zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-07-28 19:58:11 +08:00
Code Issues Packages Projects Releases Wiki Activity

refactor: improve paddle ocr logic (#13467)

Browse Source 
### What problem does this PR solve?

improve paddle ocr logic

### Type of change
- [x] Refactoring
This commit is contained in:
Stephen Hu
2026-03-09 14:16:57 +08:00
committed by GitHub
parent 3ce236c4e3
commit d0465ba909
1 changed files with 16 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
deepdoc/parser/paddleocr_parser.py
View File

@@ -63,7 +63,6 @@ def _remove_images_from_markdown(markdown: str) -> str:
class PaddleOCRVLConfig:
"""Configuration for PaddleOCR-VL algorithm."""
use_doc_orientation_classify: Optional[bool] = False
use_doc_orientation_classify: Optional[bool] = False
use_doc_unwarping: Optional[bool] = False
use_layout_detection: Optional[bool] = None
@@ -533,21 +532,25 @@ class PaddleOCRParser(RAGFlowPdfParser):
return None, None
return
height = 0
total_height = 0
max_width = 0
img_sizes = []
for img in imgs:
height += img.size[1] + GAP
height = int(height)
width = int(np.max([i.size[0] for i in imgs]))
pic = Image.new("RGB", (width, height), (245, 245, 245))
height = 0
for ii, img in enumerate(imgs):
if ii == 0 or ii + 1 == len(imgs):
w, h = img.size
img_sizes.append((w, h))
max_width = max(max_width, w)
total_height += h + GAP
pic = Image.new("RGB", (max_width, int(total_height)), (245, 245, 245))
current_height = 0
imgs_count = len(imgs)
for ii, (img, (w, h)) in enumerate(zip(imgs, img_sizes)):
if ii == 0 or ii + 1 == imgs_count:
img = img.convert("RGBA")
overlay = Image.new("RGBA", img.size, (0, 0, 0, 0))
overlay.putalpha(128)
overlay = Image.new("RGBA", img.size, (0, 0, 0, 128))
img = Image.alpha_composite(img, overlay).convert("RGB")
pic.paste(img, (0, int(height)))
height += img.size[1] + GAP
pic.paste(img, (0, int(current_height)))
current_height += h + GAP
if need_position:
return pic, positions