microsoft/markitdown · error · ValueError

Malformed data URI, missing ',' separator

Error message

Malformed data URI, missing ',' separator

What it means

RFC 2397 data URIs are '<mediatype>[;base64],<data>'; parse_data_uri() splits on the first comma and treats an empty separator (no comma at all) as malformed, raising this ValueError. The check uses uri.partition(',') so a URI that ends at the header ('data:text/plain') with no ',' and no payload triggers it.

Source

Thrown at packages/markitdown/src/markitdown/_uri_utils.py:25

def file_uri_to_path(file_uri: str) -> Tuple[str | None, str]:
    """Convert a file URI to a local file path"""
    parsed = urlparse(file_uri)
    if parsed.scheme != "file":
        raise ValueError(f"Not a file URL: {file_uri}")

    netloc = parsed.netloc if parsed.netloc else None
    path = os.path.abspath(url2pathname(parsed.path))
    return netloc, path


def parse_data_uri(uri: str) -> Tuple[str | None, Dict[str, str], bytes]:
    if not uri.startswith("data:"):
        raise ValueError("Not a data URI")

    header, _, data = uri.partition(",")
    if not _:
        raise ValueError("Malformed data URI, missing ',' separator")

    meta = header[5:]  # Strip 'data:'
    parts = meta.split(";")

    is_base64 = False
    # Ends with base64?
    if parts[-1] == "base64":
        parts.pop()
        is_base64 = True

    mime_type = None  # Normally this would default to text/plain but we won't assume
    if len(parts) and len(parts[0]) > 0:
        # First part is the mime type
        mime_type = parts.pop(0)

    attributes: Dict[str, str] = {}
    for part in parts:
        # Handle key=value pairs in the middle

View on GitHub (pinned to fd239d5d2b)

Solutions

1. Fix the generator to always emit '<mime>[;base64],<data>' with the comma
2. Validate user-supplied data URIs with a regex before conversion
3. If the payload is empty on purpose, send 'data:text/plain,'

Example fix

# before
md.convert_uri("data:application/pdf")  # ValueError: missing ','

# after
md.convert_uri("data:application/pdf;base64,JVBERi0...")

import re

DATA_URI_RE = re.compile(r"^data:[^,]*;base64,[A-Za-z0-9+/=]+$|^data:[^,]*,.+$")

def is_wellformed_data_uri(uri: str) -> bool:
    return uri.startswith("data:") and "," in uri

try:
    result = md.convert_uri(data_uri)
except ValueError as e:
    if "missing ','" in str(e):
        raise ValueError("data URI truncated: header present but no payload") from e
    raise

When it happens

Trigger: md.convert_uri('data:text/plain') or 'data:application/pdf' — header present but the comma and data section missing; also truncated copies of data URIs.

Common situations: Template strings or string concatenation that drops the payload, clipboard truncation of very long base64 URIs, or building data URIs with a missing ',' separator during generation.

Understand the failure class

• Parsing and encoding errors: unexpected token, malformed input — why parsers reject input and how to find the real culprit.

Related errors

• Not a data URI
• Unsupported file URI: {uri}. Netloc must be empty or localho
• Unsupported URI scheme: {uri.split(':')[0]}. Supported schem
• Not a file URL: {file_uri}
• No channel found in RSS feed

AI-assisted analysis of microsoft/markitdown@fd239d5d2b (2026-08-14). Data as JSON: /api/errors/066bad2bfab35227. Report an issue: GitHub.