clavinclient.py

import json

import geojson
import requests


class ClavinClient(object):
    def __init__(self, clavin_endpoint=None):
        if clavin_endpoint is None:
            #self.clavin_endpoint = "http://54.186.35.170:9090/api/v0/geotag"
            self.clavin_endpoint = "http://127.0.0.1:9090/api/v0/geotag"
            # TODO: Remove this eventually.. replace with our public endpoint when it's on 24/7?
            # Even then.. come from a config file somewhere that's user-overridable.

    @staticmethod
    def uniq(list):
        return {}.fromkeys(list).keys()

    def geocode_dynadocument(self, doc):
        feature = geojson.Feature()
        headers = {'Content-Type': 'text/plain'}
        # Send document.contents via POST to CLAVIN
        try:
            resp_json = requests.post(self.clavin_endpoint, headers=headers, data=doc.contents.encode('utf-8'))
        except AttributeError:
            return None

        # Load from JSON
        clavinresp = json.loads(resp_json._content, "utf-8")

        # Get resolved locations
        places = list()
        point_list = list()
        for location in clavinresp['resolvedLocations']:
            try:
                places.append(location['matchedName'])
                coords = (location['geoname']['longitude'], location['geoname']['latitude'])
                point_list.append(coords)
            except:
                pass
        # FIXME: hardcoded
        kvasir_prefix = "http://127.0.0.1:5000/articles/"
        # TODO: Extend this to support some notion of weighting vice just uniquing the places.
        # E.g. if it mentions a place 10 times it's probably a stronger relation
        point_list = self.uniq(point_list)
        places = self.uniq(places)
        if len(point_list) > 1:
            feature.geometry = geojson.MultiPoint(point_list)
        elif len(point_list) == 1:
            feature.geometry = geojson.MultiPoint([point_list[0]])
        else:
            feature.geometry = geojson.MultiPoint([geojson.Point(coordinates=(0, 0))])
        doc_url = kvasir_prefix + doc.dockey
        feature.properties['title'] = doc.title
        feature.properties['url'] = doc_url
        feature.properties['date'] = unicode(doc.date)
        feature.properties['dskey'] = doc.dskey
        feature.properties['places'] = ", ".join(places)
        print feature.properties['places']
        #feature.properties = {'title': document.title, 'link': document.url, 'places': places,
        #                      'name': '<A href=\"%s\">%s</A>' % (document.url, document.title)}
        return feature

    def geocode_document(self, document):
        feature = geojson.Feature()
        feature.properties = list()
        headers = {'Content-Type': 'text/plain'}
        # Send document.contents via POST to CLAVIN
        resp_json = requests.post(self.clavin_endpoint, headers=headers, data=document.contents)

        # Load from JSON
        clavinresp = json.loads(resp_json._content, "utf-8")
        # Get resolved locations

        places = list()
        point_list = list()
        for location in clavinresp['resolvedLocations']:
            try:
                places.append(location['matchedName'])
                coords = (location['geoname']['longitude'], location['geoname']['latitude'])
                point_list.append(coords)
            except:
                pass
        # TODO: Extend this to support some notion of weighting vice just uniquing the places.
        # E.g. if it mentions a place 10 times it's probably a stronger relation
        point_list = self.uniq(point_list)
        places = self.uniq(places)
        if point_list:
            feature.geometry = geojson.MultiPoint(point_list)
        else:
            feature.geometry = geojson.Point(coordinates=(0, 0))
        feature.properties = {'title': document.title, 'link': document.link, 'places': places,
                              'name': '<A href=\"%s\">%s</A>' % (document.link, document.title)}
        return feature