Create JSON from FFVL http://www.balisemeteo.com datas

balise_ffvl.py

#! /usr/bin/env python
#
# Parse balisemeteo.com
# create JSONs 
#
# jm@base-jump.org
#       2015
####################
import codecs, sys, os
import time
import xml.dom
import xml.dom.minidom
import urllib

#######################
## Init things
#######################
# get balise num from command line
# or use default
try:
    numBalise = sys.argv[1]
except:
    numBalise = 100 # default balise number. 100 = Col du Joly


# URL wap page
balise_url = "http://www.balisemeteo.com/wap/bd_balise.php?idBalise=" + str(numBalise)

## JSON files
balise_file = "[Full path to JSON dir]/balise_" + str(numBalise) + ".json"
tmp_ext = ".tmp"


## return first value
## (clean up values)
def val( str):
    temp = str.split(" ")
    return temp[0]

# get HTML data
try:
    xmldoc = xml.dom.minidom.parse(urllib.urlopen(balise_url))
    do_balise = True
except:
    print "Error loading HTML file " + balise_url
    do_balise = False

# If HTML loaded
if do_balise:
    #######################
    ## LET'S GO !
    #######################
    print " Create " + balise_file
    paragraphs = xmldoc.getElementsByTagName('p')
    keys = {}
    ## open UTF8 temp file
    with codecs.open(balise_file + tmp_ext, mode="w", encoding="utf-8") as file:
        # start json
        file.write(u"{\"BALISE\":\"" + str(numBalise) + u"\",\n")
        first = True
        # parse each <p> in wap page
        for paragraph in paragraphs:
            # parse text in current paragraph
            for text in paragraph.childNodes:
                line = str(text.nodeValue).strip()
                # if line not empty and <> 'None'
                # start parsing key / values
                if line != "None" and len(line)>0:
                    ## first line is date and time
                    ## no key
                    if first:
                        col = line.split("-")
                        key = col[0].strip()
                        file.write(u"\"DATE\":\"" + key + u"\",\n")
                        file.write(u"\"TIME\":\"" + col[1].strip() + u"\",\n")
                        first = False
                    # parse key / value
                    # or key [value1, value2]
                    else:
                        col = line.split(":")
                        # check if key does'nt already exist in array of keys
                        # if already exist add index, for example: 
                        #   vitesse = average wind speed
                        #   vitesse1 = max wind speed
                        #   vitesse2 = min wind speed
                        key = col[0].strip()
                        try:
                            i = keys[key]
                            keys[key] += 1
                            key = key + str(keys[key])                                
                        except:
                            keys[key] = 0

                        file.write(u"\"" + key + u"\":")
                        try :
                            # chek if one value or two separate by an hyphen
                            args = col[1].find("-")
                            if args > 0:
                                sub = col[1].split("-")
                                # check if hyphen is not a minus ! (temperature)
                                if sub[0].strip():
                                    file.write(u"[\"" + val(sub[0].strip()) + u"\",\"" + val(sub[1].strip()) + u"\"],\n")
                                else:
                                    file.write(u"\"" + val(col[1].strip()) + u"\",\n")
                            else:
                                file.write(u"\"" + val(col[1].strip()) + u"\",\n")
                        except IndexError:
                            # do nothing ...
                            t=0
        # end JSON, write date of JSON file creation
        file.write(u"\"DATEJSON\":\"" + time.strftime("%Y-%m-%d %H:%M") + u"\"}")    

    ## rename temp file
    os.rename(balise_file + tmp_ext, balise_file)

sample_balise_100.json

{"BALISE":"100",
"DATE":"20/03/2015",
"TIME":"16:17",
"Direction":["NO","315"],
"Vitesse":"0",
"Direction1":["NNO","337"],
"Vitesse1":"7",
"Vitesse2":"0",
"Temp":"10",
"DATEJSON":"2015-03-20 16:30"}

Bug fix :
check if the hyphen is not a minus (temperature).
If first arg after splitting is empty -> the hyphen is in fact a minus (temperature)
if first arg not empty -> make an array with 2 values (wind speed, direction, ...)