Feat(join): move join to its own lib

pdf_oralia/join.py

@@ -0,0 +1,30 @@
+import logging
+from pathlib import Path
+
+import pandas as pd
+
+
+def extract_excel_to_dfs(directory, df_names=["charge", "locataire"]):
+    p = Path(directory)
+    dfs = {name: [] for name in df_names}
+
+    for file in p.glob("*.xlsx"):
+        year, month, immeuble, table = file.stem.split("_")
+        df = pd.read_excel(file, dtype={"lot": str}).assign(
+            annee=year, mois=month, immeuble=immeuble[:3]
+        )
+        dfs[table].append(df)
+
+    return dfs
+
+
+def join_excel(directory, dest, df_names=["charge", "locataire"]):
+    dfs = extract_excel_to_dfs(directory, df_names)
+    destinations = {}
+    for tablename, datas in dfs.items():
+        df = pd.concat(datas)
+        destination = Path(dest) / f"{tablename}.xlsx"
+        df.to_excel(destination, index=False)
+        destinations[tablename] = destination
+        logging.info(f"{destination} written")
+    return destinations

pdf_oralia/scripts.py

@@ -6,6 +6,7 @@ import click
 import pandas as pd
 
 from .extract import extract_save
+from .join import join_excel
 
 logging_config = dict(
     version=1,
@@ -66,19 +67,4 @@ def all(src, dest):
 @click.option("--src", help="Tous les fichiers dans src", default="./")
 @click.option("--dest", help="Où mettre les fichiers produits", default="")
 def join(src, dest):
-    p = Path(src)
+    join_excel(src, dest, df_names=["charge", "locataire"])
-    dfs = {
-        "charge": [],
-        "locataire": [],
-    }
-    for file in p.glob("*.xlsx"):
-        year, month, immeuble, table = file.stem.split("_")
-        df = pd.read_excel(file, dtype={"lot": str}).assign(
-            annee=year, mois=month, immeuble=immeuble[:3]
-        )
-        dfs[table].append(df)
-    for tablename, datas in dfs.items():
-        df = pd.concat(datas)
-        destination = Path(dest) / f"{tablename}.xlsx"
-        df.to_excel(destination, index=False)
-        logging.info(f"{destination} written")