fix: Anthropic responses with multiple text blocks lose all but the first (#1255)

* fix: concatenate all Anthropic text blocks, not just the first

* test: Anthropic response parsing concatenates text blocks
This commit is contained in:
Afonso Coutinho
2026-06-02 16:57:20 +01:00
committed by GitHub
parent a901992d03
commit a04553013d
2 changed files with 39 additions and 5 deletions

View File

@@ -562,11 +562,18 @@ def _build_anthropic_headers(headers):
return h return h
def _parse_anthropic_response(data: dict) -> str: def _parse_anthropic_response(data: dict) -> str:
"""Extract text from Anthropic response.""" """Extract text from an Anthropic response.
for block in data.get("content", []):
if block.get("type") == "text": The Messages API `content` is an array that can hold more than one text
return block.get("text", "") block (e.g. text split around a tool_use block, or citation-segmented
return "" text). Concatenate them all instead of returning only the first, which
silently dropped the rest of the reply.
"""
return "".join(
block.get("text", "")
for block in data.get("content", [])
if isinstance(block, dict) and block.get("type") == "text"
)
def _sanitize_llm_messages(messages: List[Dict]) -> List[Dict]: def _sanitize_llm_messages(messages: List[Dict]) -> List[Dict]:

View File

@@ -0,0 +1,27 @@
"""Tests for _parse_anthropic_response (src/llm_core.py)."""
from src.llm_core import _parse_anthropic_response
def test_concatenates_multiple_text_blocks():
# Regression: only the first text block was returned, dropping the rest.
data = {"content": [
{"type": "text", "text": "Part A "},
{"type": "tool_use", "id": "t1", "name": "x", "input": {}},
{"type": "text", "text": "Part B"},
]}
assert _parse_anthropic_response(data) == "Part A Part B"
def test_skips_non_text_blocks():
data = {"content": [
{"type": "thinking", "thinking": "..."},
{"type": "text", "text": "answer"},
]}
assert _parse_anthropic_response(data) == "answer"
def test_single_block_and_empty():
assert _parse_anthropic_response({"content": [{"type": "text", "text": "hi"}]}) == "hi"
assert _parse_anthropic_response({"content": []}) == ""
assert _parse_anthropic_response({}) == ""