Skip to content

ezgh ocr process

Read the text in a document (PDF, PNG, JPEG, TIFF, WebP) with a processor.

Updated View as Markdown
ezgh ocr process <file|-> --processor <processor> [--pages 1-3] [-o text|json|yaml] [--out file] [--strict] [flags]

Process a document online: up to 10 MiB and 5 pages, answered within a minute. “-” reads the document from stdin. Its type comes from its content. ezgh waits up to 2 minutes for the answer (the upload and the processing), unless --timeout says otherwise.

The text of each page that succeeded is printed (with a “— page N —” line before each when there’s more than one); -o json prints the API’s answer as it came, -o yaml the same as YAML. --out writes the output to a file (readable only by you) instead of stdout.

A page can fail while others succeed (it timed out, or the model ran out of room): its error goes to stderr, the command still exits 0, and only the pages that succeeded are billed. --strict makes any failed page exit 1.

Processing isn’t idempotent: every request runs the model again and is billed again. So ezgh only retries when the service says it had no capacity to start (503 capacity_unavailable, after its Retry-After), never after an answer or a request that got none. A quota refusal exits 6.

Part of ezgh ocr.

Examples

ezgh ocr process receipt.jpg --processor receipts
ezgh ocr process scan.pdf --processor forms --pages 1-2 -o json --out scan.json
curl -s https://example.com/form.pdf | ezgh ocr process - --processor forms

Flags

Flag Type Default Description
--out string write the output to this file instead of stdout
--pages string only these pages, e.g. 1-3,5 (default: every page)
--processor string the processor, by ID, slug or name (required)
--strict boolean exit 1 if any page failed

This command also takes the global flags.

Navigation

Type to search…

↑↓ navigate↵ selectEsc close