llmdocparser 0.1.5

Using LLM to parse PDF and get better chunk for retrieval

LLMDocParser

A package for parsing PDFs and analyzing their content using LLMs.

This package is an improvement based on the concept of gptpdf.

Method

gptpdf uses PyMuPDF to parse PDFs, identifying both text and non-text regions. It then merges or filters the text regions based on certain rules, and inputs the final results into a multimodal model for parsing. This method is particularly effective.

Based on this concept, I made some minor improvements.

Main Process

Using a layout analysis model, each page of the PDF is parsed to identify the type of each region, which includes Text, Title, Figure, Figure caption, Table, Table caption, Header, Footer, Reference, and Equation. The coordinates of each region are also obtained.

Layout Analysis Result Example:

[{'header': ((101, 66, 436, 102), 0)},
 {'header': ((1038, 81, 1088, 95), 1)},
 {'title': ((106, 215, 947, 284), 2)},
 {'text': ((101, 319, 835, 390), 3)},
 {'text': ((100, 565, 579, 933), 4)},
 {'text': ((100, 967, 573, 1025), 5)},
 {'text': ((121, 1055, 276, 1091), 6)},
 {'reference': ((101, 1124, 562, 1429), 7)},
 {'text': ((610, 565, 1089, 930), 8)},
 {'text': ((613, 976, 1006, 1045), 9)},
 {'title': ((612, 1114, 726, 1129), 10)},
 {'text': ((611, 1165, 1089, 1431), 11)},
 {'title': ((1011, 1471, 1084, 1492), 12)}]

This result includes the type, coordinates, and reading order of each region. By using this result, more precise rules can be set to parse the PDF.

Finally, input the images of the corresponding regions into a multimodal model, such as GPT-4o or Qwen-VL, to directly obtain text blocks that are friendly to RAG solutions.

img_path	type	page_no	filename	content	filepath
{absolute_path}/page_1_title.png	Title	1	attention is all you need	[Text Block 1]	{file_absolute_path}
{absolute_path}/page_1_text.png	Text	1	attention is all you need	[Text Block 2]	{file_absolute_path}
{absolute_path}/page_2_figure.png	Figure	2	attention is all you need	[Text Block 3]	{file_absolute_path}
{absolute_path}/page_2_figure_caption.png	Figure caption	2	attention is all you need	[Text Block 4]	{file_absolute_path}
{absolute_path}/page_3_table.png	Table	3	attention is all you need	[Text Block 5]	{file_absolute_path}
{absolute_path}/page_3_table_caption.png	Table caption	3	attention is all you need	[Text Block 6]	{file_absolute_path}
{absolute_path}/page_1_header.png	Header	1	attention is all you need	[Text Block 7]	{file_absolute_path}
{absolute_path}/page_2_footer.png	Footer	2	attention is all you need	[Text Block 8]	{file_absolute_path}
{absolute_path}/page_3_reference.png	Reference	3	attention is all you need	[Text Block 9]	{file_absolute_path}
{absolute_path}/page_1_equation.png	Equation	1	attention is all you need	[Text Block 10]	{file_absolute_path}

See more in llm_parser.py main function.

Installation

pip install llmdocparser

Usage

from llmdocparser.llm_parser import get_image_content

content = get_image_content(
    llm_type="azure",
    pdf_path="path/to/your/pdf",
    output_dir="path/to/output/directory",
    max_concurrency=5,
    azure_deployment="azure-gpt-4o",
    azure_endpoint="your_azure_endpoint",
    api_key="your_api_key",
    api_version="your_api_version"
)
print(content)

Parameters

• llm_type: str

  The options are azure, openai, dashscope.

• pdf_path: str

  Path to the PDF file.

• output_dir: str

  Output directory to store all parsed images.

• max_concurrency: int

  Number of GPT parsing worker threads. Batch calling details: Batch Support

If using Azure, the azure_deployment and azure_endpoint parameters need to be passed; otherwise, only the API key needs to be provided.

• base_url: str

  OpenAI Compatible Server url. Detail: OpenAI-Compatible Server

Cost

Using the 'Attention Is All You Need' paper for analysis, the model chosen is GPT-4o, costing as follows:

Total Tokens: 44063
Prompt Tokens: 33812
Completion Tokens: 10251
Total Cost (USD): $0.322825

Average cost per page: $0.0215

Star History