ScrapeGraphAI/Scrapegraph-ai · error · AnalysisError

Semantic analysis failed: {str(e)}

Error message

Semantic analysis failed: {str(e)}

What it means

Catch-all failure of the semantic-focused LLM analysis chain; the original exception is appended to the message. Indicates prompt/LLM failure rather than a state-shape problem.

Source

Thrown at scrapegraphai/utils/code_error_analysis.py:335

        prompt = PromptTemplate(
            template=get_optimal_analysis_template("semantic"),
            input_variables=["generated_code", "differences", "explanation"],
        )
        chain = prompt | llm_model | StrOutputParser()

        # Execute chain with validated inputs
        return chain.invoke(
            {
                "generated_code": validated_state.generated_code,
                "differences": json.dumps(comparison_result["differences"], indent=2),
                "explanation": comparison_result["explanation"],
            }
        )

    except KeyError as e:
        raise InvalidStateError(f"Missing required key: {e}")
    except Exception as e:
        raise AnalysisError(f"Semantic analysis failed: {str(e)}")

View on GitHub (pinned to 532dfffbf6)

Solutions

  1. Inspect str(e); if serialization, coerce differences to plain dicts/lists before calling
  2. Verify differences/explanation are JSON-safe (json.dumps test)
  3. Check llm_model auth/connectivity; retry on transient provider errors

Example fix

# before
semantic_focused_analysis(state, {"differences": llm_obj, "explanation": ""}, llm)
# after
semantic_focused_analysis(state, {"differences": llm_obj if isinstance(llm_obj, (list, dict)) else str(llm_obj), "explanation": ""}, llm)
Defensive patterns

Strategy: try-catch

Validate before calling

import json
json.dumps(comparison_result["differences"])  # raises early if not serializable

Type guard

def differences_serializable(cr) -> bool:
    try:
        json.dumps(cr.get("differences"))
        return True
    except TypeError:
        return False

Try / catch

from scrapegraphai.utils.code_error_analysis import AnalysisError
try:
    semantic_focused_analysis(state, comparison_result, llm_model)
except AnalysisError as e:
    logger.error("semantic analysis failed: %s", e)
    raise

Prevention

When it happens

Trigger: semantic_focused_analysis / semantic_comparison_loop invoked with failing llm_model (auth, quota, network) or when json.dumps of comparison_result['differences'] raises because differences is not JSON-serializable.

Common situations: differences containing non-serializable objects (e.g. LangChain return objects), provider rate limits, invalid API key, timeouts on long code snippets.

Related errors

AI-assisted analysis of ScrapeGraphAI/Scrapegraph-ai@532dfffbf6 (2026-08-28). Data as JSON: /api/errors/7b021cec08355d47. Report an issue: GitHub.