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

# # Counting lines of code in a git repo.

# In[1]:


import importnb
from pathlib import Path
from pandas import *
import mimetypes
from toolz.curried import *
from poser import *
from toolz.curried import *


# In[2]:


mimetypes.add_type('application/x-ipynb+json', '.ipynb')
mimetypes.add_type('text/markdown', '.md')


# In[3]:


def files_by_type():
    contents = list(filter(
        compose(all, juxt(
            Path.is_file, 
            lambda x: all(
                s not in str(x)for s in ('__pycache__', 'checkpoint')
            ))), 
        Path(importnb.__file__).parent.parent.rglob('*')))
    return Series(pipe(
        contents, map(compose(first, mimetypes.guess_type, str))
    ), index=contents, name='mime').to_frame().reset_index().set_index('mime').pipe(
        lambda df: df.groupby(df.index).agg(compose(list, unique))
    )['index'].apply(Series)


# In[4]:


def count_lines(str):
    significant, empty, comment = 0, 0, 0
    for line in str.splitlines():
        if line.strip():
            if line.lstrip().startswith('#'): comment +=1 
            else: significant += 1
        else: empty += 1
    return Series(dict(zip(('significant', 'empty', 'comment'), (significant, empty, comment))))


# In[8]:


import json
from pandas import concat as Concat
from IPython.utils.capture import capture_output


# In[9]:


def nb_info(df):
    """Number of cells maybe"""
    notebooks = df.loc['application/x-ipynb+json'].dropna().apply(compose(
        Series, get('cells'), json.loads, Path.read_text
    )).set_index(df.loc['application/x-ipynb+json'].dropna().values).stack().apply(
        lambda x: Series({'cell_type': x['cell_type'], 'source': '\n'.join(x.get('source'))})
    ).set_index('cell_type', append=True).reorder_levels((2,0,1))
    notebooks = notebooks.reset_index(0, drop=True).join(
        notebooks.loc['code']['source'].apply(count_lines)
    ).set_index(notebooks.index).unstack(-1)
    return notebooks['significant'].sum(axis=1).unstack(0).fillna(0)['code']


# In[ ]:





# In[12]:


from IPython import get_ipython


# In[13]:


def iterate_over_the_project(project, max=10, iter=20, **loc):
    repo, project = project
    if not Path(project).exists():
        get_ipython().system('git clone $repo')
    get_ipython().system('cd $project && git pull')
    for i in range(max):
        with capture_output() as time:
            get_ipython().system('cd $project && git log -1 --format=%cd --date=local')
        loc[to_datetime(time.stdout)] = compose(nb_info, files_by_type)()
        get_ipython().system('cd $project && git reset --hard HEAD~$iter')
    get_ipython().system('cd $project && git pull')
    return loc


# In[11]:


def _test_with_importnb():
    assert iterate_over_the_project(('https://github.com/deathbeds/importnb', 'importnb'))


# In[16]:


if __name__ == '__main__':
    get_ipython().run_line_magic('matplotlib', 'inline')
    Concat(loc).unstack(-1).fillna(0).sum(axis=1).plot()


# In[ ]: