Extract spec sections

Section content extraction · model content-extractor

POST/v1/specs/content-extract

Queue extraction of section text (and optional LLM-structured requirement fields) for CSI-style sections from the same PDF as a completed TOC parse job. Job model: content-extractor. X-API-Key and Content-Type: application/json; response is 202 with a new job id, then poll GET /v1/jobs/{job_id}.

Returns: A job to check202 Accepted
Usage: Document indexingper page of the sections extracted
Results in: Section contentresult.sections
Auth: API keyX-API-Key header

Prerequisite

Needs a completed TOC jobRequires a toc-parser job for your account with status complete. Create it with POST /v1/specs/parse/document. The canonical path is /v1/specs/parse/document — not /v1/specs/parse.

How it runs

Send a completed TOC job_id
Get a job_id
Check your results or receive a webhook

Request

Headers: X-API-Key, Content-Type: application/json. Body: ContentExtractRequest.

Field	Description
`job_id`string · UUIDRequired	Must be a completed toc-parser job id for this account.
`section_codes`string[]Optional	Explicit 6-digit section codes.
`division_codes`string[]Optional	Expands to all 6-digit section codes under each division using result.divisions on the TOC job. Sections are pulled only from divisions present in the stored TOC result.
`webhook_url`string · URLOptional	Worker delivers on developer, pro, and enterprise only.

Header	Description
`X-API-Key`Required	Your API key. A missing or invalid key returns 401.
`Content-Type`Required	`application/json`

Section codesNon-digits stripped; first 6 digits must form a full code (e.g. "10 29 00" → 102900). Invalid input → 422 INVALID_SECTION_CODE.

Division codesNon-digits stripped; last 2 digits padded to match division (e.g. 01, 1). Sections are taken only from divisions present in the stored TOC result. After expansion, codes are deduplicated (order preserved). If nothing resolves → 422 NO_SECTIONS_RESOLVED.

Responses

The response job_id is a new content-extractor job. The request body job_id is the source TOC job; when complete, result.source_job_id points back to that TOC job. Poll GET /v1/jobs/{job_id} — see GET /v1/jobs/{job_id}.

Field	Description
`job_id`string · UUID	New content-extractor job id.
`status`string	Always queued on this response.
`poll_url`string	Path only — prepend `https://api.anchorgrid.ai`.

When status is complete and model is content-extractor, result is returned per section.

Field	Description
`source_job_id`string · UUID	The toc-parser job uuid from the request.
`sections`array	One object per requested section.
`sections[].section_code`string	6-digit code.
`sections[].section_formatted`string	e.g. 10 29 00
`sections[].section_title`string \| null	Title from the table of contents; null when the request used document_id.
`sections[].division_code`string	Division portion.
`sections[].division_title`string \| null	Division title from the table of contents; null when the request used document_id.
`sections[].found`boolean	false with empty content if the section could not be located.
`sections[].content`string	Full extracted text when found.
`sections[].start_page`integer \| null	1-based PDF page where the section starts; null if not found.
`sections[].end_page`integer \| null	Last page carrying the section code in its header or footer.
`sections[].section_pages`integer[]	Pages carrying the section code in their header or footer.
`sections[].products`array	Structured when SpecNormalizer succeeds; else [].
`sections[].required_items`array	Same.
`sections[].compliance_requirements`array	Same.
`sections[].mounting_rules`array	Same.
`sections[].furnishing_rules`array	Same.
`sections_requested`integer	Count requested.
`sections_found`integer	Sections with content located.
`model_version`string	e.g. `content-extractor-v1.0.0`
`processing_time_ms`integer	Wall time for the task.

Structured arrays are filled when SpecNormalizer succeeds on non-empty content; on failure they default to empty arrays (logged server-side). GET /v1/jobs does not post-process content-extractor results (only door-detector is filtered on read).

Usage & rate limits

Each job counts as one document-indexing page. Allowance behavior is the same as parse/document: 402 when a trial allowance is used up, 429 when a monthly allowance is; RPM 429 from middleware. Rate limit: tier RPM (job-submit bucket).

Errors (synchronous, before queueing)

Codes appear in handler detail; Intelligence may still genericize JSON bodies.

Status	What it means
401	Missing or invalid API key.
402	Trial allowance used up.
404JOB_NOT_FOUND	No row with that id + account + model toc-parser.
422	Pydantic validation: missing both section_codes and division_codes, bad UUID, etc.
422JOB_NOT_COMPLETE	TOC job still queued, processing, or failed.
422JOB_NO_FILE	TOC job missing input_s3_key / bucket.
422INVALID_SECTION_CODE	Cannot normalize to 6 digits.
422NO_SECTIONS_RESOLVED	Expansion + explicit codes yielded nothing.
429	Monthly quota or RPM limit.

See Errors for HTTP exception mapping.

Typical flow

POST /v1/documents → document_id
POST /v1/specs/parse/document → TOC job_id
GET /v1/jobs/{toc_job_id} until complete → read result.divisions / section codes
POST /v1/specs/content-extract with body job_id (TOC job) and section_codes and/or division_codes
GET /v1/jobs/{content_job_id} until complete → read result.sections

Examples

curl -X POST https://api.anchorgrid.ai/v1/specs/content-extract \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "job_id": "7c9e6679-7425-40de-944b-e07fc1f90ae7",
    "section_codes": ["102900", "033000"],
    "division_codes": ["03"]
  }'

Response · 202

{
  "job_id": "9faf778a-9636-52f0-b66d-e29fe3b12cf8",
  "status": "queued",
  "poll_url": "/v1/jobs/9faf778a-9636-52f0-b66d-e29fe3b12cf8"
}

Completed job · result

{
  "source_job_id": "7c9e6679-7425-40de-944b-e07fc1f90ae7",
  "sections": [
    {
      "section_code": "102900",
      "section_formatted": "10 29 00",
      "section_title": "TOILET, BATH, AND LAUNDRY ACCESSORIES",
      "division_code": "10",
      "division_title": "SPECIALTIES",
      "found": true,
      "content": "…full extracted text…",
      "start_page": 412,
      "end_page": 418,
      "section_pages": [
        412,
        413,
        414,
        418
      ],
      "products": [],
      "required_items": [],
      "compliance_requirements": [],
      "mounting_rules": [],
      "furnishing_rules": []
    }
  ],
  "sections_requested": 1,
  "sections_found": 1,
  "model_version": "content-extractor-v1.0.0",
  "processing_time_ms": 45000
}

Cookies & analytics