import json import re import yaml import glob print '''

This view lists scrapers with yaml-combatible comments (containing the string "nuug-postliste-endyaml" like the following in their description 

<!-- nuug-postliste-yaml -->
YAML-tagger:<br>
Type: kommune<br>
Status: finished<br>
Name: Lillesteinsmyr kommune<br>
Format: PDF<br>
Datatype: ePhorte<br>
Run: daily<br>
<!-- nuug-postliste-endyaml -->

''' counter = {} for scrapername in glob.glob("scrapersources/postlist*"): # print scrapername with open(scrapername, 'r') as scraperfile: data = scraperfile.read() if -1 == data.find("YAML-tagger:"): continue data = re.sub(r"\n\n.*", "", data, flags=re.DOTALL) data = re.sub("^.*YAML-tagger:\n", "", data, flags=re.DOTALL) data = data.replace("# ", "") # print data data = yaml.load(data) if data['Type'] in counter: counter[data['Type']] = counter[data['Type']] + 1 else: counter[data['Type']] = 1 if 'Run' in data: Run = data['Run'] else: Run = 'unknown' if 'Format' in data: Format = data['Format'] else: Format = 'unknown' if 'Datatype' in data: Type = data['Datatype'] else: Type = 'unknown' print '' % \ (data['Name'],data['Type'],data['Status'], Run, Format, Type, "unknown") print '''
Nametypestatusscheduleformatdatatypecreated
%s%s%s%s%s%s%s
''' for key in counter: print '' % (key, counter[key]) print '
typecount
%s%d
' num_kommune = float(429) num_fylke = float(19) print '' try: print "" % \ ((float(counter['kommune'])/float(num_kommune))*100, counter['kommune'], num_kommune) except KeyError: pass try: print "" % \ ((float(counter['fylkeskommune'])/float(num_fylke))*100, counter['fylkeskommune'], num_fylke) except KeyError: pass #http://stackoverflow.com/questions/7561026/jquery-tablesorter-parser-for-datetime-in-mm-dd-yyyy-hhmi-am-format #http://stackoverflow.com/questions/1707840/date-sorting-problem-with-jquery-tablesorter print '''
TypeProsent
Kommune%.2f%% (%d av %d)
Fylkeskommune%.2f%% (%d av %d)
'''