{ "id": "llm/function-call-arguments-truncated-in-streaming", "signature": "json.decoder.JSONDecodeError: Expecting property name enclosed in double quotes: line 1 column 1024 (char 1023) when parsing streaming function call arguments", "signature_zh": "json.decoder.JSONDecodeError: 期望属性名称用双引号括起来：在解析流式函数调用参数时，第 1 行第 1024 列（字符 1023）", "regex": "json\\.decoder\\.JSONDecodeError: Expecting property name enclosed in double quotes.*streaming function call arguments", "domain": "llm", "category": "protocol_error", "subcategory": null, "root_cause": "When streaming function calls, the API sends the arguments in chunks. If the argument string is long, a chunk boundary can split a JSON token (e.g., a string value or key), causing the accumulated arguments to be invalid JSON when parsed incrementally.", "root_cause_type": "generic", "root_cause_zh": "当流式传输函数调用时，API 会分块发送参数。如果参数字符串很长，块边界可能会分割 JSON 令牌（例如，字符串值或键），导致在增量解析时累积的参数成为无效 JSON。", "versions": [ { "version": "openai==1.30.0", "introduced": null, "deprecated": null, "removed": null, "behavior_change": null, "status": "active" }, { "version": "anthropic==0.25.0", "introduced": null, "deprecated": null, "removed": null, "behavior_change": null, "status": "active" }, { "version": "mistralai==0.1.0", "introduced": null, "deprecated": null, "removed": null, "behavior_change": null, "status": "active" } ], "os_specific": {}, "dead_ends": [ { "action": "Using json.loads() on each individual chunk instead of accumulating", "why_fails": "Individual chunks are not valid JSON; they are partial fragments. json.loads() will always fail on incomplete chunks.", "fail_rate": 0.95, "condition": "", "sources": [] }, { "action": "Increasing max_tokens to reduce chunking", "why_fails": "max_tokens controls the output length, not the chunk size. The API still sends chunks of arbitrary size regardless of max_tokens.", "fail_rate": 0.8, "condition": "", "sources": [] }, { "action": "Setting stream_options={'include_usage': True}", "why_fails": "This option controls whether usage information is included in the stream, it has no effect on how function call arguments are chunked or parsed.", "fail_rate": 0.9, "condition": "", "sources": [] } ], "workarounds": [ { "action": "Accumulate all chunks for a function call before attempting to parse. Wait for the 'finish_reason' to be 'stop' or 'function_call' before parsing the complete arguments string. Example:\n\nfull_args = \"\"\nfor chunk in stream:\n delta = chunk.choices[0].delta\n if delta.function_call and delta.function_call.arguments:\n full_args += delta.function_call.arguments\n if chunk.choices[0].finish_reason:\n break\nimport json\nparsed_args = json.loads(full_args)", "success_rate": 0.95, "how": "Accumulate all chunks for a function call before attempting to parse. Wait for the 'finish_reason' to be 'stop' or 'function_call' before parsing the complete arguments string. Example:\n\nfull_args = \"\"\nfor chunk in stream:\n delta = chunk.choices[0].delta\n if delta.function_call and delta.function_call.arguments:\n full_args += delta.function_call.arguments\n if chunk.choices[0].finish_reason:\n break\nimport json\nparsed_args = json.loads(full_args)", "condition": "", "sources": [] }, { "action": "Use the 'incremental parsing' approach with a JSON repair library like 'json-repair' or 'json5' that can handle partial JSON. Example:\n\nimport json5\npartial_args = \"\"\nfor chunk in stream:\n delta = chunk.choices[0].delta\n if delta.function_call and delta.function_call.arguments:\n partial_args += delta.function_call.arguments\n try:\n parsed = json5.loads(partial_args)\n # use parsed incrementally\n except Exception:\n pass", "success_rate": 0.82, "how": "Use the 'incremental parsing' approach with a JSON repair library like 'json-repair' or 'json5' that can handle partial JSON. Example:\n\nimport json5\npartial_args = \"\"\nfor chunk in stream:\n delta = chunk.choices[0].delta\n if delta.function_call and delta.function_call.arguments:\n partial_args += delta.function_call.arguments\n try:\n parsed = json5.loads(partial_args)\n # use parsed incrementally\n except Exception:\n pass", "condition": "", "sources": [] }, { "action": "Set the API parameter 'stream_options' to {'include_usage': True} and use the 'usage' field to detect the end of function call arguments, then parse the complete string.", "success_rate": 0.75, "how": "Set the API parameter 'stream_options' to {'include_usage': True} and use the 'usage' field to detect the end of function call arguments, then parse the complete string.", "condition": "", "sources": [] } ], "workarounds_zh": [ "在尝试解析之前，累积函数调用的所有块。等待 'finish_reason' 为 'stop' 或 'function_call' 后再解析完整的参数字符串。示例：\n\nfull_args = \"\"\nfor chunk in stream:\n delta = chunk.choices[0].delta\n if delta.function_call and delta.function_call.arguments:\n full_args += delta.function_call.arguments\n if chunk.choices[0].finish_reason:\n break\nimport json\nparsed_args = json.loads(full_args)", "使用 '增量解析' 方法，配合 JSON 修复库（如 'json-repair' 或 'json5'）处理部分 JSON。示例：\n\nimport json5\npartial_args = \"\"\nfor chunk in stream:\n delta = chunk.choices[0].delta\n if delta.function_call and delta.function_call.arguments:\n partial_args += delta.function_call.arguments\n try:\n parsed = json5.loads(partial_args)\n # 增量使用解析结果\n except Exception:\n pass", "设置 API 参数 'stream_options' 为 {'include_usage': True}，并使用 'usage' 字段检测函数调用参数的结束，然后解析完整字符串。" ], "transition_graph": { "leads_to": [], "preceded_by": [], "frequently_confused_with": [] }, "official_doc_url": "https://platform.openai.com/docs/guides/function-calling#streaming-function-calls", "official_doc_section": null, "error_code": null, "verification_tier": "ai_generated", "confidence": 0.89, "fix_success_rate": 0.85, "resolvable": "true", "first_seen": "2024-04-01", "last_confirmed": "2024-06-01", "last_updated": "2024-06-01", "evidence_count": 1, "tags": [], "locale": "en", "aliases": [] }