[] accessor	no	yes
redact / redact_text	no	yes
encrypt	no	yes
sign / verify_signatures	no	yes
fill_form / form_fields	no	yes
render_page / render_pages	no	yes
add_watermark / add_annotation	no	yes
split / reorder / rotate	no	yes
optimize	no	yes
validate_pdf_a / validate_pdf_ua	no	yes

If you need PDF-specific features, use paperjam.open() on a PDF file (which returns Document) or convert the document first with convert_to("pdf") and then open the result.