feat(file-tool): add djvu conversion and extraction actions

services/router/tool_manager.py

@@ -13,6 +13,7 @@ import hashlib
 import base64
 import csv
 import tempfile
+import subprocess
 import httpx
 from typing import Dict, List, Any, Optional
 from dataclasses import dataclass
@@ -330,6 +331,7 @@ TOOL_DEFINITIONS = [
                             "pptx_create", "pptx_update",
                             "ods_create", "ods_update", "parquet_create", "parquet_update",
                             "csv_create", "csv_update", "pdf_fill", "pdf_merge", "pdf_split", "pdf_update",
+                            "djvu_to_pdf", "djvu_extract_text",
                             "json_export", "yaml_export", "zip_bundle",
                             "text_create", "text_update", "markdown_create", "markdown_update",
                             "xml_export", "html_export",
@@ -671,6 +673,10 @@ class ToolManager:
             return self._file_pdf_update(args)
         if action == "pdf_fill":
             return self._file_pdf_fill(args)
+        if action == "djvu_to_pdf":
+            return self._file_djvu_to_pdf(args)
+        if action == "djvu_extract_text":
+            return self._file_djvu_extract_text(args)
 
         return ToolResult(success=False, result=None, error=f"Action not implemented yet: {action}")
 
@@ -2102,6 +2108,89 @@ class ToolManager:
             file_name=file_name,
             file_mime="application/pdf",
         )
+
+    def _file_djvu_to_pdf(self, args: Dict[str, Any]) -> ToolResult:
+        src_b64 = args.get("file_base64")
+        if not src_b64:
+            return ToolResult(success=False, result=None, error="file_base64 is required for djvu_to_pdf")
+        file_name = self._sanitize_file_name(args.get("file_name"), "converted.pdf", force_ext=".pdf")
+
+        timeout_sec = max(5, min(int(args.get("timeout_sec") or 60), 300))
+        with tempfile.TemporaryDirectory(prefix="djvu2pdf_") as tmpdir:
+            src = os.path.join(tmpdir, "input.djvu")
+            out_pdf = os.path.join(tmpdir, "output.pdf")
+            with open(src, "wb") as f:
+                f.write(self._bytes_from_b64(src_b64))
+
+            try:
+                proc = subprocess.run(
+                    ["ddjvu", "-format=pdf", src, out_pdf],
+                    capture_output=True,
+                    text=True,
+                    timeout=timeout_sec,
+                    check=False,
+                )
+            except FileNotFoundError:
+                return ToolResult(success=False, result=None, error="ddjvu not found in runtime image")
+            except subprocess.TimeoutExpired:
+                return ToolResult(success=False, result=None, error=f"DJVU conversion timed out ({timeout_sec}s)")
+
+            if proc.returncode != 0 or not os.path.exists(out_pdf):
+                stderr = (proc.stderr or "").strip()
+                return ToolResult(success=False, result=None, error=f"ddjvu failed: {stderr or 'unknown error'}")
+
+            data = open(out_pdf, "rb").read()
+            if not data:
+                return ToolResult(success=False, result=None, error="ddjvu produced empty PDF")
+
+        return ToolResult(
+            success=True,
+            result={"message": f"DJVU converted to PDF: {file_name}"},
+            file_base64=self._b64_from_bytes(data),
+            file_name=file_name,
+            file_mime="application/pdf",
+        )
+
+    def _file_djvu_extract_text(self, args: Dict[str, Any]) -> ToolResult:
+        src_b64 = args.get("file_base64")
+        if not src_b64:
+            return ToolResult(success=False, result=None, error="file_base64 is required for djvu_extract_text")
+        file_name = self._sanitize_file_name(args.get("file_name"), "extracted.txt", force_ext=".txt")
+        timeout_sec = max(5, min(int(args.get("timeout_sec") or 60), 300))
+
+        with tempfile.TemporaryDirectory(prefix="djvutxt_") as tmpdir:
+            src = os.path.join(tmpdir, "input.djvu")
+            with open(src, "wb") as f:
+                f.write(self._bytes_from_b64(src_b64))
+            try:
+                proc = subprocess.run(
+                    ["djvutxt", src],
+                    capture_output=True,
+                    text=True,
+                    timeout=timeout_sec,
+                    check=False,
+                )
+            except FileNotFoundError:
+                return ToolResult(success=False, result=None, error="djvutxt not found in runtime image")
+            except subprocess.TimeoutExpired:
+                return ToolResult(success=False, result=None, error=f"DJVU text extraction timed out ({timeout_sec}s)")
+
+            if proc.returncode != 0:
+                stderr = (proc.stderr or "").strip()
+                return ToolResult(success=False, result=None, error=f"djvutxt failed: {stderr or 'unknown error'}")
+            text = proc.stdout or ""
+            msg = f"DJVU text extracted: {file_name}"
+            if not text.strip():
+                msg = f"DJVU has no extractable text layer, returned empty text file: {file_name}"
+
+        payload = text.encode("utf-8")
+        return ToolResult(
+            success=True,
+            result={"message": msg},
+            file_base64=self._b64_from_bytes(payload),
+            file_name=file_name,
+            file_mime="text/plain",
+        )
     
     async def _memory_search(self, args: Dict, agent_id: str = None, chat_id: str = None, user_id: str = None) -> ToolResult:
         """Search in Qdrant vector memory using Router's memory_retrieval - PRIORITY 1"""