Fix 7: change the default FOURNISSEUR

pdf_oralia/pages/charge.py

@@ -17,6 +17,7 @@ DF_TYPES = {
     "annee": str,
     "lot": str,
 }
+DEFAULT_FOURNISSEUR = "ROSIER MODICA MOTTEROZ SA"
 
 
 def is_it(page_text):
@@ -62,8 +63,8 @@ def extract(table, additionnal_fields: dict = {}):
             for k, v in additionnal_fields.items():
                 r[k] = v
 
-            if "honoraire" in row[RECAPITULATIF_DES_OPERATIONS]:
-                r["Fournisseur"] = "IMI GERANCE"
+            if "honoraire" in row[RECAPITULATIF_DES_OPERATIONS].lower():
+                r["Fournisseur"] = DEFAULT_FOURNISSEUR
 
             extracted.append(r)
 
@@ -83,6 +84,5 @@ def table2df(tables):
     df = pd.concat(dfs)
 
     df["immeuble"] = df["immeuble"].apply(lambda x: x[0].capitalize())
-    print(df.columns)
     df["lot"] = df["RECAPITULATIF DES OPERATIONS"].apply(get_lot)
-    return df.astype(DF_TYPES, errors="ignore")
+    return df.astype(DF_TYPES)

pdf_oralia/pages/locataire.py

@@ -1,3 +1,4 @@
+import numpy as np
 import pandas as pd
 
 DF_TYPES = {
@@ -33,7 +34,7 @@ def is_drop(row):
 
 
 def extract(table, additionnal_fields: dict = {}):
-    """Turn table to dictionary with additionnal fields"""
+    """Turn table to dictionary with additional fields"""
     extracted = []
     header = table[0]
     for row in table[1:]:
@@ -138,8 +139,6 @@ def join_row(table):
                     }
                 )
                 joined.append(row)
-            else:
-                pass
 
     return joined
 
@@ -159,4 +158,9 @@ def table2df(tables):
     df["immeuble"] = df["immeuble"].apply(lambda x: x[0].capitalize())
     df["Type"] = df["Type"].apply(clean_type)
 
-    return df.astype(DF_TYPES, errors="ignore")
+    numeric_cols = [k for k, v in DF_TYPES.items() if v == float]
+    df[numeric_cols] = df[numeric_cols].replace("", np.nan)
+
+    df = df.drop(df[(df["Locataires"] == "") & (df["Période"] == "")].index)
+
+    return df.astype(DF_TYPES)