Update app.py

app.py

@@ -100,7 +100,7 @@ def convert_pdf_to_excel(pdf_file):
                         year = ""
 
                 if 'Part Name' in data[index].strip():
-                    each_table_data.append(data[index+1])
+                    each_table_data.append(data[index+1].replace("Sched",""))
                 
                 if 'Inventory Category' in data[index].strip():
                     each_table_data.append(data[index+1].replace('Receive Type',""))
@@ -125,9 +125,9 @@ def convert_pdf_to_excel(pdf_file):
     extracted_file = "Data Extracted.xlsx"
     data_for_mapping = "Data Mapping.xlsx"
 
-    extracted_data_for_mapping = whole_data.drop('Date Qty',axis=1)
-    extracted_data_for_mapping = extracted_data_for_mapping.drop_duplicates(subset=["Part No.","Part Color Code","Part Name","Inventory Category"])
-    extracted_data_for_mapping.columns = ['Customer Part no as per pdf','Customer Part color  as per pdf','Customer Part name as per pdf',"Inventory Category as per pdf"]
+    extracted_data_for_mapping = whole_data.drop(['Inventory Category','Date Qty'],axis=1)
+    extracted_data_for_mapping = extracted_data_for_mapping.drop_duplicates(subset=["Part No.","Part Color Code","Part Name"])
+    extracted_data_for_mapping.columns = ['Customer Part no as per pdf','Customer Part color  as per pdf','Customer Part name as per pdf']
     extracted_data_for_mapping['Item Code'] = ""
 
     whole_data.to_excel(extracted_file, index=False)
@@ -141,7 +141,7 @@ def map_data_to_template(excel_file, mapping_file):
     mapping_data = mapping_data.drop_duplicates(subset=['Customer Part no as per pdf','Customer Part name as per pdf','Customer Part color  as per pdf','Item Code'])
     mapping_data.to_excel(MAPPING_FILENAME,index=False)
     save_mapping_file_to_drive()
-    mapping_data = mapping_data.rename(columns = {'Customer Part no as per pdf':'Part No.'})
+    mapping_data = mapping_data.rename(columns = {'Customer Part no as per pdf':'Part No.','Customer Part name as per pdf':'Part Name','Customer Part color  as per pdf':'Part Color Code'})
 
     # Perform mapping
     extracted_data['Date Qty'] = extracted_data['Date Qty'].apply(lambda x: ast.literal_eval(x))
@@ -149,7 +149,8 @@ def map_data_to_template(excel_file, mapping_file):
     extracted_data[['SchDate','Qty']]= pd.DataFrame(extracted_data['Date Qty'].to_list(), index= extracted_data.index)
     extracted_data = extracted_data.drop('Date Qty',axis=1)
     extracted_data = extracted_data[~extracted_data['SchDate'].isna()]
-    mapped_data = extracted_data.merge(mapping_data, on =['Part No.'],how='outer')[['Item Code','SchDate','Qty','Inventory Category']]
+
+    mapped_data = extracted_data.merge(mapping_data, on =["Part No.","Part Name","Part Color Code"],how='outer')[['Item Code','SchDate','Qty','Inventory Category']]
     mapped_data = mapped_data[~mapped_data["SchDate"].isna()]
     mapped_data = mapped_data[~mapped_data["SchDate"].str.strip().isin(["",None])] 				
 
@@ -282,7 +283,7 @@ def main():
                 mapping_data_from_drive.columns = [i.strip() for i in mapping_data_from_drive.columns if "inventory category" not in i.lower()]
                 mapping_data_from_drive = mapping_data_from_drive.drop_duplicates(subset=['Customer Part no as per pdf','Customer Part name as per pdf','Customer Part color  as per pdf','Item Code'])
 
-                extracted_data_for_mapping = extracted_data_for_mapping[['Customer Part no as per pdf','Customer Part name as per pdf','Customer Part color  as per pdf',"Inventory Category as per pdf"]].merge(mapping_data_from_drive, on = ['Customer Part no as per pdf','Customer Part name as per pdf','Customer Part color  as per pdf'], how='outer')
+                extracted_data_for_mapping = extracted_data_for_mapping[['Customer Part no as per pdf','Customer Part name as per pdf','Customer Part color  as per pdf']].merge(mapping_data_from_drive, on = ['Customer Part no as per pdf','Customer Part name as per pdf','Customer Part color  as per pdf'], how='outer')
                 extracted_data_for_mapping.to_excel(data_for_mapping,index=False)
             except Exception as e:
                 st.error("Error in the Mapping Master file on Cloud. " + str(e))