Skip to content

fix: safely handle non-string message content in tokens command #681

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dwash96 merged 1 commit into from
Sep 18, 2026
Merged

fix: safely handle non-string message content in tokens command #681

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
94 changes: 51 additions & 43 deletions cecli/commands/tokens.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
from typing import List
from typing import Any, Dict, List, Optional

from cecli.commands.utils.base_command import BaseCommand
from cecli.commands.utils.helpers import format_command_result
Expand All @@ -9,6 +9,48 @@ class TokensCommand(BaseCommand):
NORM_NAME = "tokens"
DESCRIPTION = "Report on the number of tokens used by the current chat context"

@classmethod
def _extract_file_name(cls, msg: Dict[str, Any]) -> Optional[str]:
"""Extract file name from a message dictionary."""
if not isinstance(msg, dict):
return None

# Check explicit image_file metadata first
fname = msg.get("image_file")
if fname:
return fname

content = msg.get("content")
if isinstance(content, str):
if content.startswith(("Original File Contents For", "Current File Contents For")):
lines = content.split("\n", 3)
if len(lines) > 1:
return lines[1].strip()
elif content.startswith("Image file: "):
return content[len("Image file: "):].strip()
elif isinstance(content, list):
for part in content:
if isinstance(part, dict):
if part.get("image_file"):
return part.get("image_file")
text = part.get("text")
if isinstance(text, str):
if text.startswith(("Original File Contents For", "Current File Contents For")):
lines = text.split("\n", 3)
if len(lines) > 1:
return lines[1].strip()
elif text.startswith("Image file: "):
return text[len("Image file: "):].strip()
elif isinstance(part, str):
if part.startswith(("Original File Contents For", "Current File Contents For")):
lines = part.split("\n", 3)
if len(lines) > 1:
return lines[1].strip()
elif part.startswith("Image file: "):
return part[len("Image file: "):].strip()

return None

@classmethod
async def execute(cls, io, coder, args, **kwargs):
res = []
Expand Down Expand Up @@ -123,27 +165,10 @@ async def execute(cls, io, coder, args, **kwargs):
# Group messages by file (each file has user and assistant messages)
file_tokens = {}
for msg in readonly_msgs:
# Extract file name from message content
content = msg.get("content", "")
if content.startswith("Original File Contents For"):
# Extract file path from "File Contents {path}:"
lines = content.split("\n", 3)
if lines:
file_line = lines[1]
fname = file_line.strip()
# Calculate tokens for this message
tokens = coder.main_model.token_count([msg])
if fname not in file_tokens:
file_tokens[fname] = 0
file_tokens[fname] += tokens
elif "image_file" in msg:
# Handle image files
fname = msg.get("image_file")
if fname:
tokens = coder.main_model.token_count([msg])
if fname not in file_tokens:
file_tokens[fname] = 0
file_tokens[fname] += tokens
fname = cls._extract_file_name(msg)
if fname:
tokens = coder.main_model.token_count([msg])
file_tokens[fname] = file_tokens.get(fname, 0) + tokens

# Add to results
for fname, tokens in file_tokens.items():
Expand All @@ -158,27 +183,10 @@ async def execute(cls, io, coder, args, **kwargs):
msgs = ConversationService.get_manager(coder).get_messages_dict(tag=tag)
if msgs:
for msg in msgs:
# Extract file name from message content
content = msg.get("content", "")
if content.startswith("Original File Contents For"):
# Extract file path from "File Contents {path}:"
lines = content.split("\n", 3)
if lines:
file_line = lines[1]
fname = file_line.strip()
# Calculate tokens for this message
tokens = coder.main_model.token_count([msg])
if fname not in editable_file_tokens:
editable_file_tokens[fname] = 0
editable_file_tokens[fname] += tokens
elif "image_file" in msg:
# Handle image files
fname = msg.get("image_file")
if fname:
tokens = coder.main_model.token_count([msg])
if fname not in editable_file_tokens:
editable_file_tokens[fname] = 0
editable_file_tokens[fname] += tokens
fname = cls._extract_file_name(msg)
if fname:
tokens = coder.main_model.token_count([msg])
editable_file_tokens[fname] = editable_file_tokens.get(fname, 0) + tokens

# Add editable files to results
for fname, tokens in editable_file_tokens.items():
Expand Down
117 changes: 117 additions & 0 deletions tests/basic/test_commands.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -420,6 +420,123 @@ async def test_cmd_tokens(self):
self.assertIn("foo.txt", console_output)
self.assertIn("bar.txt", console_output)

async def test_cmd_tokens_with_image_list_content(self):
from cecli.commands.tokens import TokensCommand
from cecli.helpers.conversation import ConversationService, MessageTag

# Direct unit tests for _extract_file_name
self.assertIsNone(TokensCommand._extract_file_name(None))
self.assertIsNone(TokensCommand._extract_file_name("not a dict"))
self.assertIsNone(TokensCommand._extract_file_name({}))
self.assertIsNone(TokensCommand._extract_file_name({"content": None}))
self.assertIsNone(TokensCommand._extract_file_name({"content": "Just a normal message"}))

# String content
self.assertEqual(
TokensCommand._extract_file_name({
"content": "Original File Contents For:\n/path/to/file.py\n\ncode..."
}),
"/path/to/file.py",
)
self.assertEqual(
TokensCommand._extract_file_name({
"content": "Current File Contents For:\n/path/to/file2.py\n\ncode..."
}),
"/path/to/file2.py",
)
self.assertEqual(
TokensCommand._extract_file_name({"content": "Image file: photo.png"}),
"photo.png",
)

# image_file key
self.assertEqual(
TokensCommand._extract_file_name({"image_file": "photo.jpg"}),
"photo.jpg",
)
self.assertEqual(
TokensCommand._extract_file_name({
"image_file": "photo.jpg",
"content": [{"type": "image_url", "image_url": {}}],
}),
"photo.jpg",
)

# List content
self.assertEqual(
TokensCommand._extract_file_name({
"content": [
{"type": "text", "text": "Image file: nested.png"},
{"type": "image_url", "image_url": {}},
]
}),
"nested.png",
)
self.assertEqual(
TokensCommand._extract_file_name({
"content": [
{"type": "text", "text": "Original File Contents For:\nmodule.py\n\ncode"}
]
}),
"module.py",
)
self.assertEqual(
TokensCommand._extract_file_name({
"content": [{"image_file": "from_part.png"}]
}),
"from_part.png",
)
self.assertEqual(
TokensCommand._extract_file_name({
"content": ["Image file: string_part.png"]
}),
"string_part.png",
)

# Integration test with Coder / Commands execution
io = InputOutput(pretty=False, fancy_input=False, yes=True)
coder = await Coder.create(self.GPT35, None, io)
commands = Commands(io, coder)

manager = ConversationService.get_manager(coder)
img_msg = {
"role": "user",
"content": [
{"type": "text", "text": "Image file: test_image.png"},
{"type": "image_url", "image_url": {"url": "data:image/png;base64,123"}},
],
"image_file": "test_image.png",
}
manager.add_message(
message_dict=img_msg,
tag=MessageTag.READONLY_FILES,
hash_key=("image_user", "test_image.png"),
)

img_msg_no_attr = {
"role": "user",
"content": [
{"type": "text", "text": "Image file: another_image.png"},
{"type": "image_url", "image_url": {"url": "data:image/png;base64,456"}},
],
}
manager.add_message(
message_dict=img_msg_no_attr,
tag=MessageTag.CHAT_FILES,
hash_key=("image_user", "another_image.png"),
)

stdout = StringIO()
sys.stdout = stdout
try:
commands.execute("tokens", "")
finally:
sys.stdout = sys.__stdout__

console_output = stdout.getvalue()
self.assertIn("test_image.png", console_output)
self.assertIn("another_image.png", console_output)

async def test_cmd_add_from_subdir(self):
repo = git.Repo.init()
repo.config_writer().set_value("user", "name", "Test User").release()
Expand Down