Initial commit: alpha version of my own, working refresh script

1 files changed, 133 insertions, 0 deletions

diff --git a/refresh b/refresh
new file mode 100755
index 0000000..c5d308f
--- /dev/null
+++ b/refresh
@@ -0,0 +1,133 @@
+#!/usr/bin/env python
+# -*- encoding: utf-8 -*-
+# vim:syntax=python
+# http://holidata.net/en-US/2016.json
+
+"""
+This script downloads data from holidata.net and writes it into files in the current directory.
+Its output formats are parseable by timewarrior (https://taskwarrior.org/docs/timewarrior/).
+
+This is very alpha. Since holidata.net has a few minor mistakes for Germany I'm writing this to
+    test.xx-YY for now (I've just reported these, so allow the guys some time :) ). The files
+    are still sourcable with that name :)
+
+Also, for now this script only downloads de-DE. You will notice the complete lack of a main()
+    definition, that is the next step: it will also carry parameters so you can download your
+    own stuff.
+
+Finally, I'm fixing the ISO-3166-2 violation for North Rhine Westphalia - holidata.net has de-NRW.
+    That is the next bug to fix for them :)
+
+tl;dr
+SCRIPT IS NOT FINISHED, THIS IS A GIT - USE ONLY IF YOU UNDERSTAND AND IF YOU ARE FROM ZE GERMANY
+"""
+
+#import os, sys, re, time, argparse
+import json, sys
+reload(sys)
+sys.setdefaultencoding('utf8')
+from datetime import datetime
+from time import sleep
+
+if sys.version_info >= (3, 0):
+    from urllib.request import urlopen
+    from urllib.error import HTTPError
+else:
+    from urllib2 import urlopen, HTTPError
+
+__hheader__ = "# File auto-generated by lirion.de's refresh,\n" \
+            "# Date: " + datetime.strftime(datetime.now(), "%Y-%m-%d %H:%M:%S %z") + "\n" \
+            "# data provided by holidata.net\n\n" \
+            "define holidays:\n"
+
+def headwrite(file):
+    """
+    This writes the default header to the file it is being passed on. Content see
+    __hheader__ right above.
+    """
+    try:   
+        open(file, 'w').close()
+    except IOError:
+        print("Cannot write to %s!", file)
+        exit(2)
+    with open(file, "a") as ofile:
+        #f = open(ofile, "a")
+        ofile.write(__hheader__)
+ 
+def hfetch(locode, year):
+    """
+    This function downloads the holiday data for a given ISO-3166 code from holidata.net
+    It will return a single string containing all of the JSON.
+    """
+    myurl = "https://holidata.net/" + mylocode + "/" + myyear + ".json"
+    try:
+        lines = urlopen(myurl).read().decode('utf-8')
+    except HTTPError as httpe:
+        if httpe.code == 404:
+            print("holidata.net does not have data for %s(%s), for %d." % (mylocale, myregion,
+                mydate))
+        else:
+            print(httpe.code, httpe.read())
+    return lines
+
+def hparse(lines, locode, year):
+    """
+    This function parses the JSON string we expect from holidata.net and writes its fragments to
+    the specific region files, if a region is given. So, e.g.: a holiday for all of the US will
+    not have a region definition and thus be written to file.en-US, whereas a holiday that is not
+    valid for all of the US but is valid for Oregon will be written to file.en-OR.
+    """
+    islocdef = { 'XY-ZZ': 1 }
+    for line in lines.split('\n'):
+        if line:
+            #sys.stdout.write(".")
+            jdata = json.loads(line)
+            mylocale = unicode(jdata['locale'])
+            mycountry = mylocale.split('-')[0]
+            myregion = unicode(jdata['region'])
+            if myregion == "":
+                myregion = mylocale.split('-')[1]
+            # holidata.net violates ISO-3166-2:DE here:
+            if myregion == "NRW":
+                myregion = "NW" # or Neustadt/Weinstr., Palz > Pott
+            myhlocale = unicode(mycountry + "-" + myregion)
+            if jdata['date'] == "":
+                print("%s has no date given for %s!", jdata['description'], jdata['locale'])
+                mydate = unicode('1970_01_01')
+            else:
+                mydate = unicode(datetime.strftime(datetime.strptime(jdata['date'], '%Y-%m-%d'),
+                    '%Y_%m_%d'))
+            myfile = "test." + myhlocale
+            if myhlocale not in islocdef:
+                headwrite(myfile)
+                with open(myfile, "a") as ofile:
+                    #f = open(ofile, "a")
+                    ofile.write("  " + mycountry + "-" + myregion + ":\n")
+                islocdef[myhlocale] = myhlocale
+            with open(myfile, "a") as ofile:
+                ofile.write("    " + mydate + " = " + unicode(jdata['description']) + "\n")
+            #print("%s (%s), %s, %s: %s" % (unicode(jdata['description']), unicode(jdata['region']),
+            #    mycountry, myregion, mydate))
+            sleep(0.1)
+            sys.stdout.flush()
+            print islocdef.keys()
+
+if __name__ == "__main__":
+    myyear = ""
+    mylocode = ""
+    if myyear == "":
+        now = datetime.now()
+        myyear = unicode(now.year)
+    if mylocode == "":
+        mylocode = "de-DE"
+    sys.stdout.write("Fetching holiday data from holidata.net...")
+    sys.stdout.flush()
+    lines = hfetch(mylocode, myyear)
+    print(" done.")
+    if lines == "":
+        print("No lines returned from holidata.net for %s!", mylocode)
+        exit(3)
+    sys.stdout.write("Parsing data")
+    sys.stdout.flush()
+    hparse(lines, mylocode, myyear)
+    print(" done.")