mirror of
https://github.com/VectifyAI/PageIndex.git
synced 2026-04-26 00:26:21 +02:00
101 lines
3.6 KiB
Python
101 lines
3.6 KiB
Python
import pymupdf
|
|
from pathlib import Path
|
|
from .protocol import ContentNode, ParsedDocument
|
|
from ..index.utils import count_tokens
|
|
|
|
# Minimum image dimension to keep (skip icons/artifacts)
|
|
_MIN_IMAGE_SIZE = 32
|
|
|
|
|
|
class PdfParser:
|
|
def supported_extensions(self) -> list[str]:
|
|
return [".pdf"]
|
|
|
|
def parse(self, file_path: str, **kwargs) -> ParsedDocument:
|
|
path = Path(file_path)
|
|
model = kwargs.get("model")
|
|
images_dir = kwargs.get("images_dir")
|
|
nodes = []
|
|
|
|
with pymupdf.open(str(path)) as doc:
|
|
for i, page in enumerate(doc):
|
|
page_num = i + 1
|
|
if images_dir:
|
|
content, images = self._extract_page_with_images(
|
|
doc, page, page_num, images_dir)
|
|
else:
|
|
content = page.get_text()
|
|
images = None
|
|
|
|
tokens = count_tokens(content, model=model)
|
|
nodes.append(ContentNode(
|
|
content=content or "",
|
|
tokens=tokens,
|
|
index=page_num,
|
|
images=images if images else None,
|
|
))
|
|
|
|
return ParsedDocument(doc_name=path.stem, nodes=nodes)
|
|
|
|
@staticmethod
|
|
def _extract_page_with_images(doc, page, page_num: int,
|
|
images_dir: str) -> tuple[str, list[dict]]:
|
|
"""Extract text and images from a page, preserving their relative order.
|
|
|
|
Uses get_text("dict") to iterate blocks in reading order.
|
|
Text blocks become text; image blocks are saved to disk and replaced
|
|
with an inline placeholder: 
|
|
"""
|
|
images_path = Path(images_dir)
|
|
images_path.mkdir(parents=True, exist_ok=True)
|
|
# Use path relative to cwd so downstream consumers can access directly
|
|
try:
|
|
rel_images_path = images_path.relative_to(Path.cwd())
|
|
except ValueError:
|
|
rel_images_path = images_path
|
|
|
|
parts: list[str] = []
|
|
images: list[dict] = []
|
|
img_idx = 0
|
|
|
|
for block in page.get_text("dict")["blocks"]:
|
|
if block["type"] == 0: # text block
|
|
lines = []
|
|
for line in block["lines"]:
|
|
spans_text = "".join(span["text"] for span in line["spans"])
|
|
lines.append(spans_text)
|
|
parts.append("\n".join(lines))
|
|
|
|
elif block["type"] == 1: # image block
|
|
width = block.get("width", 0)
|
|
height = block.get("height", 0)
|
|
if width < _MIN_IMAGE_SIZE or height < _MIN_IMAGE_SIZE:
|
|
continue
|
|
|
|
image_bytes = block.get("image")
|
|
ext = block.get("ext", "png")
|
|
if not image_bytes:
|
|
continue
|
|
|
|
try:
|
|
pix = pymupdf.Pixmap(image_bytes)
|
|
if pix.n > 4:
|
|
pix = pymupdf.Pixmap(pymupdf.csRGB, pix)
|
|
filename = f"p{page_num}_img{img_idx}.png"
|
|
save_path = images_path / filename
|
|
pix.save(str(save_path))
|
|
pix = None
|
|
except Exception:
|
|
continue
|
|
|
|
rel_path = str(rel_images_path / filename)
|
|
images.append({
|
|
"path": rel_path,
|
|
"width": width,
|
|
"height": height,
|
|
})
|
|
parts.append(f"")
|
|
img_idx += 1
|
|
|
|
content = "\n".join(parts)
|
|
return content, images
|