fix: pass max_num_pages and max_file_size to Docling convert (C3)

Defense-in-depth: even if upload validation passes, Docling itself
now enforces page count and file size limits. Configurable via
MAX_PAGE_COUNT and MAX_FILE_SIZE env vars (0 = unlimited).

Ref #57 (C3)

document-parser/infra/local_converter.py

@@ -222,7 +222,12 @@ def _convert_sync(file_path: str, options: ConversionOptions) -> ConversionResul
         )
     try:
         conv = _select_converter(options)
-        result = conv.convert(file_path)
+        kwargs: dict = {}
+        if settings.max_page_count > 0:
+            kwargs["max_num_pages"] = settings.max_page_count
+        if settings.max_file_size > 0:
+            kwargs["max_file_size"] = settings.max_file_size
+        result = conv.convert(file_path, **kwargs)
     finally:
         _converter_lock.release()
 

document-parser/infra/settings.py

@@ -16,7 +16,8 @@ class Settings:
     conversion_timeout: int = 900
     document_timeout: float = 120.0  # Docling-level per-document timeout (seconds)
     max_concurrent_analyses: int = 3
-    max_page_count: int = 0  # 0 = unlimited
+    max_page_count: int = 0  # 0 = unlimited (upload validation)
+    max_file_size: int = 0  # 0 = unlimited (Docling-level, bytes)
     upload_dir: str = "./uploads"
     db_path: str = "./data/docling_studio.db"
     cors_origins: list[str] = field(
@@ -37,6 +38,7 @@ class Settings:
             document_timeout=float(os.environ.get("DOCUMENT_TIMEOUT", "120.0")),
             max_concurrent_analyses=int(os.environ.get("MAX_CONCURRENT_ANALYSES", "3")),
             max_page_count=int(os.environ.get("MAX_PAGE_COUNT", "0")),
+            max_file_size=int(os.environ.get("MAX_FILE_SIZE", "0")),
             upload_dir=os.environ.get("UPLOAD_DIR", "./uploads"),
             db_path=os.environ.get("DB_PATH", "./data/docling_studio.db"),
             cors_origins=[o.strip() for o in cors_raw.split(",")],