feat(api): pagination + search + PDF/DOCX extract in folder agent tools

tests/test_folder_agent_tool.py

@@ -4,7 +4,10 @@ from unittest.mock import AsyncMock, patch
 
 import pytest
 
-from app.agents.folder_agent import read_project_folder_file
+from app.agents.folder_agent import (
+    read_project_folder_file,
+    search_project_folder_file,
+)
 
 pytestmark = pytest.mark.asyncio
 
@@ -12,10 +15,11 @@ pytestmark = pytest.mark.asyncio
 async def test_happy_path():
     with patch(
         "app.agents.folder_agent.execute_on_client",
-        new=AsyncMock(return_value={"content": "file body"}),
+        new=AsyncMock(return_value={"content": "file body", "kind": "text", "totalSize": 9}),
     ):
         out = await read_project_folder_file.ainvoke({"project_id": "p1", "relative_path": "docs/x.md"})
-    assert out == "file body"
+    assert "file body" in out
+    assert "kind=text" in out
 
 
 async def test_traversal_rejected():
@@ -31,7 +35,105 @@ async def test_absolute_path_rejected():
 async def test_missing_file():
     with patch(
         "app.agents.folder_agent.execute_on_client",
-        new=AsyncMock(return_value={"content": ""}),
+        new=AsyncMock(return_value={"content": "", "kind": "missing", "totalSize": 0}),
     ):
         out = await read_project_folder_file.ainvoke({"project_id": "p1", "relative_path": "ghost.md"})
     assert "not found" in out.lower()
+
+
+async def test_pagination_signals_more_available():
+    # Electron returned the first slice, totalSize larger than slice length.
+    with patch(
+        "app.agents.folder_agent.execute_on_client",
+        new=AsyncMock(return_value={"content": "first chunk", "kind": "text", "totalSize": 1000}),
+    ):
+        out = await read_project_folder_file.ainvoke({
+            "project_id": "p1",
+            "relative_path": "big.txt",
+            "offset": 0,
+            "length": 11,
+        })
+    assert "first chunk" in out
+    assert "More content available" in out
+    assert "offset=11" in out
+
+
+async def test_pdf_extracted_then_sliced(monkeypatch):
+    from app.agents import folder_agent
+    monkeypatch.setattr(folder_agent, "_extract_pdf_text", lambda b: "ABC " * 100)
+    with patch(
+        "app.agents.folder_agent.execute_on_client",
+        new=AsyncMock(return_value={"content": "JVBERi0xLg==", "kind": "pdf", "totalSize": 12}),
+    ):
+        out = await read_project_folder_file.ainvoke({
+            "project_id": "p1",
+            "relative_path": "doc.pdf",
+            "offset": 0,
+            "length": 8,
+        })
+    assert "kind=pdf" in out
+    assert "ABC ABC " in out
+    assert "More content available" in out
+
+
+async def test_image_returns_placeholder():
+    with patch(
+        "app.agents.folder_agent.execute_on_client",
+        new=AsyncMock(return_value={"content": "iVBORw0K", "kind": "image", "totalSize": 1024}),
+    ):
+        out = await read_project_folder_file.ainvoke({"project_id": "p1", "relative_path": "logo.png"})
+    assert "image" in out.lower()
+
+
+async def test_search_finds_match_with_context():
+    body = "alpha\nbeta\nthe needle is here\ngamma\ndelta"
+    with patch(
+        "app.agents.folder_agent.execute_on_client",
+        new=AsyncMock(return_value={"content": body, "kind": "text", "totalSize": len(body)}),
+    ):
+        out = await search_project_folder_file.ainvoke({
+            "project_id": "p1",
+            "relative_path": "log.txt",
+            "query": "needle",
+            "context_lines": 1,
+        })
+    assert "needle" in out
+    assert "matches=1" in out
+    # Context lines included
+    assert "beta" in out
+    assert "gamma" in out
+
+
+async def test_search_no_match():
+    with patch(
+        "app.agents.folder_agent.execute_on_client",
+        new=AsyncMock(return_value={"content": "nothing here", "kind": "text", "totalSize": 12}),
+    ):
+        out = await search_project_folder_file.ainvoke({
+            "project_id": "p1",
+            "relative_path": "x.txt",
+            "query": "zzz",
+        })
+    assert "No matches" in out
+
+
+async def test_search_rejects_traversal():
+    out = await search_project_folder_file.ainvoke({
+        "project_id": "p1",
+        "relative_path": "../etc/passwd",
+        "query": "root",
+    })
+    assert out == "Access denied"
+
+
+async def test_search_image_rejected():
+    with patch(
+        "app.agents.folder_agent.execute_on_client",
+        new=AsyncMock(return_value={"content": "b64data", "kind": "image", "totalSize": 100}),
+    ):
+        out = await search_project_folder_file.ainvoke({
+            "project_id": "p1",
+            "relative_path": "logo.png",
+            "query": "anything",
+        })
+    assert "Cannot search" in out