Create an animated heat map .gif from your lastfm scrobbler history

README.md

Lastfm Heat Map gif

Description

Use as input file the output of lastfm-export.py. If you don't want to download one you could use mine here.

heatmaps.py generate the heat maps images on the current directory.
heatmap-makegif.sh creates the final gif and these directories smalldir/outputgif/ also it does all the gif compression for you.

Usage

# ls exported_tracks.txt
exported_tracks.txt
# python2 heatmaps.py
# bash heatmap-makegif.sh
# display smalldir/outputgif/beat_o3.gif

heatmap-makegif.sh

#!/bin/bash

hash convert || { echo 'Error: convert was not found on PATH, please install ImageMagick'; exit 1; }
hash gifsicle || { echo 'Warning: gifsicle not found on PATH, you might want to install it to obtain a optmized gif file.'; }

set -x

rm -rf smalldir
mkdir smalldir/

for i in ./beat*; do
  convert $i -resize %40 smalldir/${i##*/}
done

cd smalldir && {
  mkdir outputgif
  # "fix" bug on the python script that generates wrong heat maps...by deleting the latest ones
  ls beat* | sort -n -k1.5,1.7n | tail -n 6 | xargs -n1 rm -v
  convert \
                 -crop -20-20 \
                 +repage \
                 -loop 0 \
                 -delay 25 \
                 -layers OptimizeTransparency \
                 -ordered-dither o8x8,14 \
                 +map \
                 $(ls beat* | sort -n -k1.5,1.7n) \
                 outputgif/beat.gif
  cd outputgif && {
    gifsicle -O3 < beat.gif > beat_o3.gif
  }
}

heatmaps.py

#!/usr/bin/env python
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# df = pd.read_csv('./exported_tracks.new.txt.gz', compression='gzip',
df = pd.read_csv('./exported_tracks.txt',
              sep='\t',
              header=None,
              names=['date','tr_name','ar_name','al_name',
                       'tr_duration','tr_listeners',
                       'tr_playcount',
                       'tr_t0','tr_t1','tr_t2','tr_t3','tr_t4',
                       'tr_mbid','ar_yf','ar_yt',
                       'ar_ctry','ar_listeners','ar_playcount',
                       'ar_t0','ar_t1','ar_t2','ar_t3','ar_t4',
                       'ar_mbid',
                       'al_releasedate','al_playcount',
                       'al_listeners','al_mbid'],
              parse_dates=False)


df.index = pd.to_datetime(df.pop('date'), unit='s')


df.tr_duration = df.tr_duration.apply(lambda x: x/60/1000)


df['al_releasedate'] = df.al_releasedate.apply( lambda x: np.nan if x == '    ' else x)
df['al_releasedate'] = pd.to_datetime(df.al_releasedate, format='    %d %b %Y, %H:%M')

b = df.groupby(pd.TimeGrouper('d')).ar_t0.apply(lambda x: x.value_counts().head(30)).unstack(level=1).fillna(0)\
  .add(df.groupby(pd.TimeGrouper('d')).ar_t1.apply(lambda x: x.value_counts().head(30)).unstack(level=1).fillna(0), fill_value=0)\
    .add(df.groupby(pd.TimeGrouper('d')).ar_t2.apply(lambda x: x.value_counts().head(30)).unstack(level=1).fillna(0), fill_value=0)\
      .add(df.groupby(pd.TimeGrouper('d')).ar_t3.apply(lambda x: x.value_counts().head(30)).unstack(level=1).fillna(0), fill_value=0)\
          .add(df.groupby(pd.TimeGrouper('d')).ar_t4.apply(lambda x: x.value_counts().head(30)).unstack(level=1).fillna(0), fill_value=0)
b_t = b.T

b_t['svalue'] = b_t.sum(axis=1).values
b_t_plot = b_t.sort_index(by='svalue', ascending=True).drop('svalue',axis=1)[-30:]#.sort_index()

current_plot = 0

cols_max = len(b_t_plot.columns)

for slice_value_tip in xrange(0,len(b_t_plot.columns),5):
  slice_value_top = slice_value_tip + 30
  df_temp = b_t_plot.iloc[:,slice_value_tip:slice_value_top]
  print slice_value_tip, slice_value_top, cols_max

  fig, ax = plt.subplots()
  heatmap = ax.pcolor(df_temp,
                      cmap=plt.cm.YlGn,
                      alpha=0.8,
                      vmin=0,
                      vmax= b_t_plot.max().max() / 4)
  
  fig = plt.gcf()
  fig.set_size_inches(7,7)
  # turn off the frame
  ax.set_frame_on(False)
  
  # put the major ticks at the middle of each cell
  ax.set_yticks(np.arange(df_temp.shape[0])+0.1, minor=False)
  ax.set_xticks(np.arange(df_temp.shape[1])+0.5, minor=False)
  #ax.set_xticks(minor=False)
  # want a more natural, table-like display
  ax.invert_yaxis()
  ax.xaxis.tick_top()
  
  # Set the labels
  
  # note I could have used nba_sort.columns but made "labels" instead
  ax.set_xticklabels(df_temp.columns.map(lambda x: x.year), minor=False, size='small') 
  ax.set_yticklabels(df_temp.index, minor=False, va='top',size='small')
  
  ax.grid(False)
  
  # rotate the 
  plt.xticks(rotation=90)
  
  # Turn off all the ticks
  ax = plt.gca()
  
  for t in ax.xaxis.get_major_ticks(): 
      t.tick1On = False 
      t.tick2On = False 
  for t in ax.yaxis.get_major_ticks(): 
      t.tick1On = False 
      t.tick2On = False  
  
  fig.savefig('./beat' + str(current_plot) + '.png')
  current_plot = current_plot + 1
  del df_temp
  plt.clf()