feat(translation_config): add doc_layout_model to TranslationConfig

- introduce doc_layout_model parameter in TranslationConfig
- load DocLayoutModel if not provided
- update il_creater to use translation_config.doc_layout_model

refactor(il_creater): improve code style and consistency

- fix PSLiteral check indentation
- add empty line for better readability

yadt/document_il/frontend/il_creater.py

@@ -20,7 +20,7 @@ def __init__(self, translation_config: TranslationConfig):
         self.progress = None
         self.current_page: il_version_1.Page = None
         self.mupdf: pymupdf.Document = None
-        self.model = DocLayoutModel.load_available()
+        self.model = translation_config.doc_layout_model
         self.docs = il_version_1.Document(page=[])
         self.stroking_color_space_name = None
         self.non_stroking_color_space_name = None
@@ -42,8 +42,8 @@ def on_passthrough_per_char(self, operator: str, args: list[str]):
         pass
 
     def parse_arg(self, arg: str):
-        if isinstance(arg,PSLiteral):
-            return f'/{arg.name}'
+        if isinstance(arg, PSLiteral):
+            return f"/{arg.name}"
         if not isinstance(arg, str):
             return str(arg)
         return arg
@@ -83,8 +83,10 @@ def on_page_start(self):
         )
         self.current_page_font_name_id_map = {}
         self.docs.page.append(self.current_page)
+
     def on_page_end(self):
         self.progress.advance(1)
+
     def on_page_crop_box(
         self,
         x0: float | int,

yadt/translation_config.py

@@ -5,6 +5,7 @@
 )
 import os
 
+from yadt.doclayout import DocLayoutModel
 from yadt.progress_monitor import ProgressMonitor
 
 
@@ -29,6 +30,7 @@ def __init__(
         short_line_split_factor: float = 0.8,  # 切分阈值系数。实际阈值为当前页所有行长度中位数*此系数
         use_rich_pbar: bool = True,  # 是否使用 rich 进度条
         progress_monitor: Optional[ProgressMonitor] = None,  # progress_monitor
+        doc_layout_model=None,
     ):
         self.input_file = input_file
         self.translator = translator
@@ -62,6 +64,10 @@ def __init__(
 
         os.makedirs(output_dir, exist_ok=True)
 
+        if doc_layout_model is None:
+            doc_layout_model = DocLayoutModel.load_available()
+        self.doc_layout_model = doc_layout_model
+
     def get_output_file_path(self, filename):
         return os.path.join(self.output_dir, filename)