Merge pull request #97 from PeARSearch/96-pearslocal-urls-from-remote…

…-instances-dont-work

Solve issue with pearslocal URLs from remote instances

app/search/cross_instance_search.py

@@ -114,8 +114,8 @@ def get_cross_instance_results(query, instances):
     best_instances = get_best_instances(query, 'en', instances, M, top_k=2)
     results = {}
     headers = {'User-Agent': app.config['USER-AGENT']}
-    for i in best_instances:
-        url = join(i["url"], 'api', 'search?q='+query)
+    for instance in best_instances:
+        url = join(instance["url"], 'api', 'search?q='+query)
         req_success = False
         try:
             t_before = time()
@@ -131,21 +131,32 @@ def get_cross_instance_results(query, instances):
             json_result = resp.json()['json_list']
             # legacy code for older instances
             if type(json_result) is list:
-                r = json_result[1]
+                remote_results = json_result[1]
             # up-to-date instances
             else:
-                r = json_result
+                remote_results = json_result
         else:
             print(f"Got non-200 status code from {url}")
-            r = {}
+            remote_results = {}
+
+        remote_results_updated = {}
+        for url, result_data in remote_results.items():
+            result_data_updated = {k: v for k, v in result_data.items()}
+            result_data_updated["x_instance_info"] = instance
+            # make sure pearslocal URLs point to the remote instance
+            remote_results_updated[url] = result_data_updated
+            if result_data["url"].startswith("pearslocal"):
+                del remote_results_updated[url]
+                url = join(instance["url"], "api", "get?url=") + result_data["url"]
+                result_data_updated["url"] = url
+                result_data_updated["share"] = url
+                remote_results_updated[url] = result_data_updated
 
-        for url, d in r.items():
-            r[url]["x_instance_info"] = i
             # The following is only temporary until all instances have been updated to return page scores
-            if 'score' not in d:
-                if any(w in d['title'] for w in query.lower().split()) or any(w in d['snippet'].lower() for w in query.lower().split()):
-                    r[url]['score'] = 2
+            if 'score' not in result_data_updated:
+                if any(w in result_data['title'] for w in query.lower().split()) or any(w in result_data['snippet'].lower() for w in query.lower().split()):
+                    result_data_updated['score'] = 2
                 else:
-                    r[url]['score'] = 0
-        results.update(r)
+                    result_data_updated['score'] = 0
+        results.update(remote_results_updated)
     return results