Docx

Author: hydropix

requirements.txt

@@ -15,6 +15,10 @@ jinja2>=3.0
 langdetect>=1.0.9
 Pillow>=10.0.0
 
+# DOCX support
+mammoth>=1.0.0
+python-docx>=1.1.0
+
 # Chatterbox TTS (GPU-accelerated local TTS) - OPTIONAL
 # These dependencies have strict numpy version requirements that may conflict.
 # Install separately if needed:

src/api/services/path_validator.py

@@ -1,6 +1,7 @@
 """
 Path validation utilities for secure file operations
 """
+import os
 from typing import Tuple
 
 
@@ -23,10 +24,24 @@ def validate_filename(filename: str) -> Tuple[bool, str]:
         if not filename:
             return False, "Filename cannot be empty"
 
-        # Prevent directory traversal
-        if '..' in filename or filename.startswith('/') or filename.startswith('\\'):
+        # Prevent directory traversal - check for path separators with '..'
+        # This allows '...' or '....' but blocks '../' or '..\' patterns
+        if filename.startswith('/') or filename.startswith('\\'):
+            return False, "Invalid filename: absolute path not allowed"
+
+        # Check for directory traversal patterns
+        # Block: ../ or ..\ (with separators)
+        if '/../' in filename or '\\..\\' in filename or '/..' in filename or '\\..' in filename:
             return False, "Invalid filename: directory traversal not allowed"
 
+        # Also check if the normalized filename contains path separators
+        # This catches cases like "foo/../bar" or "foo/bar"
+        if os.path.sep in filename or ('/' in filename or '\\' in filename):
+            # Exception: allow if it's just the filename itself (no actual traversal)
+            # Use os.path.basename to check if it's a pure filename
+            if os.path.basename(filename) != filename:
+                return False, "Invalid filename: path separators not allowed"
+
         # Check filename length
         if len(filename) > PathValidator.MAX_FILENAME_LENGTH:
             return False, f"Filename too long (max {PathValidator.MAX_FILENAME_LENGTH} characters)"

src/core/adapters/translate_file.py

@@ -137,6 +137,40 @@ async def translate_file(
         )
         return True  # Legacy function doesn't return success status
 
+    # DOCX translation using EPUB pipeline (Phase 1 implementation)
+    # Similar to EPUB, DOCX requires HTML chunking, tag preservation, etc.
+    # This reuses the EPUB pipeline for rapid deployment
+    if ext == '.docx':
+        from src.core.docx.translator import translate_docx_file
+        from src.core.llm import create_llm_provider
+
+        # Create LLM client
+        llm_client = create_llm_provider(
+            provider_type=llm_provider,
+            endpoint=llm_api_endpoint,
+            model=model_name,
+            gemini_api_key=gemini_api_key,
+            openai_api_key=openai_api_key,
+            openrouter_api_key=openrouter_api_key
+        )
+
+        result = await translate_docx_file(
+            input_filepath=input_filepath,
+            output_filepath=output_filepath,
+            source_language=source_language,
+            target_language=target_language,
+            model_name=model_name,
+            llm_client=llm_client,
+            max_tokens_per_chunk=context_window or 450,
+            log_callback=log_callback,
+            progress_callback=progress_callback,
+            stats_callback=stats_callback,
+            prompt_options=prompt_options,
+            max_retries=1,
+            context_manager=None
+        )
+        return result['success']
+
     # Map file extensions to adapters
     adapter_map = {
         '.txt': TxtAdapter,
@@ -146,7 +180,7 @@ async def translate_file(
 
     adapter_class = adapter_map.get(ext)
     if not adapter_class:
-        supported = ', '.join(['.txt', '.srt', '.epub'])  # Include .epub in supported formats
+        supported = ', '.join(['.txt', '.srt', '.epub', '.docx'])  # Include .epub and .docx in supported formats
         raise UnsupportedFormatError(
             f"Unsupported file format: {ext}. Supported formats: {supported}"
         )
@@ -214,6 +248,7 @@ def get_file_type_from_path(filepath: str) -> str:
         '.txt': 'txt',
         '.srt': 'srt',
         '.epub': 'epub',
+        '.docx': 'docx',
     }
 
     return type_map.get(ext, 'unknown')
@@ -266,6 +301,7 @@ async def build_translated_output(
         'txt': TxtAdapter,
         'srt': SrtAdapter,
         'epub': EpubAdapter,
+        # Note: docx doesn't support checkpoint reconstruction yet
     }
 
     adapter_class = adapter_map.get(file_type)

src/core/common/__init__.py

@@ -0,0 +1,18 @@
+"""
+Common translation components.
+
+This module contains reusable components for translation:
+- GenericTranslationOrchestrator: Unified pipeline for all formats
+- TranslationAdapter: Interface for format-specific adapters
+- TranslationMetrics: Shared metrics tracking
+"""
+
+from .translation_orchestrator import (
+    GenericTranslationOrchestrator,
+    TranslationAdapter
+)
+
+__all__ = [
+    'GenericTranslationOrchestrator',
+    'TranslationAdapter',
+]