ExtractousEx (ExtractousEx v0.2.1)

View Source

ExtractousEx provides text extraction from various document formats using the Extractous Rust library.

Supports PDF, Microsoft Office documents, HTML, plain text, CSV, JSON, and many other formats.

Extraction Methods

Each method has a corresponding bang version (e.g., extract_from_file!/2) that raises on error.

Options

All extraction methods accept a keyword list of options:

General options (top-level):

  • :xml - when true, returns structured XML output. When false (default), returns plain text.
  • :max_length - maximum length of extracted text in characters. Default is 500,000.
  • :encoding - encoding for text extraction. Supported values: "UTF-8" (default), "UTF-16BE", "US-ASCII".

PDF options (under :pdf key):

  • :ocr_strategy - "NO_OCR", "AUTO", "OCR_ONLY", or "OCR_AND_TEXT_EXTRACTION"
  • :extract_annotation_text - extract text from PDF annotations
  • :extract_inline_images - extract inline images
  • :extract_unique_inline_images_only - extract only unique inline images
  • :extract_marked_content - extract marked content

Office options (under :office key):

  • :include_shape_based_content - include shape-based content
  • :include_slide_notes - include slide notes
  • :include_slide_master_content - include slide master content
  • :concatenate_phonetic_runs - concatenate phonetic runs
  • :include_headers_and_footers - include headers and footers
  • :include_deleted_content - include deleted content
  • :include_move_from_content - include move-from content
  • :include_missing_rows - include missing rows
  • :extract_macros - extract macros
  • :extract_all_alternatives_from_msg - extract all alternatives from MSG files

OCR options (under :ocr key):

  • :language - Tesseract language (e.g., "eng", "deu", "eng+deu")
  • :timeout_seconds - OCR timeout in seconds
  • :density - image density for OCR
  • :depth - image depth for OCR
  • :apply_rotation - apply rotation during OCR
  • :enable_image_preprocessing - enable image preprocessing for OCR

Examples

# Simple usage
ExtractousEx.extract_from_file("document.pdf")
{:ok, %{content: "Document text...", metadata: %{}}}

# With basic options
ExtractousEx.extract_from_file("document.pdf", xml: true, max_length: 100_000)
{:ok, %{content: "<xml>...</xml>", metadata: %{}}}

# With advanced options
ExtractousEx.extract_from_file("document.pdf",
  max_length: 1_000_000,
  xml: true,
  pdf: [
    ocr_strategy: "AUTO",
    extract_annotation_text: true
  ],
  ocr: [
    language: "eng+deu",
    timeout_seconds: 60
  ]
)
{:ok, %{content: "Document text...", metadata: %{}}}

Summary

Functions

extract_from_bytes(bytes, opts \\ [])

Extracts text and metadata from binary data.

extract_from_bytes!(bytes, opts \\ [])

Extracts text from bytes, raising on error.

extract_from_file(file_path, opts \\ [])

Extracts text and metadata from a file.

extract_from_file!(file_path, opts \\ [])

Extracts text from a file, raising on error.

extract_from_url(url, opts \\ [])

Extracts text and metadata from a URL.

extract_from_url!(url, opts \\ [])

Extracts text from a URL, raising on error.

Functions

extract_from_bytes(bytes, opts \\ [])

@spec extract_from_bytes(
  binary(),
  keyword()
) :: {:ok, %{content: String.t(), metadata: map()}} | {:error, String.t()}

Extracts text and metadata from binary data.

Options

Accepts a keyword list of options. See module documentation for full details.

Examples

# Extract from PDF bytes
{:ok, pdf_data} = File.read("document.pdf")
ExtractousEx.extract_from_bytes(pdf_data)
{:ok, %{content: "Document text...", metadata: %{}}}

# With options
ExtractousEx.extract_from_bytes(data,
  xml: true,
  max_length: 100_000
)
{:ok, %{content: "<xml>...</xml>", metadata: %{}}}

Returns

Returns {:ok, result} on success where result is a map with:

  • :content - The extracted text content
  • :metadata - Document metadata as a map

Returns {:error, reason} on failure.

extract_from_bytes!(bytes, opts \\ [])

@spec extract_from_bytes!(
  binary(),
  keyword()
) :: %{content: String.t(), metadata: map()}

Extracts text from bytes, raising on error.

Same as extract_from_bytes/2 but raises an exception on failure.

Examples

{:ok, data} = File.read("document.pdf")
ExtractousEx.extract_from_bytes!(data)
%{content: "Document text...", metadata: %{}}

extract_from_file(file_path, opts \\ [])

@spec extract_from_file(
  String.t(),
  keyword()
) :: {:ok, %{content: String.t(), metadata: map()}} | {:error, String.t()}

Extracts text and metadata from a file.

Options

Accepts a keyword list of options. See module documentation for full details.

Examples

# Simple usage
ExtractousEx.extract_from_file("document.pdf")
{:ok, %{content: "Document text...", metadata: %{}}}

# With options
ExtractousEx.extract_from_file("document.pdf", xml: true)
{:ok, %{content: "<xml>...</xml>", metadata: %{}}}

# With nested options
ExtractousEx.extract_from_file("document.pdf",
  xml: true,
  pdf: [ocr_strategy: "NO_OCR"]
)
{:ok, %{content: "Document text...", metadata: %{}}}

Returns

Returns {:ok, result} on success where result is a map with:

  • :content - The extracted text content
  • :metadata - Document metadata as a map

Returns {:error, reason} on failure.

extract_from_file!(file_path, opts \\ [])

@spec extract_from_file!(
  String.t(),
  keyword()
) :: %{content: String.t(), metadata: map()}

Extracts text from a file, raising on error.

Same as extract_from_file/2 but raises an exception on failure.

Examples

ExtractousEx.extract_from_file!("document.pdf")
%{content: "Document text...", metadata: %{}}

extract_from_url(url, opts \\ [])

@spec extract_from_url(
  String.t(),
  keyword()
) :: {:ok, %{content: String.t(), metadata: map()}} | {:error, String.t()}

Extracts text and metadata from a URL.

Options

Accepts a keyword list of options. See module documentation for full details.

Examples

# Extract from a URL
ExtractousEx.extract_from_url("https://example.com/document.pdf")
{:ok, %{content: "Document text...", metadata: %{}}}

# With options
ExtractousEx.extract_from_url("https://example.com/doc.html",
  xml: true,
  office: [include_headers_and_footers: true]
)
{:ok, %{content: "<html>...</html>", metadata: %{}}}

Returns

Returns {:ok, result} on success where result is a map with:

  • :content - The extracted text content
  • :metadata - Document metadata as a map

Returns {:error, reason} on failure.

extract_from_url!(url, opts \\ [])

@spec extract_from_url!(
  String.t(),
  keyword()
) :: %{content: String.t(), metadata: map()}

Extracts text from a URL, raising on error.

Same as extract_from_url/2 but raises an exception on failure.

Examples

ExtractousEx.extract_from_url!("https://example.com/document.pdf")
%{content: "Document text...", metadata: %{}}