Implementing a new Bing Translate implementation

stress_test_bing.py

@@ -0,0 +1,12 @@
+from translatepy.translators.bing import BingTranslate
+from time import sleep
+
+dl = BingTranslate()
+while True:
+    result = dl.translate("hello", "russian")
+    print(result)
+
+    result1 = dl.translate("hello, world!", "russian")
+    print(result1)
+
+    dl.clean_cache()

translatepy/translators/bing.py

@@ -1,8 +1,13 @@
-from typing import Union
+"""
+This implementation was made specifically for translatepy from 'Zhymabek Roman', based on 'Anime no Sekai' version.
+"""
 
+import re
+import json
 import requests
 
 import pyuseragents
+
 from translatepy.translators.base import BaseTranslator
 from translatepy.exceptions import UnsupportedMethod
 from translatepy.utils.annotations import Tuple
@@ -20,8 +25,6 @@
     "Connection": "keep-alive"
 }
 
-PARAMS = {'IID': 'translator.5033.3'}
-
 # TODO: read documentation: https://docs.microsoft.com/ru-ru/azure/cognitive-services/translator/language-support
 
 class Example():
@@ -59,11 +62,49 @@ def __repr__(self) -> str:
         return str(self.source)
 
 
+class BingSessionManager():
+    def __init__(self):
+        self._parse_authorization_data()
+
+    def _parse_authorization_data(self):
+        _page = requests.get("https://www.bing.com/translator").text
+        _parsed_IG = re.findall('IG:"(.*?)"', _page)
+        _parsed_IID = re.findall('data-iid="(.*?)"', _page)
+        _parsed_helper_info = re.findall("params_RichTranslateHelper = (.*?);", _page)
+
+        _normalized_key = json.loads(_parsed_helper_info[0])[0]
+        _normalized_token = json.loads(_parsed_helper_info[0])[1]
+
+        self.ig = _parsed_IG[0]
+        self.iid = _parsed_IID[0]
+        self.key = _normalized_key
+        self.token = _normalized_token
+
+    def send(self, url, data):
+        while True:
+            _params = {'IG': self.ig, 'IID': "self.iid.{}".format(1)}
+            _data = {'token': self.token, 'key': self.key}
+            _data.update(data)
+
+            request = requests.post(url, params=_params, data=_data, headers=HEADERS)
+            response = request.json()
+
+            if isinstance(response, dict):
+                if response.get("statusCode", 200) == 400:
+                    self._parse_authorization_data()
+                    continue
+
+            return response
+
+
 class BingTranslate(BaseTranslator):
     """
     A Python implementation of Microsoft Bing Translation's APIs
     """
 
+    def __init__(self):
+        self.session_manager = BingSessionManager()
+
     def _translate(self, text: str, destination_language: str, source_language: str) -> str:
         """
         Translates the given text to the given language
@@ -77,14 +118,8 @@ def _translate(self, text: str, destination_language: str, source_language: str)
             Tuple(str, str) --> tuple with source_lang, translation
 
         """
-        ic(text)
-        ic(destination_language)
-        ic(source_language)
-        request = requests.post("https://www.bing.com/ttranslatev3", headers=HEADERS, params=PARAMS, data={'text': text, 'fromLang': source_language, 'to': destination_language})
-        response = request.json()
-        ic(response)
-        if request.status_code < 400:
-            return response[0]["translations"][0]["text"]
+        response = self.session_manager.send("https://www.bing.com/ttranslatev3", data={'text': text, 'fromLang': source_language, 'to': destination_language})
+        return response[0]["translations"][0]["text"]
 
     def _example(self, text, destination_language, source_language, translation) -> str:
         """
@@ -102,13 +137,11 @@ def _example(self, text, destination_language, source_language, translation) ->
         if translation is None:
             source_language, translation = self.translate(text, destination_language, source_language)
 
-            if source_language == "auto-detect":
-                source_language = self._language(text)
+        if source_language == "auto-detect":
+            source_language = self._language(text)
 
-        request = requests.post("https://www.bing.com/texamplev3", headers=HEADERS, params=PARAMS, data={'text': text.lower(), 'from': source_language, 'to': destination_language, 'translation': translation.lower()})
-        response = requests.json()
-        if request.status_code < 400:
-            return [Example(example) for example in response[0]["examples"]]
+        response = self.session_manager.send("https://www.bing.com/texamplev3", data={'text': text.lower(), 'from': source_language, 'to': destination_language, 'translation': translation.lower()})
+        return [Example(example) for example in response[0]["examples"]]
 
     def _spellcheck(self, text: str, source_language: str) -> str:
         """
@@ -123,14 +156,12 @@ def _spellcheck(self, text: str, source_language: str) -> str:
         """
         if source_language == "auto-detect":
             source_language = self._language(text)
-        request = requests.post("https://www.bing.com/tspellcheckv3", headers=HEADERS, params=PARAMS, data={'text': text, 'fromLang': source_language})
-        response = request.json()
-        ic(response)
-        if request.status_code < 400:
-            result = response["correctedText"]
-            if result == "":
-                return text
-            return result
+
+        response = self.session_manager.send("https://www.bing.com/tspellcheckv3", data={'text': text, 'fromLang': source_language})
+        result = response["correctedText"]
+        if result == "":
+            return text
+        return result
 
     def _language(self, text: str) -> str:
         """
@@ -143,25 +174,19 @@ def _language(self, text: str) -> str:
             str --> the language code
 
         """
-        request = requests.post("https://www.bing.com/ttranslatev3", headers=HEADERS, params=PARAMS, data={'text': text, 'fromLang': "auto-detect", 'to': "en"})
-        response = request.json()
-        ic(response)
-        if request.status_code < 400:
-            return response[0]["detectedLanguage"]["language"]
+        response = self.session_manager.send("https://www.bing.com/ttranslatev3", data={'text': text, 'fromLang': "auto-detect", 'to': "en"})
+        return response[0]["detectedLanguage"]["language"]
 
     def _transliterate(self, text: str, destination_language: str, source_language: str):
         # TODO: alternative implementation, won't work
-        # request = requests.post("https://www.bing.com/ttransliteratev3", headers=HEADERS, params=PARAMS, data={'text': text, 'language': source_language, 'toScript': destination_language})
-
-        request = requests.post("https://www.bing.com/ttranslatev3", headers=HEADERS, params=PARAMS, data={'text': text, 'fromLang': source_language, 'to': destination_language})
-        response = request.json()
-        ic(response)
-        if request.status_code < 400:
-            # XXX: Not a predictable response from Bing Translate
-            try:
-                return response[1]["inputTransliteration"]
-            except IndexError:
-                return response[0]["translations"][0]["transliteration"]["text"]
+        # request = requests.post("https://www.bing.com/ttransliteratev3", data={'text': text, 'language': source_language, 'toScript': destination_language})
+
+        response = self.session_manager.send("https://www.bing.com/ttranslatev3", data={'text': text, 'fromLang': source_language, 'to': destination_language})
+        # XXX: Not a predictable response from Bing Translate
+        try:
+            return response[1]["inputTransliteration"]
+        except IndexError:
+            return response[0]["translations"][0]["transliteration"]["text"]
 
     def _dictionary(self, text: str, destination_language: str, source_language: str):
         # TODO: Implement