# Extract Information to Table

Parse material, extract fields, and write structured results to a database table. Use it to turn forms, resumes, or business documents into records.

## Prepare your resources

Prepare source material, JSON Schema, a result database, table name, and output volume. Check columns and types when using an existing table. The template creates a missing table from field definitions or reuses an existing table with the same name.

This template saves one combined extraction into a table. If each resume or receipt should become one row, validate a single file first, then design batch processing.

## Use the template

On the workflow creation page, start from a template, search for **Extract Information to Table**, and load it onto the canvas.

Complete the required fields in the parameter form.

| Setting | Purpose |
| --- | --- |
| Source material | Documents, forms, email, or folders to parse |
| Parsing tier | Controls parsing speed, quality, and cost |
| Extraction schema | JSON Schema defining fields to extract and write |
| Result database | Database for extracted records |
| Result table name | Reuse an existing table with this name in the selected database, or create it if absent |
| Output location | Catalog location for extraction results |

Save the parameters, then start a manual run. The system saves the current workflow before submitting execution.

## Complete processing flow

Arrows show execution order; configured bindings supply each step’s input.

![Extract Information to Table](../images/extraction-to-table-flow.svg)

## Configure fields and the table

| Setting | What to check |
| --- | --- |
| Extraction schema | Field names and types match target columns |
| Instructions | Meanings, formats, and missing-value rules |
| Layout-sensitive fields | Names of fields whose position matters |
| Result database and table | Correct destination; avoid an incompatible existing table |
| Output location | File destination, checked separately from table records |

Extraction uses `n_to_1`. It combines inputs into one JSON result and binds `result` to the table node's `raw_json`. The template has no automatic one-file-per-row loop.

Define structures for multi-value fields, such as education arrays. A schema accepting arrays does not mean any existing column can store them. Check the target and actual write result; see [Save to MOI Table](../nodes/data-io/save-table.md).

## Check table and file results

1. Inspect parsed content using one source file.
2. Verify extracted fields and values.
3. Check `database`, `table_name`, `written_columns`, and `affected_rows` on the writing node.
4. Compare records in the result table with the original.
5. Inspect saved results, parsed files, and final source registration.

Table writing precedes file saving and registration. If a later step fails, records may already exist; inspect them before rerunning.

## Example: store one resume

Define name, education, and work experience fields, including the structure of multiple entries. Select one resume, configure a dedicated result table, and run. Check completeness, written columns, and affected rows. See [Extract resume information](resume-extraction.md).

## Common problems

| Symptom | Check and action |
| --- | --- |
| Extraction succeeds but writing fails | Check the database, columns, and field types |
| Several files yield one result | Extraction combines inputs; per-file processing needs a different flow |
| Nested fields cannot be written | Check schema, column types, and accepted node inputs |
| A failed run left table records | Locate whether failure occurred after writing before repeating it |
| Only files are needed | Use [Document Information Extraction](document-extraction.md) |

## Related pages

- [Workflow templates](index.md)
- [Run and debug](../run-debug.md)