80dbf0f703
- Add profile source modules: apply_profile, extraction, extraction_loader, field_extractor, match_eval - Add profiles CLI subcommand (profiles_cmd.rs) - Update all 9 built-in profile YAMLs (invoice, receipt, contract, scientific_paper, slide_deck, form, bank_statement, legal_filing, book_chapter) - Add 50 invoice fixture PDFs - Add 2 receipt fixture PDFs Part of: pdftract-3a310 (Phase 7.10 coordinator)
81 lines
1.8 KiB
YAML
81 lines
1.8 KiB
YAML
# Receipt extraction profile
|
|
# Matches point-of-sale or purchase receipts with items and payment method
|
|
name: receipt
|
|
description: Point-of-sale or purchase receipt with items, payment method
|
|
priority: 45
|
|
|
|
match:
|
|
all:
|
|
- any:
|
|
- text_contains:
|
|
patterns: ["receipt", "store receipt", "register receipt", "transaction receipt"]
|
|
- text_contains:
|
|
patterns: ["total sold", "change due", "cash credit", "card payment"]
|
|
- structural:
|
|
has_table: true
|
|
has_form_field: false
|
|
has_math: false
|
|
page_count:
|
|
min: 1
|
|
max: 2
|
|
|
|
extraction:
|
|
reading_order: line_dominant
|
|
table_detection: default
|
|
readability_threshold: 0.5
|
|
include_invisible: false
|
|
include_headers_footers: false
|
|
force_ocr: false
|
|
min_block_chars: 0
|
|
|
|
fields:
|
|
merchant:
|
|
type: string
|
|
extraction:
|
|
region: top_quarter
|
|
pick: largest_font
|
|
parse: string
|
|
|
|
date:
|
|
type: date
|
|
extraction:
|
|
regex: "\\d{1,2}[/-]\\d{1,2}[/-]\\d{2,4}"
|
|
parse: date
|
|
|
|
total:
|
|
type: decimal
|
|
extraction:
|
|
regex: "([\\d,]+\\.\\d{2})"
|
|
near: ["Total", "Amount Due", "Balance"]
|
|
max_distance_pt: 80
|
|
parse: decimal
|
|
|
|
tax:
|
|
type: decimal
|
|
extraction:
|
|
regex: "([\\d,]+\\.\\d{2})"
|
|
near: ["Tax", "VAT"]
|
|
max_distance_pt: 80
|
|
parse: decimal
|
|
|
|
items:
|
|
type: array
|
|
extraction:
|
|
table_region: largest_table
|
|
schema:
|
|
- name: name
|
|
type: string
|
|
required: true
|
|
- name: quantity
|
|
type: decimal
|
|
required: false
|
|
- name: price
|
|
type: decimal
|
|
required: false
|
|
fallback: []
|
|
|
|
payment_method:
|
|
type: string
|
|
extraction:
|
|
regex: "(cash|credit|debit|visa|mastercard|amex|discover|check|cheque)"
|
|
parse: string
|