Docs

Reference

Release Notes

Chatbot SDK - REST API

REST API Reference

Complete reference for every HTTP endpoint exposed by the four pdf-autofillr module servers. All requests and responses use JSON. Each module runs as an independent FastAPI server on its own port — or inprocess inside your application.

4 serversPorts: 8000 / 8001 / 8002 / 8003Content-Type: application/jsonNo auth by default

ModuleServer commandDefault port

chatbotchatbot-server8000

doc_uploaddoc-upload-server8001

mapperpdf-mapper-server8002

ragragpdf-server8003

MethodPathPurpose

POST/chatbot/chatSend a message - one call per conversation turn

GET/chatbot/session/{uid}/{sid}Get final collected JSON when session complete

GET/chatbot/session/{uid}/{sid}/fill-reportGet field fill statistics for a completed session

DELETE/chatbot/session/{uid}/{sid}Delete all files for a session

GET/healthServer health check - storage + filler mode

GET/API info - version and endpoint map

chatbot-server

Chatbot Server — Port 8000

Start with chatbot-server --host 0.0.0.0 --port 8000. All endpoints are prefixed with /chatbot.

Send a message

The main endpoint. Call once per conversation turn. Send an empty string for message on the first turn to receive the greeting. The bot processes the message, advances the conversation state machine, and returns its reply. When session_complete is true, filled_data contains the complete flat dict of all collected fields.

Get session data

Returns the final collected data for a completed session. Only call this after session_complete = true has appeared in a chat response. Returns HTTP 404 if the session does not exist or has not reached COMPLETE state yet.

Get fill report

Returns field fill statistics for a completed session. Add ?format=text to receive a human-readable text report instead of JSON.

Delete session

Permanently deletes all files for a session - state, conversation log, output, fill report, and PDF workflow logs. This action is irreversible. For S3 storage, all associated S3 objects are deleted.

Health check

Returns server health status. Use this to verify the server is running and confirm which storage backend and PDF filler mode are active. Does not require any authentication.

API info

Returns the SDK version number and a map of all available endpoints. Useful for verifying the installed version and discovering endpoint paths programmatically.

doc-upload-server

Doc Upload Server — Port 8001

Start with doc-upload-server --host 0.0.0.0 --port 8001. Accepts document uploads (PDF, Word, Excel) and extracts field values using an LLM, then optionally fills a blank PDF via the mapper module.

MethodPathPurpose

POST/doc-upload/uploadUpload a document and start field extraction (returns job_id)

GET/doc-upload/job/{job_id}Get extraction status and output for a job

GET/doc-upload/job/{job_id}/reportGet field fill report for a completed job

GET/healthServer health check

CLI equivalent: doc-upload-cli --document path/to/file.pdf --schema configs/form_keys.json --report. The --report flag prints a fill summary after extraction.

pdf-mapper-server

Mapper Server — Port 8002

Start with pdf-mapper-server --host 0.0.0.0 --port 8002. Enable by setting MAPPER_API_URL=http://localhost:8002 in .env. By default, mapper runs inprocess.

MethodPathPurpose

POST/mapper/extractExtract fillable fields from a blank PDF

POST/mapper/mapSemantically map collected data to extracted PDF fields

POST/mapper/embedEmbed mapped values into the PDF

POST/mapper/fillFill a blank PDF with collected data (run-all in one call)

GET/healthServer health check

CLI equivalent: pdf-mapper run-all --pdf data/input/blank_form.pdf --data collected.json. Individual pipeline steps can be run separately: extract, map, embed, fill.

ragpdf-server

RAG Server — Port 8003

Start with ragpdf-server (or uvicorn ragpdf.entrypoints.fastapi_app:app --port 8003). Enable by setting RAG_MODE=http and RAG_API_URL=http://localhost:8003 in .env.

MethodPathPurpose

POST/rag/predictGet field value predictions for a new session

POST/rag/feedbackSubmit post-fill corrections to update the vector database

GET/rag/metrics/globalGlobal prediction accuracy metrics across all sessions

GET/rag/metrics/categoryMetrics filtered by field category

GET/rag/system-infoVector database status, embedding backend, and configuration

POST/rag/init-vectorsRe-initialise the vector database from a source file

GET/healthServer health check

CLI equivalents: ragpdf system-info, ragpdf metrics --type global, ragpdf feedback --user u1 --session s1 --pdf p1 --errors errors.json, ragpdf init-vectors --force.