Re-inject exif metadata into photos in a facebook archive

inject_exif.py

#!/usr/bin/python
from HTMLParser import HTMLParser
from PIL import Image
import piexif
import datetime
import sys
from fractions import Fraction

def rational(s):
  f = Fraction(s).limit_denominator(10000000)
  return (f.numerator, f.denominator)

def dms(s):
  dd = float(s)
  is_positive = dd >= 0
  dd = abs(dd)
  minutes,seconds = divmod(dd*3600,60)
  degrees,minutes = divmod(minutes,60)
  degrees = degrees if is_positive else 360 - degrees
  return (rational(degrees),
          rational(minutes),
          rational(seconds))

def exif_date(s):
  return datetime.datetime.fromtimestamp(
    int(s)).strftime('%Y:%m:%d %H:%M:%S')

class ArchiveParser(HTMLParser):
  def __init__(self):
    HTMLParser.__init__(self)
    self.contents = False
    self.div = 0
    self.path = ''
    self.key = ''
    self.isKey = False
    self.isValue = False
    self.exif_data = dict()

  def handle_starttag(self, tag, attrs):
    attr = dict(attrs)
    if self.contents:
      if 'class' in attr and attr['class'] == 'block':
        self.div = 0
        self.path = ''

      if tag == 'div':
        self.div += 1

      if self.div > 0:
        if tag == 'img':
          if 'src' in attr:
            self.path = attr['src']
        elif tag == 'th':
          self.isKey = True
        elif tag == 'td':
          self.isValue = True
    else:
      if 'class' in attr and attr['class'] == 'contents':
        self.contents = True
        
  def handle_endtag(self, tag):
    if self.contents:
      if tag == 'div':
        self.div -= 1
        if self.div == -1:
          self.contents = False
        if self.div == 0:
          # write exif data here
          if self.path:
            img = Image.open(self.path)
            if 'exif' in img.info:
              exif_dict = piexif.load(img.info["exif"])
            else:
              exif_dict = {'0th': dict(), 'Exif': dict(), 'GPS': dict()}

            for k,v in self.exif_data.iteritems():
              if k == 'Taken':
                exif_dict['Exif'][piexif.ExifIFD.DateTimeOriginal] = exif_date(v)
              elif k == 'Camera Make':
                exif_dict['0th'][piexif.ImageIFD.Make] = v
              elif k == 'Camera Model':
                exif_dict['0th'][piexif.ImageIFD.Model] = v
              elif k == 'Orientation':
                exif_dict['0th'][piexif.ImageIFD.Orientation] = int(v)
              elif k == 'Exposure':
                exif_dict['Exif'][piexif.ExifIFD.ExposureTime] = rational(v)
              elif k == 'F-Stop':
                exif_dict['Exif'][piexif.ExifIFD.FNumber] = rational(v)
              elif k == 'ISO Speed':
                exif_dict['Exif'][piexif.ExifIFD.ISOSpeed] = int(v)
              elif k == 'Focal Length':
                exif_dict['Exif'][piexif.ExifIFD.FocalLength] = rational(v)
              elif k == 'Latitude':
                exif_dict['GPS'][piexif.GPSIFD.GPSLatitude] = dms(v)
              elif k == 'Longitude':
                exif_dict['GPS'][piexif.GPSIFD.GPSLongitude] = dms(v)
              elif k == 'Modified':
                exif_dict['0th'][piexif.ImageIFD.DateTime] = exif_date(v)
              else:
                print 'Unhandled Key: ' + k + ': ' + v
            exif_bytes = piexif.dump(exif_dict)
            img.save(self.path, exif=exif_bytes)

          self.path = ''
          self.key = ''
          self.isKey = False
          self.isValue = False
          self.exif_data = dict()
      elif tag == 'th':
        self.isKey = False
      elif tag == 'td':
        self.isValue = False

  def handle_data(self, data):
    if self.isKey:
      self.key = data
    elif self.isValue and self.key:
      self.exif_data[self.key] = data

# Parse a facebook 'index.htm' file and re-inject the exif data
if len(sys.argv) > 1:
  parser = ArchiveParser()
  with open(sys.argv[1], 'r') as fp:
    content = fp.read()
    parser.feed(content)