modified: excel_to_json.py

renamed:    exel_datein/2024/06/Tagesbericht Beispiel.xlsx -> exel_datein/2024/06/13.xlsx
	new file:   exel_datein/2024/09/4.xlsx
This commit is contained in:
SimolZimol
2025-06-18 21:13:41 +02:00
parent d181b69a17
commit 1e490c23c2
3 changed files with 10 additions and 7 deletions

View File

@@ -4,24 +4,27 @@ import os
import re
root_dir = "exel_datein"
date_pattern = re.compile(r'(\d{4})[\\/](\d{2})')
# Passe das Pattern an: Jahr/Monat/Tag
date_pattern = re.compile(r'(\d{4})[\\/](\d{2})[\\/](\d{1,2})')
all_notes = []
for dirpath, _, filenames in os.walk(root_dir):
match = date_pattern.search(dirpath)
if not match:
continue
jahr, monat = match.groups()
for filename in filenames:
if filename.lower().endswith(('.xlsx', '.xls')):
excel_path = os.path.join(dirpath, filename)
# Versuche, Jahr, Monat und Tag aus dem Pfad zu extrahieren
match = date_pattern.search(excel_path)
if match:
jahr, monat, tag = match.groups()
else:
jahr = monat = tag = "unbekannt"
try:
df = pd.read_excel(excel_path, sheet_name=0, usecols=[0, 1], names=["Kunde", "Info"])
df = df.dropna(subset=["Kunde", "Info"])
for _, row in df.iterrows():
# Format: Kunde: Info (Jahr: xxxx, Monat: xx)
note = f"{row['Kunde']}: {row['Info']} (Jahr: {jahr}, Monat: {monat})"
# Format: Kunde: Info (Jahr: xxxx, Monat: xx, Tag: xx)
note = f"{row['Kunde']}: {row['Info']} (Jahr: {jahr}, Monat: {monat}, Tag: {tag})"
all_notes.append(note)
print(f"Verarbeitet: {excel_path}")
except Exception as e:

BIN
exel_datein/2024/09/4.xlsx Normal file

Binary file not shown.