Fix SSE route dependency and align architecture docs
This commit is contained in:
ash66
@@ -1,4 +1,4 @@
|
||||
"""PDF文档解析 - 使用PyMuPDF基础解析"""
|
||||
"""Provide service-layer logic for pdf parser."""
|
||||
|
||||
import fitz # PyMuPDF
|
||||
from typing import List, Dict, Optional, Tuple
|
||||
@@ -9,17 +9,17 @@ import re
|
||||
|
||||
@dataclass
|
||||
class PDFPageContent:
|
||||
"""PDF页面内容"""
|
||||
"""Represent the P D F Page Content type."""
|
||||
page_number: int
|
||||
text: str
|
||||
tables: List[str] = field(default_factory=list)
|
||||
images: List[str] = field(default_factory=list) # 图片路径列表
|
||||
images: List[str] = field(default_factory=list) # Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
blocks: List[Dict] = field(default_factory=list)
|
||||
|
||||
|
||||
@dataclass
|
||||
class PDFDocumentContent:
|
||||
"""PDF文档完整内容"""
|
||||
"""Represent the P D F Document Content type."""
|
||||
file_path: str
|
||||
total_pages: int
|
||||
pages: List[PDFPageContent]
|
||||
@@ -28,23 +28,14 @@ class PDFDocumentContent:
|
||||
|
||||
|
||||
class PDFParser:
|
||||
"""PDF文档解析器 - 基于PyMuPDF"""
|
||||
"""Provide the P D F Parser parser."""
|
||||
|
||||
def __init__(self):
|
||||
"""Initialize the P D F Parser instance."""
|
||||
self.pdf = None
|
||||
|
||||
def parse(self, file_path: str, extract_tables: bool = True, extract_images: bool = False) -> PDFDocumentContent:
|
||||
"""
|
||||
解析PDF文档
|
||||
|
||||
Args:
|
||||
file_path: PDF文件路径
|
||||
extract_tables: 是否提取表格
|
||||
extract_images: 是否提取图片
|
||||
|
||||
Returns:
|
||||
PDFDocumentContent: 解析后的文档内容
|
||||
"""
|
||||
"""Handle parse for the P D F Parser instance."""
|
||||
logger.info(f"开始解析PDF文档: {file_path}")
|
||||
|
||||
try:
|
||||
@@ -55,16 +46,16 @@ class PDFParser:
|
||||
pages=[]
|
||||
)
|
||||
|
||||
# 提取文档元数据
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
doc_content.metadata = self._extract_metadata()
|
||||
|
||||
# 逐页解析
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
for page_num in range(self.pdf.page_count):
|
||||
page = self.pdf[page_num]
|
||||
page_content = self._parse_page(page, page_num + 1, extract_tables, extract_images)
|
||||
doc_content.pages.append(page_content)
|
||||
|
||||
# 生成Markdown格式文本
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
doc_content.markdown_text = self._generate_markdown(doc_content)
|
||||
|
||||
self.pdf.close()
|
||||
@@ -77,7 +68,7 @@ class PDFParser:
|
||||
raise
|
||||
|
||||
def _extract_metadata(self) -> Dict[str, str]:
|
||||
"""提取PDF元数据"""
|
||||
"""Handle extract metadata for this module for the P D F Parser instance."""
|
||||
metadata = {}
|
||||
try:
|
||||
meta = self.pdf.metadata
|
||||
@@ -97,23 +88,23 @@ class PDFParser:
|
||||
|
||||
def _parse_page(self, page: fitz.Page, page_num: int,
|
||||
extract_tables: bool, extract_images: bool) -> PDFPageContent:
|
||||
"""解析单页内容"""
|
||||
"""Handle parse page for this module for the P D F Parser instance."""
|
||||
page_content = PDFPageContent(page_number=page_num, text="")
|
||||
|
||||
# 提取文本块（保留结构）
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
blocks = page.get_text("dict", flags=fitz.TEXT_PRESERVE_WHITESPACE)["blocks"]
|
||||
page_content.blocks = blocks
|
||||
|
||||
# 提取纯文本
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
text = page.get_text("text", flags=fitz.TEXT_PRESERVE_WHITESPACE)
|
||||
page_content.text = text.strip()
|
||||
|
||||
# 提取表格（使用PyMuPDF的表格提取功能）
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
if extract_tables:
|
||||
tables = self._extract_tables_from_page(page)
|
||||
page_content.tables = tables
|
||||
|
||||
# 提取图片
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
if extract_images:
|
||||
images = self._extract_images_from_page(page, page_num)
|
||||
page_content.images = images
|
||||
@@ -121,25 +112,22 @@ class PDFParser:
|
||||
return page_content
|
||||
|
||||
def _extract_tables_from_page(self, page: fitz.Page) -> List[str]:
|
||||
"""
|
||||
从页面提取表格（基于文本块分析）
|
||||
注意：PyMuPDF基础版表格提取能力有限，复杂表格建议使用MinerU
|
||||
"""
|
||||
"""Handle extract tables from page for this module for the P D F Parser instance."""
|
||||
tables = []
|
||||
|
||||
try:
|
||||
# 使用PyMuPDF的表格提取方法（2.4+版本）
|
||||
# 对于更复杂的表格，需要在mineru_parser中使用更高级的方法
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
tabs = page.find_tables()
|
||||
if tabs:
|
||||
for tab in tabs:
|
||||
table_text = tab.extract()
|
||||
# 将表格转换为Markdown格式
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
markdown_table = self._table_to_markdown(table_text)
|
||||
tables.append(markdown_table)
|
||||
|
||||
except AttributeError:
|
||||
# 旧版本PyMuPDF没有表格提取功能
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
logger.warning("PyMuPDF版本不支持表格提取，请升级到2.4+版本")
|
||||
except Exception as e:
|
||||
logger.warning(f"表格提取失败: {e}")
|
||||
@@ -147,28 +135,28 @@ class PDFParser:
|
||||
return tables
|
||||
|
||||
def _table_to_markdown(self, table_data: List[List[str]]) -> str:
|
||||
"""将表格数据转换为Markdown格式"""
|
||||
"""Handle table to markdown for this module for the P D F Parser instance."""
|
||||
if not table_data or len(table_data) < 1:
|
||||
return ""
|
||||
|
||||
lines = []
|
||||
# 表头
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
if len(table_data) >= 1:
|
||||
header = table_data[0]
|
||||
lines.append("| " + " | ".join(str(cell).strip() for cell in header) + " |")
|
||||
lines.append("| " + " | ".join("---" for _ in header) + " |")
|
||||
|
||||
# 数据行
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
for row in table_data[1:]:
|
||||
lines.append("| " + " | ".join(str(cell).strip() for cell in row) + " |")
|
||||
|
||||
return "\n".join(lines)
|
||||
|
||||
def _extract_images_from_page(self, page: fitz.Page, page_num: int) -> List[str]:
|
||||
"""提取页面图片"""
|
||||
"""Handle extract images from page for this module for the P D F Parser instance."""
|
||||
images = []
|
||||
# 图片提取功能（可选实现）
|
||||
# 这里仅记录图片信息，实际图片需要额外保存
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
try:
|
||||
image_list = page.get_images()
|
||||
for img_index, img in enumerate(image_list):
|
||||
@@ -179,52 +167,52 @@ class PDFParser:
|
||||
return images
|
||||
|
||||
def _generate_markdown(self, doc_content: PDFDocumentContent) -> str:
|
||||
"""生成Markdown格式文本"""
|
||||
"""Handle generate markdown for this module for the P D F Parser instance."""
|
||||
lines = []
|
||||
|
||||
# 文档标题
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
title = doc_content.metadata.get("title", "")
|
||||
if title:
|
||||
lines.append(f"# {title}\n")
|
||||
else:
|
||||
lines.append(f"# {doc_content.file_path}\n")
|
||||
|
||||
# 元数据信息
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
lines.append("\n## 文档信息\n")
|
||||
for key, value in doc_content.metadata.items():
|
||||
if value and key in ["author", "subject", "keywords", "creation_date"]:
|
||||
lines.append(f"- **{key}**: {value}")
|
||||
|
||||
# 正文内容
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
lines.append("\n## 正文\n")
|
||||
|
||||
for page in doc_content.pages:
|
||||
# 页码标记
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
lines.append(f"\n---\n**第 {page.page_number} 页**\n")
|
||||
|
||||
# 处理文本内容，识别标题结构
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
text = self._process_page_text(page.text, page.blocks)
|
||||
lines.append(text)
|
||||
|
||||
# 添加表格
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
for table in page.tables:
|
||||
lines.append("\n" + table + "\n")
|
||||
|
||||
return "\n".join(lines)
|
||||
|
||||
def _process_page_text(self, text: str, blocks: List[Dict]) -> str:
|
||||
"""处理页面文本，识别标题结构"""
|
||||
# 基于字体大小识别标题
|
||||
"""Handle process page text for this module for the P D F Parser instance."""
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
processed_text = text
|
||||
|
||||
# 尝试识别标题（基于字号）
|
||||
# 法规文档通常有明确的层级结构：章、节、条
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
processed_text = self._detect_headers(text, blocks)
|
||||
|
||||
return processed_text
|
||||
|
||||
def _detect_headers(self, text: str, blocks: List[Dict]) -> str:
|
||||
"""检测并标记标题（基于字号或内容模式）"""
|
||||
"""Handle detect headers for this module for the P D F Parser instance."""
|
||||
lines = text.split("\n")
|
||||
processed_lines = []
|
||||
|
||||
@@ -233,8 +221,8 @@ class PDFParser:
|
||||
if not line:
|
||||
continue
|
||||
|
||||
# 法规标题模式检测
|
||||
# 第一章、第X章、第X节、第X条等
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
if re.match(r'^第[一二三四五六七八九十百]+章\s', line):
|
||||
processed_lines.append(f"\n## {line}\n")
|
||||
elif re.match(r'^第[一二三四五六七八九十百]+节\s', line):
|
||||
@@ -242,7 +230,7 @@ class PDFParser:
|
||||
elif re.match(r'^第[一二三四五六七八九十百]+条\s', line):
|
||||
processed_lines.append(f"\n#### {line}\n")
|
||||
elif re.match(r'^[一二三四五六七八九十]+\s*[、.]', line):
|
||||
# 条款子项
|
||||
# Keep service responsibilities explicit so downstream behavior stays predictable.
|
||||
processed_lines.append(f"- {line}")
|
||||
else:
|
||||
processed_lines.append(line)
|
||||
@@ -250,18 +238,18 @@ class PDFParser:
|
||||
return "\n".join(processed_lines)
|
||||
|
||||
def parse_to_markdown(self, file_path: str) -> str:
|
||||
"""直接解析并返回Markdown文本"""
|
||||
"""Parse to markdown for the P D F Parser instance."""
|
||||
doc_content = self.parse(file_path)
|
||||
return doc_content.markdown_text
|
||||
|
||||
|
||||
def parse_pdf(file_path: str, **kwargs) -> PDFDocumentContent:
|
||||
"""便捷函数：解析PDF文档"""
|
||||
"""Parse pdf."""
|
||||
parser = PDFParser()
|
||||
return parser.parse(file_path, **kwargs)
|
||||
|
||||
|
||||
def parse_pdf_to_markdown(file_path: str) -> str:
|
||||
"""便捷函数：解析PDF并返回Markdown"""
|
||||
"""Parse pdf to markdown."""
|
||||
parser = PDFParser()
|
||||
return parser.parse_to_markdown(file_path)
|
||||
|
||||
Reference in New Issue
Block a user