wc-timeseries.sh

#!/bin/bash

# wc-timeseries.sh
# Plot word count over time for a latex project under git version control.

# Iterate over git commits
# Append date, word count and commit sha1 to a csv file
echo date,count,sha1 > /tmp/wc-timeseries.csv
for line in $(git log --decorate=no --reverse --format="format:%H,%aI")
do
    sha1=$(echo $line | cut -f 1 -d ,)
    date=$(echo $line | cut -f 2 -d ,)
    git checkout --force --quiet $sha1
    wordcount=$(texcount $(find . -iname "*.tex") | grep "Words in text: " | tail -n 1 | cut -f 4 -d " ")
    echo $date,$wordcount,$sha1
done >> /tmp/wc-timeseries.csv 2> /tmp/wc-timeseries-stderr.txt

# Plot timeseries
python -c """
import pandas as pd
import matplotlib.pyplot as plt
df = pd.read_csv('/tmp/wc-timeseries.csv', index_col=0)
df.index = pd.to_datetime(df.index)
df['count'].plot(clip_on=False, zorder=20)
plt.ylim(0, plt.gca().get_ylim()[1])
plt.ylabel('Word count')
plt.xlabel('Date')
plt.savefig('wc-timeseries.png', dpi=200)
plt.close()
"""

# Back to the original state of the repo
git checkout master