BabelDOC 0.3.23__tar.gz → 0.3.24__tar.gz

{babeldoc-0.3.23 → babeldoc-0.3.24}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: BabelDOC
-Version: 0.3.23
+Version: 0.3.24
 Summary: Yet Another Document Translator
 Project-URL: Homepage, https://github.com/funstory-ai/BabelDOC
 Project-URL: Issues, https://github.com/funstory-ai/BabelDOC/issues

babeldoc-0.3.24/babeldoc/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.3.24"

{babeldoc-0.3.23 → babeldoc-0.3.24}/babeldoc/const.py

@@ -3,7 +3,7 @@ import shutil
 import subprocess
 from pathlib import Path
 
-__version__ = "0.3.23"
+__version__ = "0.3.24"
 
 CACHE_FOLDER = Path.home() / ".cache" / "babeldoc"
 

{babeldoc-0.3.23 → babeldoc-0.3.24}/babeldoc/document_il/backend/pdf_creater.py

@@ -777,7 +777,10 @@ class PDFCreater:
 
     def restore_media_box(self, doc: pymupdf.Document, mediabox_data: dict) -> None:
         for pageno, mediabox in mediabox_data.items():
-            doc.xref_set_key(doc[pageno].xref, "MediaBox", mediabox[1])
+            try:
+                doc.xref_set_key(doc[pageno].xref, "MediaBox", mediabox[1])
+            except Exception:
+                logger.exception(f"restore media box failed on page: {pageno}")
 
     def write(
         self, translation_config: TranslationConfig, check_font_exists: bool = False
@@ -945,7 +948,10 @@ class PDFCreater:
                     )
 
                 pbar.advance()
-            self.restore_media_box(pdf, self.mediabox_data)
+            try:
+                self.restore_media_box(pdf, self.mediabox_data)
+            except Exception:
+                logger.exception("restore media box failed")
             with self.translation_config.progress_monitor.stage_start(
                 SAVE_PDF_STAGE_NAME,
                 2,

{babeldoc-0.3.23 → babeldoc-0.3.24}/babeldoc/document_il/midend/paragraph_finder.py

@@ -167,6 +167,8 @@ class ParagraphFinder:
             # image characters are not needed
             page.pdf_character = []
 
+        self.fix_overlapping_paragraphs(page)
+
     def is_isolated_formula(self, char: PdfCharacter):
         return char.char_unicode in (
             "(cid:122)",
@@ -510,3 +512,105 @@ class ParagraphFinder:
                     break
                 j += 1
             i += 1
+
+    @staticmethod
+    def is_bbox_contain_in_vertical(bbox1: Box, bbox2: Box) -> bool:
+        """Check if one bounding box is completely contained within the other."""
+        # Check if bbox1 is contained in bbox2
+        bbox1_in_bbox2 = bbox1.y >= bbox2.y and bbox1.y2 <= bbox2.y2
+        # Check if bbox2 is contained in bbox1
+        bbox2_in_bbox1 = bbox2.y >= bbox1.y and bbox2.y2 <= bbox1.y2
+        return bbox1_in_bbox2 or bbox2_in_bbox1
+
+    def fix_overlapping_paragraphs(self, page: Page):
+        """
+        Adjusts the bounding boxes of paragraphs on a page to resolve vertical overlaps.
+
+        Iteratively checks pairs of paragraphs and adjusts their vertical boundaries
+        (y and y2) if they overlap, aiming to place the boundary at the midpoint
+        of the vertical overlap.
+        """
+        paragraphs = page.pdf_paragraph
+        if not paragraphs or len(paragraphs) < 2:
+            return
+
+        max_iterations = len(paragraphs) * len(paragraphs)  # Safety break
+        iterations = 0
+
+        while iterations < max_iterations:
+            iterations += 1
+            overlap_found_in_pass = False
+
+            for i in range(len(paragraphs)):
+                for j in range(i + 1, len(paragraphs)):
+                    para1 = paragraphs[i]
+                    para2 = paragraphs[j]
+
+                    if para1.box is None or para2.box is None:
+                        continue
+
+                    if para1.xobj_id != para2.xobj_id:
+                        continue
+
+                    # Check for overlap using the existing method
+                    if self.bbox_overlap(para1.box, para2.box):
+                        if self.is_bbox_contain_in_vertical(para1.box, para2.box):
+                            continue
+                        # Calculate vertical overlap details
+                        overlap_y_start = max(para1.box.y, para2.box.y)
+                        overlap_y_end = min(para1.box.y2, para2.box.y2)
+                        overlap_height = overlap_y_end - overlap_y_start
+
+                        # Calculate horizontal overlap details
+                        overlap_x_start = max(para1.box.x, para2.box.x)
+                        overlap_x_end = min(para1.box.x2, para2.box.x2)
+                        overlap_width = overlap_x_end - overlap_x_start
+
+                        # Ensure there's a real 2D overlap, focusing on vertical adjustment
+                        if overlap_height > 1e-6 and overlap_width > 1e-6:
+                            overlap_found_in_pass = True
+
+                            # Determine which paragraph is visually higher
+                            if para1.box.y2 > para2.box.y and para1.box.y < para2.box.y:
+                                lower_para = para1
+                                higher_para = para2
+                            # Handle cases where y values are identical (or very close)
+                            # Prefer the one with smaller y2 as the higher one, or break tie arbitrarily
+                            elif para1.box.y2 < para2.box.y2:
+                                lower_para = para1
+                                higher_para = para2
+                            else:
+                                lower_para = para2
+                                higher_para = para1
+
+                            # Calculate the midpoint of the vertical overlap
+                            mid_y = overlap_y_start + overlap_height / 2
+
+                            # Adjust boxes, ensuring they remain valid (y2 > y)
+                            if mid_y > higher_para.box.y and mid_y < lower_para.box.y2:
+                                higher_para.box.y = mid_y + 1
+                                lower_para.box.y2 = mid_y - 1
+                            else:
+                                # This might happen if one box is fully contained vertically
+                                # within another, or due to floating point issues.
+                                # Log a warning and skip adjustment for this pair in this iteration.
+                                # A more complex strategy might be needed for full containment.
+                                logger.warning(
+                                    "Could not resolve overlap between paragraphs"
+                                    f" {higher_para.debug_id} and {lower_para.debug_id}"
+                                    " using simple midpoint strategy."
+                                    f" Midpoint: {mid_y},"
+                                    f" Higher Box: {higher_para.box},"
+                                    f" Lower Box: {lower_para.box}"
+                                )
+
+            # If no overlaps were found and adjusted in this pass, we're done.
+            if not overlap_found_in_pass:
+                break
+
+        if iterations == max_iterations:
+            logger.warning(
+                f"Maximum iterations ({max_iterations}) reached in"
+                f" fix_overlapping_paragraphs for page {page.page_number}."
+                " Some overlaps might remain."
+            )

{babeldoc-0.3.23 → babeldoc-0.3.24}/babeldoc/document_il/midend/typesetting.py

@@ -645,7 +645,7 @@ class Typesetting:
         current_x = box.x
         current_y = box.y2 - avg_height
         box = copy.deepcopy(box)
-        box.y -= avg_height * (line_spacing - 1)
+        # box.y -= avg_height * (line_spacing - 1.01)
         line_height = 0
         current_line_heights = []  # 存储当前行所有元素的高度
 
@@ -722,7 +722,8 @@ class Typesetting:
                 current_line_heights = []  # 清空当前行高度列表
 
                 # 检查是否超出底部边界
-                if current_y - unit_height < box.y:
+                # if current_y - unit_height < box.y:
+                if current_y < box.y:
                     all_units_fit = False
                     break
 
@@ -750,8 +751,9 @@ class Typesetting:
         # 处理最后一行的行高
         if current_line_heights:
             mode_height = statistics.mode(current_line_heights)
-            max_height = max(current_line_heights)
-            line_height = max(mode_height, max_height)
+            # max_height = max(current_line_heights)
+            # line_height = max(mode_height, max_height)
+            line_height = mode_height
 
         return typeset_units, all_units_fit
 

{babeldoc-0.3.23 → babeldoc-0.3.24}/babeldoc/main.py

@@ -23,7 +23,7 @@ from babeldoc.translation_config import TranslationConfig
 from babeldoc.translation_config import WatermarkOutputMode
 
 logger = logging.getLogger(__name__)
-__version__ = "0.3.23"
+__version__ = "0.3.24"
 
 
 def create_parser():

{babeldoc-0.3.23 → babeldoc-0.3.24}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "BabelDOC"
-version = "0.3.23"
+version = "0.3.24"
 description = "Yet Another Document Translator"
 license = "AGPL-3.0"
 readme = "README.md"
@@ -149,7 +149,7 @@ pythonpath = [".", "src"]
 testpaths = ["tests"]
 
 [bumpver]
-current_version = "0.3.23"
+current_version = "0.3.24"
 version_pattern = "MAJOR.MINOR.PATCH[.PYTAGNUM]"
 
 [bumpver.file_patterns]

{babeldoc-0.3.23 → babeldoc-0.3.24}/uv.lock

@@ -48,7 +48,7 @@ wheels = [
 
 [[package]]
 name = "babeldoc"
-version = "0.3.23"
+version = "0.3.24"
 source = { editable = "." }
 dependencies = [
     { name = "bitstring" },

babeldoc-0.3.23/babeldoc/__init__.py

@@ -1 +0,0 @@
-__version__ = "0.3.23"