Update scripture_compare.py

scripture_compare.py

@@ -1,3 +1,4 @@
+import gradio
 import urllib.request, json
 from transformers.utils import logging
 logging.set_verbosity_error()
@@ -15,6 +16,8 @@ COMPARISON_VERSION = 'SEP' # Will use the Septuagint "SEP" version for most comp
 # types
 REFERENCES = 'Gen 1'
 
+model = SentenceTransformer("all-MiniLM-L6-v2")
+
 
 # Candidate helper class that holds a scripture, the comparison text and the resulting match score.
 class Candidate:
@@ -36,24 +39,10 @@ class Candidate:
     def __repr__(self):
         return str(self)
 
-def compare(model, candidate):
-    embeddings1 = model.encode(candidate.standardText, convert_to_tensor=True)
-    #print(embeddings1)
-    embeddings2 = model.encode(candidate.compareText, convert_to_tensor=True)
-    #print(embeddings2)
-
-    cosine_scores = util.cos_sim(embeddings1, embeddings2)
-    #print(cosine_scores)
-
-    candidate.score = cosine_scores[0][0]
-
-
-if __name__ == '__main__':
-
+def compare(reference):
     candidates = []
-    candidateMap = {}
-    # first connect to the api and get the standard translation
-    with urllib.request.urlopen(SEARCH_URL.format(urllib.parse.quote(REFERENCES), STANDARD_VERSION)) as url:
+    # Connect to the api and get the standard translation
+    with urllib.request.urlopen(SEARCH_URL.format(urllib.parse.quote(reference), STANDARD_VERSION)) as url:
         response = json.load(url)
         # print(response)
 
@@ -62,9 +51,9 @@ if __name__ == '__main__':
             candidate = Candidate(result['book'], result['chapter'], result['verse'], result['text'])
             candidates.append(candidate)
             candidateMap[candidate.reference()] = candidate
-
+            
     # Then fetch the comparison text
-    with urllib.request.urlopen(SEARCH_URL.format(urllib.parse.quote(REFERENCES), COMPARISON_VERSION)) as url:
+    with urllib.request.urlopen(SEARCH_URL.format(urllib.parse.quote(reference), COMPARISON_VERSION)) as url:
         response = json.load(url)
         # print(response)
 
@@ -78,8 +67,19 @@ if __name__ == '__main__':
 
             candidate.compareText = result['text']
 
-    model = SentenceTransformer("all-MiniLM-L6-v2")
-
+        
+    standardTexts = []
+    compareTexts = []
     for candidate in candidates:
-        compare(model, candidate)
-        print(candidate)
+        standardTexts.append(candidate.standardText)
+        compareTexts.append(candidate.compareText)
+        
+    embeddings1 = model.encode(standardTexts, convert_to_tensor=True)
+    embeddings2 = model.encode(compareTexts, convert_to_tensor=True)
+    cosine_scores = util.cos_sim(embeddings1, embeddings2)
+
+    for i in range(len(candidates)):
+        candidates[i].score = cosine_scores[i][i]
+        print(candidates[i])
+
+    return candidates