From bb88bbeb24dfed168bf19d9244eb95c682b38d7a Mon Sep 17 00:00:00 2001
From: Thomas Wade <gigatech55@gmail.com>
Date: Sat, 29 Sep 2018 13:29:08 +0930
Subject: [PATCH] Convert human-readable numbers to ints

---
 scrape.py | 30 ++++++++++++++++++++++--------
 1 file changed, 22 insertions(+), 8 deletions(-)

diff --git a/scrape.py b/scrape.py
index bdade95..d5bdee5 100755
--- a/scrape.py
+++ b/scrape.py
@@ -121,25 +121,25 @@ def Get_Basic_Info(user):
     # Grab their track count
     track_count_elem = soup.find('a', href='/' + user.url_username + '/tracks', class_='infoStats__statLink')
     if track_count_elem:
-        user.track_count = str(track_count_elem.div.string)
+        user.track_count = Human_Str_To_Int(str(track_count_elem.div.string))
     print('    Tracks: ' + str(user.track_count))
 
     # Grab their following count
     following_count_elem = soup.find('a', href='/' + user.url_username + '/following', class_='infoStats__statLink')
     if following_count_elem:
-        user.following_count = str(following_count_elem.div.string)
+        user.following_count = Human_Str_To_Int(str(following_count_elem.div.string))
     print('    Following: ' + str(user.following_count))
 
     # Grab their follower count
     follower_count_elem = soup.find('a', href='/' + user.url_username + '/followers', class_='infoStats__statLink')
     if follower_count_elem:
-        user.follower_count = str(follower_count_elem.div.string)
+        user.follower_count = Human_Str_To_Int(str(follower_count_elem.div.string))
     print('    Followers: ' + str(user.follower_count))
 
     # Grab their like count
     like_count_elem = soup.find(class_='sidebarHeader__actualTitle')
     if like_count_elem:
-        user.like_count = str(like_count_elem.string.split(' ')[0])
+        user.like_count = Human_Str_To_Int(str(like_count_elem.string.split(' ')[0]))
     print('    Likes: ' + str(user.like_count))
 
 # Gets a list of fully-populated Track objects
@@ -212,19 +212,19 @@ def Get_Tracks_Info(url, limit = 100):
             for elem in play_comment_elems:
                 num, unit = str(elem['title']).split(' ')
                 if unit == 'plays':
-                    track.plays = num
+                    track.plays = Human_Str_To_Int(num)
                 elif unit == 'comments':
-                    track.comments = num
+                    track.comments = Human_Str_To_Int(num)
 
         # Grab the like count
         like_elem = track_elem.find(class_='sc-button-like')
         if like_elem:
-            track.likes = str(like_elem.string.strip())
+            track.likes = Human_Str_To_Int(str(like_elem.string.strip()))
 
         # Grab the repost count
         repost_elem = track_elem.find(class_='sc-button-repost')
         if repost_elem:
-            track.reposts = str(repost_elem.string.strip())
+            track.reposts = Human_Str_To_Int(str(repost_elem.string.strip()))
 
         # Finally add the track to the dictionary
         tracks[track.url] = track
@@ -276,6 +276,20 @@ def Get_Follows(url, limit = 100):
 
     return users.values()
 
+def Human_Str_To_Int(string):
+    num = None
+    string = string.lower().replace(',', '')
+    if 'k' in string:
+        string = string.replace('k', '')
+        num = int(float(string) * 1000)
+    elif 'm' in string:
+        string = string.replace('m', '')
+        num = int(float(string) * 1000000)
+    else:
+        num = int(string)
+
+    return num
+
 ##### Scrape ################################################
 
 user_dict = {}