Tencent Cloud General Invoice OCR Advanced Edition

Problem Solved

In reimbursement, reconciliation, and tax archive workflows, VAT invoices often arrive as images, URLs, or PDFs. Manual entry of invoice codes, tax IDs, amounts, and line items is error-prone. This skill calls the Tencent Cloud advanced general invoice OCR API and converts invoice content into structured JSON, targeting special VAT invoices, general VAT invoices, electronic invoices, and fully digital invoices.

How It Works

• Input options: accepts ImageBase64 or ImageUrl; provide at least one, with ImageUrl taking precedence. Set IsPdf=true to enable PDF recognition.
• Field coverage: extracts invoice code, invoice number, issue date, check code, buyer/seller names and tax IDs, bank details, amounts, tax amounts, total amount, issuer, reviewer, remarks, and more.
• Line-item structure: maps service or goods name, specification, unit, quantity, unit price, amount without tax, tax rate, tax amount, and tax classification code into VatInvoiceItem for bookkeeping and validation.
• Fully digital invoice support: internally classifies the document first, then maps the result into consistent TextVatInvoice and VatInvoiceItem output.
• Coordinate assistance: some fields include Polygon points for locating values in the source image or supporting manual review.

Boundaries and Notes

• Encoded file size must be under 10M, and pixel size should be between 20 and 10000. Supported formats include PNG, JPG, JPEG, and PDF when enabled; GIF is not supported.
• PdfPageNumber applies only when IsPdf=true, defaulting to page 1.
• The default request rate limit is 10 requests/second, so batch jobs should be throttled.
• Errors indicate service not enabled, arrears, exhausted resource packages, blur, or decode failures; storing files in Tencent Cloud COS can reduce download failures.