From 48672c979e514ccd28ef3426c37a11de135cd8aa Mon Sep 17 00:00:00 2001
From: Bo Peng <bpeng@smsassist.com>
Date: Fri, 15 Jun 2018 12:40:39 -0500
Subject: [PATCH] Update SpaCy server python script

---
 BotSharp.Core/Engines/SpaCy/server.py | 59 ++++++++++++++++++++++++---
 1 file changed, 54 insertions(+), 5 deletions(-)

diff --git a/BotSharp.Core/Engines/SpaCy/server.py b/BotSharp.Core/Engines/SpaCy/server.py
index e57cec2c..37d47bcf 100644
--- a/BotSharp.Core/Engines/SpaCy/server.py
+++ b/BotSharp.Core/Engines/SpaCy/server.py
@@ -1,6 +1,7 @@
 from bottle import route, run, request
 from spacy.tokenizer import Tokenizer
 from spacy.pipeline import EntityRecognizer
+from spacy.pipeline import TextCategorizer
 import spacy
 
 nlp = spacy.load('en')
@@ -19,7 +20,7 @@ def tokenize():
         print(token)
         list.append({'text': token.text, 'offset': token.idx})
     return {'tokens': list}
-    
+
 @route('/featurize')
 def tokenize():
     doc = nlp(request.query.text)
@@ -29,7 +30,7 @@ def tokenize():
         print(vec)
         list.append(str(vec.real))
     return {'vectors': list}
-    
+
 @route('/entitize')
 def entitize():
     doc = nlp(request.query.text)
@@ -38,6 +39,54 @@ def entitize():
     for entity in doc.ents:
         print(entity)
         list.append(entity)
-    return {'entities': list}    
-    
-run(host='0.0.0.0', port=5005, debug=True)
\ No newline at end of file
+    return {'entities': list}
+
+@route('/textcategorizer', method='POST')
+def textcategorizer():
+    texts = request.json["Texts"]
+    golds = request.json["Golds"]
+    labels = request.json["Labels"]
+    print(labels)
+    train_data = []
+    for index in range(len(texts)):
+        tuple =(texts[index], golds[index])
+        train_data.append(tuple)
+
+    print("training data body is: {0}".format(train_data))
+    textcat = nlp.create_pipe('textcat')
+    nlp.add_pipe(textcat, last=True)
+    for label in labels:
+        textcat.add_label(label)
+    optimizer = nlp.begin_training()
+    for itn in range(2):
+        for doc, gold in train_data:
+            nlp.update([doc], [gold], sgd=optimizer)
+
+    textcat.to_disk('./textcat_try')
+
+
+
+    return {'ModelName':'textcat_try'}
+
+@route('/predict')
+def predict():
+    textcat = TextCategorizer(nlp.vocab)
+    textcat.from_disk('./textcat_try')
+
+    nlp.add_pipe(textcat, last=True)
+
+    doc = nlp(request.query.text)
+
+    #scores = textcat.predict([request.query.text])
+    #print(scores)
+
+    list = []
+    for label, confidence in doc.cats:
+        print(label)
+        list.append({'Label': label, 'Confidence': confidence})
+
+    print (list)
+
+    return {'Labels': list}
+
+run(host='0.0.0.0', port=5005, debug=True)