datawhalechina/hello-agents · error · ValueError

Selected table_id '{requested_table_id}' was not found in th

Error message

Selected table_id '{requested_table_id}' was not found in the extracted candidate tables.

What it means

After pdfplumber extracts candidate tables, if a selected_table_id was supplied, ingest_document looks it up among the extracted records; no match raises ValueError "Selected table_id '...' was not found in the extracted candidate tables." The id must be one of the table_ids reported by a prior preview/ingestion run for that exact PDF.

Source

Thrown at Co-creation-projects/healer-666-Academic-Data-Agent/src/data_analysis_agent/document_ingestion.py:351

        raise ValueError("V1 暂不支持 vision_fallback，请先使用文本型 PDF 或手动裁剪目标表格。")

    extracted_tables_dir = data_dir / "extracted_tables"
    cleaned_data_path = (data_dir / "cleaned_data.csv").resolve()
    parsed_document_path = (data_dir / "parsed_document.json").resolve()

    full_text, records = _extract_pdf_payload(
        source_path,
        max_pdf_pages=max_pdf_pages,
        max_candidate_tables=max_candidate_tables,
        extracted_tables_dir=extracted_tables_dir,
    )
    background_literature_context = _extract_background_context(full_text)
    requested_table_id = str(selected_table_id or "").strip()
    requested_record = None
    if requested_table_id:
        requested_record = next((record for record in records if record.table_id == requested_table_id), None)
        if requested_record is None:
            raise ValueError(
                f"Selected table_id '{requested_table_id}' was not found in the extracted candidate tables."
            )
        if not requested_record.numeric_columns:
            raise ValueError(
                f"Selected table_id '{requested_table_id}' does not contain any numeric columns and cannot be analyzed."
            )
    primary_record = requested_record or _select_primary_table(records)
    warnings: list[str] = []

    if primary_record is None:
        summary = (
            "PDF 解析失败：未提取到满足主表路由规则的结构化表格。"
            "V1 暂不支持复杂多表路由或扫描件恢复，请手动裁剪 PDF 或改上传目标表格。"
        )
        parsed_payload = _serialize_parsed_document(
            source_pdf=source_path,
            background_literature_context=background_literature_context,
            full_text_excerpt=full_text[:2000],

View on GitHub (pinned to 606a07d341)

Solutions

Re-run preview_pdf_tables (or check the candidate_table_summaries in the previous result) on the same file with the same page/table limits to get fresh table_ids.
Pass the exact id string returned by that run, unmodified.
If you don't need a specific table, omit selected_table_id so `_select_primary_table` picks one automatically.
Keep ingestion parameters (max_pdf_pages, max_candidate_tables) identical between the preview that produced the id and the ingest that consumes it.

Defensive patterns

Strategy: validation

Validate before calling

def selected_id_exists(preview_result, table_id: str) -> bool:
    ids = {c["table_id"] for c in preview_result.candidate_table_summaries}
    return table_id in ids

# one flow: preview then ingest with the SAME parameters
preview = preview_pdf_tables(pdf_path, max_pdf_pages=P, max_candidate_tables=T)
if not selected_id_exists(preview, chosen_id):
    raise ValueError(f"{chosen_id!r} not among {[c['table_id'] for c in preview.candidate_table_summaries]}")

Try / catch

try:
    result = ingest_document(pdf_path, data_dir, logs_dir, selected_table_id=tid,
                             max_pdf_pages=P, max_candidate_tables=T)
except ValueError as e:
    if "not found in the extracted candidate tables" in str(e):
        result = ingest_document(pdf_path, data_dir, logs_dir,
                                 max_pdf_pages=P, max_candidate_tables=T)  # auto-select primary table
    else:
        raise

Prevention

Always preview and select within the same parameter set (max_pdf_pages, max_candidate_tables).
Treat table_ids as ephemeral per-extraction: re-preview whenever the PDF or parameters change.
Store the preview result alongside the id you pass so you can refresh it in one step.

When it happens

Trigger: Passing a table_id from an earlier run against a different PDF, from a run with different max_pdf_pages/max_candidate_tables (which changes extraction and thus ids), a typo'd id, or an id whose table failed extraction this time.

Common situations: Multi-turn workflows where the user previews tables, edits parameters, then re-selects by stale id; PDFs regenerated upstream so table ordering/ids shift; ids copied with whitespace or case changes.

Related errors

AI-assisted analysis of datawhalechina/hello-agents@606a07d341 (2026-08-14). Data as JSON: /api/errors/f7f8506bf8abbde5. Report an issue: GitHub.