fixed functor-frames

src/backend_flask/app.py

@@ -235,28 +235,6 @@ def api_token():
 
 # FRAMES ----------------------------.
 
-def prepare_frames(ret_frames):
-    # append sentences
-    for frame in ret_frames:
-        unique_sids = {".".join(x.split(".")[:-1]): x for x in frame.tids}
-        # frame.sentences = []
-        frame.aggr_sent = {}
-        # sid, tid==hw
-        for sid, tid in unique_sids.items():
-            # hwl = vallex.get_token(tid)["lemma"]
-            hwl = frame.hw
-            tmp_idx = len(frame.sentences)
-            if hwl not in frame.aggr_sent:
-                frame.aggr_sent[hwl] = []
-            frame.aggr_sent[hwl].append(tmp_idx)
-    # return (n-frames, rendered template)
-    # json frames
-    json_ret = {"frames": []}
-    for frame in ret_frames:
-        json_ret["frames"].append(DC(frame.to_json()))
-    return json.dumps(json_ret)
-
-
 # input: hw, reduct_function
 @app.route("/api/frames")
 def api_get_frames():
@@ -280,8 +258,34 @@ def api_get_frames():
     frames = [x for x in frames if x.hw == hw]
 
     ret_frames = RF(frames, valdb[SENSEMAP_COLL])
-    return prepare_frames(ret_frames)
 
+    json_ret = {"frames": []}
+    for frame in ret_frames:
+        json_ret["frames"].append(frame.to_json())
+    return json.dumps(json_ret)
+    # return prepare_frames(ret_frames)
+
+def _aggregate_by_hw(ret_frames):
+
+    def _tid_to_lemma(tid, sentence):
+        # slow and hackish
+        for pair in sentence:
+            if pair[0] == tid:
+                return pair[1]["lemma"]
+        return None
+
+    # append sentences
+    for frame in ret_frames:
+        # unique_sids = {".".join(x.split(".")[:-1]): x for x in frame.tids}
+        frame.aggr_sent = {}  # map of headword: [sentence indexes]
+        # sid, tid==hw
+        for i, tid in enumerate(frame.tids):
+            # hwl = vallex.get_token(tid)["lemma"]
+            hwl = _tid_to_lemma(tid, frame.sentences[i])
+            if hwl not in frame.aggr_sent:
+                frame.aggr_sent[hwl] = []
+            frame.aggr_sent[hwl].append(i)
+    return ret_frames
 
 # input: functor, reduce_function
 @app.route("/api/functor-frames")
@@ -302,12 +306,17 @@ def api_get_functor_frames():
     for ent in cur:
         frames += frames_from_db_entry(ent)  # pre-process this step for prod TODO
 
-    for f in frames:
-        print(f.to_json())
+    # filter by relevant functor
+    frames = [x for x in frames if functor in x.get_functors()]
 
     # raw_frames = vallex.functors_index[functor]  # TODO
     ret_frames = RF(frames, valdb[SENSEMAP_COLL])
-    return prepare_frames(ret_frames)
+    ret_frames = _aggregate_by_hw(ret_frames)
+
+    json_ret = {"frames": []}
+    for frame in ret_frames:
+        json_ret["frames"].append(DC(frame.to_json()))
+    return json.dumps(json_ret)
 
 # FRAMES ----------------------------^