saadiiii commited on
Commit
aa8cb15
·
1 Parent(s): 76f1044

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +25 -25
app.py CHANGED
@@ -7,7 +7,7 @@ import pymongo
7
  import torch
8
  import spacy
9
  from spacy import displacy
10
- from pymongo import MongoClient
11
 
12
  from transformers import AutoTokenizer
13
  tokenizer = AutoTokenizer.from_pretrained("law-ai/InLegalBERT")
@@ -248,30 +248,30 @@ def judgtext_analysis(text):
248
  doc.ents = ents
249
 
250
  #logic for repository
251
- cluster = MongoClient("mongodb+srv://testuser:test123@ner-gradio.mgng1wv.mongodb.net/?retryWrites=true&w=majority")
252
- db = cluster["nerdb"]
253
- collection = db["named_entities"]
254
- content = displacy.render(doc, style='ent')
255
- extsoup = bs(content, 'html.parser')
256
-
257
- txtlist=[]
258
- entlist=[]
259
- for h in extsoup.findAll('div'):
260
- mark = h.findAll('mark')
261
- span = h.findAll('span')
262
- for i in mark:
263
- txt = i.find(text=True)
264
- txt = txt.replace("\n","");
265
- txt = txt.strip();
266
- txtlist.append(str(txt))
267
- for i in span:
268
- ent = i.find(text=True)
269
- entlist.append(str(ent))
270
- zipped = zip(txtlist, entlist)
271
- ziplist = list(zipped)
272
-
273
- post = {"api_docid":docid, "document_text":finaltext, "named_entities":ziplist, "entities_indices": finalentities}
274
- collection.insert_one(post)
275
 
276
  html = displacy.render(doc, style="ent", page = True)
277
  html = (
 
7
  import torch
8
  import spacy
9
  from spacy import displacy
10
+ # from pymongo import MongoClient
11
 
12
  from transformers import AutoTokenizer
13
  tokenizer = AutoTokenizer.from_pretrained("law-ai/InLegalBERT")
 
248
  doc.ents = ents
249
 
250
  #logic for repository
251
+ # cluster = MongoClient("mongodb+srv://testuser:test123@ner-gradio.mgng1wv.mongodb.net/?retryWrites=true&w=majority")
252
+ # db = cluster["nerdb"]
253
+ # collection = db["named_entities"]
254
+ content = displacy.render(doc, style='ent')
255
+ # extsoup = bs(content, 'html.parser')
256
+
257
+ # txtlist=[]
258
+ # entlist=[]
259
+ # for h in extsoup.findAll('div'):
260
+ # mark = h.findAll('mark')
261
+ # span = h.findAll('span')
262
+ # for i in mark:
263
+ # txt = i.find(text=True)
264
+ # txt = txt.replace("\n","");
265
+ # txt = txt.strip();
266
+ # txtlist.append(str(txt))
267
+ # for i in span:
268
+ # ent = i.find(text=True)
269
+ # entlist.append(str(ent))
270
+ # zipped = zip(txtlist, entlist)
271
+ # ziplist = list(zipped)
272
+
273
+ # post = {"api_docid":docid, "document_text":finaltext, "named_entities":ziplist, "entities_indices": finalentities}
274
+ # collection.insert_one(post)
275
 
276
  html = displacy.render(doc, style="ent", page = True)
277
  html = (