=
commited on
Commit
·
77f147b
1
Parent(s):
05a3c2c
started working on qcri
Browse files- deep_translator/constants.py +2 -1
- deep_translator/qcri.py +145 -0
deep_translator/constants.py
CHANGED
|
@@ -5,7 +5,8 @@ BASE_URLS = {
|
|
| 5 |
"PONS": "https://en.pons.com/translate/",
|
| 6 |
"YANDEX": "https://translate.yandex.net/api/{version}/tr.json/{endpoint}",
|
| 7 |
"LINGUEE": "https://www.linguee.com/",
|
| 8 |
-
"MYMEMORY": "http://api.mymemory.translated.net/get"
|
|
|
|
| 9 |
}
|
| 10 |
|
| 11 |
GOOGLE_CODES_TO_LANGUAGES = {
|
|
|
|
| 5 |
"PONS": "https://en.pons.com/translate/",
|
| 6 |
"YANDEX": "https://translate.yandex.net/api/{version}/tr.json/{endpoint}",
|
| 7 |
"LINGUEE": "https://www.linguee.com/",
|
| 8 |
+
"MYMEMORY": "http://api.mymemory.translated.net/get",
|
| 9 |
+
"QCRI": "https://mt.qcri.org/api/v1/{endpoint}?"
|
| 10 |
}
|
| 11 |
|
| 12 |
GOOGLE_CODES_TO_LANGUAGES = {
|
deep_translator/qcri.py
ADDED
|
@@ -0,0 +1,145 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
import requests
|
| 3 |
+
from deep_translator.constants import BASE_URLS
|
| 4 |
+
from deep_translator.exceptions import (RequestError,
|
| 5 |
+
YandexDefaultException, TranslationNotFound, TooManyRequests)
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
class YandexTranslator(object):
|
| 9 |
+
"""
|
| 10 |
+
class that wraps functions, which use the yandex translator under the hood to translate word(s)
|
| 11 |
+
"""
|
| 12 |
+
|
| 13 |
+
def __init__(self, api_key=None):
|
| 14 |
+
"""
|
| 15 |
+
@param api_key: your qrci api key. Get one for free here https://mt.qcri.org/api/v1/ref
|
| 16 |
+
"""
|
| 17 |
+
|
| 18 |
+
# key: "72e9e2cc7c992db4dcbdd6fb9f91a0d1"
|
| 19 |
+
if not api_key:
|
| 20 |
+
raise YandexDefaultException(401)
|
| 21 |
+
self.__base_url = BASE_URLS.get("QRCI")
|
| 22 |
+
|
| 23 |
+
self.api_key = api_key
|
| 24 |
+
self.api_version = "v1.5"
|
| 25 |
+
self.api_endpoints = {
|
| 26 |
+
"langs": "getLangs",
|
| 27 |
+
"detect": "detect",
|
| 28 |
+
"translate": "translate",
|
| 29 |
+
}
|
| 30 |
+
endpoints = {
|
| 31 |
+
"get_languages": "getLanguagePairs",
|
| 32 |
+
"get_domains": "getDomains",
|
| 33 |
+
"translate": "translate",
|
| 34 |
+
|
| 35 |
+
}
|
| 36 |
+
|
| 37 |
+
params = {
|
| 38 |
+
"key": api_key
|
| 39 |
+
}
|
| 40 |
+
pairs = requests.get(self.__base_url.format(endpoint=endpoints["get_languages"]), params=params)
|
| 41 |
+
|
| 42 |
+
print(pairs.url, pairs.text)
|
| 43 |
+
|
| 44 |
+
def get_supported_languages(self):
|
| 45 |
+
return set(x.split("-")[0] for x in self.dirs)
|
| 46 |
+
|
| 47 |
+
@property
|
| 48 |
+
def languages(self):
|
| 49 |
+
return self.get_supported_languages()
|
| 50 |
+
|
| 51 |
+
@property
|
| 52 |
+
def dirs(self, proxies=None):
|
| 53 |
+
|
| 54 |
+
try:
|
| 55 |
+
url = self.__base_url.format(version=self.api_version, endpoint="getLangs")
|
| 56 |
+
print("url: ", url)
|
| 57 |
+
response = requests.get(url, params={"key": self.api_key}, proxies=proxies)
|
| 58 |
+
except requests.exceptions.ConnectionError:
|
| 59 |
+
raise YandexDefaultException(503)
|
| 60 |
+
else:
|
| 61 |
+
data = response.json()
|
| 62 |
+
|
| 63 |
+
if response.status_code != 200:
|
| 64 |
+
raise YandexDefaultException(response.status_code)
|
| 65 |
+
return data.get("dirs")
|
| 66 |
+
|
| 67 |
+
def detect(self, text, proxies=None):
|
| 68 |
+
response = None
|
| 69 |
+
params = {
|
| 70 |
+
"text": text,
|
| 71 |
+
"format": "plain",
|
| 72 |
+
"key": self.api_key,
|
| 73 |
+
}
|
| 74 |
+
try:
|
| 75 |
+
url = self.__base_url.format(version=self.api_version, endpoint="detect")
|
| 76 |
+
response = requests.post(url, data=params, proxies=proxies)
|
| 77 |
+
|
| 78 |
+
except RequestError:
|
| 79 |
+
raise
|
| 80 |
+
except ConnectionError:
|
| 81 |
+
raise YandexDefaultException(503)
|
| 82 |
+
except ValueError:
|
| 83 |
+
raise YandexDefaultException(response.status_code)
|
| 84 |
+
else:
|
| 85 |
+
response = response.json()
|
| 86 |
+
language = response['lang']
|
| 87 |
+
status_code = response['code']
|
| 88 |
+
if status_code != 200:
|
| 89 |
+
raise RequestError()
|
| 90 |
+
elif not language:
|
| 91 |
+
raise YandexDefaultException(501)
|
| 92 |
+
return language
|
| 93 |
+
|
| 94 |
+
def translate(self, source, target, text, proxies=None):
|
| 95 |
+
params = {
|
| 96 |
+
"text": text,
|
| 97 |
+
"format": "plain",
|
| 98 |
+
"lang": target if source == "auto" else "{}-{}".format(source, target),
|
| 99 |
+
"key": self.api_key
|
| 100 |
+
}
|
| 101 |
+
try:
|
| 102 |
+
url = self.__base_url.format(version=self.api_version, endpoint="translate")
|
| 103 |
+
response = requests.post(url, data=params, proxies=proxies)
|
| 104 |
+
except ConnectionError:
|
| 105 |
+
raise YandexDefaultException(503)
|
| 106 |
+
else:
|
| 107 |
+
response = response.json()
|
| 108 |
+
|
| 109 |
+
if response['code'] == 429:
|
| 110 |
+
raise TooManyRequests()
|
| 111 |
+
|
| 112 |
+
if response['code'] != 200:
|
| 113 |
+
raise YandexDefaultException(response['code'])
|
| 114 |
+
|
| 115 |
+
if not response['text']:
|
| 116 |
+
raise TranslationNotFound()
|
| 117 |
+
|
| 118 |
+
return response['text']
|
| 119 |
+
|
| 120 |
+
def translate_file(self, source, target, path):
|
| 121 |
+
"""
|
| 122 |
+
translate from a file
|
| 123 |
+
@param source: source language
|
| 124 |
+
@param target: target language
|
| 125 |
+
@param path: path to file
|
| 126 |
+
@return: translated text
|
| 127 |
+
"""
|
| 128 |
+
try:
|
| 129 |
+
with open(path) as f:
|
| 130 |
+
text = f.read()
|
| 131 |
+
|
| 132 |
+
return self.translate(source, target, text)
|
| 133 |
+
except Exception as e:
|
| 134 |
+
raise e
|
| 135 |
+
|
| 136 |
+
def translate_batch(self, source, target, batch):
|
| 137 |
+
"""
|
| 138 |
+
translate a batch of texts
|
| 139 |
+
@param source: source language
|
| 140 |
+
@param target: target language
|
| 141 |
+
@param batch: list of texts to translate
|
| 142 |
+
@return: list of translations
|
| 143 |
+
"""
|
| 144 |
+
return [self.translate(source, target, text) for text in batch]
|
| 145 |
+
|