batch_map_concepts_from_csv raises ValueError on every row

[hyun-hyang]

Summary

batch_map_concepts_from_csv is broken on every input row. Since 67fc34a (expand find_omop_concept to return multiple candidate concepts with metadata), find_omop_concept returns

{"candidates": [...], "search_metadata": {...}}        # success
{"error": "..."}                                       # no results

but batch_map_concepts_from_csv still merges that dict directly into a csv.DictWriter row whose fieldnames list expects flat keys (concept_id, code, name, class, concept, validity, domain, vocab, url, reason).

csv.DictWriter defaults to extrasaction='raise', so the merged candidates / search_metadata keys (or the error key) trigger:

ValueError: dict contains fields not in fieldnames: 'candidates', 'search_metadata'

The tool fails on the first row of any input CSV.

Location

omop_mcp/src/omop_mcp/server.py

Lines 179 to 208 in 001c91b

	async def batch_map_concepts_from_csv(csv_path: str) -> str:
	"""
	Process a CSV file of keywords, mapping each row and returning a CSV with mapping results appended as new columns.
	"""
	output = io.StringIO()
	with open(csv_path, newline="", encoding="utf-8") as infile:
	reader = csv.DictReader(infile)
	fieldnames = reader.fieldnames + [
	"concept_id",
	"code",
	"name",
	"class",
	"concept",
	"validity",
	"domain",
	"vocab",
	"url",
	"reason",
	]
	writer = csv.DictWriter(output, fieldnames=fieldnames)
	writer.writeheader()
	for row in reader:
	keyword = row.get("keyword", "")
	omop_field = row.get("omop_field", "")
	omop_table = row.get("omop_table", "")
	result = await find_omop_concept(keyword, omop_table, omop_field)
	# Merge result into row
	row.update(result)
	writer.writerow(row)
	return output.getvalue()

Repro

# input.csv
# keyword,omop_table,omop_field
# diabetes,condition_occurrence,condition_concept_id
import asyncio, csv, io
from omop_mcp.server import batch_map_concepts_from_csv
asyncio.run(batch_map_concepts_from_csv("input.csv"))
# ValueError: dict contains fields not in fieldnames: 'candidates', 'search_metadata'

Fix

PR: #10