OCR.space

OCR.space

This connector reads text from images and PDFs with the OCR.space OCR API (by a9t9): receipts, invoices, delivery notes, scanned letters, screenshots and photos of documents, from a public URL. It can also turn a scan into a searchable PDF.

Setup (about 1 minute)

1. Request a free API key at https://ocr.space/ocrapi/freekey (it arrives by e-mail), or use the key of a PRO plan.
2. Paste it into OCR_SPACE_API_KEY. It is sent in the apikey header.
3. Install: the check reads a small sample image, which counts as one request.

Tools

• ocr_space_read_url: the text of every page of an image or PDF. Use is_table true for receipts and tables (keeps lines in reading order), scale true for small or blurry images, detect_orientation true for rotated photos.
• ocr_space_make_searchable_pdf: a searchable PDF of a scan, as a download link (SearchablePDFURL) that expires after about an hour, plus the text.

Engines: engine 2 (default) is fast, detects the language itself and handles numbers and special characters well. Engine 3 is the most accurate (200+ languages, handwriting, complex layouts) but slower and limited to 1,000 pages a month on the free plan. language auto works with both; set a code (ger, fre, ita, spa, chs, jpn...) when auto-detection picks the wrong one.

Files: the URL must be reachable from the internet without a login; a link to a private drive or e-mail attachment fails with E217 (could not download). Free plan: 1 MB per file, 3 PDF pages, 25,000 requests a month, 500 a day per IP. PRO: 5 MB and more pages.

Reading the answer: pages[].text is the text, with line breaks. OCRExitCode 1 means all pages were read, 2 some, 3 or 4 none; ErrorMessage says why. OCR.space reports processing errors with HTTP 200 and IsErroredOnProcessing true; the connector turns those into a failed call carrying that message.

Errors: 403 E555 means a wrong API key; 400 E400 no file given or E204 a wrong parameter; E217 a URL that cannot be downloaded; 429 or E500 the monthly or daily limit.