fix: share one 80-char cap for image and table asset filenames

apps/worker/app/services/common/file_utils.py

@@ -4,6 +4,10 @@
 
 import pandas as pd
 
+# Shared cap for cosmetic asset filename stems (images/tables). Keeps OS
+# basename limits safe while preserving a short debug-friendly context.
+MAX_ASSET_FILE_NAME_CHARS = 80
+
 
 def clean_file(path_, mode="remove", cols=None):
     """

apps/worker/app/services/document_parser/formats/markdown/deferred_summary.py

@@ -29,7 +29,7 @@
 from shared.services.ai.summary.engine import summarize
 from shared.services.ai.summary.model import AssetSummary, BodySummary
 from shared.utils.chunk_refs import build_chunk_ref
-from app.services.common.file_utils import path_handle
+from app.services.common.file_utils import MAX_ASSET_FILE_NAME_CHARS, path_handle
 
 # Each deferred task now carries the engine's typed contract straight through to
 # the apply step (audit §4.5): assets → AssetSummary, text → BodySummary. The row
@@ -280,7 +280,10 @@ def _apply_image_summary_result(
     image_dir = original_task.image_dir
     old_img_name = original_task.image_name
     image_suffix = original_task.image_suffix
-    safe_title = path_handle(str(img_title), mode="clean_single")
+    cleaned_title = path_handle(str(img_title), mode="clean_single")
+    if not isinstance(cleaned_title, str) or not cleaned_title:
+        return
+    safe_title = cleaned_title[:MAX_ASSET_FILE_NAME_CHARS]
     img_num_match = re.match(r"image-(\d+)", str(old_img_name))
     img_num = (
         img_num_match.group(1)
@@ -290,6 +293,8 @@ def _apply_image_summary_result(
         else "0"
     )
     new_img_name = path_handle(f"image-{img_num}-{safe_title}", mode="clean_single")
+    if not isinstance(new_img_name, str) or not new_img_name:
+        return
     old_path = os.path.join(image_dir, f"{old_img_name}{image_suffix}")
     new_path = os.path.join(image_dir, f"{new_img_name}{image_suffix}")
     if old_path == new_path or not os.path.exists(old_path):

apps/worker/app/services/document_parser/formats/markdown/image_asset.py

@@ -19,7 +19,7 @@
 from loguru import logger
 
 from shared.utils.chunk_refs import build_chunk_ref
-from app.services.common.file_utils import path_handle
+from app.services.common.file_utils import MAX_ASSET_FILE_NAME_CHARS, path_handle
 
 
 @dataclass(frozen=True)
@@ -135,7 +135,7 @@ def build_markdown_image_asset(
 def build_markdown_image_name(*, image_count: int, last_context: str) -> str:
     image_name_context = path_handle(last_context.strip(), mode="clean_single")
     if image_name_context:
-        image_name_context = image_name_context[:60]
+        image_name_context = image_name_context[:MAX_ASSET_FILE_NAME_CHARS]
         return f"image-{image_count}-{image_name_context}"
     return f"image-{image_count}"
 

apps/worker/app/services/document_parser/tables/table_text_parser.py

@@ -8,7 +8,7 @@
 import pandas as pd
 from bs4 import BeautifulSoup, Tag
 
-_MAX_TABLE_NAME_CHARS = 80
+from app.services.common.file_utils import MAX_ASSET_FILE_NAME_CHARS
 
 
 def sanitize_table_name_from_header(raw_header_text: str) -> str:
@@ -31,8 +31,8 @@ def sanitize_table_name_from_header(raw_header_text: str) -> str:
 
     meaningful = [field for field in unique if _is_meaningful_token(field)]
     result = " ".join(meaningful)
-    if len(result) > _MAX_TABLE_NAME_CHARS:
-        result = result[:_MAX_TABLE_NAME_CHARS].rstrip()
+    if len(result) > MAX_ASSET_FILE_NAME_CHARS:
+        result = result[:MAX_ASSET_FILE_NAME_CHARS].rstrip()
     return result