#!/usr/bin/env python
# coding: utf-8

# # Is my life boring?
# 
# When my life is interesting I take a lot of photos. Lets see if I can plot these photos to see if my life is getting more interesting. 

# In[1]:


base_dir = "/home/visgean/Dropbox/**/*"


# In[2]:


import glob
import os

def get_extension(filename):
    filename, file_extension = os.path.splitext(filename)
    return file_extension.lower()


picture_extensions = ['.jpg', '.jpeg', '.png']
pictures = list(filter(
    lambda f: get_extension(f) in picture_extensions,
    glob.iglob(base_dir, recursive=True)
))


# In[3]:


print('# of pics:', len(pictures))


# In[4]:


import exifread

with open(pictures[40], 'rb') as f:
    tags = exifread.process_file(f)
    
print(tags.keys())


# In[5]:


date = tags['Image DateTime']


# In[6]:


print(date.values)


# In[7]:


from datetime import datetime
datetime.strptime(str(date.values), '%Y:%m:%d %H:%M:%S')


# In[8]:


def parse_date(exif_info):
    if not exif_info:
        return None    
    date = exif_info.get('Image DateTime')
    if not date:
        return None
    try:
        return datetime.strptime(str(date.values), '%Y:%m:%d %H:%M:%S').date()
    except:
        print(date)


# In[9]:


def get_exif(filename):
    try:
        with open(filename, 'rb') as f:
            return exifread.process_file(f)
    except:
        return None


# In[10]:


exif_data = list(map(get_exif, pictures)) # i want to iterate through these multiple times


# In[11]:


dates = list(map(parse_date, exif_data))


# In[12]:


from collections import Counter

photos_per_day = Counter(dates)


# In[13]:


print('Missing dates', photos_per_day[None])
del photos_per_day[None]


# In[14]:


print('Missing dates', photos_per_day[None])


# In[15]:


from bokeh.plotting import figure, show, output_notebook
output_notebook()


# In[16]:


time_graph = figure(
    title="Pics over time", 
    background_fill_color="#E8DDCB",
    y_axis_label='# of pics', 
    x_axis_label='Time',
    x_axis_type="datetime"
)

sorted_vals = photos_per_day.most_common()

time_graph.circle([x[0] for x in sorted_vals], [x[1] for x in sorted_vals])
show(time_graph)


# As you can see that is a bit messy so lets try to collect it by month

# In[17]:


photos_per_month = Counter([datetime(d.year, d.month, 1) for d in dates if d])


# In[18]:


time_graph = figure(
    title="Pics per month", 
    background_fill_color="#E8DDCB",
    y_axis_label='# of pics', 
    x_axis_label='Time',
    x_axis_type="datetime"
)

sorted_vals = photos_per_month.most_common()

time_graph.circle([x[0] for x in sorted_vals], [x[1] for x in sorted_vals])
show(time_graph)


# In[21]:


print(len(list(filter(None, dates))))


# I was only able to parse about 60% of my photos - rest of them comes from facebook or dont have dates at all :/ But its still nice to see that I am taking more and more photos.