From c82fdd19dfe43561b07b9082d3a4a0d13fbb3bc5 Mon Sep 17 00:00:00 2001
From: tanmay-punekar <52330061+tanmay-punekar@users.noreply.github.com>
Date: Wed, 1 Apr 2020 18:24:28 +0530
Subject: [PATCH 1/4] Add files via upload

---
 newspaper/resources/text/stopwords-mr.txt | 99 +++++++++++++++++++++++
 1 file changed, 99 insertions(+)
 create mode 100644 newspaper/resources/text/stopwords-mr.txt

diff --git a/newspaper/resources/text/stopwords-mr.txt b/newspaper/resources/text/stopwords-mr.txt
new file mode 100644
index 000000000..27949fafb
--- /dev/null
+++ b/newspaper/resources/text/stopwords-mr.txt
@@ -0,0 +1,99 @@
+अधिक
+अनेक
+अशी
+असलयाचे
+असलेल्या
+असा
+असून
+असे
+आज
+आणि
+आता
+आपल्या
+आला
+आली
+आले
+आहे
+आहेत
+एक
+एका
+कमी
+करणयात
+करून
+का
+काम
+काय
+काही
+किवा
+की
+केला
+केली
+केले
+कोटी
+गेल्या
+घेऊन
+जात
+झाला
+झाली
+झाले
+झालेल्या
+टा
+डॉ
+तर
+तरी
+तसेच
+ता
+ती
+तीन
+ते
+तो
+त्या
+त्याचा
+त्याची
+त्याच्या
+त्याना
+त्यानी
+त्यामुळे
+त्री
+दिली
+दोन
+न
+नाही
+निर्ण्य
+पण
+पम
+परयतन
+पाटील
+म
+मात्र
+माहिती
+मी
+मुबी
+म्हणजे
+म्हणाले
+म्हणून
+या
+याचा
+याची
+याच्या
+याना
+यानी
+येणार
+येत
+येथील
+येथे
+लाख
+व
+व्यकत
+सर्व
+सागित्ले
+सुरू
+हजार
+हा
+ही
+हे
+होणार
+होत
+होता
+होती
+होते
\ No newline at end of file

From 08592dd2ce729e9867667b3ee52e5d5f841fc089 Mon Sep 17 00:00:00 2001
From: tanmay-punekar <52330061+tanmay-punekar@users.noreply.github.com>
Date: Wed, 1 Apr 2020 18:29:33 +0530
Subject: [PATCH 2/4] Update article.py

---
 newspaper/article.py | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/newspaper/article.py b/newspaper/article.py
index df0d9c435..789e8515d 100644
--- a/newspaper/article.py
+++ b/newspaper/article.py
@@ -12,6 +12,7 @@
 
 import requests
 
+from googletrans import Translator
 from . import images
 from . import network
 from . import nlp
@@ -368,6 +369,12 @@ def is_media_news(self):
             if s in self.url:
                 return True
         return False
+      
+     def translater(self,dest = 'en'):
+        self.dest = dest
+        translator = Translator()
+        result = translator.translate(self.text,self.dest)
+        return result.text
 
     def nlp(self):
         """Keyword extraction wrapper

From a6087251e6a50badd06a95474f2dda1a8f629290 Mon Sep 17 00:00:00 2001
From: tanmay-punekar <52330061+tanmay-punekar@users.noreply.github.com>
Date: Wed, 1 Apr 2020 18:31:30 +0530
Subject: [PATCH 3/4] Update utils.py

---
 newspaper/utils.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/newspaper/utils.py b/newspaper/utils.py
index bfa441482..f0ede42cb 100644
--- a/newspaper/utils.py
+++ b/newspaper/utils.py
@@ -393,6 +393,7 @@ def print_available_languages():
         'uk': 'Ukrainian',
         'vi': 'Vietnamese',
         'zh': 'Chinese',
+        'mr': 'Marathi'
     }
 
     codes = get_available_languages()

From b5262a7d6d8b09e89d295e2ed95be949a6b6b30b Mon Sep 17 00:00:00 2001
From: tanmay-punekar <52330061+tanmay-punekar@users.noreply.github.com>
Date: Wed, 1 Apr 2020 18:38:03 +0530
Subject: [PATCH 4/4] Update text.py

---
 newspaper/text.py | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/newspaper/text.py b/newspaper/text.py
index 23b4c6b1e..1046f46dd 100644
--- a/newspaper/text.py
+++ b/newspaper/text.py
@@ -182,6 +182,30 @@ def get_stopword_count(self, content):
         ws.set_stopword_count(len(overlapping_stopwords))
         ws.set_stop_words(overlapping_stopwords)
         return ws
+    
+class StopWordsMarathi(StopWords):
+    """Marathi segmentation
+    """
+    def __init__(self, language='mr'):
+        super(StopWordsHindi, self).__init__(language='mr')
+
+    def get_stopword_count(self, content):
+        if not content:
+            return WordStats()
+        ws = WordStats()
+        stripped_input = self.remove_punctuation(content)
+        candidate_words = self.candidate_words(stripped_input)
+        overlapping_stopwords = []
+        c = 0
+        for w in candidate_words:
+            c += 1
+            for stop_word in self.STOP_WORDS:
+                overlapping_stopwords.append(stop_word)
+
+        ws.set_word_count(c)
+        ws.set_stopword_count(len(overlapping_stopwords))
+        ws.set_stop_words(overlapping_stopwords)
+        return ws
 
 
 class StopWordsJapanese(StopWords):