kai/invoice-master-poc-v2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix: Enable substring matching for OCR, InvoiceNumber, Bankgiro, Plusgiro

Browse Source 
Previously substring matching was only enabled for date fields, causing
OCR values embedded in longer tokens like "Fakturanummer: 2465027205"
to not be matched.

Changes:
- Extended Strategy 4 (substring match) to numeric fields
- Updated _find_substring_matches to support OCR, InvoiceNumber, Bankgiro, Plusgiro

This should significantly improve match rates for these fields.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Yaojia Wang
2026-01-10 17:49:27 +01:00
parent 8938661850
commit dd69fbe9ed
1 changed files with 12 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
src/matcher/field_matcher.py
View File

@@ -106,8 +106,9 @@ class FieldMatcher:
fuzzy_matches = self._find_fuzzy_matches(page_tokens, value, field_name)
matches.extend(fuzzy_matches)
# Strategy 4: Substring match (for dates embedded in longer text)
if field_name in ('InvoiceDate', 'InvoiceDueDate'):
# Strategy 4: Substring match (for values embedded in longer text)
# e.g., "Fakturanummer: 2465027205" should match OCR value "2465027205"
if field_name in ('InvoiceDate', 'InvoiceDueDate', 'InvoiceNumber', 'OCR', 'Bankgiro', 'Plusgiro'):
substring_matches = self._find_substring_matches(page_tokens, value, field_name)
matches.extend(substring_matches)
@@ -240,16 +241,19 @@ class FieldMatcher:
"""
Find value as a substring within longer tokens.
Handles cases like 'Fakturadatum: 2026-01-09' where the date
is embedded in a longer text string.
Handles cases like:
- 'Fakturadatum: 2026-01-09' where the date is embedded
- 'Fakturanummer: 2465027205' where OCR/invoice number is embedded
- 'OCR: 1234567890' where reference number is embedded
Uses lower score (0.75) than exact match to prefer exact matches.
Only matches if the value appears as a distinct segment (not part of a number).
Uses lower score (0.75-0.85) than exact match to prefer exact matches.
Only matches if the value appears as a distinct segment (not part of a larger number).
"""
matches = []
# Only use for date fields - other fields risk false positives
if field_name not in ('InvoiceDate', 'InvoiceDueDate'):
# Supported fields for substring matching
supported_fields = ('InvoiceDate', 'InvoiceDueDate', 'InvoiceNumber', 'OCR', 'Bankgiro', 'Plusgiro')
if field_name not in supported_fields:
return matches
for token in tokens: