🧾 Rename Japan e-bookkeeping receipts and invoices with Gemini and Google Drive
⚡ 8 views · 🧾 Invoice Processing
Description
Quick overview
Helps Japanese businesses digitizing paper invoices and receipts under Japan’s Electronic Bookkeeping Act. Files in a Google Drive INPUT folder are read with Google Gemini OCR, then copied to OUTPUT with a compliant searchable filename (date_partner_amount_type) or to SKIP for human review. Originals are never modified.
How it works
- Runs manually to start processing.
- Lists up to the configured maximum number of files in a Google Drive INPUT folder and iterates through them one at a time.
- Downloads each file from Google Drive, detects its MIME type, base64-encodes the content, and prepares a Gemini request payload.
- Sends supported images and PDFs to the Google Gemini generateContent API to extract document count, transaction date, partner, amount, and document type as JSON.
- Builds a compliant filename in the format YYYYMMDD_partner_amount_type.ext when the response indicates a single valid document with required fields.
- Copies the original file to a Google Drive OUTPUT folder using the new filename, or copies it to a SKIP folder unchanged when the file type is unsupported or the extraction is ambiguous (non-document/multiple documents/missing fields).
Setup
- Create three Google Drive folders for INPUT, OUTPUT, and SKIP, then paste their folder IDs into the configuration step and set the per-run maxFiles value.
- Add a Google Drive OAuth2 credential and select it in all Google Drive steps.
- Add your Google Gemini API key (from Google AI Studio) to the x-goog-api-key header in the Gemini HTTP request step.
- Place invoices/receipts (supported: PDF, JPG/JPEG, PNG, WEBP, HEIC/HEIF, TIFF) into the INPUT folder and run the workflow.
Requirements
- A Google Drive account and credential
- A Google Gemini API key from Google AI Studio
Customization
- Change maxFiles in the Config node to process more files per run
- Edit the Gemini prompt to extract additional fields
- Enable Retry On Fail on the Gemini node to absorb temporary rate limits
🔗 Nodes Used
HTTP Request, Google Drive
📥 Import
Download workflow.json and import into n8n:
Workflow menu → Import from File