sudacode/owocr
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Avoid reordering horizontal text if it overlaps horizontally

Browse Source
This commit is contained in:
AuroraWright
2025-10-31 23:17:33 +01:00
parent 2f37909167
commit 3d28e2600c
1 changed files with 20 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
owocr/run.py
View File

@@ -792,10 +792,10 @@ class TextFiltering:
return result_data return result_data
paragraphs_with_lines = [p for p in result_data.paragraphs if p.lines] paragraphs_with_lines = [p for p in result_data.paragraphs if p.lines]
ordered_paragraphs = self._order_paragraphs_by_orientation_and_overlap(paragraphs_with_lines) ordered_paragraphs = self._order_paragraphs(paragraphs_with_lines)
for paragraph in ordered_paragraphs: for paragraph in ordered_paragraphs:
paragraph.lines = self._order_lines_by_paragraph_orientation( paragraph.lines = self._order_lines(
paragraph.lines, paragraph.lines,
self._is_paragraph_vertical(paragraph) self._is_paragraph_vertical(paragraph)
) )
@@ -805,7 +805,7 @@ class TextFiltering:
paragraphs=ordered_paragraphs paragraphs=ordered_paragraphs
) )
def _order_lines_by_paragraph_orientation(self, lines, is_paragraph_vertical): def _order_lines(self, lines, is_paragraph_vertical):
if len(lines) <= 1: if len(lines) <= 1:
return lines return lines
@@ -833,8 +833,14 @@ class TextFiltering:
if line_i.bounding_box.center_x < line_j.bounding_box.center_x: if line_i.bounding_box.center_x < line_j.bounding_box.center_x:
should_swap = True should_swap = True
else: else:
# For horizontal paragraphs: order left to right (center_x ascending) # For horizontal paragraphs: check horizontal overlap first
if line_i.bounding_box.center_x > line_j.bounding_box.center_x: horizontal_overlap = self._check_horizontal_overlap(
line_i.bounding_box,
line_j.bounding_box
)
# Only swap if there's NO horizontal overlap
if horizontal_overlap == 0 and line_i.bounding_box.center_x > line_j.bounding_box.center_x:
should_swap = True should_swap = True
if should_swap: if should_swap:
@@ -842,7 +848,7 @@ class TextFiltering:
return ordered_lines return ordered_lines
def _order_paragraphs_by_orientation_and_overlap(self, paragraphs): def _order_paragraphs(self, paragraphs):
if len(paragraphs) <= 1: if len(paragraphs) <= 1:
return paragraphs return paragraphs
@@ -881,8 +887,14 @@ class TextFiltering:
if para_i.bounding_box.center_x > para_j.bounding_box.center_x: if para_i.bounding_box.center_x > para_j.bounding_box.center_x:
should_swap = True should_swap = True
else: else:
# Both horizontal: order left to right (center_x ascending) # Both horizontal: check horizontal overlap first
if para_i.bounding_box.center_x > para_j.bounding_box.center_x: horizontal_overlap = self._check_horizontal_overlap(
para_i.bounding_box,
para_j.bounding_box
)
# Only swap if there's NO horizontal overlap
if horizontal_overlap == 0 and para_i.bounding_box.center_x > para_j.bounding_box.center_x:
should_swap = True should_swap = True
if should_swap: if should_swap: