Skip to content

OSM data load correctly but API result is not correct #169

Open

Description

@geopamplona

Hi

I am building a full instance of Pelias using Dockers and following the documentation.

I have built the complete stack with the following pieces: libpostal, placeholder, api, pip, elasticsearch 5.6.12 and the OSM importer.


Here's what I did 😇

Libpostal load and run correctly without dependencies.
Placeholder :

  • Download store.sqlite3.gz from s3.amazonaws.com/pelias-data.nextzen.org/placeholder
  • Extract data
  • Download WhosonFirst data
  • Load all data correctly; I dont have errros
    Are these steps Ok for run placeholder? or I am missing some steps?

API runs OK and detect al services.

PIP:

  • Download whosonfirst data
  • Start service correctly

OSM importer:

  • Load PBF file correctly and the data is contrasted correctly against PIP
  • I see logs in PIP container; seem OK

At this point, I have load all OSM data in Elasticsearch correctly.


Here's what I got 🙀

I have data correctly load in Elastic

imagen

When I try request data to API, I have this result:
imagen

So, I have 2 questions:

  • What is the function of Placeholder and Libpostal? Should they be used together or have different purposes? or different applications?.
    In broad strokes I understand what each piece does, but I don't understand how libpostal intervenes or how important it is.
  • Data imported from OSM is loaded correctly, however the result is not as expected; At least from my ignorance.
    I expected a result like the following GeocodeEarth example or Mapzen
    imagen
    imagen

I see that the information loaded in the database, includes the street number, and makes sense, however the quality of the result is lower than in the examples.
Autocomplete text is waiting for a full address, however in the examples this does not happen, right?

On the other hand, it is necessary to cure the database to eliminate the information it stores, to achieve a result similar to the examples?

I understand that the example of GeocodeEarth is a complete example of a system in production, and is the maximum quality of results that I can achieve using Pelias?

If you think more information is needed, I can share more details

Thanks !

Versions:

  • Libpostal: pelias/libpostal-service:master-2019-08-06-809f79af5728af35aeb7c472497387b13c154cec
  • Placeholder: pelias/placeholder:master-2019-12-17-a05a26268c1684622f207eb95ab3b2f837a456e1
  • API: pelias/api:master-2019-11-04-fc999ffb06ce3ccca950d96f59c0ccc5c02dfe3e
  • PIP: pelias/pip-service:master-2019-12-18-24a6e31f42b07adec218a6f5a1aebf45105d681e
  • OSM importer: pelias/openstreetmap:master-2019-09-13-b1a69740ee6d34b98290d5554d80b692beddb7ad

Activity

  1. missinglink commented on Jan 21, 2020

    @missinglink
    Member

    Can you please post an example HTTP URL where the response differs on your version from what you are seeing elsewhere?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions