G+ summary

parser.py

from bs4 import BeautifulSoup
import re
import sys

def include_all_class(classes):
  def check_class(tag):
    result = True
    for c in classes:
      if not tag.has_key('class') or not c in tag['class']:
        result = False
        break
    return result
  return check_class

def show_summary(file_name):
  total_post = 0
  reshare_post = 0
  original_post = 0
  public_post = 0
  limit_post = 0
  public_original_post = 0
  limit_original_post = 0

  soup = BeautifulSoup(open(file_name))
  posts = soup.find_all(id=re.compile('update-'))
  total_post = len(posts)
  for post in posts:
    header = post.find('header')
    time_label = header.span.span.a
    view = header.span.contents[2].string
    if view == "Mobile" or view[:7] == "+1'd on" or view == "Google Reader" \
      or view == "Hangout":
      view = header.span.contents[4].string
    is_public = True
    if view[:6] == "Public" or view == "Community":
      public_post += 1
    else:
      limit_post += 1
      is_public = False
    check_reshare1 = post.find_all(include_all_class(["uGbq2"]))
    check_reshare2 = post.find_all(include_all_class(["DTMmlc"]))
    if check_reshare1 or check_reshare2:
      reshare_post += 1
    else:
      original_post += 1
      if is_public:
        public_original_post += 1
      else:
        limit_original_post += 1

  print("summary:")
  print("  Total : {0}".format(total_post))
  print("    Original: {0}".format(original_post))
  print("    Reshared: {0}".format(reshare_post))
  print("  Public: {0}".format(public_post))
  print("    Original: {0}".format(public_original_post))
  print("    Reshared: {0}".format(public_post - public_original_post))
  print("  Limit : {0}".format(limit_post))
  print("    Original: {0}".format(limit_original_post))
  print("    Reshared: {0}".format(limit_post - limit_original_post))

if __name__ == "__main__":
  file_name = "source"
  if len(sys.argv) > 1:
    file_name = sys.argv[1]
  show_summary(file_name)