textanalysis.py

# textanalysis.py
# 
# Analyzing iphone text message csv dump from iBackup Viewer 
# (http://www.imactools.com/iphonebackupviewer/)
#
# Requirements: 
# - WordCloud (http://github.com/amueller/word_cloud)
#
# How to use:
# 1. Generate csv dump via iphonebackupviewer and point to it in __main__
# 1a. Delete the first line of the CSV dump, it's a header.
# 2. Change numbers param in __main__ to be the 2 phone numbers you're 
#    interested in  (or just 1 is okay too) and the relative file path of the 
#    csv to this program.
# 3. Choose a TrueType font on your system to use with the wordcloud generator
#    and point to it with the font variable in __main__ 
#
# OR write your own program that sets these params for the local system.
# This is just a couple-hour hack for fun. DWTFYW license.
#
# Original Author: Sam Liu <sam@ambushnetworks.com>

from wordcloud import WordCloud, STOPWORDS
import datetime
import os
import re


class WCGenerator():
  # Word cloud generator from text messages between two people!
  def __init__(self, input_file='', words_to_remove=None, 
               font='', 
               numbers=None):
    # Note: Replace font parameter with the path of a font on your system.
    self.input_file = input_file
    self.words_to_remove = words_to_remove
    self.font = font
    self.numbers = numbers
 
  def generate_word_clouds(self):
    f = open(self.input_file)
    print "Reading from " + self.input_file + "..."
    for line in f.readlines():
      values = line.split(',')
      number = values[0]

      datestring = ''.join(values[1:4])
      datestring = datestring.strip('"')
      timestamp = datetime.datetime.strptime(datestring, '%b %d %Y %H:%M %p');
      # You can look at a specific month or day or anything using the timestamp. 
      # I parsed the format provided by ibackupviewer for your convenience.
      # Example:
      # if timestamp.month != 7:
      #   continue
      
      # Stick all the text bodies into one string, removing unwanted words 
      # from being put into the word cloud.
      if number in self.numbers:
        msg = values[5]
        if self.words_to_remove:
          remove = '|'.join(words_to_remove)
          regex = re.compile(r'('+remove+')', flags=re.IGNORECASE)
          msg = regex.sub("", msg)
        self.numbers[number]['text'] += msg 
      else:
        print "ERROR: " + number + " was not a valid number."

    # Make wordcloud from each string and save image with filename using the 
    # number.
    d = os.path.dirname(__file__)
    wc = WordCloud(font_path=self.font)
    for number in self.numbers:
      if not self.numbers[number]['text']: 
        continue
      wc.generate(self.numbers[number]['text'])
      filename = number[1:]  # Remove the prepended '+' for the filename.
      filepath = os.path.join(d, filename + '.png')
      wc.to_file(filepath)
      print "Wrote file: " + filepath 
    
if __name__ == '__main__':
  # TODO(samcliu): Use argparse to take cmdline args.
  numbers = {
    '+16505551234' : { 'name': 'Name1', 'text': '' },
    '+16505554321' : { 'name': 'Name2', 'text': '' },
  }
  input_file = 'putyourfilename.csv'
  # Assuming you like DroidSansMono and have it installed...
  font = '/Users/youruser/Library/Fonts/DroidSansMono.ttf'

  wc_generator = WCGenerator(numbers=numbers, input_file=input_file, 
                             font=font)
  wc_generator.generate_word_clouds()