BabelDOC 0.3.43__tar.gz → 0.3.50__tar.gz

babeldoc-0.3.50/.github/PULL_REQUEST_TEMPLATE.md

@@ -0,0 +1,65 @@
+### PR Title
+
+<!-- Please fill in a concise and clear PR title below -->
+[PR] <Your concise title here>
+
+### Related Issue(s)
+
+<!-- If this PR closes or is related to one or more issues, please list them here (e.g., #37) -->
+<!-- e.g.: Closes #37, Relates to #42 -->
+
+### Motivation and Context
+
+<!-- Why is this change required? What problem does it solve? -->
+<!-- If it fixes an open issue, please link to the issue here. -->
+
+### Summary of Changes
+
+<!-- What does this PR introduce or fix? Please describe concisely. -->
+
+### PR Type
+
+<!-- What kind of change is this? Please select one or more -->
+- [ ] ✨ Enhancement
+- [ ] 🐛 Bug Fix
+- [ ] 📚 Documentation
+- [ ] 🏗️ Refactor
+- [ ] 🧪 Test
+- [ ] 🧹 Chore
+
+### Breaking Changes
+
+<!-- Does this PR introduce any breaking changes? If so, please describe them. -->
+<!-- - [ ] Yes, this PR introduces breaking changes.
+<!-- - [ ] No, this PR does not introduce breaking changes. -->
+<!-- Detailed description of breaking changes (if any): -->
+
+### Contributor Checklist
+
+- [ ] I have fully read and understood the **[CONTRIBUTING.md](https://funstory-ai.github.io/BabelDOC/CONTRIBUTING/)** guide.
+- [ ] I have performed a self-review of my own code.
+- [ ] My changes follow the project's code style and guidelines
+- [ ] I have linked the related issue(s) in the description above (if applicable)
+- [ ] I have updated relevant documentation (if applicable)
+- [ ] I have added necessary tests that prove my fix is effective or that my feature works (if applicable)
+- [ ] All new and existing tests passed locally with my changes
+- [ ] My changes generate no new warnings or errors
+- [ ] I understand that due to limited maintainer resources, only small PRs are accepted. Suggestions with proof-of-concept patches are appreciated, and my patch may be rewritten if necessary.
+
+### Testing Instructions
+
+<!-- Please provide clear and concise step-by-step instructions on how to test your changes. -->
+<!-- e.g.: -->
+<!-- 1. Check out this branch. -->
+<!-- 2. Run `...` to install dependencies. -->
+<!-- 3. Run `...` to start the application/run the script. -->
+<!-- 4. Navigate to `...` or observe `...` -->
+<!-- 5. Verify that `...` (expected outcome). -->
+
+### Screenshots (if applicable)
+
+<!-- If your changes include UI modifications, please add screenshots or GIFs to show the before and after. -->
+
+### Additional Notes
+
+<!-- Is there anything else the reviewer should know? For example, any dependencies, or potential impacts. --> 

{babeldoc-0.3.43 → babeldoc-0.3.50}/.github/workflows/test.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.10", "3.11", "3.12"]
+        python-version: ["3.10", "3.11", "3.12", "3.13"]
 
     steps:
       - uses: actions/checkout@v4

{babeldoc-0.3.43 → babeldoc-0.3.50}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: BabelDOC
-Version: 0.3.43
+Version: 0.3.50
 Summary: Yet Another Document Translator
 Project-URL: Homepage, https://github.com/funstory-ai/BabelDOC
 Project-URL: Issues, https://github.com/funstory-ai/BabelDOC/issues
@@ -11,7 +11,7 @@ License-File: LICENSE
 Keywords: PDF
 Classifier: Operating System :: OS Independent
 Classifier: Programming Language :: Python :: 3
-Requires-Python: <3.13,>=3.10
+Requires-Python: <3.14,>=3.10
 Requires-Dist: bitstring>=4.3.0
 Requires-Dist: configargparse>=1.7
 Requires-Dist: freetype-py>=2.5.1
@@ -19,7 +19,7 @@ Requires-Dist: httpx[socks]>=0.27.0
 Requires-Dist: huggingface-hub>=0.27.0
 Requires-Dist: msgpack>=1.1.0
 Requires-Dist: numpy>=2.0.2
-Requires-Dist: onnx>=1.17.0
+Requires-Dist: onnx>=1.18.0
 Requires-Dist: onnxruntime>=1.16.1
 Requires-Dist: openai>=1.59.3
 Requires-Dist: opencv-python-headless>=4.10.0.84
@@ -205,6 +205,7 @@ uv run babeldoc --files example.pdf --files example2.pdf --openai --openai-model
 - `--translate-table-text`: Translate table text (experimental, default: False)
 - `--skip-scanned-detection`: Skip scanned document detection (default: False). When using split translation, only the first part performs detection if not skipped.
 - `--ocr-workaround`: Use OCR workaround (default: False). When enabled, the tool will use OCR to detect text and fill background for scanned PDF.
+- `--working-dir`: Working directory for translation. If not set, use temp directory.
 
 > [!TIP]
 > - Both `--skip-clean` and `--dual-translate-first` may help improve compatibility with some PDF readers

{babeldoc-0.3.43 → babeldoc-0.3.50}/README.md

@@ -156,6 +156,7 @@ uv run babeldoc --files example.pdf --files example2.pdf --openai --openai-model
 - `--translate-table-text`: Translate table text (experimental, default: False)
 - `--skip-scanned-detection`: Skip scanned document detection (default: False). When using split translation, only the first part performs detection if not skipped.
 - `--ocr-workaround`: Use OCR workaround (default: False). When enabled, the tool will use OCR to detect text and fill background for scanned PDF.
+- `--working-dir`: Working directory for translation. If not set, use temp directory.
 
 > [!TIP]
 > - Both `--skip-clean` and `--dual-translate-first` may help improve compatibility with some PDF readers

babeldoc-0.3.50/babeldoc/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.3.50"

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/const.py

@@ -3,7 +3,7 @@ import shutil
 import subprocess
 from pathlib import Path
 
-__version__ = "0.3.43"
+__version__ = "0.3.50"
 
 CACHE_FOLDER = Path.home() / ".cache" / "babeldoc"
 

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/document_il/backend/pdf_creater.py

@@ -386,20 +386,22 @@ class PDFCreater:
             # Get pages from both PDFs
             orig_page = original_pdf[page_id]
             trans_page = translated_pdf[page_id]
-
-            # Calculate total width and use max height
+            rotate_angle = orig_page.rotation
             total_width = orig_page.rect.width + trans_page.rect.width
             max_height = max(orig_page.rect.height, trans_page.rect.height)
-
-            # Create new page with combined width
-            dual_page = dual.new_page(width=total_width, height=max_height)
-
-            # Define rectangles for left and right sides
             left_width = (
                 orig_page.rect.width
                 if not translation_config.dual_translate_first
                 else trans_page.rect.width
             )
+
+            orig_page.set_rotation(0)
+            trans_page.set_rotation(0)
+
+            # Create new page with combined width
+            dual_page = dual.new_page(width=total_width, height=max_height)
+
+            # Define rectangles for left and right sides
             rect_left = pymupdf.Rect(0, 0, left_width, max_height)
             rect_right = pymupdf.Rect(left_width, 0, total_width, max_height)
 
@@ -414,6 +416,7 @@ class PDFCreater:
                     original_pdf,
                     page_id,
                     keep_proportion=True,
+                    rotate=-rotate_angle,
                 )
             except Exception as e:
                 logger.warning(
@@ -429,6 +432,7 @@ class PDFCreater:
                     translated_pdf,
                     page_id,
                     keep_proportion=True,
+                    rotate=-rotate_angle,
                 )
             except Exception as e:
                 logger.warning(

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/document_il/midend/il_translator.py

@@ -649,7 +649,7 @@ class ILTranslator:
                     and re.match(f"^{p.left_regex_pattern}", matched_text)
                 )
                 text = re.match(
-                    f"^{placeholder.left_regex_pattern}(.+){placeholder.right_regex_pattern}$",
+                    f"^{placeholder.left_regex_pattern}(.*){placeholder.right_regex_pattern}$",
                     matched_text,
                 ).group(1)
 

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/document_il/midend/il_translator_llm_only.py

@@ -267,37 +267,25 @@ class ILTranslatorLLMOnly:
                     obj["formula_placeholders_hint"] = placeholders_hint
                 json_format_input.append(obj)
 
-            json_format_input = json.dumps(
+            json_format_input_str = json.dumps(
                 json_format_input, ensure_ascii=False, indent=2
             )
+
+            # Start building the new prompt
+            llm_prompt_parts = []
+
+            # 1. #role
+            llm_prompt_parts.append("#role")
             if self.translation_config.custom_system_prompt:
-                llm_input = [self.translation_config.custom_system_prompt]
+                llm_prompt_parts.append(self.translation_config.custom_system_prompt)
             else:
-                llm_input = [
+                llm_prompt_parts.append(
                     f"You are a professional and reliable machine translation engine responsible for translating the input text into {self.translation_config.lang_out}."
-                ]
-
-            llm_hint = []
-            if title_paragraph:
-                llm_hint.append(
-                    f"The first title in the full text: {title_paragraph.unicode}"
-                )
-            if (
-                local_title_paragraph
-                and local_title_paragraph.debug_id != title_paragraph.debug_id
-            ):
-                llm_hint.append(
-                    f"The most similar title in the full text: {local_title_paragraph.unicode}"
                 )
 
-            if llm_hint:
-                llm_input.append(
-                    "When translating, please refer to the following information to improve translation quality:"
-                )
-                for i, line in enumerate(llm_hint):
-                    llm_input.append(f"{i}. {line}")
-            llm_input.append("When translating, please follow the following rules:")
-            # Create a structured prompt template for LLM translation
+            # 2. ##rules
+            llm_prompt_parts.append("\n##rules")
+            # Dynamically get placeholder examples
             rich_text_left_placeholder = (
                 self.translate_engine.get_rich_text_left_placeholder(1)
             )
@@ -308,66 +296,102 @@ class ILTranslatorLLMOnly:
             )
             if isinstance(rich_text_right_placeholder, tuple):
                 rich_text_right_placeholder = rich_text_right_placeholder[0]
-            llm_input.append(
-                f'1. Do not translate style tags, such as "{rich_text_left_placeholder}xxx{rich_text_right_placeholder}"!'
-            )
-
             formula_placeholder = self.translate_engine.get_formular_placeholder(3)
             if isinstance(formula_placeholder, tuple):
                 formula_placeholder = formula_placeholder[0]
-            llm_input.append(
+
+            llm_prompt_parts.append(
+                f'1. Do not translate style tags, such as "{rich_text_left_placeholder}xxx{rich_text_right_placeholder}"'
+            )
+            llm_prompt_parts.append(
                 f'2. Do not translate formula placeholders, such as "{formula_placeholder}". The system will automatically replace the placeholders with the corresponding formulas.'
             )
-            llm_input.append(
+            llm_prompt_parts.append(
                 "3. If there is no need to translate (such as proper nouns, codes, etc.), then return the original text."
             )
-            # Create a structured prompt template for LLM translation
-            prompt_template = (
-                f"""
-    4. You will be given a JSON formatted input containing entries with "id" and "input" fields. 
-    
-    For each entry in the JSON, translate the contents of the "input" field into {self.translation_config.lang_out}.
-    Write the translation back into the "output" field for that entry.
-    
-    """
-                + """
-    Here is an example of the expected format:
-    
-    <example>
-    ```json
-    Input:
-    {
-        "id": 1,
-        "input": "Source",
-        "layout_label": "plain text",
-        // this is optional
-        "formula_placeholders_hint": {
-            "placeholder1": "hint1",
-            "placeholder2": "hint2"
-        }
-    }
-    ```
-    Output:
-    ```json
-    {
-        "id": 1,
-        "output": "Translation"
-    }
-    ```
-    </example>
-    
-    Please return the translated json directly without wrapping ```json``` tag or include any additional information.
-    """
+
+            # 3. ##Input
+            llm_prompt_parts.append("\n##Input")
+            llm_prompt_parts.append(
+                'You will be given a JSON formatted input containing entries with "id" and "input" fields.'
+            )
+
+            # 4. ##Output
+            llm_prompt_parts.append("\n##Output")
+            llm_prompt_parts.append(
+                f'For each entry in the JSON, translate the contents of the "input" field into {self.translation_config.lang_out}.'
             )
-            llm_input.append(prompt_template)
-            llm_input.append(
-                f"""Here is the input:
-    
-    ```json
-    {json_format_input}
-    ```"""
+            llm_prompt_parts.append(
+                'Write the translation back into the "output" field for that entry.'
             )
-            final_input = "\n".join(llm_input).strip()
+            llm_prompt_parts.append(
+                "Please return the translated json directly without wrapping ```json``` tag or include any additional information."
+            )
+
+            # 5. ##example
+            llm_prompt_parts.append("\n##example")
+            llm_prompt_parts.append("Here is an example of the expected format:")
+            llm_prompt_parts.append("")  # Blank line
+            llm_prompt_parts.append("<example>")
+            llm_prompt_parts.append("```json")
+            llm_prompt_parts.append("Input:")
+            llm_prompt_parts.append("{")
+            llm_prompt_parts.append('    "id": 1,')
+            llm_prompt_parts.append('    "input": "Source",')
+            llm_prompt_parts.append('    "layout_label": "plain text",')
+            llm_prompt_parts.append("    // this is optional")
+            llm_prompt_parts.append('    "formula_placeholders_hint": {')
+            llm_prompt_parts.append('        "placeholder1": "hint1",')
+            llm_prompt_parts.append('        "placeholder2": "hint2"')
+            llm_prompt_parts.append("    }")
+            llm_prompt_parts.append("}")
+            llm_prompt_parts.append("```")
+            llm_prompt_parts.append("Output:")
+            llm_prompt_parts.append("```json")
+            llm_prompt_parts.append("{")
+            llm_prompt_parts.append('    "id": 1,')
+            llm_prompt_parts.append('    "output": "Translation"')
+            llm_prompt_parts.append("}")
+            llm_prompt_parts.append("```")
+            llm_prompt_parts.append("</example>")
+
+            # 6. ##Others (contextual hints)
+            other_hints = []
+            hint_idx = 0
+            if title_paragraph:
+                other_hints.append(
+                    f"{hint_idx}. The first title in the full text: {title_paragraph.unicode}"
+                )
+                hint_idx += 1
+
+            if local_title_paragraph:
+                # Check if it's different from the global title_paragraph before adding
+                is_different_from_global = True
+                if title_paragraph:
+                    if local_title_paragraph.debug_id == title_paragraph.debug_id:
+                        is_different_from_global = False
+
+                if is_different_from_global:
+                    other_hints.append(
+                        f"{hint_idx}. The most similar title in the full text: {local_title_paragraph.unicode}"
+                    )
+                    hint_idx += 1
+
+            if other_hints:
+                llm_prompt_parts.append("\n##Others")
+                llm_prompt_parts.append(
+                    "When translating, please refer to the following information to improve translation quality:"
+                )
+                llm_prompt_parts.extend(other_hints)
+
+            llm_prompt_parts.append("\n## Actual Input")
+
+            # Combine all parts for the main prompt
+            main_prompt_content = "\n".join(llm_prompt_parts)
+
+            # Append the actual JSON input string at the end, without markdown fence
+            final_input = main_prompt_content + "\n\n" + json_format_input_str
+
             for llm_translate_tracker in llm_translate_trackers:
                 llm_translate_tracker.set_input(final_input)
             llm_output = self.translate_engine.llm_translate(
@@ -382,6 +406,11 @@ class ILTranslatorLLMOnly:
 
             parsed_output = json.loads(llm_output)
 
+            if isinstance(parsed_output, dict) and parsed_output.get(
+                "output", parsed_output.get("input", False)
+            ):
+                parsed_output = [parsed_output]
+
             translation_results = {
                 item["id"]: item.get("output", item.get("input"))
                 for item in parsed_output

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/document_il/midend/layout_parser.py

@@ -6,6 +6,7 @@ import numpy as np
 from pymupdf import Document
 
 from babeldoc.document_il import il_version_1
+from babeldoc.document_il.utils.mupdf_helper import get_no_rotation_img
 from babeldoc.document_il.utils.style_helper import GREEN
 from babeldoc.translation_config import TranslationConfig
 
@@ -124,7 +125,8 @@ class LayoutParser:
                     # Convert coordinate system from picture to il
                     # system to the il coordinate system
                     x0, y0, x1, y1 = layout.xyxy
-                    pix = mupdf_doc[page.page_number].get_pixmap()
+                    pix = get_no_rotation_img(mupdf_doc[page.page_number])
+                    # pix = mupdf_doc[page.page_number].get_pixmap()
                     h, w = pix.height, pix.width
                     x0, y0, x1, y1 = (
                         np.clip(int(x0 - 1), 0, w - 1),

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/document_il/midend/styles_and_formulas.py

@@ -607,6 +607,9 @@ class StylesAndFormulas:
             r"|.*Mincho.*"
             r"|.*Amerigo.*"
             r"|.*MSGloriolaIIStd.*"
+            r"|.*CMU.+"
+            r"|.*LinLibertine.*"
+            r"|.*txsys.*"
             r")$"
         )
         precise_formula_font_pattern = (
@@ -644,7 +647,7 @@ class StylesAndFormulas:
             r"|.*KpMath.*"
             r"|.*Lete Sans Math.*"
             r"|.*LeteSansMath.*"
-            r"|.*LinLibertineO.*"
+            # r"|.*LinLibertineO.*"
             r"|.*Linux Libertine O.*"
             r"|.*LibertinusMath-Regular.*"
             r"|.*Libertinus Math.*"

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/document_il/midend/table_parser.py

@@ -6,6 +6,7 @@ import numpy as np
 from pymupdf import Document
 
 from babeldoc.document_il import il_version_1
+from babeldoc.document_il.utils.mupdf_helper import get_no_rotation_img
 from babeldoc.document_il.utils.style_helper import GREEN
 from babeldoc.translation_config import TranslationConfig
 
@@ -136,7 +137,8 @@ class TableParser:
                     # Convert coordinate system from picture to il
                     # system to the il coordinate system
                     x0, y0, x1, y1 = layout.xyxy
-                    pix = mupdf_doc[page.page_number].get_pixmap()
+                    # pix = mupdf_doc[page.page_number].get_pixmap()
+                    pix = get_no_rotation_img(mupdf_doc[page.page_number])
                     h, w = pix.height, pix.width
                     x0, y0, x1, y1 = (
                         np.clip(int(x0 - 1), 0, w - 1),

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/document_il/translator/translator.py

@@ -255,6 +255,7 @@ class OpenAITranslator(BaseTranslator):
         response = self.client.chat.completions.create(
             model=self.model,
             **self.options,
+            max_tokens=2048,
             messages=[
                 {
                     "role": "user",

babeldoc-0.3.50/babeldoc/document_il/utils/mupdf_helper.py

@@ -0,0 +1,10 @@
+import pymupdf
+
+
+def get_no_rotation_img(page: pymupdf.Page):
+    # return page.get_pixmap(dpi=72)
+    original_rotation = page.rotation
+    page.set_rotation(0)
+    pix = page.get_pixmap(dpi=72)
+    page.set_rotation(original_rotation)
+    return pix

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/docvision/doclayout.py

@@ -9,6 +9,8 @@ from collections.abc import Generator
 import cv2
 import numpy as np
 
+from babeldoc.document_il.utils.mupdf_helper import get_no_rotation_img
+
 try:
     import onnx
     import onnxruntime
@@ -282,7 +284,8 @@ class OnnxModel(DocLayoutModel):
         for page in pages:
             translate_config.raise_if_cancelled()
             with self.lock:
-                pix = mupdf_doc[page.page_number].get_pixmap(dpi=72)
+                # pix = mupdf_doc[page.page_number].get_pixmap(dpi=72)
+                pix = get_no_rotation_img(mupdf_doc[page.page_number])
             image = np.fromstring(pix.samples, np.uint8).reshape(
                 pix.height,
                 pix.width,

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/docvision/rpc_doclayout.py

@@ -14,6 +14,7 @@ from tenacity import stop_after_attempt
 from tenacity import wait_exponential
 
 import babeldoc
+from babeldoc.document_il.utils.mupdf_helper import get_no_rotation_img
 from babeldoc.docvision.doclayout import DocLayoutModel
 from babeldoc.docvision.doclayout import YoloBox
 from babeldoc.docvision.doclayout import YoloResult
@@ -256,7 +257,8 @@ class RpcDocLayoutModel(DocLayoutModel):
     ):
         translate_config.raise_if_cancelled()
         with self.lock:
-            pix = mupdf_doc[page.page_number].get_pixmap(dpi=72)
+            # pix = mupdf_doc[page.page_number].get_pixmap(dpi=72)
+            pix = get_no_rotation_img(mupdf_doc[page.page_number])
         image = np.fromstring(pix.samples, np.uint8).reshape(
             pix.height,
             pix.width,

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/docvision/table_detection/rapidocr.py

@@ -6,6 +6,7 @@ from collections.abc import Generator
 import cv2
 import numpy as np
 from babeldoc.assets.assets import get_table_detection_rapidocr_model_path
+from babeldoc.document_il.utils.mupdf_helper import get_no_rotation_img
 from babeldoc.docvision.doclayout import YoloBox
 from babeldoc.docvision.doclayout import YoloResult
 from rapidocr_onnxruntime import RapidOCR
@@ -236,7 +237,8 @@ class RapidOCRModel:
         for page in pages:
             translate_config.raise_if_cancelled()
             with self.lock:
-                pix = mupdf_doc[page.page_number].get_pixmap(dpi=72)
+                # pix = mupdf_doc[page.page_number].get_pixmap(dpi=72)
+                pix = get_no_rotation_img(mupdf_doc[page.page_number])
             image = np.fromstring(pix.samples, np.uint8).reshape(
                 pix.height,
                 pix.width,

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/high_level.py

@@ -423,9 +423,10 @@ def fix_filter(doc):
         contents = page.get_contents()
         if len(contents) > 1:
             page_streams = [doc.xref_stream(i) for i in contents]
-            head = contents[0]
-            doc.update_stream(head, b" ".join(page_streams))
-            doc.xref_set_key(page.xref, "Contents", f"{head} 0 R")
+            r = doc.get_new_xref()
+            doc.update_object(r, "<<>>")
+            doc.update_stream(r, b" ".join(page_streams))
+            doc.xref_set_key(page.xref, "Contents", f"{r} 0 R")
 
 
 def do_translate(
@@ -643,8 +644,11 @@ def _do_translate_single(
             "input.decompressed.pdf",
         )
         # Fix null xref in PDF file
-        fix_filter(doc_input)
-        fix_null_xref(doc_input)
+        try:
+            fix_filter(doc_input)
+            fix_null_xref(doc_input)
+        except Exception:
+            logger.exception("auto fix failed, please check the pdf file")
         doc_input.save(output_path, expand=True, pretty=True)
         del doc_input
 
@@ -654,8 +658,11 @@ def _do_translate_single(
     resfont = "china-ss"
 
     # Fix null xref in PDF file
-    fix_filter(doc_pdf2zh)
-    fix_null_xref(doc_pdf2zh)
+    try:
+        fix_filter(doc_pdf2zh)
+        fix_null_xref(doc_pdf2zh)
+    except Exception:
+        logger.exception("auto fix failed, please check the pdf file")
 
     mediabox_data = fix_media_box(doc_pdf2zh)
 

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/main.py

@@ -23,7 +23,7 @@ from babeldoc.translation_config import TranslationConfig
 from babeldoc.translation_config import WatermarkOutputMode
 
 logger = logging.getLogger(__name__)
-__version__ = "0.3.43"
+__version__ = "0.3.50"
 
 
 def create_parser():
@@ -70,6 +70,11 @@ def create_parser():
         default=None,
         help="Restore offline assets package from the specified file",
     )
+    parser.add_argument(
+        "--working-dir",
+        default=None,
+        help="Working directory for translation. If not set, use temp directory.",
+    )
     # translation option argument group
     translation_group = parser.add_argument_group(
         "Translation",
@@ -320,7 +325,7 @@ async def main():
         if not Path(file).exists():
             logger.error(f"文件不存在：{file}")
             exit(1)
-        if not file.endswith(".pdf"):
+        if not file.lower().endswith(".pdf"):
             logger.error(f"文件不是 PDF 文件：{file}")
             exit(1)
         pending_files.append(file)
@@ -339,6 +344,21 @@ async def main():
     else:
         args.output = None
 
+    if args.working_dir:
+        working_dir = Path(args.working_dir)
+        if not working_dir.exists():
+            logger.info(f"工作目录不存在，创建：{working_dir}")
+            try:
+                working_dir.mkdir(parents=True, exist_ok=True)
+            except OSError:
+                logger.critical(
+                    f"Failed to create working directory at {working_dir}",
+                    exc_info=True,
+                )
+                exit(1)
+    else:
+        working_dir = None
+
     watermark_output_mode = WatermarkOutputMode.Watermarked
     if args.no_watermark:
         watermark_output_mode = WatermarkOutputMode.NoWatermark
@@ -390,6 +410,7 @@ async def main():
             skip_scanned_detection=args.skip_scanned_detection,
             ocr_workaround=args.ocr_workaround,
             custom_system_prompt=args.custom_system_prompt,
+            working_dir=working_dir,
         )
 
         # Create progress handler

{babeldoc-0.3.43 → babeldoc-0.3.50}/babeldoc/pdfinterp.py

@@ -377,6 +377,11 @@ class PDFPageInterpreterEx(PDFPageInterpreter):
             ctm = (0, 1, -1, 0, y1, -x0)
         else:
             ctm = (1, 0, 0, 1, -x0, -y0)
+        # ctm_for_ops = copy.copy(ctm)
+        ctm_for_ops = (1, 0, 0, 1, -x0, -y0)
+        ctm = (1, 0, 0, 1, -x0, -y0)
+        if page.rotate == 90 or page.rotate == 270:
+            (x0, y0, x1, y1) = (y0, x1, y1, x0)
         self.il_creater.on_page_start()
         self.il_creater.on_page_crop_box(x0, y0, x1, y1)
         self.device.begin_page(page, ctm)
@@ -391,7 +396,8 @@ class PDFPageInterpreterEx(PDFPageInterpreter):
         # )
         # for obj in page.contents:
         #     self.obj_patch[obj.objid] = ""
-        return f"q {ops_base} Q 1 0 0 1 {x0} {y0} cm"
+        return f"q {ops_base} Q {' '.join(f'{x:f}' for x in ctm_for_ops)} cm"
+        # return f"q {ops_base} Q 1 0 0 1 {x0} {y0} cm"
 
     def render_contents(
         self,