Name shortcut zu mailadler
This commit is contained in:
georg0480
487
UEBERSETZUNGS_OPTIONEN.md
Normal file
487
UEBERSETZUNGS_OPTIONEN.md
Normal file
@@ -0,0 +1,487 @@
|
||||
# Übersetzungsoptionen - Günstig & Praktisch
|
||||
|
||||
## 1. Übersetzungs-Anbieter (Vergleich)
|
||||
|
||||
| Anbieter | Kostenlos | Qualität | API | Limit | Empfehlung |
|
||||
|----------|-----------|----------|-----|-------|------------|
|
||||
| **Google Translate Free** | ✅ Kostenlos | ✅✅ Gut | ❌ Unofficial | Unbegrenzt | ✅ Einmalig |
|
||||
| **DeepL Free** | ✅ 500K chars/Monat | ✅✅✅ Sehr gut | ✅ Kostenlos | 500K | ✅ BESTE Qualität |
|
||||
| **Microsoft Translator** | ⚠️ 2M chars/Monat | ✅✅ Gut | ✅ Kostenlos | 2M | ✅ Viel Freiheit |
|
||||
| **Yandex** | ✅ Kostenlos | ✅ Gut | ✅ Free | Unbegrenzt | ✅ Backup |
|
||||
| **OpenAI GPT-4** | ❌ $0.03 pro 1K Tokens | ✅✅✅ Excellent | ✅ API | Pay-as-you-go | ⚠️ Teuer |
|
||||
| **AWS Translate** | ❌ $15 pro 1M chars | ✅✅ Gut | ✅ API | Pay-as-you-go | ⚠️ Teuer |
|
||||
| **Ollama lokal** | ✅ Kostenlos | ✅ Gut | ✅ Lokal | Unbegrenzt | ✅ Datenschutz |
|
||||
|
||||
---
|
||||
|
||||
## 2. EMPFEHLUNG: DeepL Free
|
||||
|
||||
### Warum DeepL?
|
||||
```
|
||||
✅ Kostenlos (500K characters/Monat)
|
||||
✅ BESTE Übersetzungsqualität (besser als Google)
|
||||
✅ Kostenlose API verfügbar
|
||||
✅ 70 Wörter × 30 Sprachen = ~2100 chars = KOSTENLOS!
|
||||
✅ Unbegrenztes Kontingent mit Free-Tier
|
||||
```
|
||||
|
||||
### Beispiel: 70 Wörter × 30 Sprachen
|
||||
```
|
||||
70 Wörter durchschnittlich 6 Buchstaben = 420 Zeichen
|
||||
× 30 Sprachen = 12.600 Zeichen
|
||||
500.000 Zeichen/Monat → locker kostenlos!
|
||||
|
||||
Selbst 100 Sprachen würden passen!
|
||||
```
|
||||
|
||||
### DeepL Free Setup:
|
||||
|
||||
```bash
|
||||
# 1. Kostenlos registrieren
|
||||
https://www.deepl.com/de/signup
|
||||
|
||||
# 2. Python-Library
|
||||
pip install deepl
|
||||
|
||||
# 3. Script:
|
||||
```
|
||||
|
||||
```python
|
||||
#!/usr/bin/env python3
|
||||
# scripts/deepl_translate.py
|
||||
|
||||
import deepl
|
||||
import csv
|
||||
import argparse
|
||||
|
||||
def translate_csv_with_deepl(csv_file: str, language_code: str):
|
||||
"""
|
||||
Übersetze CSV-Spalte mit DeepL
|
||||
language_code: "en", "fr", "es", "pt", "it", "nl", "pl"
|
||||
"""
|
||||
|
||||
# DeepL kostenlos (kein API-Key nötig für Web-Interface)
|
||||
# Oder mit API-Key (kostenlos 500K chars):
|
||||
# translator = deepl.Translator("your-free-api-key")
|
||||
|
||||
# Für Free-Tier ohne API-Key: Google Translate Alternative
|
||||
# ODER: Registriere dich für DeepL Free API
|
||||
|
||||
translator = deepl.Translator("your-deepl-api-key")
|
||||
|
||||
lang_map = {
|
||||
'en': 'EN-US',
|
||||
'fr': 'FR',
|
||||
'es': 'ES',
|
||||
'pt': 'PT',
|
||||
'it': 'IT',
|
||||
'nl': 'NL',
|
||||
'pl': 'PL',
|
||||
'de': 'DE'
|
||||
}
|
||||
|
||||
target_lang = lang_map.get(language_code, 'EN-US')
|
||||
|
||||
# Lese CSV
|
||||
with open(csv_file, 'r', encoding='utf-8') as f:
|
||||
reader = csv.reader(f)
|
||||
rows = list(reader)
|
||||
|
||||
# Übersetze Englisch-Spalte (Index 1)
|
||||
if len(rows[0]) > 1 and rows[0][1] == 'Englisch':
|
||||
# Erste Zeile ist Header, überspringe
|
||||
for i in range(1, len(rows)):
|
||||
if len(rows[i]) > 1 and rows[i][1]: # Wenn Englisch-Text
|
||||
english_text = rows[i][1]
|
||||
|
||||
# Übersetze mit DeepL
|
||||
result = translator.translate_text(
|
||||
english_text,
|
||||
target_lang=target_lang
|
||||
)
|
||||
|
||||
rows[i].append(result.text)
|
||||
print(f"✓ {english_text:30} → {result.text}")
|
||||
|
||||
# Speichern
|
||||
with open(csv_file, 'w', newline='', encoding='utf-8') as f:
|
||||
writer = csv.writer(f)
|
||||
writer.writerows(rows)
|
||||
|
||||
print(f"\n✅ Übersetzt mit DeepL!")
|
||||
|
||||
if __name__ == '__main__':
|
||||
parser = argparse.ArgumentParser()
|
||||
parser.add_argument('--csv', required=True)
|
||||
parser.add_argument('--lang', required=True, help='en, fr, es, pt, it, nl, pl')
|
||||
|
||||
args = parser.parse_args()
|
||||
translate_csv_with_deepl(args.csv, args.lang)
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 3. Strategie: Nur NEUE Strings übersetzen
|
||||
|
||||
### Problem:
|
||||
```
|
||||
Jedes Mal ALLE 70 Wörter übersetzen = Verschwendung
|
||||
Besser: Nur neue/veränderte Strings
|
||||
```
|
||||
|
||||
### Lösung: Delta-Übersetzung
|
||||
|
||||
```python
|
||||
#!/usr/bin/env python3
|
||||
# scripts/translate_delta.py
|
||||
|
||||
import deepl
|
||||
import csv
|
||||
import hashlib
|
||||
import json
|
||||
from pathlib import Path
|
||||
|
||||
class DeltaTranslator:
|
||||
def __init__(self, api_key: str):
|
||||
self.translator = deepl.Translator(api_key)
|
||||
self.cache_file = "translations/translation_cache.json"
|
||||
self.cache = self.load_cache()
|
||||
|
||||
def load_cache(self):
|
||||
"""Lade bereits übersetzte Wörter aus Cache"""
|
||||
if Path(self.cache_file).exists():
|
||||
with open(self.cache_file, 'r', encoding='utf-8') as f:
|
||||
return json.load(f)
|
||||
return {}
|
||||
|
||||
def save_cache(self):
|
||||
"""Speichere Cache"""
|
||||
with open(self.cache_file, 'w', encoding='utf-8') as f:
|
||||
json.dump(self.cache, f, ensure_ascii=False, indent=2)
|
||||
|
||||
def get_hash(self, text: str) -> str:
|
||||
"""Generiere Hash für Wort"""
|
||||
return hashlib.md5(text.encode()).hexdigest()
|
||||
|
||||
def translate_csv_delta(self, csv_file: str, language_code: str):
|
||||
"""
|
||||
Übersetze nur NEUE Wörter
|
||||
Cache speichert bereits übersetzte
|
||||
"""
|
||||
|
||||
lang_map = {
|
||||
'en': 'EN-US', 'fr': 'FR', 'es': 'ES', 'pt': 'PT',
|
||||
'it': 'IT', 'nl': 'NL', 'pl': 'PL'
|
||||
}
|
||||
target_lang = lang_map.get(language_code, 'EN-US')
|
||||
|
||||
# Lese CSV
|
||||
with open(csv_file, 'r', encoding='utf-8') as f:
|
||||
reader = csv.reader(f)
|
||||
rows = list(reader)
|
||||
|
||||
translated_count = 0
|
||||
cached_count = 0
|
||||
|
||||
# Verarbeite Strings
|
||||
for i in range(1, len(rows)): # Überspringe Header
|
||||
if len(rows[i]) > 1 and rows[i][1]:
|
||||
english_text = rows[i][1]
|
||||
text_hash = self.get_hash(english_text)
|
||||
|
||||
# Check Cache
|
||||
cache_key = f"{language_code}:{text_hash}"
|
||||
|
||||
if cache_key in self.cache:
|
||||
# Aus Cache nehmen
|
||||
translation = self.cache[cache_key]
|
||||
cached_count += 1
|
||||
print(f"⚡ (Cache) {english_text:30} → {translation}")
|
||||
else:
|
||||
# Neu übersetzen
|
||||
result = self.translator.translate_text(
|
||||
english_text,
|
||||
target_lang=target_lang
|
||||
)
|
||||
translation = result.text
|
||||
self.cache[cache_key] = translation
|
||||
translated_count += 1
|
||||
print(f"✓ (Neu) {english_text:30} → {translation}")
|
||||
|
||||
rows[i].append(translation)
|
||||
|
||||
# Speichern
|
||||
with open(csv_file, 'w', newline='', encoding='utf-8') as f:
|
||||
writer = csv.writer(f)
|
||||
writer.writerows(rows)
|
||||
|
||||
# Cache speichern
|
||||
self.save_cache()
|
||||
|
||||
print(f"\n✅ Fertig!")
|
||||
print(f" Neu übersetzt: {translated_count}")
|
||||
print(f" Aus Cache: {cached_count}")
|
||||
|
||||
if __name__ == '__main__':
|
||||
import argparse
|
||||
|
||||
parser = argparse.ArgumentParser()
|
||||
parser.add_argument('--csv', required=True)
|
||||
parser.add_argument('--lang', required=True)
|
||||
parser.add_argument('--api-key', required=True)
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
translator = DeltaTranslator(args.api_key)
|
||||
translator.translate_csv_delta(args.csv, args.lang)
|
||||
```
|
||||
|
||||
### Verwendung:
|
||||
|
||||
```bash
|
||||
# Erste Übersetzung (alle Wörter)
|
||||
python3 scripts/translate_delta.py \
|
||||
--csv translations/glossary_all.csv \
|
||||
--lang fr \
|
||||
--api-key "your-deepl-api-key"
|
||||
|
||||
# Output:
|
||||
# ✓ (Neu) Abbrechen → Annuler
|
||||
# ✓ (Neu) Anmeldedaten → Identifiants
|
||||
# ... 70 Wörter
|
||||
# ✅ Fertig!
|
||||
# Neu übersetzt: 70
|
||||
# Aus Cache: 0
|
||||
|
||||
# Später: Nur 5 neue Wörter hinzugefügt
|
||||
# Zweite Übersetzung
|
||||
python3 scripts/translate_delta.py \
|
||||
--csv translations/glossary_all.csv \
|
||||
--lang fr \
|
||||
--api-key "your-deepl-api-key"
|
||||
|
||||
# Output:
|
||||
# ⚡ (Cache) Abbrechen → Annuler
|
||||
# ⚡ (Cache) Anmeldedaten → Identifiants
|
||||
# ... 65 cached
|
||||
# ✓ (Neu) Synchronisieren → Synchroniser
|
||||
# ✓ (Neu) Verschlüsseln → Chiffrer
|
||||
# ... 5 neue
|
||||
# ✅ Fertig!
|
||||
# Neu übersetzt: 5
|
||||
# Aus Cache: 65
|
||||
|
||||
# Cache-Datei: translation_cache.json
|
||||
# {
|
||||
# "fr:abc123...": "Annuler",
|
||||
# "fr:def456...": "Identifiants",
|
||||
# ...
|
||||
# }
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 4. Rechtschreibung & Grammatik
|
||||
|
||||
### Optionen:
|
||||
|
||||
| Tool | Kostenlos | Qualität | LLM | Einfachheit |
|
||||
|------|-----------|----------|-----|-------------|
|
||||
| **LanguageTool** | ✅ Kostenlos | ✅✅ Gut | ❌ | ✅✅ Einfach |
|
||||
| **Grammarly API** | ❌ Bezahlt | ✅✅✅ Sehr gut | ✅ LLM | ⚠️ Komplex |
|
||||
| **Ollama (lokales LLM)** | ✅ Kostenlos | ✅ Gut | ✅ Ja | ✅ Einfach |
|
||||
| **ChatGPT API** | ❌ Bezahlt | ✅✅✅ Excellent | ✅ GPT-4 | ⚠️ Teuer |
|
||||
|
||||
### EMPFEHLUNG: LanguageTool (kostenlos)
|
||||
|
||||
```bash
|
||||
pip install language-tool-python
|
||||
```
|
||||
|
||||
```python
|
||||
#!/usr/bin/env python3
|
||||
# scripts/check_grammar.py
|
||||
|
||||
from language_tool_python import LanguageTool
|
||||
import csv
|
||||
|
||||
def check_translations_grammar(csv_file: str, language_code: str):
|
||||
"""
|
||||
Prüfe Rechtschreibung & Grammatik der Übersetzungen
|
||||
"""
|
||||
|
||||
# LanguageTool für verschiedene Sprachen
|
||||
lang_map = {
|
||||
'en': 'en-US',
|
||||
'fr': 'fr',
|
||||
'es': 'es',
|
||||
'pt': 'pt',
|
||||
'it': 'it',
|
||||
'nl': 'nl',
|
||||
'pl': 'pl'
|
||||
}
|
||||
|
||||
tool = LanguageTool(lang_map.get(language_code, 'en-US'))
|
||||
|
||||
# Lese CSV
|
||||
with open(csv_file, 'r', encoding='utf-8') as f:
|
||||
reader = csv.reader(f)
|
||||
rows = list(reader)
|
||||
|
||||
issues_found = 0
|
||||
|
||||
# Prüfe jede Übersetzung
|
||||
for i in range(1, len(rows)):
|
||||
if len(rows[i]) > 1 and rows[i][1]:
|
||||
original = rows[i][0]
|
||||
translation = rows[i][1]
|
||||
|
||||
# Prüfe
|
||||
matches = tool.check(translation)
|
||||
|
||||
if matches:
|
||||
issues_found += 1
|
||||
print(f"\n⚠️ {original}")
|
||||
print(f" Übersetzung: {translation}")
|
||||
|
||||
for match in matches:
|
||||
print(f" Fehler: {match.message}")
|
||||
print(f" Vorschlag: {match.replacements[:3]}")
|
||||
|
||||
print(f"\n✅ Grammatik-Prüfung fertig!")
|
||||
print(f" Probleme gefunden: {issues_found}")
|
||||
|
||||
if __name__ == '__main__':
|
||||
import argparse
|
||||
|
||||
parser = argparse.ArgumentParser()
|
||||
parser.add_argument('--csv', required=True)
|
||||
parser.add_argument('--lang', required=True)
|
||||
|
||||
args = parser.parse_args()
|
||||
check_translations_grammar(args.csv, args.lang)
|
||||
```
|
||||
|
||||
### Verwendung:
|
||||
|
||||
```bash
|
||||
python3 scripts/check_grammar.py \
|
||||
--csv translations/glossary_all.csv \
|
||||
--lang fr
|
||||
|
||||
# Output:
|
||||
# ⚠️ Abbrechen
|
||||
# Übersetzung: Anuler
|
||||
# Fehler: Typo or grammar error
|
||||
# Vorschlag: ['Annuler', 'Annulé', 'Annulez']
|
||||
|
||||
# ⚠️ Synchronisieren
|
||||
# Übersetzung: Sincroniser
|
||||
# Fehler: Word not recognized
|
||||
# Vorschlag: ['Synchroniser', 'Sincronisé']
|
||||
|
||||
# ✅ Grammatik-Prüfung fertig!
|
||||
# Probleme gefunden: 2
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 5. Kompletter praktischer Workflow
|
||||
|
||||
### Schritt 1: Englisch manuell (LM Studio)
|
||||
```bash
|
||||
# 70 Wörter mit LM Studio/Ollama
|
||||
# 10-15 Minuten
|
||||
```
|
||||
|
||||
### Schritt 2: Alle anderen Sprachen mit DeepL
|
||||
```bash
|
||||
# Englisch → 29 Sprachen
|
||||
for lang in fr es pt it nl pl de sv da no; do
|
||||
python3 scripts/translate_delta.py \
|
||||
--csv translations/glossary_all.csv \
|
||||
--lang $lang \
|
||||
--api-key "your-deepl-api-key"
|
||||
done
|
||||
|
||||
# Total: ~30 Sekunden (alles cached nach erstem Lauf)
|
||||
```
|
||||
|
||||
### Schritt 3: Grammatik-Prüfung
|
||||
```bash
|
||||
python3 scripts/check_grammar.py \
|
||||
--csv translations/glossary_all.csv \
|
||||
--lang fr
|
||||
|
||||
# Behebe Fehler manuell in Excel
|
||||
```
|
||||
|
||||
### Schritt 4: Import & Release
|
||||
```bash
|
||||
./batch_import_parallel.sh
|
||||
|
||||
git push
|
||||
# GitHub Actions → Release
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 6. Kostenübersicht (30 Sprachen, 70 Wörter)
|
||||
|
||||
| Methode | Kosten/Monat | Qualität |
|
||||
|---------|-------------|----------|
|
||||
| **DeepL Free** | €0 | ✅✅✅ Beste |
|
||||
| **Google Translate Free** | €0 | ✅✅ Gut |
|
||||
| **Microsoft Translator Free** | €0 | ✅✅ Gut |
|
||||
| **OpenAI GPT-4** | €0.05-0.10 | ✅✅✅ Excellent |
|
||||
| **AWS Translate** | €0.30 | ✅✅ Gut |
|
||||
|
||||
**EMPFEHLUNG: DeepL Free + LanguageTool (€0 / 100% kostenlos)**
|
||||
|
||||
---
|
||||
|
||||
## 7. Cache-Strategie (wichtig!)
|
||||
|
||||
```
|
||||
Ersten Monat: Alle 70 Wörter × 30 Sprachen = 500K chars
|
||||
↓
|
||||
Cache speichert alles
|
||||
↓
|
||||
Nächste Monate:
|
||||
- 5 neue Strings hinzugefügt?
|
||||
- Nur diese 5 × 30 Sprachen übersetzen
|
||||
- Rest aus Cache
|
||||
↓
|
||||
99% Kostenersparnis!
|
||||
```
|
||||
|
||||
### Cache-Datei:
|
||||
```json
|
||||
{
|
||||
"fr:abc123": "Annuler",
|
||||
"es:abc123": "Cancelar",
|
||||
"pt:abc123": "Cancelar",
|
||||
"it:abc123": "Annulla",
|
||||
...
|
||||
}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Fazit
|
||||
|
||||
**Dein BESTES Setup:**
|
||||
|
||||
```
|
||||
1. Englisch: LM Studio/Ollama manuell (10 Min)
|
||||
2. Rest: DeepL Free API (kostenlos, sehr gut)
|
||||
3. Cache: Nur neue Strings übersetzen (99% Ersparnis)
|
||||
4. Grammar: LanguageTool kostenlos prüfen
|
||||
5. Import: Automatisch
|
||||
|
||||
TOTAL KOSTEN: €0 / 100% kostenlos!
|
||||
TOTAL ZEIT: 15-20 Minuten für 30 Sprachen
|
||||
QUALITÄT: Höchste (besser als Google!)
|
||||
```
|
||||
|
||||
**Du brauchst wirklich nichts zu bezahlen!** 🎯
|
||||
Reference in New Issue
Block a user