#! /usr/bin/env python

""" read daily dat fromShannon Airport at
    http://www.met.ie/climate/daily-data.asp
    Data returned contains 7 or 8 elements:
    0       1        2    3    4        5                6         7
    Date	Rainfall Temp Temp Sunshine Gusts            Windspeed gmin
            (mm)	 Max  Min  (hours)  (if >= 34 knots) (knots)   (Centigrade)
"""

import sys
from   HTMLParser import HTMLParser

def read_html_data(mydata):
    """ Extract relevant information from
        http://www.met.ie/climate/daily-data.asp data stream.
        We are interested in the daily data from Shannon Airport.
        The data is embedded in a longs stream in side various tables.
        The interesting bit starts with the word "gmin".
        The next table row contains the measured data, starting with the
        date in the format %d/%m/%Y.
    """

    class MyHTMLParser(HTMLParser):
        """my HTML browser. We have to subclass the
           tags 'table', 'tr', 'th' and 'td'.
        """
        def __init__(self):
            "Standard init plus extras."
            HTMLParser.__init__(self)
            self._go       = False
            self.table     = []
            self.row       = []
            self.tags      = {
                'tr'    : 1,
                'table' : 1,
                'td'    : 1,
            }
            self.result  = []

        def handle_starttag(self, tag, attrs):
            "start tag: set state, call method via getattr."
            #print "handle_starttag %s" % tag
            self._state = tag
            if tag in self.tags:
                try:
                    getattr(self, tag + '_start')(attrs)
                except AttributeError:
                    pass

        def handle_endtag(self, tag):
            "end tag: reset state, call method via getattr."
            #print "handle_endtag !%s!" % tag
            self._state = ""
            if tag in self.tags:
                try:
                    getattr(self, tag + '_end')()
                except AttributeError:
                    pass

        def tr_start(self, attrs):
            "Handle a new row: empty row element"
            print "tr_start"
            self.row = []

        def tr_end(self):
            "End of a row."
            # attach row to self.table
            if self._go:
                self.table.append(self.row)
                self.row = []
            print "tr_end"

        def td_start(self, attrs):
            if self._go:
                print "td_start"
            self.td_data = True
            self.found_td_data = False

        def td_end(self):
            self.td_data = False
            if not self.found_td_data:
                self.row.append('')
            if self._go:
                print "td_end"

        def table_end(self):
            "Reset table state: save result."
            if self._go:
                self.result = self.table[1]
                self._go = False
            print "table_end"

        def handle_data(self, data):
            """Check for data."""
            # we must be triggered by "gmin"
            ###data = data.strip()
            if data == 'gmin':
                self._go = True
            if self._go and self.td_data:
                print "data=!%s!" % data
            if self._go and self.td_data:
                self.row.append(data)
                self.found_td_data = True

    html  = MyHTMLParser()
    html  . feed(mydata)
    html  . close()
    return html.result

def get_data(filename):
    handle = open(filename, 'r')
    data = handle.read()
    handle.close()
    return read_html_data(data)

def main(filename):
        gimme = get_data(filename)
        print ';'.join(gimme)

if __name__ == '__main__':
    main(sys.argv[1])