got hting working

SearchResults.py

@@ -64,6 +64,7 @@ class SearchResults:
         Returns:
             a SearchResults object
         '''
+        print(filters)
         self.__filters = filters
         self.__required_words = []
         self.__required_words += self.get_required_words(query)
@@ -115,21 +116,11 @@ class SearchResults:
                 months = num / 720
             if months > self.__filters['date_updated']:
                 return False
-        if self.__filters['q_verif']:
-            try:
-                if 'Quality Tested' not in outer_article.verif:
-                    return full_expanse
-            except AttributeError:
-                pass
-        if self.__filters['e_verif']:
-            try:
-                if 'Expert Co-Authored' not in outer_article.verif:
-                    return False
-            except AttributeError:
-                pass
         if 'num_views' in self.__filters:
             if self.__filters['num_views'] > outer_article.views:
                 return False
+        if 'Main-Page' in outer_article.url:
+            return False
         return True
             
 
@@ -151,11 +142,12 @@ class SearchResults:
         else:
             forbidden_words += match.group().split()
             forbidden_words[0] = forbidden_words[0][1:]
-        print('forbidden words',forbidden_words)
         return forbidden_words
 
 
     def passes_all_filters(self, article):
+        if 'Main-Page' in article.url:
+            return False
         if 'child_safe' in self.__filters:
             if self.__filters['child_safe']:
                 self.__forbidden_words += inappropriate_words_lst
@@ -169,7 +161,7 @@ class SearchResults:
             if int(article.num_voters) < self.__filters['num_voters']:
                 return False
         if 'num_sources' in self.__filters:
-            if len(article.sources) < self.__filters['num_sources']:
+            if len(article.sources) < int(self.__filters['num_sources']):
                 return False
         if 'num_views' in self.__filters:
             if article.num_views < self.__filters['num_views']:
@@ -179,9 +171,6 @@ class SearchResults:
                 W = w[0].upper() + w[1:]
                 wl = w[0].lower() + w[1:]
                 if W not in article.words and wl not in article.words:
-                    print(article.title)
-                    print(W,wl)
-                    print('not there')
                     return False
         if len(self.__forbidden_words) != 0:
             for w in self.__forbidden_words:
@@ -189,6 +178,21 @@ class SearchResults:
                 wl = w[0].lower() + w[1:]
                 if W in article.words or wl in article.words:
                     return False
+        if self.__filters['q_verif']:
+            try:
+                if 'Quality Tested' not in article.verif:
+                    return False
+            except AttributeError:
+                pass
+        if self.__filters['e_verif']:
+            try:
+                if 'Expert Co-authored' not in article.verif:
+                    return False
+            except AttributeError:
+                pass
+        if self.__filters['language'] != '':
+            if self.__filters['language'] not in article.languages:
+                return False
         return True
 
 
@@ -233,20 +237,17 @@ class Article:
             self.get_sp_text_data(self.__soup)
         self.url = url
         self.num_voters, self.pct_helpful = self.get_helpful(self.__soup)
-        print(self.title,'votes',self.num_voters,'percent',self.pct_helpful)
         self.rating = self.find_rating(self.__soup)
         self.languages = self.find_languages(self.__soup)
+        self.verif = self.expert_data(self.__soup)
         self.view = self.__repr__()
         self.words = self.scrape_words(self.url, self.__soup)
-        '''
-        self.salient = self.find_salient_ngrams(self.words,\
-                       1, False, 0.5)
-        '''
         self.tags = self.find_common_words(self.words)
         self.link_family = self.all_peripheral_links(self.__soup, self.url)
         self.related_articles = []
         self.related_categories = []
 
+
     def get_page_soup(self, url):
         response = requests.get(url)
         return BeautifulSoup(response.content, 'html.parser')
@@ -259,10 +260,27 @@ class Article:
                    divs.find_all('a', class_ = 'external free')]
         except AttributeError:
             sources = []
-        print('sources',sources)
         return sources
 
 
+    def expert_data(self, soup):
+        """
+        Take a soup object and return a string if it has an
+        expert co-author or if it has been quality tested
+
+        Input: soup object
+        Returns: String
+        """
+        expert = soup.find("div", class_ = "sp_coauthor_label")
+        verif = ''
+        if expert != None:
+            if "Co-authored" in expert.text:
+                verif += "Expert Co-authored"
+            elif "Tested" in expert.text:
+                verif += "Quality Tested"
+        return verif
+
+
     def get_author(self, soup):
         div = soup.find('a', class_ = "sp_namelink")
         try:
@@ -277,9 +295,14 @@ class Article:
         lst = []
         span = soup.find_all('span', class_='sp_text_data')
         for x in span:
+            print(x)
             if x.text not in lst:
                 lst.append(x.text)
-        lst[-1] = int(''.join(lst[-1].split(',')))
+        print(lst)
+        try:
+            lst[-1] = int(''.join(lst[-1].split(',')))
+        except IndexError:
+            return ['','','']
         return lst
 
 
@@ -350,6 +373,7 @@ class Article:
         Input: soup object
         Returns: String containing the text on the article
         """
+        print(url)
         title = soup.find("h1", id = "section_0").text
         description = soup.find("div", class_ = "mf-section-0").text
         steps = soup.find_all("div", class_ = "step")
@@ -515,10 +539,7 @@ class Article:
 
 
     def __repr__(self):
-        print('repr')
-        info = self.title
-        if self.url != None:
-            info += '\n' + self.url
+        info = ''
         if self.author != None:
             info += '\nMain Author: ' + self.author
             if self.author_bio != None:
@@ -526,7 +547,12 @@ class Article:
         else:
             info += '\nNo listed main author'
         info += '\nCoauthored by ' + self.num_coauthors + ' additional authors'
-        info += '\n Viewed by ' +str(self.num_views)
+        if self.verif != '':
+            info += '\nVerification: '
+            verifs = [v for v in ('Expert Co-authored', 'Quality Tested')\
+                      if v in self.verif]
+            info += ', \n'.join(verifs)
+        info += '\nViewed by ' + str(self.num_views)
         if self.num_voters != 0:
             info += '\n' + str(self.pct_helpful) + '% of ' + str(self.num_voters)
             info += ' voters found this article helpful'
@@ -535,10 +561,8 @@ class Article:
         info += '\nLast Updated on ' + self.date_updated
         info += '\nAlso available in '
         info += ', '.join(self.languages)
-        if self.sources != None:
-            info += '\nSources cited: \n'
-
-            info += '\nSources Cited: ' + ', \n'.join(self.sources)
-        else:
+        if self.sources == None:
             info += '\nNo Sources Cited'
         return info
+
+        
\ No newline at end of file

WikiHowHow.py → WikiWikiHow.py

@@ -2,41 +2,16 @@
 import sys
 import PySimpleGUI as sg
 import SearchResults
- 
- 
- 
- 
- 
-def callback_function1():
- 
-  sg.popup('In Callback Function 1')
-  print('In the callback function 1')
- 
-def callback_function2():
+import webbrowser
 
-   sg.popup('In Callback Function 2')
-   print('In the callback function 2')
-
-
-'''
-
-layout = [[sg.Text('Wikihow Tool')],
-         [sg.Text('Search Here'), sg.Input(key = 'Query'), sg.Button('Search')],
-         [sg.Text('Filters', font = ('Helvetica', 15, 'underline'))],
-         [sg.Text('ChildSafety'), sg.CBox('On', key = 'child_safe'), sg.Spin([i for i in range(53)], initial_value=0, key = "date_updated"), sg.Text('Max time (months) since last update')], 
-         [sg.Text('Expert Co-authored'), sg.CBox('', key = 'e_verif'), sg.Text('Quality Tested'), sg.CBox('', key = 'q_verif')],
-         [sg.Text('Minimum % viewers who found it helpful'), sg.Slider(range=(0,100), default_value=50, size=(20,15), orientation='horizontal', font=('Helvetica', 12), key = 'pct_helpful')],
-         [sg.Spin([i for i in range(100)], initial_value=0, key = "num_voters"), sg.Text('Number of people who voted on helpfulness')],
-         [sg.Text('Categories'), sg.CBox('House & Garden', key = 'House'), sg.CBox('Computers & Internet', key = 'CP & I'), sg.CBox('Art and Literature', key = 'Arts'), sg.CBox('Social and Organizational Skills', key = 'Social keys')],
-         [sg.Spin([i for i in range(100)], initial_value=0, key = "num_sources"), sg.Text('Number of sources cited')],
-         [sg.Spin([i for i in range(100)], initial_value=0, key = "num_views"), sg.Text('Views')],
-         [sg.Combo(['English', 'Español', 'Português', 'Italiano', 'Français', 'русский язык', 'Deutsch', '简体中文', 'Nederlands', 'Čeština', 'Bahasa Indonesia', '日本語', 'हिंदी', 'ภาษาไทย',' العَرَبِيةُ', 'Tiếng Việt', '한국말', 'Türkçe',' فارسی']), sg.Text('Language of the Article')],
-         [sg.StatusBar( text=f'| Filters On: {None}', size=(30,1), pad=(0,0), text_color='black', background_color='white', relief=sg.RELIEF_FLAT, justification='left', visible=True, key='status_bar' )]]
-
-'''
+ 
 
 def interact(filters = {}):
+    '''
+    
+    '''
  
+
    layout = [[sg.Text('Wikihow Tool')],
         [sg.Text('Search Here'), sg.Input(key = 'Query'), sg.Button('Search')],
         [sg.Text('Filters', font = ('Helvetica', 15, 'underline'))],
@@ -46,14 +21,13 @@ def interact(filters = {}):
         [sg.Spin([i for i in range(100)], initial_value=0, key = "num_voters"), sg.Text('Number of people who voted on helpfulness')],
         [sg.Spin([i for i in range(100)], initial_value=0, key = "num_sources"), sg.Text('Number of sources cited')],
         [sg.Spin([i for i in range(100)], initial_value=0, key = "num_views"), sg.Text('Views')],
-        [sg.Combo(['English', 'Español', 'Português', 'Italiano', 'Français', 'русский язык', 'Deutsch', '简体中文', 'Nederlands', 'Čeština', 'Bahasa Indonesia', '日本語', 'हिंदी', 'ภาษาไทย',' العَرَبِيةُ', 'Tiếng Việt', '한국말', 'Türkçe',' فارسی']), sg.Text('Available in')]]
+        [sg.Combo(['English', 'Español', 'Português', 'Italiano', 'Français', 'русский язык', 'Deutsch', '简体中文', 'Nederlands', 'Čeština', 'Bahasa', '日本語', 'हिंदी', 'ภาษาไทย',' العَرَبِيةُ', 'Tiếng Việt', '한국말', 'Türkçe',' فارسی'], key = 'language'), sg.Text('Available in')]]
  
    #layout_x = layout
    window = sg.Window('WikiHow Search and Filters', layout)
    results_pg_active = False
  
-   while True:
-       print(results_pg_active)            
+   while True:           
        event, values = window.read()
        if event == sg.WIN_CLOSED:
            break   
@@ -61,31 +35,37 @@ def interact(filters = {}):
            results_pg_active = True
            filters = values
            wh_query = values['Query']
-           print(event, values)
            display_results(filters, results_pg_active)
            results_pg_active = False
    window.close()
  
  
 def display_results(filters, results_pg_active):
-   print(results_pg_active)
    query = filters['Query']
    input_filters = dict(filters)
    if filters['date_updated'] == 0:
        del input_filters['date_updated']
    results = SearchResults.SearchResults(query, input_filters)
-   print(results)
-   layout2 = [[sg.Text('Results')],
-               [sg.Text(results)],
-               [sg.Text(str(filters))]]
-   layout2 = [[sg.Column(layout2, scrollable=True)]]
-   win2 = sg.Window('Window 2', layout2)
- 
-   if results_pg_active:
+   layout2 = [[sg.Text('Results:\n\n')]]
+   for a in results.articles:
+       layout2.append([sg.Text(a.title)])
+       if a.url != None:
+           layout2.append([sg.Text(a.url, enable_events=True)])
+       layout2.append([sg.Text(a.view)])
+       for s in a.sources:
+          layout2.append([sg.Text(s, enable_events=True)])
+       layout2.append([sg.Text('\n\n')])
+   layout2.append([sg.Text(str(filters))])
+   layout3 = [[sg.Column(layout2, scrollable=True)]]
+   win2 = sg.Window('Window 2', layout3)
+ 
+   while results_pg_active:
        ev2, vals2 = win2.read(timeout=100)
        if ev2 == sg.WIN_CLOSED or ev2 == 'Exit':
            results_pg_active  = False
-           win2.close()
+       elif 'http' in ev2:
+           webbrowser.open(ev2, new = 2)
+   win2.close()