Finish MVP for transliterations

code is reasonably well-written
checked that we get seemingly good data back
adding README
adding data

transliterations/src/__init__.py

@@ -0,0 +1,2 @@
+from wikidata_api_calls import *
+from find_entities import *

transliterations/src/defaults.py

@@ -0,0 +1 @@
+user_agent = "COVID-19 Digital Observatory, a Community Data Science Collective project. (https://github.com/CommunityDataScienceCollective/COVID-19_Digital_Observatory)"

transliterations/src/wikidata_api_calls.py

@@ -0,0 +1,35 @@
+# File defines functions for making api calls to find translations and transliterations for key terms.
+import mwapi
+import requests
+import sys
+import time
+from defaults import user_agent
+
+def get_wikidata_api():
+    session = mwapi.Session(host="https://wikidata.org/w/api.php", user_agent=user_agent)
+    return session
+
+def search_wikidata(session, term, *args, **kwargs):
+    search_results = session.get(action='query',
+                                 list='search',
+                                 srsearch=term,
+#                                 srqiprofile='popular_inclinks_pv',
+                                 srlimit='max',
+                                 srnamespace=0,
+                                 *args,
+                                 **kwargs)
+
+
+    query = search_results.get('query', None)
+    results = query.get('search', None)
+
+    if results is None:
+        raise mwapi.session.APIError(f"No results for query: {term}")
+
+    return results
+
+def run_sparql_query(q):
+    results = requests.get("https://query.wikidata.org/bigdata/namespace/wdq/sparql",params={"query":q,"format":"json"})
+    time.sleep(2)
+    return results
+

transliterations/src/wikidata_search.py

@@ -0,0 +1,65 @@
+# generate a list of wikidata entities related to keywords
+from os import path
+from sys import stdout
+from wikidata_api_calls import search_wikidata, get_wikidata_api
+import csv
+
+class Wikidata_ResultSet:
+    def __init__(self):
+        self.results = []
+
+    def extend(self, term, results):
+        self.results.extend([Wikidata_Result(term, result, i)
+                                    for i, result in enumerate(results)])
+
+    def to_csv(self, outfile=None):
+        if outfile is None:
+            of = stdout
+
+        else:
+            of = open(outfile,'w',newline='')
+
+        writer = csv.writer(of)
+        writer.writerow(Wikidata_Result.__slots__)
+        writer.writerows(map(Wikidata_Result.to_list, self.results))
+
+
+class Wikidata_Result:
+    # store unique entities found in the search results, the position in the search result, and the date
+    __slots__=['search_term','entityid','pageid','search_position','timestamp']
+
+    def __init__(self,
+                 term,
+                 search_result,
+                 position):
+
+        self.search_term = term.strip()
+        self.entityid = search_result['title']
+        self.pageid = int(search_result['pageid'])
+        self.search_position = int(position)
+        self.timestamp = search_result['timestamp']
+
+    def to_list(self):
+        return [self.search_term,
+                self.entityid,
+                self.pageid,
+                self.search_position,
+                self.timestamp]
+    
+def run_wikidata_searches(terms_file = '../data/input/base_terms.txt', outfile="../data/output/wikidata_search_results.csv"):
+
+    resultset = Wikidata_ResultSet()
+    for term in open(terms_file,'r'):
+        api = get_wikidata_api()
+        search_results = search_wikidata(api, term)
+        resultset.extend(term, search_results)
+
+    resultset.to_csv(outfile)
+
+
+    ## search each of the base terms in wikidata
+
+    # store unique entities found in the search results, the position in the search result, and the date
+
+if __name__ == "__main__":
+    run_wikidata_searches()

transliterations/src/wikidata_transliterations.py

@@ -0,0 +1,79 @@
+from wikidata_api_calls import run_sparql_query
+from itertools import chain, islice
+import csv
+from json import JSONDecodeError
+
+class LabelData:
+    __slots__ = ['entityid','label','langcode','is_alt']
+
+    def __init__(self, wd_res, entityid, is_alt):
+        obj = wd_res.get('label',None)
+        self.label = obj.get('value',None)
+        self.langcode = obj.get('xml:lang',None)
+        self.entityid = entityid
+        self.is_alt = is_alt
+
+    def to_list(self):
+        return [self.entityid,
+                self.label,
+                self.langcode,
+                self.is_alt]
+
+
+def GetAllLabels(in_csv, outfile, topN):
+
+    def load_entity_ids(in_csv, topN=5):
+        with open(in_csv,'r',newline='') as infile:
+            reader = csv.DictReader(infile)
+            for row in reader:
+                if int(row['search_position']) < topN:
+                    yield row["entityid"]
+
+    ids = set(load_entity_ids(in_csv, topN))
+
+    labeldata = chain(* map(GetEntityLabels, ids))
+
+    with open(outfile, 'w', newline='') as of:
+        writer = csv.writer(of)
+        writer.writerow(LabelData.__slots__)
+        writer.writerows(map(LabelData.to_list,labeldata))
+
+    
+def GetEntityLabels(entityid):
+
+    def run_query_and_parse(query, entityid, is_alt):
+        results = run_sparql_query(query % entityid)
+        try:
+            jobj = results.json()
+            res = jobj.get('results',None)
+            if res is not None:
+                res = res.get('bindings',None)
+            if res is None:
+                raise requests.APIError(f"got invalid response from wikidata for {query % entityid}")
+            for info in res:
+                yield LabelData(info, entityid, is_alt)
+
+        except JSONDecodeError as e:
+            print(e)
+            print(query % entityid)
+            
+
+    label_base_query = """
+    SELECT DISTINCT ?label WHERE {
+    wd:%s rdfs:label ?label;
+    }"""
+
+    altLabel_base_query = """
+    SELECT DISTINCT ?label WHERE {
+    wd:%s skos:altLabel ?label;
+    }"""
+
+    label_results = run_query_and_parse(label_base_query, entityid, is_alt=False)
+
+    altLabel_results = run_query_and_parse(altLabel_base_query, entityid, is_alt=True)
+
+    return chain(label_results, altLabel_results)
+        
+
+if __name__ == "__main__":
+    GetAllLabels("../data/output/wikidata_search_results.csv","../data/output/wikidata_entity_labels.csv", topN=20)