fix case sensitiveness on autocomplete endpoint

requirements.txt

@@ -10,3 +10,4 @@ psycopg2-binary==2.8.5
 requests==2.24.0
 SQLAlchemy==1.3.17
 Werkzeug==1.0.1
+logzero==1.5.0

views/autocomplete.py

@@ -1,4 +1,7 @@
+from logzero import logger
+
 from views import *
+from views.exceptions import PhenopolisException
 
 
 @application.route('/<language>/autocomplete/<query_type>/<query>')
@@ -7,66 +10,77 @@
 @application.route('/autocomplete/<query>')
 @requires_auth
 def autocomplete(query, query_type=''):
-   if query_type: query_type=query_type+':'
-   print(query_type, query)
-   patient_results=[]
-   gene_results=[]
-   hpo_results=[]
-   regex="%"+re.escape(query)+"%"
-   c=postgres_cursor()
-   if query_type in ['gene:','']: 
-       c.execute("select * from genes where gene_name_upper like '%s' limit 20"%regex)
-       headers=[h[0] for h in c.description]
-       gene_hits=[dict(zip(headers,r)) for r in c.fetchall()]
-       if query_type=='gene:':
-           gene_results = [x['gene_name_upper'] for x in gene_hits]
-       else:
-           gene_results = ['gene:'+x['gene_name_upper'] for x in gene_hits]
-   if query_type in ['phenotype:','']: 
-       c.execute("select * from hpo where hpo_name like '%s' limit 20"%regex)
-       headers=[h[0] for h in c.description]
-       hpo_hits=[dict(zip(headers,r)) for r in c.fetchall()]
-       if query_type=='phenotype:':
-           hpo_results = [x['hpo_name'] for x in hpo_hits]
-       else:
-           hpo_results = ['phenotype:'+x['hpo_name'] for x in hpo_hits]
-   if query_type in ['patient:','']:
-       c.execute("""
+
+    logger.debug("Autocomplete query '{}' and query type '{}'".format(query, query_type))
+
+    cursor = postgres_cursor()
+    regex_query = "%" + re.escape(query.upper()) + "%"
+    if query_type == 'gene':
+        results = _search_genes(cursor, regex_query)
+    elif query_type == 'phenotype':
+        results = _search_phenotypes(cursor, regex_query)
+    elif query_type == 'patient':
+        results = _search_patients(cursor, regex_query)
+    elif query_type == '':
+        results = ["gene:" + x for x in _search_genes(cursor, regex_query)] + \
+                  ["phenotype:" + x for x in _search_phenotypes(cursor, regex_query)] + \
+                  ["patient:" + x for x in _search_patients(cursor, regex_query)]
+    else:
+        message = "Autocomplete request with unsupported query type '{}'".format(query_type)
+        logger.error(message)
+        raise PhenopolisException(message)
+    cursor.close()
+
+    # removes possible duplicates and chooses 20 suggestions
+    suggestions = list(itertools.islice(list(set(results)), 0, 20))
+    return Response(json.dumps(suggestions), mimetype='application/json')
+
+
+def _search_patients(cursor, query):
+    cursor.execute("""
                select
                i.external_id, i.internal_id
                from individuals i,
                users_individuals ui
                where
                ui.internal_id=i.internal_id
                and
-               ui.user='%s'
+               ui.user='{user}'
                and
-               i.internal_id like '%s' limit 20
-               """%(session['user'],regex,))
-       patient_hits=[dict(zip([h[0] for h in c.description],r)) for r in c.fetchall()]
-       if query_type=='patient:':
-           patient_results = [x['internal_id'] for x in patient_hits]
-       else:
-           patient_results = ['patient:'+x['internal_id'] for x in patient_hits]
-   #c.execute('select * from variants where "#CHROM"=? and POS=? and REF=? and ALT=? limit 20',regex.split('-'))
-   #headers=[h[0] for h in c.description]
-   #variant_hits=[dict(zip(headers,r)) for r in c.fetchall()]
-   #variant_results = ['variant:'+x['variant_id'] for x in variant_hits]
-   results = list(set(patient_results+gene_results+hpo_results))
-   suggestions = list(itertools.islice(results, 0, 20))
-   return Response(json.dumps(suggestions),  mimetype='application/json')
+               i.internal_id like '{query}' limit 20
+               """.format(user=session['user'], query=query))
+    patient_hits = [dict(zip([h[0] for h in cursor.description], r)) for r in cursor.fetchall()]
+    return [x['internal_id'] for x in patient_hits]
 
-@application.route('/best_guess/<query>')
-@requires_auth
-def best_guess(query=''):
-     print(query)
-     if query.startswith('gene:'): return jsonify(redirect='/gene/{}'.format(query.replace('gene:','')))
-     elif query.startswith('patient:') or query.startswith('PH'): return jsonify(redirect='/individual/{}'.format(query.replace('patient:','')))
-     elif query.startswith('phenotype:'): return jsonify(redirect='/hpo/{}'.format(query.replace('phenotype:','')))
-     elif query.startswith('variant:'): return jsonify(redirect='/variant/{}'.format(query.replace('variant:','')))
-     elif '-' in query and len(query.split('-'))==4: return jsonify(redirect='/variant/{}'.format(query.replace('variant:','')))
-     return jsonify(message='Could not find search query'), 420
+
+def _search_phenotypes(cursor, query):
+    cursor.execute("select * from hpo where UPPER(hpo_name) like '{query}' limit 20".format(query=query))
+    headers = [h[0] for h in cursor.description]
+    hpo_hits = [dict(zip(headers, r)) for r in cursor.fetchall()]
+    return [x['hpo_name'] for x in hpo_hits]
 
 
+def _search_genes(cursor, query):
+    cursor.execute("select * from genes where gene_name_upper like '{query}' or other_names like '{query}' limit 20".
+                   format(query=query))
+    headers = [h[0] for h in cursor.description]
+    gene_hits = [dict(zip(headers, r)) for r in cursor.fetchall()]
+    # while the search is performed on the upper cased gene name, it returns the original gene name
+    return [x['gene_name'] for x in gene_hits]
 
 
+@application.route('/best_guess/<query>')
+@requires_auth
+def best_guess(query=''):
+    print(query)
+    if query.startswith('gene:'):
+        return jsonify(redirect='/gene/{}'.format(query.replace('gene:', '')))
+    elif query.startswith('patient:') or query.startswith('PH'):
+        return jsonify(redirect='/individual/{}'.format(query.replace('patient:', '')))
+    elif query.startswith('phenotype:'):
+        return jsonify(redirect='/hpo/{}'.format(query.replace('phenotype:', '')))
+    elif query.startswith('variant:'):
+        return jsonify(redirect='/variant/{}'.format(query.replace('variant:', '')))
+    elif '-' in query and len(query.split('-')) == 4:
+        return jsonify(redirect='/variant/{}'.format(query.replace('variant:', '')))
+    return jsonify(message='Could not find search query'), 420

views/exceptions.py

@@ -0,0 +1,5 @@
+
+
+class PhenopolisException(Exception):
+
+    pass