# coding: utf-8

# debianforum.de donation plot
# written by TRex, modified by Meillo


from sys import argv

if len(argv) != 2:
    print "Usage: "+ argv[0] +" OUTFILE.png"
    exit(1)
outfile = argv[1]

# In[1]:

from datetime import datetime

years = range(2003, datetime.now().year + 1)
donations_url = "https://wiki.debianforum.de/Debianforum.de/Spenden"
expenses_url = "https://wiki.debianforum.de/Debianforum.de/Ausgaben"


# In[2]:

import urllib
urls = {year: (donations_url + '/' + str(year) if year < years[-1] else donations_url) for year in years}

y = dict()
for year, url in urls.items():
    f = urllib.urlopen(url)
    y[year] = f.read()
    
expense_html = urllib.urlopen(expenses_url).read()


# In[3]:

def parse_html_to_list(html):
    """
    Parse HTML table to list of tuples, expecting three cells with date, note, amount in each row.
    """
    data = []
    parser = etree.HTMLParser()
    tree = etree.parse(StringIO(html.decode('utf-8')), parser)
    rows = tree.xpath('//table//tr')
    for index, row in enumerate(rows):
        cells = row.getchildren()
        if len(cells) == 3 and index > 0:
            date = cells[0].text.strip()
            try:
                date = datetime.strptime(date, "%d.%m.%Y")
            except Exception as e:
                continue
            note = cells[1].xpath(".//a/text()")
            if not note:
                note = cells[1].xpath(".//text()")
            amount = float(cells[2].text.strip().replace(",","."))

            data.append((date, note[0].strip(), amount))
    return data
    


# In[4]:

from io import StringIO, BytesIO
from lxml import etree


donations = []

for year, donate_html in y.items():
    yearly_list = parse_html_to_list(donate_html)
    donations += yearly_list
donations.sort()



# In[5]:

expenses = parse_html_to_list(expense_html)


# In[6]:

sum([x[2] for x in expenses])


# In[7]:

sum([x[2] for x in donations])


# In[8]:

sum([x[2] for x in donations]) - sum([x[2] for x in expenses])


# Donations/month

# In[9]:

from itertools import groupby
import numpy as np
import matplotlib
matplotlib.use('agg')
import matplotlib.pyplot as plt
from datetime import date, timedelta
#get_ipython().magic(u'matplotlib inline')

def _groupy(item):
    return item[0].year, item[0].month

def get_date_grouped_list(list_):
    agg_list = []
    for ( (year, month), items ) in groupby( list_, _groupy ):
        agg_list.append( ( date(year, month, 1), sum([x[2] for x in items])) )
    return agg_list

donations_prefix = [(date(donations[0][0].year, x+1, 1), '', 0) for x in range(0, donations[0][0].month-1)] # cheating. Now even better.
agg_donations = get_date_grouped_list(donations_prefix + donations)
agg_expenses = get_date_grouped_list(donations_prefix + expenses)

agg_donations.sort()
agg_expenses.sort()

fig = plt.figure(figsize=(15,7))
# donations
plt.plot([x[0] for x in agg_donations], [x[1] for x in agg_donations], color="green", label="Spenden")
# expenses
plt.plot([x[0] for x in agg_expenses], [x[1] for x in agg_expenses], color="red", label="Ausgaben")

# spartopf
cash_available = []
cash = 0
d = dict(agg_donations)
# shitty workaround: this month is missing. probably a bug in the source table.
for k,v in agg_expenses + [(date(2009, 3, 1), 0)]:
    donation = d.pop(k, 0)
    cash = cash - v + donation
    
    cash_available.append((k, cash))
cash_available.sort()

plt.plot([x[0] for x in cash_available], [x[1] for x in cash_available], color="blue", label="Spendenkonto")
plt.xlabel("Datum")
plt.ylabel("Euro")
plt.legend()


plt.savefig(outfile)
plt.close(fig)


# In[10]:

exit()

# csv thing
for x in agg_plot:
    print "%s;%.2f" % (x[0].strftime("%Y-%m-%d"), x[1])


# In[ ]:

# csv thing
for x in donations:
    print "%s;\"%s\";%.2f" % (x[0].strftime("%Y-%m-%d"), x[1], x[2])