acrowth commited on
Commit
4415c8e
·
1 Parent(s): eeae230

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +5 -2
app.py CHANGED
@@ -3,12 +3,15 @@ import pandas as pd
3
  import gradio as gr
4
 
5
  df=pd.read_parquet('df.parquet')
6
- #df2=pd.read_parquet('df2.parquet')
7
- #df=df.merge(df2)
8
  dirname='lot3'
9
 
10
  df['fileclean']=df.file.str.replace(f'.*{dirname}/[^/]+/','').str.replace('[\(\)]','').str.replace('/[^/]+$','').str.replace('/',' ').str.replace('-',' ').str.replace(' 0+',' ')
11
  candidats=pd.read_parquet('candidats.parquet')
 
 
 
 
 
12
 
13
  document_store = InMemoryDocumentStore(use_bm25=True)
14
  docs=df.drop_duplicates(subset=['fileclean']).rename(columns={'fileclean':'content'}).to_dict(orient='records')
 
3
  import gradio as gr
4
 
5
  df=pd.read_parquet('df.parquet')
 
 
6
  dirname='lot3'
7
 
8
  df['fileclean']=df.file.str.replace(f'.*{dirname}/[^/]+/','').str.replace('[\(\)]','').str.replace('/[^/]+$','').str.replace('/',' ').str.replace('-',' ').str.replace(' 0+',' ')
9
  candidats=pd.read_parquet('candidats.parquet')
10
+ df2=pd.read_parquet('df2.parquet')
11
+ for c in df2.columns:
12
+ candidats[c]=candidats[c].astype(str)
13
+ df2[c]=df2[c].astype(str)
14
+ candidats=candidats.merge(df2)
15
 
16
  document_store = InMemoryDocumentStore(use_bm25=True)
17
  docs=df.drop_duplicates(subset=['fileclean']).rename(columns={'fileclean':'content'}).to_dict(orient='records')