https://parser.demo.geocode.earth/demo/#indianapolis%2C%20in <img width="949" height="749" alt="Image" src="https://github.com/user-attachments/assets/93a3516b-1ddd-4dbf-a8ed-b8636550b80b" /> The parser doesn't seem to recognize "IN" for Indiana as a region. You can see it on the live demo site, as linked above. Another example w/ a full random address I picked on the map: https://parser.demo.geocode.earth/demo/#1441%20Brown%20Ave%2C%20Whiting%2C%20IN%2046394 <img width="1010" height="193" alt="Image" src="https://github.com/user-attachments/assets/5fdbc802-c334-4294-a1ba-5bb23aa6d451" /> <img width="894" height="852" alt="Image" src="https://github.com/user-attachments/assets/302d06a2-b395-47f2-854a-62c2d8bf0937" />
Activity
missinglink commentedon Jul 8, 2025
Hmm interesting, it seems to work fine for other US state abbreviations.
I don't have the time to look into this right now, but I checked that the dictionaries do include the abbreviation:
https://github.com/pelias/parser/tree/master/resources/whosonfirst/dictionaries/region
missinglink commentedon Jul 8, 2025
Agh so I did end up having a quick look at it, the cause is this line due to the token
INbeing classified as an English Stop Word.One potential solution is to exclude
inas a stopword by adding a config file containing the negation!in:cat resources/pelias/dictionaries/libpostal/en/stopwords.txt !in================================================================ SOLUTIONS (1ms) ---------------------------------------------------------------- (0.93) ➜ [ { housenumber: '1441' }, { street: 'Brown Ave' }, { locality: 'Whiting' }, { region: 'IN' }, { postcode: '46394' } ]missinglink commentedon Sep 22, 2025
This was resolved in #195 by adding
!intoresources/pelias/dictionaries/libpostal/en/stopwords.txtSee also: #195 (review)
feat(fr/es/en): parser improvements for French/Spanish (pelias#195)