#!/usr/bin/env sh
set -eu
: "${PDFSCRIBE_API_BASE:?set to https://api.pdfscribe.ai/v1}"
: "${PDFSCRIBE_API_KEY:?set a scoped bearer key}"
: "${PDFSCRIBE_FILE:?set a PDF or image path}"
command -v jq >/dev/null
poll_timeout=${PDFSCRIBE_POLL_TIMEOUT_SECONDS:-300}
case "$poll_timeout" in ''|*[!0-9]*) echo 'PDFSCRIBE_POLL_TIMEOUT_SECONDS must be 1 to 3600' >&2; exit 1;; esac
if [ "$poll_timeout" -lt 1 ] || [ "$poll_timeout" -gt 3600 ]; then echo 'PDFSCRIBE_POLL_TIMEOUT_SECONDS must be 1 to 3600' >&2; exit 1; fi
request_timeout=30
size=$(wc -c < "$PDFSCRIBE_FILE" | tr -d ' ')
content_type=${PDFSCRIBE_CONTENT_TYPE:-application/pdf}
upload_idempotency_key="quickstart-upload-$(date +%s)-$$"
api() {
  if curl -fs --globoff --max-time "$request_timeout" -H "Authorization: Bearer $PDFSCRIBE_API_KEY" "$@"; then return 0; else curl_status=$?; fi
  if [ "$curl_status" -eq 28 ] && [ "${polling_budget_limited:-false}" = true ]; then
    printf 'Polling timed out; resume checking the existing extraction ID: %s\n' "$extraction_id" >&2
  else
    echo 'API request failed; check the existing extraction before retrying.' >&2
  fi
  return "$curl_status"
}
create=$(api -H 'Content-Type: application/json' -H "Idempotency-Key: $upload_idempotency_key" -d "{\"content_type\":\"$content_type\",\"size_bytes\":$size}" "$PDFSCRIBE_API_BASE/uploads")
url=$(printf %s "$create" | jq -r '.upload.url')
method=$(printf %s "$create" | jq -r '.upload.method')
upload_id=$(printf %s "$create" | jq -r '.id')
# Escape every value for curl config, keeping signed URLs out of process arguments.
# The bearer key is never forwarded to the signed storage destination.
printf %s "$create" | jq -r --arg file "$PDFSCRIBE_FILE" '
  "url = " + (.upload.url | @json),
  "request = " + (.upload.method | @json),
  "upload-file = " + ($file | @json),
  (.upload.headers | to_entries[] | .key as $key | .value[] | "header = " + (($key + ": " + .) | @json))
' | curl -fs --globoff --max-time 30 --config - >/dev/null || { echo 'Signed upload failed.' >&2; exit 1; }
api -X POST "$PDFSCRIBE_API_BASE/uploads/$upload_id/complete" >/dev/null
idempotency_key="quickstart-$(date +%s)-$$"
job=$(api -H 'Content-Type: application/json' -H "Idempotency-Key: $idempotency_key" -d "{\"upload_id\":\"$upload_id\",\"mode\":\"ocr\",\"pages\":\"all\",\"output\":{\"layout\":\"ordered\",\"include\":[\"text\",\"tables\"],\"table_format\":\"markdown\",\"confidence\":\"block\"}}" "$PDFSCRIBE_API_BASE/extractions")
extraction_id=$(printf %s "$job" | jq -r .id)
printf 'Accepted extraction ID: %s\n' "$extraction_id" >&2
deadline=$(( $(date +%s) + poll_timeout ))
while :; do
  remaining=$(( deadline - $(date +%s) ))
  if [ "$remaining" -le 0 ]; then printf 'Polling timed out; resume checking the existing extraction ID: %s\n' "$extraction_id" >&2; exit 1; fi
  request_timeout=$remaining
  polling_budget_limited=true
  if [ "$request_timeout" -gt 30 ]; then request_timeout=30; polling_budget_limited=false; fi
  status=$(api "$PDFSCRIBE_API_BASE/extractions/$extraction_id")
  state=$(printf %s "$status" | jq -r .status)
  case "$state" in
    succeeded|partially_succeeded) break;;
    failed|cancelled|expired) printf 'Extraction ended with status: %s\n' "$state" >&2; exit 1;;
  esac
  sleep 1
done
polling_budget_limited=false
request_timeout=30
api "$PDFSCRIBE_API_BASE/extractions/$extraction_id/result?view=ordered&include=text,tables&table_format=markdown&confidence=block" | jq .
api "$PDFSCRIBE_API_BASE/credits" | jq .
api "$PDFSCRIBE_API_BASE/usage?page=1" | jq .