Document table extraction

POST /api/document-table-extraction5 quota units per call · cache hits free

Find the rows and cells of a printed/scanned table using OpenCV morphological line detection. Returns JSON cell rectangles ready for downstream OCR.

Detects the cells of a **ruled-grid table** in a document image using OpenCV morphological line detection. The pipeline: 1. Convert to grayscale + adaptive-threshold (binarise). 2. Extract horizontal lines (morph-open with a long horizontal kernel) and vertical lines (long vertical kernel). 3. Find connected components on `~(horiz ∪ vert)` — each component is a candidate cell. 4. Filter by area; sort top→bottom, then group rows by y-clustering; sort each row left→right. **Output:** JSON with `rows` (array of rows, each an array of cell rectangles) plus a flat `cells` list. Use the returned boxes to crop each cell and feed it through `document-ocr` for text extraction. When `output=annotated-png` is set, returns a PNG with detected cells outlined in red — useful for visually verifying the detection before kicking off a per-cell OCR loop. **Limitations:** - Only works on tables with **visible grid lines**. Borderless tables (e.g. invoices with whitespace-separated columns) need a different algorithm — use ML table-detection for those. - Tilted scans should be deskewed first via `image-deskew`.

Inputs

Name	Type	Default	Description
image*	file	—	Document image bytes (PNG/JPEG).
minCellArea	number (16…1000000)	200	Minimum cell area in pixels² — filters out noise components.
maxCellArea	number (16…100000000)	1000000	Maximum cell area in pixels² — filters out the whole-page background component.
lineKernelDivisor	number (5…200)	30	Line-extraction kernel length = image_dim / divisor. Smaller = more permissive (detects shorter lines).
rowClusterTolerance	number (1…200)	12	Pixels of y-difference within which cells are considered the same row.
output	enum (json \| annotated-png)	"json"	JSON cells (default) or a PNG with cell rectangles drawn for visual debug.

Response

Modes: json, binary. Cache: yes (24h TTL).

Every response carries x-cache (HIT / MISS / BYPASS), x-cache-signature (stable across identical inputs), and the rate-limit headers listed below.

Quota & limits

Cost per call	5 units against the 10,000-unit monthly quota. Cache hits are free — only cache misses decrement.
Burst limit	200 requests per rolling minute, shared across all tools.
Max request size	`10 MiB` (base64 inflates file payloads ~33% — the limit applies to the decoded bytes).
Max response size	`25 MiB`
Timeout	30s wall clock.
Rate-limit headers	`X-RateLimit-Limit`, `X-RateLimit-Remaining`, `X-RateLimit-Burst-Limit` on every response; `X-Quota-Warning` once 80% of the monthly quota is spent; `Retry-After` on 429s.

Errors

Errors are JSON with an error string; schema failures add a details object (Zod's flattened field errors).

Status	When	Body
400	Input fails the tool's schema (wrong type, missing required field, malformed base64) or the file bytes can't be parsed as the expected format.	{ "error": "invalid inputs", "details": { "fieldErrors": { … } } }
401	Missing or invalid X-Api-Key / X-Account-Id headers (or no session when called from the browser).	{ "error": "unauthorized" }
404	Unknown tool id, or calling /api/* on the website hostname instead of api.xtract.bot.	{ "error": "not found" }
413	Request body over the tool's max request size, or an image header declaring more megapixels than the tool's pixel budget.	{ "error": "…exceeds maxRequestBytes…" }
415	Content-Type isn't application/json on the json-body transport.	{ "error": "unsupported content-type; expected application/json" }
429	Monthly quota or the 200/min burst limit exhausted. Check Retry-After and the X-RateLimit-* headers.	{ "error": "monthly quota exceeded", "monthRemaining": 0, … }
5xx	Conversion engine failure. Safe to retry; if it persists, the input is hitting a bug — please report it.	{ "error": "…" }

Code samples

Built from the default example.

# Download or substitute the example input:
#   curl -O https://xtract.bot/examples/document-split-pages/sample.png
IMAGE=$(base64 -w0 < sample.png)

curl -X POST https://api.xtract.bot/api/document-table-extraction \
  -H "Content-Type: application/json" \
  -H "Accept: application/json" \
  -H "X-Account-Id: $XTRACT_ACCOUNT_ID" \
  -H "X-Api-Key: $XTRACT_API_KEY" \
  -d '{
  "minCellArea": 200,
  "lineKernelDivisor": 30,
  "rowClusterTolerance": 12,
  "output": "json",
  "image": "'"$IMAGE"'"
}'