diff options
Diffstat (limited to 'parsers/logbooks.py')
| -rw-r--r-- | parsers/logbooks.py | 10 |
1 files changed, 10 insertions, 0 deletions
diff --git a/parsers/logbooks.py b/parsers/logbooks.py index 224b2cd..012d771 100644 --- a/parsers/logbooks.py +++ b/parsers/logbooks.py @@ -14,6 +14,7 @@ import csv import re import datetime import os +from fuzzywuzzy import fuzz from utils import save_carefully @@ -42,6 +43,15 @@ def GetTripPersons(trippeople, expedition, logtime_underground): print(" - No name match for: '%s'" % tripperson) message = "No name match for: '%s' in year '%s'" % (tripperson, expedition.year) models.DataIssue.objects.create(parser='logbooks', message=message) + print(' - Lets try something fuzzy') + fuzzy_matches = {} + for person in GetPersonExpeditionNameLookup(expedition): + fuzz_num = fuzz.ratio(tripperson.lower(), person) + if fuzz_num > 50: + #print(" - %s -> %s = %d" % (tripperson.lower(), person, fuzz_num)) + fuzzy_matches[person] = fuzz_num + for i in sorted(fuzzy_matches.items(), key = lambda kv:(kv[1]), reverse=True): + print(' - %s -> %s' % (i[0], i[1])) res.append((personyear, logtime_underground)) if mul: author = personyear |