fix write_file fenced source artifacts

This commit is contained in:
pewdiepie-archdaemon
2026-09-18 13:35:44 +00:00
parent fb9558439c
commit cdd28ed9b2
2 changed files with 59 additions and 0 deletions
+26
View File
@@ -25,6 +25,31 @@ _BINARY_ARTIFACT_SUFFIXES = _STRUCTURED_DOCUMENT_SUFFIXES | frozenset({
".png", ".wav", ".webm", ".webp", ".zip",
})
# Models frequently put source artifacts in a Markdown code fence even when a
# tool schema asks for the raw file body. Persisting that fence makes HTML,
# CSS, JavaScript, and source files invalid. Restrict normalization to
# code-like targets so a user can still write a literal fence to Markdown.
_FENCED_SOURCE_SUFFIXES = frozenset({
".css", ".csv", ".html", ".htm", ".js", ".json", ".jsx", ".mjs",
".py", ".sh", ".sql", ".svg", ".ts", ".tsx", ".xml", ".yaml", ".yml",
})
def _unwrap_fenced_source_body(body: str, path: str) -> str:
"""Remove an accidental outer Markdown fence from a source artifact.
An opening fence is enough to normalize: generation can end during a tool
call while its argument remains otherwise usable, and retaining the fence
corrupts the artifact. This only applies to source-like file extensions.
"""
if os.path.splitext(path)[1].casefold() not in _FENCED_SOURCE_SUFFIXES:
return body
match = re.match(r"^(\s*)```[^\r\n]*\r?\n", body)
if not match:
return body
unwrapped = body[match.end():]
return re.sub(r"\r?\n```\s*$", "", unwrapped)
def _glob_to_regex(pat: str) -> "re.Pattern":
"""Translate a forward-slash glob (**, *, ?) into a compiled regex.
@@ -247,6 +272,7 @@ class WriteFileTool:
path = _resolve_tool_path(raw_path)
except ValueError as e:
return {"error": f"write_file: {e}", "exit_code": 1}
body = _unwrap_fenced_source_body(body, path)
# A frequent multimodal artifact failure is writing SVG markup to a
# path whose extension promises a raster image. The file exists, so
# ordinary artifact checks pass, but image judges cannot decode it.
@@ -1,3 +1,5 @@
import json
import pytest
from src.agent_tools.filesystem_tools import EditFileTool, ReadFileTool, WriteFileTool
@@ -90,3 +92,34 @@ async def test_write_file_still_rewrites_existing_text_file(tmp_path, monkeypatc
assert result["exit_code"] == 0
assert target.read_text(encoding="utf-8") == "new"
@pytest.mark.asyncio
async def test_write_file_unwraps_markdown_fence_for_html_artifact(tmp_path, monkeypatch):
import src.tool_execution as tool_execution
target = tmp_path / "output.html"
monkeypatch.setattr(tool_execution, "_resolve_tool_path", lambda _path: str(target))
result = await WriteFileTool().execute(
'{"path":"output.html","content":"```html\\n<div>clock</div>\\n```"}', {}
)
assert result["exit_code"] == 0
assert target.read_text(encoding="utf-8") == "<div>clock</div>"
@pytest.mark.asyncio
async def test_write_file_preserves_literal_fence_for_markdown(tmp_path, monkeypatch):
import src.tool_execution as tool_execution
target = tmp_path / "notes.md"
monkeypatch.setattr(tool_execution, "_resolve_tool_path", lambda _path: str(target))
content = "```python\\nprint('literal example')\\n```"
result = await WriteFileTool().execute(
json.dumps({"path": "notes.md", "content": content}), {}
)
assert result["exit_code"] == 0
assert target.read_text(encoding="utf-8") == content