fix(worker): prevent long internal parser filenames

apps/worker/app/services/document_parser/support/filename_limits.py

@@ -0,0 +1,28 @@
+"""Filename safety helpers for parser working files."""
+
+import os
+from hashlib import sha256
+
+
+MAX_INTERNAL_FILENAME_BYTES = 240
+INTERNAL_FILENAME_HASH_LENGTH = 12
+
+
+def truncate_internal_filename(filename: str) -> str:
+    """Keep parser filenames below Linux NAME_MAX while retaining identity."""
+    if len(os.fsencode(filename)) <= MAX_INTERNAL_FILENAME_BYTES:
+        return filename
+
+    name_root, name_ext = os.path.splitext(filename)
+    filename_hash = sha256(filename.encode("utf-8")).hexdigest()[
+        :INTERNAL_FILENAME_HASH_LENGTH
+    ]
+    suffix = f"-{filename_hash}{name_ext}"
+    available_root_bytes = MAX_INTERNAL_FILENAME_BYTES - len(os.fsencode(suffix))
+    if available_root_bytes <= 0:
+        return f"document-{filename_hash}.bin"
+
+    truncated_root = name_root.encode("utf-8")[:available_root_bytes].decode(
+        "utf-8", errors="ignore"
+    )
+    return f"{truncated_root}{suffix}"

apps/worker/app/services/document_parser/support/internal_parse_name.py

@@ -5,6 +5,9 @@
 from dataclasses import dataclass
 
 from app.services.common.file_utils import path_handle
+from app.services.document_parser.support.filename_limits import (
+    truncate_internal_filename,
+)
 
 
 @dataclass(frozen=True)
@@ -39,7 +42,7 @@ def normalize_internal_parse_name(
     effective_root = name_root or "document"
     internal_name = f"{effective_root}{effective_ext}"
 
-    return (
+    normalized_name = (
         internal_name.replace("(", "-")
         .replace(")", "-")
         .replace("[", "-")
@@ -53,6 +56,7 @@ def normalize_internal_parse_name(
         .replace(chr(0x2015), "-")
         .replace(chr(0x2212), "-")
     )
+    return truncate_internal_filename(normalized_name)
 
 
 def prepare_internal_parse_input(

apps/worker/tests/unit/test_internal_parse_name.py

@@ -0,0 +1,34 @@
+from __future__ import annotations
+
+import os
+from pathlib import Path
+
+from app.services.document_parser.support.internal_parse_name import (
+    prepare_internal_parse_input,
+)
+
+
+def test_prepare_internal_parse_input_handles_long_encoded_filename(
+    tmp_path,
+) -> None:
+    temporary_file_path = tmp_path / "temporary.pdf"
+    temporary_file_path.write_bytes(b"pdf")
+    encoded_filename = (
+        "%E9%99%84%E4%BB%B65.%E5%8D%97%E4%BA%AC%E4%BF%A1%E6%81%AF%E5%B7%A5%E7%A8%8B"
+        * 20
+        + ".pdf"
+    )
+
+    prepared_input = prepare_internal_parse_input(
+        str(temporary_file_path),
+        encoded_filename,
+        fallback_ext=".pdf",
+        prefer_fallback_ext=True,
+    )
+    prepared_file_path = Path(prepared_input.file_path)
+    file_exists = prepared_file_path.exists()
+    file_contents = prepared_file_path.read_bytes()
+
+    assert len(os.fsencode(prepared_input.internal_filename)) <= 240
+    assert file_exists
+    assert file_contents == b"pdf"