=
commited on
Commit
·
418aed1
1
Parent(s):
9a69a65
added linguee support
Browse files- HISTORY.rst +1 -1
- deep_translator/constants.py +36 -2
- deep_translator/google_trans.py +3 -4
- deep_translator/linguee.py +77 -0
- deep_translator/pons.py +1 -2
HISTORY.rst
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
History
|
| 3 |
=======
|
| 4 |
|
| 5 |
-
0.1.
|
| 6 |
---------------------
|
| 7 |
|
|
|
|
| 2 |
History
|
| 3 |
=======
|
| 4 |
|
| 5 |
+
0.1.4 stable release
|
| 6 |
---------------------
|
| 7 |
|
deep_translator/constants.py
CHANGED
|
@@ -1,7 +1,10 @@
|
|
| 1 |
|
|
|
|
| 2 |
BASE_URLS = {
|
| 3 |
"GOOGLE_TRANSLATE": "https://translate.google.com/m",
|
| 4 |
-
"PONS": "https://en.pons.com/translate/"
|
|
|
|
|
|
|
| 5 |
}
|
| 6 |
|
| 7 |
GOOGLE_CODES_TO_LANGUAGES = {
|
|
@@ -115,7 +118,6 @@ GOOGLE_CODES_TO_LANGUAGES = {
|
|
| 115 |
|
| 116 |
GOOGLE_LANGUAGES_TO_CODES = {v: k for k, v in GOOGLE_CODES_TO_LANGUAGES.items()}
|
| 117 |
|
| 118 |
-
|
| 119 |
PONS_CODES_TO_LANGUAGES = {
|
| 120 |
'ar': 'arabic',
|
| 121 |
'bg': 'bulgarian',
|
|
@@ -142,3 +144,35 @@ PONS_CODES_TO_LANGUAGES = {
|
|
| 142 |
}
|
| 143 |
|
| 144 |
PONS_LANGUAGES_TO_CODES = {v: k for k, v in PONS_CODES_TO_LANGUAGES.items()}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
|
| 2 |
+
|
| 3 |
BASE_URLS = {
|
| 4 |
"GOOGLE_TRANSLATE": "https://translate.google.com/m",
|
| 5 |
+
"PONS": "https://en.pons.com/translate/",
|
| 6 |
+
"YANDEX": "https://translate.yandex.com/",
|
| 7 |
+
"LINGUEE": "https://www.linguee.com/"
|
| 8 |
}
|
| 9 |
|
| 10 |
GOOGLE_CODES_TO_LANGUAGES = {
|
|
|
|
| 118 |
|
| 119 |
GOOGLE_LANGUAGES_TO_CODES = {v: k for k, v in GOOGLE_CODES_TO_LANGUAGES.items()}
|
| 120 |
|
|
|
|
| 121 |
PONS_CODES_TO_LANGUAGES = {
|
| 122 |
'ar': 'arabic',
|
| 123 |
'bg': 'bulgarian',
|
|
|
|
| 144 |
}
|
| 145 |
|
| 146 |
PONS_LANGUAGES_TO_CODES = {v: k for k, v in PONS_CODES_TO_LANGUAGES.items()}
|
| 147 |
+
|
| 148 |
+
LINGUEE_LANGUAGES_TO_CODES = {
|
| 149 |
+
"maltese": "mt",
|
| 150 |
+
"english": "en",
|
| 151 |
+
"german": "de",
|
| 152 |
+
"bulgarian": "bg",
|
| 153 |
+
"polish": "pl",
|
| 154 |
+
"portuguese": "pt",
|
| 155 |
+
"hungarian": "hu",
|
| 156 |
+
"romanian": "ro",
|
| 157 |
+
"russian": "ru",
|
| 158 |
+
#"serbian": "sr",
|
| 159 |
+
"dutch": "nl",
|
| 160 |
+
"slovakian": "sk",
|
| 161 |
+
"greek": "el",
|
| 162 |
+
"slovenian": "sl",
|
| 163 |
+
"danish": "da",
|
| 164 |
+
"italian": "it",
|
| 165 |
+
"spanish": "es",
|
| 166 |
+
"finnish": "fi",
|
| 167 |
+
"chinese": "zh",
|
| 168 |
+
"french": "fr",
|
| 169 |
+
#"croatian": "hr",
|
| 170 |
+
"czech": "cs",
|
| 171 |
+
"laotian": "lo",
|
| 172 |
+
"swedish": "sv",
|
| 173 |
+
"latvian": "lv",
|
| 174 |
+
"estonian": "et",
|
| 175 |
+
"japanese": "ja"
|
| 176 |
+
}
|
| 177 |
+
|
| 178 |
+
LINGUEE_CODE_TO_LANGUAGE = {v: k for k, v in LINGUEE_LANGUAGES_TO_CODES.items()}
|
deep_translator/google_trans.py
CHANGED
|
@@ -1,4 +1,3 @@
|
|
| 1 |
-
from abc import ABC
|
| 2 |
|
| 3 |
from deep_translator.constants import BASE_URLS, GOOGLE_LANGUAGES_TO_CODES
|
| 4 |
from deep_translator.exceptions import LanguageNotSupportedException, ElementNotFoundInGetRequest, NotValidPayload, NotValidLength
|
|
@@ -7,7 +6,7 @@ from bs4 import BeautifulSoup
|
|
| 7 |
import requests
|
| 8 |
|
| 9 |
|
| 10 |
-
class GoogleTranslator(BaseTranslator
|
| 11 |
"""
|
| 12 |
class that uses google translate to translate texts
|
| 13 |
"""
|
|
@@ -65,8 +64,8 @@ class GoogleTranslator(BaseTranslator, ABC):
|
|
| 65 |
if self.payload_key:
|
| 66 |
self._url_params[self.payload_key] = payload
|
| 67 |
|
| 68 |
-
|
| 69 |
-
soup = BeautifulSoup(
|
| 70 |
element = soup.find(self._element_tag, self._element_query)
|
| 71 |
if not element:
|
| 72 |
raise ElementNotFoundInGetRequest(element)
|
|
|
|
|
|
|
| 1 |
|
| 2 |
from deep_translator.constants import BASE_URLS, GOOGLE_LANGUAGES_TO_CODES
|
| 3 |
from deep_translator.exceptions import LanguageNotSupportedException, ElementNotFoundInGetRequest, NotValidPayload, NotValidLength
|
|
|
|
| 6 |
import requests
|
| 7 |
|
| 8 |
|
| 9 |
+
class GoogleTranslator(BaseTranslator):
|
| 10 |
"""
|
| 11 |
class that uses google translate to translate texts
|
| 12 |
"""
|
|
|
|
| 64 |
if self.payload_key:
|
| 65 |
self._url_params[self.payload_key] = payload
|
| 66 |
|
| 67 |
+
response = requests.get(self.__base_url, params=self._url_params)
|
| 68 |
+
soup = BeautifulSoup(response.text, 'html.parser')
|
| 69 |
element = soup.find(self._element_tag, self._element_query)
|
| 70 |
if not element:
|
| 71 |
raise ElementNotFoundInGetRequest(element)
|
deep_translator/linguee.py
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from deep_translator.constants import BASE_URLS, LINGUEE_LANGUAGES_TO_CODES, LINGUEE_CODE_TO_LANGUAGE
|
| 2 |
+
from deep_translator.exceptions import LanguageNotSupportedException, ElementNotFoundInGetRequest, NotValidPayload, NotValidLength
|
| 3 |
+
from deep_translator.parent import BaseTranslator
|
| 4 |
+
from bs4 import BeautifulSoup
|
| 5 |
+
import requests
|
| 6 |
+
from requests.utils import quote
|
| 7 |
+
|
| 8 |
+
|
| 9 |
+
class LingueeTranslator(BaseTranslator):
|
| 10 |
+
def __init__(self, source, target):
|
| 11 |
+
"""
|
| 12 |
+
@param source: source language to translate from
|
| 13 |
+
@param target: target language to translate to
|
| 14 |
+
"""
|
| 15 |
+
self.__base_url = BASE_URLS.get("LINGUEE")
|
| 16 |
+
|
| 17 |
+
if self.is_language_supported(source, target):
|
| 18 |
+
self._source, self._target = self._map_language_to_code(source.lower(), target.lower())
|
| 19 |
+
|
| 20 |
+
super().__init__(base_url=self.__base_url,
|
| 21 |
+
source=self._source,
|
| 22 |
+
target=self._target,
|
| 23 |
+
element_tag='a',
|
| 24 |
+
element_query={'class': 'dictLink featured'},
|
| 25 |
+
payload_key=None, # key of payload in the url
|
| 26 |
+
)
|
| 27 |
+
|
| 28 |
+
def _map_language_to_code(self, *languages, **kwargs):
|
| 29 |
+
"""
|
| 30 |
+
@param language: type of language
|
| 31 |
+
@return: mapped value of the language or raise an exception if the language is not supported
|
| 32 |
+
"""
|
| 33 |
+
for language in languages:
|
| 34 |
+
if language in LINGUEE_LANGUAGES_TO_CODES.values():
|
| 35 |
+
yield LINGUEE_CODE_TO_LANGUAGE[language]
|
| 36 |
+
elif language in LINGUEE_LANGUAGES_TO_CODES.keys():
|
| 37 |
+
yield language
|
| 38 |
+
else:
|
| 39 |
+
raise LanguageNotSupportedException(language)
|
| 40 |
+
|
| 41 |
+
def is_language_supported(self, *languages, **kwargs):
|
| 42 |
+
for lang in languages:
|
| 43 |
+
if lang not in LINGUEE_LANGUAGES_TO_CODES.keys():
|
| 44 |
+
if lang not in LINGUEE_LANGUAGES_TO_CODES.values():
|
| 45 |
+
raise LanguageNotSupportedException(lang)
|
| 46 |
+
return True
|
| 47 |
+
|
| 48 |
+
def translate(self, word, **kwargs):
|
| 49 |
+
|
| 50 |
+
if self._validate_payload(word):
|
| 51 |
+
# %s-%s/translation/%s.html
|
| 52 |
+
url = "{}{}-{}/translation/{}.html".format(self.__base_url, self._source, self._target, quote(word))
|
| 53 |
+
response = requests.get(url)
|
| 54 |
+
soup = BeautifulSoup(response.text, 'html.parser')
|
| 55 |
+
elements = soup.find_all(self._element_tag, self._element_query)
|
| 56 |
+
if not elements:
|
| 57 |
+
raise ElementNotFoundInGetRequest(elements)
|
| 58 |
+
|
| 59 |
+
if 'return_all' in kwargs and kwargs.get('return_all'):
|
| 60 |
+
return [el.get_text(strip=True) for el in elements]
|
| 61 |
+
else:
|
| 62 |
+
return elements[0].get_text(strip=True)
|
| 63 |
+
|
| 64 |
+
def translate_words(self, words, **kwargs):
|
| 65 |
+
if not words:
|
| 66 |
+
raise NotValidPayload(words)
|
| 67 |
+
|
| 68 |
+
translated_words = []
|
| 69 |
+
for word in words:
|
| 70 |
+
translated_words.append(self.translate(payload=word))
|
| 71 |
+
return translated_words
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
if __name__ == '__main__':
|
| 75 |
+
res = LingueeTranslator(source="en", target="de").translate(word='cute', return_all=False)
|
| 76 |
+
print(res)
|
| 77 |
+
|
deep_translator/pons.py
CHANGED
|
@@ -36,7 +36,7 @@ class PonsTranslator(BaseTranslator):
|
|
| 36 |
@return: mapped value of the language or raise an exception if the language is not supported
|
| 37 |
"""
|
| 38 |
for language in languages:
|
| 39 |
-
if language in PONS_LANGUAGES_TO_CODES.values()
|
| 40 |
yield PONS_CODES_TO_LANGUAGES[language]
|
| 41 |
elif language in PONS_LANGUAGES_TO_CODES.keys():
|
| 42 |
yield language
|
|
@@ -94,5 +94,4 @@ if __name__ == '__main__':
|
|
| 94 |
# ])
|
| 95 |
# res = PonsTranslator(source="en", target="ar").translate(payload='good')
|
| 96 |
res = PonsTranslator(source="en", target="ar").translate_words(words=('good', 'cute', 'angry'))
|
| 97 |
-
|
| 98 |
print(res)
|
|
|
|
| 36 |
@return: mapped value of the language or raise an exception if the language is not supported
|
| 37 |
"""
|
| 38 |
for language in languages:
|
| 39 |
+
if language in PONS_LANGUAGES_TO_CODES.values():
|
| 40 |
yield PONS_CODES_TO_LANGUAGES[language]
|
| 41 |
elif language in PONS_LANGUAGES_TO_CODES.keys():
|
| 42 |
yield language
|
|
|
|
| 94 |
# ])
|
| 95 |
# res = PonsTranslator(source="en", target="ar").translate(payload='good')
|
| 96 |
res = PonsTranslator(source="en", target="ar").translate_words(words=('good', 'cute', 'angry'))
|
|
|
|
| 97 |
print(res)
|