Merge pull request #92 from funstory-ai/develop

feat(progress_monitor): add configurable report interval and threading support

README.md

@@ -148,6 +148,7 @@ uv run babeldoc --bing --files example.pdf --files example2.pdf
 
 - `--output`, `-o`: Output directory for translated files. If not set, use current working directory.
 - `--debug`, `-d`: Enable debug logging level and export detailed intermediate results in `~/.cache/yadt/working`.
+- `--report-interval`: Progress report interval in seconds (default: 0.1).
 
 ### Configuration File
 

babeldoc/__init__.py

@@ -1 +1 @@
-__version__ = "0.1.6"
+__version__ = "0.1.7"

babeldoc/document_il/backend/pdf_creater.py

@@ -14,7 +14,7 @@
 
 
 class PDFCreater:
-    stage_name = "创建PDF文件"
+    stage_name = "Create PDF file"
 
     def __init__(
         self,

babeldoc/document_il/frontend/il_creater.py

@@ -17,7 +17,7 @@
 
 
 class ILCreater:
-    stage_name = "解析PDF并创建中间表示"
+    stage_name = "Parse PDF and Create Intermediate Representation"
 
     def __init__(self, translation_config: TranslationConfig):
         self.progress = None

babeldoc/document_il/midend/add_debug_information.py

@@ -11,7 +11,7 @@
 
 
 class AddDebugInformation:
-    stage_name = "添加DEBUG信息"
+    stage_name = "Add Debug Information"
 
     def __init__(self, translation_config: TranslationConfig):
         self.translation_config = translation_config

babeldoc/document_il/midend/il_translator.py

@@ -112,7 +112,7 @@ def set_output(self, output: str):
 
 
 class ILTranslator:
-    stage_name = "翻译段落"
+    stage_name = "Translate Paragraphs"
 
     def __init__(
         self,

babeldoc/document_il/midend/layout_parser.py

@@ -13,7 +13,7 @@
 
 
 class LayoutParser:
-    stage_name = "解析页面布局"
+    stage_name = "Parse Page Layout"
 
     def __init__(self, translation_config: TranslationConfig):
         self.translation_config = translation_config

babeldoc/document_il/midend/paragraph_finder.py

@@ -26,7 +26,7 @@ def generate_base58_id(length: int = 5) -> str:
 
 
 class ParagraphFinder:
-    stage_name = "解析段落"
+    stage_name = "Parse Paragraphs"
 
     def __init__(self, translation_config: TranslationConfig):
         self.translation_config = translation_config

babeldoc/document_il/midend/remove_descent.py

@@ -9,7 +9,7 @@
 
 
 class RemoveDescent:
-    stage_name = "移除字体下沉"
+    stage_name = "Remove Char Descent"
 
     def __init__(self, translation_config: TranslationConfig):
         self.translation_config = translation_config

babeldoc/document_il/midend/styles_and_formulas.py

@@ -22,7 +22,7 @@
 
 
 class StylesAndFormulas:
-    stage_name = "解析公式与样式"
+    stage_name = "Parse Formulas and Styles"
 
     def __init__(self, translation_config: TranslationConfig):
         self.translation_config = translation_config

babeldoc/document_il/midend/typesetting.py

@@ -355,7 +355,7 @@ def render(self) -> [PdfCharacter]:
 
 
 class Typesetting:
-    stage_name = "排版"
+    stage_name = "Typesetting"
 
     def __init__(self, translation_config: TranslationConfig):
         self.font_mapper = FontMapper(translation_config)

babeldoc/document_il/utils/fontmap.py

@@ -13,7 +13,7 @@
 
 
 class FontMapper:
-    stage_name = "添加字体"
+    stage_name = "Add Fonts"
 
     def __init__(self, translation_config: TranslationConfig):
         self.font_names = [

babeldoc/high_level.py

@@ -287,6 +287,7 @@ async def async_translate(translation_config: TranslationConfig):
         finish_event=finish_event,
         cancel_event=cancel_event,
         loop=loop,
+        report_interval=translation_config.report_interval,
     ) as pm:
         future = loop.run_in_executor(None, do_translate, pm, translation_config)
         try:

babeldoc/main.py

@@ -20,7 +20,7 @@
 from babeldoc.translation_config import TranslationConfig
 
 logger = logging.getLogger(__name__)
-__version__ = "0.1.6"
+__version__ = "0.1.7"
 
 
 def create_parser():
@@ -170,6 +170,12 @@ def create_parser():
         action="store_true",
         help="Enable all compatibility enhancement options (equivalent to --skip-clean --dual-translate-first --disable-rich-text-translate)",
     )
+    translation_params.add_argument(
+        "--report-interval",
+        type=float,
+        default=0.1,
+        help="Progress report interval in seconds (default: 0.1)",
+    )
     service_params = translation_params.add_mutually_exclusive_group()
     service_params.add_argument(
         "--openai",
@@ -414,6 +420,7 @@ async def main():
             dual_translate_first=args.dual_translate_first,
             disable_rich_text_translate=args.disable_rich_text_translate,
             enhance_compatibility=args.enhance_compatibility,
+            report_interval=args.report_interval,
         )
 
         # Create progress handler

babeldoc/progress_monitor.py

@@ -24,13 +24,26 @@ def __init__(
         self.progress_change_callback = progress_change_callback
         self.finish_callback = finish_callback
         self.report_interval = report_interval
+        logger.debug(f"report_interval: {self.report_interval}")
         self.last_report_time = 0
         self.finish_stage_count = 0
         self.finish_event = finish_event
         self.cancel_event = cancel_event
         self.loop = loop
         if finish_event and not loop:
             raise ValueError("finish_event requires a loop")
+        if self.progress_change_callback:
+            self.progress_change_callback(
+                type="stage_summary",
+                stages=[
+                    {
+                        "name": name,
+                        "percent": 1.0 / len(stages),
+                    }
+                    for name in stages
+                ],
+            )
+        self.lock = threading.Lock()
 
     def stage_start(self, stage_name: str, total: int):
         stage = self.stage[stage_name]
@@ -49,6 +62,7 @@ def stage_start(self, stage_name: str, total: int):
                 stage_current=0,
                 stage_total=total,
             )
+        self.last_report_time = 0.0
         return stage
 
     def __enter__(self):
@@ -66,7 +80,7 @@ def on_finish(self):
             self.finish_callback(type="error", error=CancelledError)
 
     def stage_done(self, stage):
-        self.last_report_time = 0
+        self.last_report_time = 0.0
         self.finish_stage_count += 1
         if (
             stage.current != stage.total
@@ -94,19 +108,20 @@ def calculate_current_progress(self, stage=None):
         return progress
 
     def stage_update(self, stage, n: int):
-        if (
-            self.progress_change_callback
-            and time.time() - self.last_report_time > self.report_interval
-        ):
-            self.progress_change_callback(
-                type="progress_update",
-                stage=stage.display_name,
-                stage_progress=stage.current * 100 / stage.total,
-                stage_current=stage.current,
-                stage_total=stage.total,
-                overall_progress=self.calculate_current_progress(stage),
-            )
-            self.last_report_time = time.time()
+        with self.lock:
+            report_time_delta = time.time() - self.last_report_time
+            if report_time_delta < self.report_interval:
+                return
+            if self.progress_change_callback:
+                self.progress_change_callback(
+                    type="progress_update",
+                    stage=stage.display_name,
+                    stage_progress=stage.current * 100 / stage.total,
+                    stage_current=stage.current,
+                    stage_total=stage.total,
+                    overall_progress=self.calculate_current_progress(stage),
+                )
+                self.last_report_time = time.time()
 
     def translate_done(self, translate_result):
         if self.finish_callback:

babeldoc/translation_config.py

@@ -35,6 +35,7 @@ def __init__(
         dual_translate_first: bool = False,
         disable_rich_text_translate: bool = False,  # 是否禁用富文本翻译
         enhance_compatibility: bool = False,  # 增强兼容性模式
+        report_interval: float = 0.1,  # Progress report interval in seconds
     ):
         self.input_file = input_file
         self.translator = translator
@@ -58,6 +59,7 @@ def __init__(
         self.disable_rich_text_translate = (
             disable_rich_text_translate or enhance_compatibility
         )
+        self.report_interval = report_interval
         if progress_monitor:
             if progress_monitor.cancel_event is None:
                 progress_monitor.cancel_event = threading.Event()

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "BabelDOC"
-version = "0.1.6"
+version = "0.1.7"
 description = "Yet Another Document Translator"
 license = "AGPL-3.0"
 readme = "README.md"
@@ -117,7 +117,7 @@ dev = [
 ]
 
 [bumpver]
-current_version = "0.1.6"
+current_version = "0.1.7"
 version_pattern = "MAJOR.MINOR.PATCH[.PYTAGNUM]"
 
 [bumpver.file_patterns]