Feat: join locataire table with pagebreak

pdf_oralia/extract.py

@@ -14,7 +14,7 @@ charge_table_settings = {
 
 def extract_from_pdf(pdf, charge_dest, location_dest):
     """Build charge_dest and location_dest xlsx file from pdf"""
-    loc_table = []
+    loc_tables = []
     for page in pdf.pages[1:]:
         page_text = page.extract_text()
         situation_loc_line = [
@@ -22,10 +22,10 @@ def extract_from_pdf(pdf, charge_dest, location_dest):
         ]
         if situation_loc_line:
             mois, annee = situation_loc_line[0].split(" ")[-2:]
-            if loc_table:
+            if loc_tables:
-                loc_table += page.extract_table()[1:]
+                loc_tables.append(page.extract_table()[1:])
             else:
-                loc_table = page.extract_table()
+                loc_tables.append(page.extract_table())
 
         elif "HONORAIRES" in page_text:
             table = page.extract_table(charge_table_settings)
@@ -33,7 +33,7 @@ def extract_from_pdf(pdf, charge_dest, location_dest):
             df_charge.to_excel(charge_dest, sheet_name="Charges", index=False)
             logging.info(f"{charge_dest} saved")
 
-    df_loc = extract_situation_loc(loc_table, mois=mois, annee=annee)
+    df_loc = extract_situation_loc(loc_tables, mois=mois, annee=annee)
     df_loc = df_loc.assign()
     df_loc.to_excel(location_dest, sheet_name="Location", index=False)
     logging.info(f"{location_dest} saved")

pdf_oralia/extract_locataire.py

@@ -1,3 +1,5 @@
+import logging
+
 import pandas as pd
 
 
@@ -18,8 +20,39 @@ def parse_above_loc(content):
     return pd.Series(row)
 
 
-def extract_situation_loc(table, mois, annee):
+def join_row(last, next):
+    row = []
+    for i in range(len(last)):
+        if last[i] and next[i]:
+            row.append(f"{last[i]}\n{next[i]}")
+        elif last[i]:
+            row.append(last[i])
+        elif next[i]:
+            row.append(next[i])
+        else:
+            row.append("")
+    return row
+
+
+def join_tables(tables):
+
+    joined = tables[0]
+
+    for t in tables[1:]:
+        last_row = joined[-1]
+        if "Totaux" not in last_row[0]:
+            first_row = t[0]
+            joined_row = join_row(last_row, first_row)
+            joined = joined[:-1] + [joined_row] + t[1:]
+        else:
+            joined += t
+
+    return joined
+
+
+def extract_situation_loc(tables, mois, annee):
     """From pdfplumber table extract locataire df"""
+    table = join_tables(tables)
     try:
         df = pd.DataFrame(table[1:], columns=table[0])
     except IndexError: