X-Git-Url: https://git.toastfreeware.priv.at/philipp/winterrodeln/wrpylib.git/blobdiff_plain/d944e761a8ffcedde65f2ceac51062f71b7c21ac..f6c4fcdf278c631e8900c06845eebfe36aa5dc51:/wrpylib/wrmwmarkup.py diff --git a/wrpylib/wrmwmarkup.py b/wrpylib/wrmwmarkup.py index c9aba4f..264c7e2 100644 --- a/wrpylib/wrmwmarkup.py +++ b/wrpylib/wrmwmarkup.py @@ -1,225 +1,158 @@ -#!/usr/bin/python2.7 -# -*- coding: iso-8859-15 -*- +#!/usr/bin/python3.4 # $Id$ # $HeadURL$ -"""This module contains winterrodeln specific functions that are prcocessing the MediaWiki markup. +"""This module contains winterrodeln specific functions that are processing the MediaWiki markup. """ import re import xml.etree.ElementTree -import formencode +import collections +import mwparserfromhell import wrpylib.wrvalidators import wrpylib.mwmarkup - -WRMAP_POINT_TYPE = {'gasthaus': 'inn', 'haltestelle': 'busstop', 'parkplatz': 'carpark', 'achtung': 'attention', 'punkt': 'point'} -WRMAP_LINE_TYPE = {'rodelbahn': 'sledrun', 'gehweg': 'walk', 'alternative': 'alternative', 'lift': 'lift', 'linie': 'line'} - -def _conv(fnct, value, fieldname): - """Internal function. - Like one of the to_xxx functions (e.g. to_bool), but adds the field name to the error message""" - try: return fnct(value) - except formencode.Invalid as e: raise formencode.Invalid(u"Conversion error in field '%s': %s" % (fieldname, unicode(e)), e.value, e.state) - - -def rodelbahnbox_to_sledrun(wikitext, sledrun=None): - """Converts a sledrun wiki page containing the {{Rodelbahnbox}} - to a sledrun. sledrun may be an instance of WrSledrunCache or an "empty" class (object()) (default). - Raises a formencode.Invalid exception if the format is not OK or the Rodelbahnbox is not found. - :return: (start, end, sledrun) tuple of the Rodelbahnbox.""" - if sledrun is None: - class Sledrun(object): pass - sledrun = Sledrun() - - # match Rodelbahnbox - start, end = wrpylib.mwmarkup.find_template(wikitext, u'Rodelbahnbox') - if start is None: raise formencode.Invalid(u"Rodelbahnbox nicht gefunden", wikitext, None) - template_title, properties = wrpylib.mwmarkup.split_template(wikitext[start:end]) - - # process properties - for key, value in properties.iteritems(): - if key == u'Position': sledrun.position_latitude, sledrun.position_longitude = _conv(wrpylib.wrvalidators.GeoNone().to_python, value, key) # '47.583333 N 15.75 E' - elif key == u'Position oben': sledrun.top_latitude, sledrun.top_longitude = _conv(wrpylib.wrvalidators.GeoNone().to_python, value, key) # '47.583333 N 15.75 E' - elif key == u'Höhe oben': sledrun.top_elevation = _conv(wrpylib.wrvalidators.UnsignedNone().to_python, value, key) # '2000' - elif key == u'Position unten': sledrun.bottom_latitude, sledrun.bottom_longitude = _conv(wrpylib.wrvalidators.GeoNone().to_python, value, key) # '47.583333 N 15.75 E' - elif key == u'Höhe unten': sledrun.bottom_elevation = _conv(wrpylib.wrvalidators.UnsignedNone().to_python, value, key) # '1200' - elif key == u'Länge': sledrun.length = _conv(wrpylib.wrvalidators.UnsignedNone().to_python, value, key) # 3500 - elif key == u'Schwierigkeit': sledrun.difficulty = _conv(wrpylib.wrvalidators.GermanDifficulty().to_python, value, key) # 'mittel' elif key == u'Lawinen': sledrun.avalanches = _conv(wrpylib.wrvalidators.GermanAvalanches().to_python, value, key) # 'kaum' - elif key == u'Lawinen': sledrun.avalanches = _conv(wrpylib.wrvalidators.GermanAvalanches().to_python, value, key) # 'kaum' - elif key == u'Betreiber': sledrun.operator = _conv(wrpylib.wrvalidators.UnicodeNone().to_python, value, key) # 'Max Mustermann' - elif key == u'Öffentliche Anreise': sledrun.public_transport = _conv(wrpylib.wrvalidators.GermanPublicTransport().to_python, value, key) # 'Mittelmäßig' - elif key == u'Aufstieg möglich': sledrun.walkup_possible = _conv(wrpylib.wrvalidators.GermanBoolNone().to_python, value, key) # 'Ja' - elif key == u'Aufstieg getrennt': sledrun.walkup_separate, sledrun.walkup_separate_comment = _conv(wrpylib.wrvalidators.GermanTristateFloatComment().to_python, value, key) # 'Ja' - elif key == u'Gehzeit': sledrun.walkup_time = _conv(wrpylib.wrvalidators.UnsignedNone().to_python, value, key) # 90 - elif key == u'Aufstiegshilfe': sledrun.lift, sledrun.lift_details = _conv(wrpylib.wrvalidators.GermanLift().to_python, value, key) # 'Gondel (unterer Teil)' - elif key == u'Beleuchtungsanlage': sledrun.night_light, sledrun.night_light_comment = _conv(wrpylib.wrvalidators.GermanTristateFloatComment().to_python, value, key) - elif key == u'Beleuchtungstage': sledrun.night_light_days, sledrun.night_light_days_comment = _conv(wrpylib.wrvalidators.UnsignedCommentNone(7).to_python, value, key) # '3 (Montag, Mittwoch, Freitag)' - elif key == u'Rodelverleih': sledrun.sled_rental, sledrun.sled_rental_comment = _conv(wrpylib.wrvalidators.SledRental().to_python, value, key) # 'Talstation Serlesbahnan' - elif key == u'Gütesiegel': sledrun.cachet = _conv(wrpylib.wrvalidators.GermanCachet().to_python, value, key) # 'Tiroler Naturrodelbahn-Gütesiegel 2009 mittel' - elif key == u'Webauskunft': sledrun.information_web = _conv(wrpylib.wrvalidators.UrlNeinNone().to_python, value, key) # 'http://www.nösslachhütte.at/page9.php' - elif key == u'Telefonauskunft': sledrun.information_phone = _conv(wrpylib.wrvalidators.PhoneCommentListNeinLoopNone(comments_are_optional=False).to_python, value, key) # '+43-664-5487520 (Mitterer Alm)' - elif key == u'Bild': sledrun.image = _conv(wrpylib.wrvalidators.UnicodeNone().to_python, value, key) - elif key == u'In Übersichtskarte': sledrun.show_in_overview = _conv(wrpylib.wrvalidators.GermanBoolNone().to_python, value, key) - elif key == u'Forumid': sledrun.forum_id = _conv(wrpylib.wrvalidators.UnsignedNeinNone().to_python, value, key) - else: raise formencode.Invalid(u"Unbekannte Eigenschaft der Rodelbahnbox: '%s' (mit Wert '%s')" % (key, value), value, None) - return start, end, sledrun - - -def sledrun_to_rodelbahnbox(sledrun, version): - """Converts a sledrun class to the {{Rodelbahnbox}} representation. - The sledrun class has to have properties like position_latitude, ... - See the table sledruncache for field (column) values. - :param sledrun: an arbitrary class that contains the right properties - :param version: a string specifying the version of the rodelbahnbox zu produce. - Version '1.3' and '1.4' are supported.""" - keys = [] - values = [] - keys.append(u'Position') - values.append(wrpylib.wrvalidators.GeoNone().from_python((sledrun.position_latitude, sledrun.position_longitude))) - keys.append(u'Position oben') - values.append(wrpylib.wrvalidators.GeoNone().from_python((sledrun.top_latitude, sledrun.top_longitude))) - keys.append(u'Höhe oben') - values.append(wrpylib.wrvalidators.UnsignedNone().from_python(sledrun.top_elevation)) - keys.append(u'Position unten') - values.append(wrpylib.wrvalidators.GeoNone().from_python((sledrun.bottom_latitude, sledrun.bottom_longitude))) - keys.append(u'Höhe unten') - values.append(wrpylib.wrvalidators.UnsignedNone().from_python(sledrun.bottom_elevation)) - keys.append(u'Länge') - values.append(wrpylib.wrvalidators.UnsignedNone().from_python(sledrun.length)) - keys.append(u'Schwierigkeit') - values.append(wrpylib.wrvalidators.GermanDifficulty().from_python(sledrun.difficulty)) - keys.append(u'Lawinen') - values.append(wrpylib.wrvalidators.GermanAvalanches().from_python(sledrun.avalanches)) - keys.append(u'Betreiber') - values.append(wrpylib.wrvalidators.UnicodeNone().from_python(sledrun.operator)) - keys.append(u'Öffentliche Anreise') - values.append(wrpylib.wrvalidators.GermanPublicTransport().from_python(sledrun.public_transport)) - if version == '1.4': - keys.append(u'Aufstieg möglich') - values.append(wrpylib.wrvalidators.GermanBoolNone().from_python(sledrun.walkup_possible)) - keys.append(u'Aufstieg getrennt') - values.append(wrpylib.wrvalidators.GermanTristateFloatComment().from_python((sledrun.walkup_separate, sledrun.walkup_separate_comment))) - keys.append(u'Gehzeit') - values.append(wrpylib.wrvalidators.UnsignedNone().from_python(sledrun.walkup_time)) - keys.append(u'Aufstiegshilfe') - values.append(wrpylib.wrvalidators.GermanLift().from_python((sledrun.lift, sledrun.lift_details))) - keys.append(u'Beleuchtungsanlage') - values.append(wrpylib.wrvalidators.GermanTristateFloatComment().from_python((sledrun.night_light, sledrun.night_light_comment))) - keys.append(u'Beleuchtungstage') - values.append(wrpylib.wrvalidators.UnsignedCommentNone(max=7).from_python((sledrun.night_light_days, sledrun.night_light_days_comment))) - keys.append(u'Rodelverleih') - values.append(wrpylib.wrvalidators.SledRental().from_python((sledrun.sled_rental, sledrun.sled_rental_comment))) - keys.append(u'Gütesiegel') - values.append(wrpylib.wrvalidators.GermanCachet().from_python(sledrun.cachet)) - keys.append(u'Webauskunft') - values.append(wrpylib.wrvalidators.UrlNeinNone().from_python(sledrun.information_web)) - keys.append(u'Telefonauskunft') - values.append(wrpylib.wrvalidators.PhoneCommentListNeinLoopNone(comments_are_optional=False).from_python(sledrun.information_phone)) - keys.append(u'Bild') - values.append(wrpylib.wrvalidators.UnicodeNone().from_python(sledrun.image)) - keys.append(u'In Übersichtskarte') - values.append(wrpylib.wrvalidators.GermanBoolNone().from_python(sledrun.show_in_overview)) - keys.append(u'Forumid') - values.append(wrpylib.wrvalidators.UnsignedNeinNone().from_python(sledrun.forum_id)) - return wrpylib.mwmarkup.create_template(u'Rodelbahnbox', [], keys, values, True, 20) - - -def gasthausbox_to_inn(wikitext, inn=None): - """Converts a inn wiki page containing a {{Gasthausbox}} to an inn. - raises a formencode.Invalid exception if an error occurs. - :return: (start, end, inn) tuple.""" - if inn is None: - class Inn(object): pass - inn = Inn() - - # Match Gasthausbox - start, end = wrpylib.mwmarkup.find_template(wikitext, u'Gasthausbox') - if start is None: raise formencode.Invalid(u"No 'Gasthausbox' found", wikitext, None) - template_title, properties = wrpylib.mwmarkup.split_template(wikitext[start:end]) - - # Process properties - for key, value in properties.iteritems(): - if key == u'Position': inn.position_latitude, inn.position_longitude = _conv(wrpylib.wrvalidators.GeoNone().to_python, value, key) # '47.583333 N 15.75 E' - elif key == u'Höhe': inn.position_elevation = _conv(wrpylib.wrvalidators.UnsignedNone().to_python, value, key) - elif key == u'Betreiber': inn.operator = _conv(wrpylib.wrvalidators.UnicodeNone().to_python, value, key) - elif key == u'Sitzplätze': inn.seats = _conv(wrpylib.wrvalidators.UnsignedNone().to_python, value, key) - elif key == u'Übernachtung': inn.overnight, inn.overnight_comment = _conv(wrpylib.wrvalidators.BoolUnicodeTupleValidator().to_python, value, key) - elif key == u'Rauchfrei': inn.nonsmoker_area, inn.smoker_area = _conv(wrpylib.wrvalidators.GermanTristateTuple().to_python, value, key) - elif key == u'Rodelverleih': inn.sled_rental, inn.sled_rental_comment = _conv(wrpylib.wrvalidators.BoolUnicodeTupleValidator().to_python, value, key) - elif key == u'Handyempfang': inn.mobile_provider = _conv(wrpylib.wrvalidators.ValueCommentListNeinLoopNone().to_python, value, key) - elif key == u'Homepage': inn.homepage = _conv(wrpylib.wrvalidators.UrlNeinNone().to_python, value, key) - elif key == u'E-Mail': inn.email_list = _conv(wrpylib.wrvalidators.EmailCommentListNeinLoopNone(allow_masked_email=True).to_python, value, key) - elif key == u'Telefon': inn.phone_list = _conv(wrpylib.wrvalidators.PhoneCommentListNeinLoopNone(comments_are_optional=True).to_python, value, key) - elif key == u'Bild': inn.image = _conv(wrpylib.wrvalidators.UnicodeNone().to_python, value, key) - elif key == u'Rodelbahnen': inn.sledding_list = _conv(wrpylib.wrvalidators.WikiPageListLoopNone().to_python, value, key) - else: raise formencode.Invalid(u"Unbekannte Eigenschaft der Gasthausbox: '%s' (mit Wert '%s')" % (key, value), value, None) - return start, end, inn +import wrpylib.wrmwdb +from wrpylib.wrvalidators import LonLat, opt_lonlat_from_str, opt_lonlat_to_str, opt_uint_from_str, opt_uint_to_str, \ + opt_str_opt_comment_enum_to_str, lift_german_to_str, webauskunft_to_str, cachet_german_to_str, \ + opt_phone_comment_enum_to_str, lift_german_from_str + + +def sledrun_from_rodelbahnbox(value, sledrun): + """Takes a Rodelbahnbox as returned by rodelbahnbox_from_str (that is, an OrderedDict) and + updates the sledrun instance with all values present in the Rodelbahnbox. Other values are not + updated. Does not validate the arguments.""" + # sledrun.page_id = None # this field is not updated because it is not present in the RodelbahnBox + # sledrun.page_title = None # this field is not updated because it is not present in the RodelbahnBox + # sledrun.name_url = None # this field is not updated because it is not present in the RodelbahnBox + sledrun.position_longitude, sledrun.position_latitude = value['Position'] + sledrun.top_longitude, sledrun.top_latitude = value['Position oben'] + sledrun.top_elevation = value['Höhe oben'] + sledrun.bottom_longitude, sledrun.bottom_latitude = value['Position unten'] + sledrun.bottom_elevation = value['Höhe unten'] + sledrun.length = value['Länge'] + sledrun.difficulty = value['Schwierigkeit'] + sledrun.avalanches = value['Lawinen'] + sledrun.operator = value['Betreiber'] + sledrun.public_transport = value['Öffentliche Anreise'] + sledrun.walkup_possible = value['Aufstieg möglich'] + sledrun.walkup_time = value['Gehzeit'] + sledrun.walkup_separate, sledrun.walkup_separate_comment = value['Aufstieg getrennt'] + sledrun.lift = None if value['Aufstiegshilfe'] is None else len(value['Aufstiegshilfe']) > 0 + sledrun.lift_details = lift_german_to_str(value['Aufstiegshilfe']) + sledrun.night_light, sledrun.night_light_comment = value['Beleuchtungsanlage'] + sledrun.night_light_days, sledrun.night_light_days_comment = value['Beleuchtungstage'] + sledrun.sled_rental = None if value['Rodelverleih'] is None else len(value['Rodelverleih']) > 0 + sledrun.sled_rental_comment = opt_str_opt_comment_enum_to_str(value['Rodelverleih']) + sledrun.cachet = cachet_german_to_str(value['Gütesiegel']) + sledrun.information_web = webauskunft_to_str(value['Webauskunft']) + sledrun.information_phone = opt_phone_comment_enum_to_str(value['Telefonauskunft']) + sledrun.image = value['Bild'] + sledrun.show_in_overview = value['In Übersichtskarte'] + sledrun.forum_id = value['Forumid'] + # sledrun.under_construction = None # this field is not updated because it is not present in the RodelbahnBox + return sledrun + + +def sledrun_to_rodelbahnbox(sledrun): + """Takes a sledrun instance that might come from the database and converts it to a OrderedDict ready + to be formatted as RodelbahnBox.""" + value = collections.OrderedDict() + value['Position'] = LonLat(sledrun.position_longitude, sledrun.position_latitude) + value['Position oben'] = LonLat(sledrun.top_longitude, sledrun.top_latitude) + value['Höhe oben'] = sledrun.top_elevation + value['Position unten'] = LonLat(sledrun.bottom_longitude, sledrun.bottom_latitude) + value['Höhe unten'] = sledrun.bottom_elevation + value['Länge'] = sledrun.length + value['Schwierigkeit'] = sledrun.difficulty + value['Lawinen'] = sledrun.avalanches + value['Betreiber'] = sledrun.operator + value['Öffentliche Anreise'] = sledrun.public_transport + value['Aufstieg möglich'] = sledrun.walkup_possible + value['Gehzeit'] = sledrun.walkup_time + value['Aufstieg getrennt'] = sledrun.walkup_separate, sledrun.walkup_separate_comment + value['Aufstiegshilfe'] = lift_german_from_str(sledrun.lift_details) + value['Beleuchtungsanlage'] = sledrun.night_light, sledrun.night_light_comment + value['Beleuchtungstage'] = sledrun.night_light_days, sledrun.night_light_days_comment + value['Rodelverleih'] = sledrun.sled_rental, sledrun.sled_rental_comment + value['Gütesiegel'] = sledrun.cachet + value['Webauskunft'] = sledrun.information_web + value['Telefonauskunft'] = sledrun.information_phone + value['Bild'] = sledrun.image + value['In Übersichtskarte'] = sledrun.show_in_overview + value['Forumid'] = sledrun.forum_id + return value + + +def inn_from_gasthausbox(value, inn): + """Converts a dict with Gasthausbox properties to a Inn class. Does no validation. + value is a dict of properties as returned by gasthausbox_from_str.""" + inn.position_longitude, inn.position_latitude = value['Position'] + inn.position_elevation = value['Höhe'] + inn.operator = value['Betreiber'] + inn.seats = value['Sitzplätze'] + inn.overnight, inn.overnight_comment = value['Übernachtung'] + inn.nonsmoker_area, inn.smoker_area = value['Rauchfrei'] + inn.sled_rental, inn.sled_rental_comment = value['Rodelverleih'] + inn.mobile_provider = value['Handyempfang'] + inn.homepage = value['Homepage'] + inn.email_list = value['E-Mail'] + inn.phone_list = value['Telefon'] + inn.image = value['Bild'] + inn.sledding_list = value['Rodelbahnen'] + return inn def inn_to_gasthausbox(inn): - """Converts the inn class to the {{Gasthausbox}} representation.""" - keys = [] - values = [] - keys.append(u'Position') - values.append(wrpylib.wrvalidators.GeoNone().from_python((inn.position_latitude, inn.position_longitude))) - keys.append(u'Höhe') - values.append(wrpylib.wrvalidators.UnsignedNone().from_python(inn.position_elevation)) - keys.append(u'Betreiber') - values.append(wrpylib.wrvalidators.UnicodeNone().from_python(inn.operator)) - keys.append(u'Sitzplätze') - values.append(wrpylib.wrvalidators.UnsignedNone().from_python(inn.seats)) - keys.append(u'Übernachtung') - values.append(wrpylib.wrvalidators.BoolUnicodeTupleValidator().from_python((inn.overnight, inn.overnight_comment))) - keys.append(u'Rauchfrei') - values.append(wrpylib.wrvalidators.GermanTristateTuple().from_python((inn.nonsmoker_area, inn.smoker_area))) - keys.append(u'Rodelverleih') - values.append(wrpylib.wrvalidators.BoolUnicodeTupleValidator().from_python((inn.sled_rental, inn.sled_rental_comment))) - keys.append(u'Handyempfang') - values.append(wrpylib.wrvalidators.ValueCommentListNeinLoopNone().from_python(inn.mobile_provider)) - keys.append(u'Homepage') - values.append(wrpylib.wrvalidators.UrlNeinNone().from_python(inn.homepage)) - keys.append(u'E-Mail') - values.append(wrpylib.wrvalidators.EmailCommentListNeinLoopNone(allow_masked_email=True).from_python(inn.email_list)) - keys.append(u'Telefon') - values.append(wrpylib.wrvalidators.PhoneCommentListNeinLoopNone(comments_are_optional=True).from_python(inn.phone_list)) - keys.append(u'Bild') - values.append(wrpylib.wrvalidators.UnicodeNone().from_python(inn.image)) - keys.append(u'Rodelbahnen') - values.append(wrpylib.wrvalidators.WikiPageListLoopNone().from_python(inn.sledding_list)) - result = [u'{{Gasthausbox'] - return wrpylib.mwmarkup.create_template(u'Gasthausbox', [], keys, values, True) - - -def find_template_latlon_ele(wikitext, template_title): - """Finds the first occurance of the '{{template_title|47.076207 N 11.453553 E|1890}}' template - and returns the tuple (start, end, lat, lon, ele) or (None, None, None, None, None) if the - template was not found. If the template has no valid format, an exception is thrown.""" - start, end = wrpylib.mwmarkup.find_template(wikitext, template_title) - if start is None: return (None,) * 5 - title, params = wrpylib.mwmarkup.split_template(wikitext[start:end]) - lat, lon = wrpylib.wrvalidators.GeoNone().to_python(params[u'1'].strip()) - ele = wrpylib.wrvalidators.UnsignedNone().to_python(params[u'2'].strip()) - return start, end, lat, lon, ele + """Converts an inn class to a dict of Gasthausbox properties. value is an Inn instance.""" + value = collections.OrderedDict() + value['Position'] = LonLat(inn.position_longitude, inn.position_latitude) + value['Höhe'] = inn.position_elevation + value['Betreiber'] = inn.operator + value['Sitzplätze'] = inn.seats + value['Übernachtung'] = (inn.overnight, inn.overnight_comment) + value['Rauchfrei'] = (inn.nonsmoker_area, inn.smoker_area) + value['Rodelverleih'] = (inn.sled_rental, inn.sled_rental_comment) + value['Handyempfang'] = inn.mobile_provider + value['Homepage'] = inn.homepage + value['E-Mail'] = inn.email_list + value['Telefon'] = inn.phone_list + value['Bild'] = inn.image + value['Rodelbahnen'] = inn.sledding_list + return value + + +def lonlat_ele_from_template(template): + """Template is a mwparserfromhell.nodes.template.Template instance. Returns (lonlat, ele).""" + lonlat = opt_lonlat_from_str(template.params[0].strip()) + ele = opt_uint_from_str(template.params[1].strip()) + return lonlat, ele + + +def latlon_ele_to_template(lonlat_ele, name): + lonlat, ele = lonlat_ele + template = mwparserfromhell.nodes.template.Template(name) + template.add(1, opt_lonlat_to_str(lonlat)) + template.add(2, opt_uint_to_str(ele)) + wrpylib.mwmarkup.format_template_oneline(template) + return template + + +class ParseError(RuntimeError): + """Exception used by some of the functions""" + pass -def create_template_latlon_ele(template_title, lat, lon, ele): - geo = wrpylib.wrvalidators.GeoNone().from_python((lat, lon)) - if len(geo) == 0: geo = u' ' - ele = wrpylib.wrvalidators.UnsignedNone().from_python(ele) - if len(ele) == 0: ele = u' ' - return wrpylib.mwmarkup.create_template(template_title, [geo, ele]) - def find_template_PositionOben(wikitext): """Same as find_template_latlon_ele with template '{{Position oben|47.076207 N 11.453553 E|1890}}'""" - return find_template_latlon_ele(wikitext, u'Position oben') + return find_template_latlon_ele(wikitext, 'Position oben') def create_template_PositionOben(lat, lon, ele): - return create_template_latlon_ele(u'Position, oben', lat, lon, ele) + return create_template_latlon_ele('Position, oben', lat, lon, ele) def find_template_PositionUnten(wikitext): """Same as find_template_latlon_ele with template '{{Position unten|47.076207 N 11.453553 E|1890}}'""" - return find_template_latlon_ele(wikitext, u'Position unten') + return find_template_latlon_ele(wikitext, 'Position unten') def find_template_unsigned(wikitext, template_title): @@ -229,50 +162,50 @@ def find_template_unsigned(wikitext, template_title): start, end = wrpylib.mwmarkup.find_template(wikitext, template_title) if start is None: return (None,) * 3 title, params = wrpylib.mwmarkup.split_template(wikitext[start:end]) - unsigned_value = wrpylib.wrvalidators.UnsignedNone().to_python(params[u'1'].strip()) + unsigned_value = wrpylib.wrvalidators.UnsignedNone().to_python(params['1'].strip()) return start, end, unsigned_value def create_template_unsigned(template_title, unsigned): unsigned = wrpylib.wrvalidators.UnsignedNone().from_python(unsigned) - if len(unsigned) == 0: unsigned = u' ' + if len(unsigned) == 0: unsigned = ' ' return wrpylib.mwmarkup.create_template(template_title, [unsigned]) def find_template_Hoehenunterschied(wikitext): - """Same as find_template_unsigned with template '{{Höhenunterschied|350}}'""" - return find_template_unsigned(wikitext, u'Höhenunterschied') + """Same as find_template_unsigned with template '{{Höhenunterschied|350}}'""" + return find_template_unsigned(wikitext, 'Höhenunterschied') def create_template_Hoehenunterschied(ele_diff): - return create_template_unsigned(u'Höhenunterschied', ele_diff) + return create_template_unsigned('Höhenunterschied', ele_diff) def find_template_Bahnlaenge(wikitext): - """Same as find_template_unsigned with template '{{Bahnlänge|4500}}'""" - return find_template_unsigned(wikitext, u'Bahnlänge') + """Same as find_template_unsigned with template '{{Bahnlänge|4500}}'""" + return find_template_unsigned(wikitext, 'Bahnlänge') def create_template_Bahnlaenge(length): - return create_template_unsigned(u'Bahnlänge', length) + return create_template_unsigned('Bahnlänge', length) def find_template_Gehzeit(wikitext): """Same as find_template_unsigned with template '{{Gehzeit|60}}'""" - return find_template_unsigned(wikitext, u'Gehzeit') + return find_template_unsigned(wikitext, 'Gehzeit') def create_template_Gehzeit(walkup_time): - return create_template_unsigned(u'Gehzeit', walkup_time) + return create_template_unsigned('Gehzeit', walkup_time) def find_template_Forumlink(wikitext): """Same as find_template_unsigned with template '{{Forumlink|26}}'""" - start, end = wrpylib.mwmarkup.find_template(wikitext, u'Forumlink') + start, end = wrpylib.mwmarkup.find_template(wikitext, 'Forumlink') if start is None: return (None,) * 3 title, params = wrpylib.mwmarkup.split_template(wikitext[start:end]) - forumid = params[u'1'].strip() - if forumid == u'': unsigned_value = None + forumid = params['1'].strip() + if forumid == '': unsigned_value = None else: unsigned_value = wrpylib.wrvalidators.UnsignedNone().to_python(forumid) return start, end, unsigned_value # return find_template_unsigned(wikitext, u'Forumlink') @@ -280,95 +213,23 @@ def find_template_Forumlink(wikitext): def find_template_Parkplatz(wikitext): """Same as find_template_latlon_ele with template '{{Parkplatz|47.076207 N 11.453553 E|1890}}'""" - return find_template_latlon_ele(wikitext, u'Parkplatz') + return find_template_latlon_ele(wikitext, 'Parkplatz') def find_template_Haltestelle(wikitext): """Finds the first occurance of the '{{Haltestelle|Ortsname|Haltestellenname|47.076207 N 11.453553 E|1890}}' template and returns the tuple (start, end, city, stop, lat, lon, ele) or (None, None, None, None, None, None, None) if the template was not found. If the template has no valid format, an exception is thrown.""" - start, end = wrpylib.mwmarkup.find_template(wikitext, u'Haltestelle') + start, end = wrpylib.mwmarkup.find_template(wikitext, 'Haltestelle') if start is None: return (None,) * 7 title, params = wrpylib.mwmarkup.split_template(wikitext[start:end]) - city = wrpylib.wrvalidators.UnicodeNone().to_python(params[u'1'].strip()) - stop = wrpylib.wrvalidators.UnicodeNone().to_python(params[u'2'].strip()) - lat, lon = wrpylib.wrvalidators.GeoNone().to_python(params[u'3'].strip()) - ele = wrpylib.wrvalidators.UnsignedNone().to_python(params[u'4'].strip()) + city = wrpylib.wrvalidators.UnicodeNone().to_python(params['1'].strip()) + stop = wrpylib.wrvalidators.UnicodeNone().to_python(params['2'].strip()) + lat, lon = wrpylib.wrvalidators.GeoNone().to_python(params['3'].strip()) + ele = wrpylib.wrvalidators.UnsignedNone().to_python(params['4'].strip()) return start, end, city, stop, lat, lon, ele -def find_all_templates(wikitext, find_func): - """Returns a list of return values of find_func that searches for a template. - Example: - >>> find_all_tempaltes(wikitext, find_template_Haltestelle) - Returns an empty list if the template was not found at all. - """ - results = [] - result = find_func(wikitext) - start, end = result[:2] - while start is not None: - results.append(result) - result = find_func(wikitext[end:]) - if result[0] is None: - start = None - else: - start = result[0] + end - end += result[1] - result = (start, end) + result[2:] - return results - - -def parse_googlemap(wikitext): - """Parses the (unicode) u'content' of the googlemap extension - out of a page. If wikitext does not contain the googlemaps extension text None is returned. - If the googlemap contains invalid formatted lines, a RuntimeError is raised. - - :param wikitext: wikitext containing the template. Example: - - wikitext = ''' - - (Parkplatz)47.114958,11.266026 - Parkplatz - - (Gasthaus) 47.114715, 11.266262, Alt Bärnbad (Gasthaus) - 6#FF014E9A - 47.114715,11.266262 - 47.114135,11.268381 - 47.113421,11.269322 - 47.11277,11.269979 - 47.112408,11.271119 - - ''' - :returns: (attributes, GeoJSON as Python datatypes) - """ - center, zoom, coords, paths = wrpylib.mwmarkup.parse_googlemap(wikitext) - json_features = [] - - # point - for point in coords: - lon, lat, symbol, title = point - properties = {'type': symbol} - if title: properties['name'] = title - json_features.append({ - 'type': 'feature', - 'geometry': {'type': 'Point', 'coordinates': [lon, lat]}, - 'properties': properties}) - - # path - for path in paths: - style, entries = path - properties = {'type': 'line'} - json_features.append({ - 'type': 'feature', - 'geometry': { - 'type': 'LineString', - 'coordinates': [[lon, lat] for lon, lat, symbol, title in entries]}, - 'properties': properties}) - - geojson = {'type': 'FeatureCollection', 'features': json_features} - return {'lon': center[0], 'lat': center[1], 'zoom': zoom}, geojson - - def parse_wrmap_coordinates(coords): '''gets a string coordinates and returns an array of lon/lat coordinate pairs, e.g. 47.12 N 11.87 E @@ -388,18 +249,24 @@ def parse_wrmap_coordinates(coords): raise RuntimeError('Wrong coordinate format: {}'.format(coords)) +WRMAP_POINT_TYPES = ['gasthaus', 'haltestelle', 'parkplatz', 'achtung', 'foto', 'verleih', 'punkt'] +WRMAP_LINE_TYPES = ['rodelbahn', 'gehweg', 'alternative', 'lift', 'anfahrt', 'linie'] + + def parse_wrmap(wikitext): - """Parses the (unicode) u'content' of the Winterrodeln wrmap extension - out of a page. If wikitext does not contain the googlemaps extension text None is returned. - If the googlemap contains invalid formatted lines, a RuntimeError is raised. + """Parses the (unicode) u'content' of the Winterrodeln wrmap extension. + If wikitext does not contain the tag or if the tag contains + invalid formatted lines, a ParseError is raised. + Use wrpylib.mwmarkup.find_tag(wikitext, 'wrmap') to find the wrmap tag within an arbitrary + wikitext before using this function. - :param wikitext: wikitext containing the template. Example: + :param wikitext: wikitext containing only the template. Example: wikitext = u''' - 47.240689 11.190454 + 47.240689 11.190454 47.245789 11.238971 - 47.245711 11.238283 + 47.245711 11.238283 47.238587 11.203360 47.244951 11.230868 @@ -407,91 +274,85 @@ def parse_wrmap(wikitext): ''' - :returns: (attributes, GeoJSON as Python datatypes) + :returns: GeoJSON as nested Python datatype """ # parse XML try: wrmap_xml = xml.etree.ElementTree.fromstring(wikitext.encode('utf-8')) except xml.etree.ElementTree.ParseError as e: row, column = e.position - raise RuntimeError("XML parse error on row {}, column {}: {}".format(row, column, e)) - # assert(in_array($tagname, array('wrmap', 'wrgmap'))); # TODO + raise ParseError("XML parse error on row {}, column {}: {}".format(row, column, e)) + if wrmap_xml.tag not in ['wrmap', 'wrgmap']: + raise ParseError('No valid tag name') # convert XML to geojson (http://www.geojson.org/geojson-spec.html) json_features = [] for feature in wrmap_xml: # determine feature type - is_point = WRMAP_POINT_TYPE.has_key(feature.tag) - is_line = WRMAP_LINE_TYPE.has_key(feature.tag) + is_point = feature.tag in WRMAP_POINT_TYPES + is_line = feature.tag in WRMAP_LINE_TYPES if (not is_point and not is_line): - raise RuntimeError('Unknown element <{}>.'.format(feature.tag)) + raise ParseError('Unknown element <{}>.'.format(feature.tag)) # point if is_point: - properties = {'type': WRMAP_POINT_TYPE[feature.tag]} - allowed_properties = set(['name', 'wiki']) + properties = {'type': feature.tag} + allowed_properties = {'name', 'wiki'} wrong_properties = set(feature.attrib.keys()) - allowed_properties if len(wrong_properties) > 0: - raise RuntimeError("The attribute '{}' is not allowed at <{}>.".format(list(wrong_properties)[0], feature.tag)) + raise ParseError("The attribute '{}' is not allowed at <{}>.".format(list(wrong_properties)[0], feature.tag)) properties.update(feature.attrib) coordinates = parse_wrmap_coordinates(feature.text) if len(coordinates) != 1: - raise RuntimeError('The element <{}> has to have exactly one coordinate pair.'.format(feature.tag)) + raise ParseError('The element <{}> has to have exactly one coordinate pair.'.format(feature.tag)) json_features.append({ - 'type': 'feature', + 'type': 'Feature', 'geometry': {'type': 'Point', 'coordinates': coordinates[0]}, 'properties': properties}) # line if is_line: - properties = {'type': WRMAP_LINE_TYPE[feature.tag]} - allowed_properties = set(['farbe', 'dicke']) + properties = {'type': feature.tag} + allowed_properties = {'farbe', 'dicke'} wrong_properties = set(feature.attrib.keys()) - allowed_properties if len(wrong_properties) > 0: - raise RuntimeError("The attribute '{}' is not allowed at <{}>.".format(list(wrong_properties)[0], feature.tag)) - if feature.attrib.has_key('farbe'): + raise ParseError("The attribute '{}' is not allowed at <{}>.".format(list(wrong_properties)[0], feature.tag)) + if 'farbe' in feature.attrib: if not re.match('#[0-9a-fA-F]{6}$', feature.attrib['farbe']): - raise RuntimeError('The attribute "farbe" has to have a format like "#a0bb43".') + raise ParseError('The attribute "farbe" has to have a format like "#a0bb43".') properties['strokeColor'] = feature.attrib['farbe'] # e.g. #a200b7 - if feature.attrib.has_key('dicke'): + if 'dicke' in feature.attrib: try: properties['strokeWidth'] = int(feature.attrib['dicke']) # e.g. 6 except ValueError: - raise RuntimeError('The attribute "dicke" has to be an integer.') + raise ParseError('The attribute "dicke" has to be an integer.') json_features.append({ - 'type': 'feature', + 'type': 'Feature', 'geometry': {'type': 'LineString', 'coordinates': parse_wrmap_coordinates(feature.text)}, 'properties': properties}) + # attributes + properties = {} + for k, v in wrmap_xml.attrib.items(): + if k in ['lat', 'lon']: + try: + properties[k] = float(v) + except ValueError: + raise ParseError('Attribute "{}" has to be a float value.'.format(k)) + elif k in ['zoom', 'width', 'height']: + try: + properties[k] = int(v) + except ValueError: + raise ParseError('Attribute "{}" has to be an integer value.'.format(k)) + else: + raise ParseError('Unknown attribute "{}".'.format(k)) + geojson = { 'type': 'FeatureCollection', - 'features': json_features} - - # attributes - attributes = {} - try: - attributes['lat'] = float(wrmap_xml.attrib.get('lat', 47.267648)) # center lat - except ValueError: - raise RuntimeError('Attribute "lat" has to be a float value.') - try: - attributes['lon'] = float(wrmap_xml.attrib.get('lon', 11.404655)) # center lon - except ValueError: - raise RuntimeError('Attribute "lon" has to be a float value.') - try: - attributes['zoom'] = int(wrmap_xml.attrib.get('zoom', 10)) # Google Zoom Level - except ValueError: - raise RuntimeError('Attribute "zoom" has to be an integer value.') - try: - attributes['width'] = int(wrmap_xml.attrib['width']) if wrmap_xml.attrib.has_key('width') else None # None corresponds to 100% - except ValueError: - raise RuntimeError('Attribute "width" has to be an integer value.') - try: - attributes['height'] = int(wrmap_xml.attrib.get('height', 450)) # map height in px - except ValueError: - raise RuntimeError('Attribute "height" has to be an integer value.') - # show_sledruns = (wrmap_xml.tag == 'wrgmap') + 'features': json_features, + 'properties': properties} - return attributes, geojson + return geojson def create_wrmap_coordinates(coords): @@ -501,26 +362,34 @@ def create_wrmap_coordinates(coords): return '\n'.join(result) -def create_wrmap(attributes, geojson): +def create_wrmap(geojson): """Creates a wikitext from geojson (as python types).""" wrmap_xml = xml.etree.ElementTree.Element('wrmap') - for k, v in attributes.iteritems(): - wrmap_xml.attrib[k] = str(v) + wrmap_xml.text = '\n\n' + for k, v in geojson['properties'].items(): + if k in ['lon', 'lat']: + wrmap_xml.attrib[k] = '{:.6f}'.format(v) + else: + wrmap_xml.attrib[k] = str(v) assert geojson['type'] == 'FeatureCollection' json_features = geojson['features'] + last_json_feature = None for json_feature in json_features: + feature_xml = xml.etree.ElementTree.SubElement(wrmap_xml, json_feature['properties']['type']) geo = json_feature['geometry'] if geo['type'] == 'Point': - wrmap_type = dict(zip(WRMAP_POINT_TYPE.values(), WRMAP_POINT_TYPE.keys())) - feature_xml = xml.etree.ElementTree.SubElement(wrmap_xml, wrmap_type[json_feature['properties']['type']]) feature_xml.text = create_wrmap_coordinates([geo['coordinates']]) - feature_xml.attrib = json_feature['properties'] - if geo['type'] == 'LineString': - wrmap_type = dict(zip(WRMAP_LINE_TYPE.values(), WRMAP_LINE_TYPE.keys())) - feature_xml = xml.etree.ElementTree.SubElement(wrmap_xml, wrmap_type[json_feature['properties']['type']]) - feature_xml.text = create_wrmap_coordinates(geo['coordinates']) - feature_xml.attrib = json_feature['properties'] - - return xml.etree.ElementTree.tostring(wrmap_xml) - + if last_json_feature is not None: + last_json_feature.tail = '\n' + else: + if last_json_feature is not None: + last_json_feature.tail = '\n\n' + feature_xml.text = '\n' + create_wrmap_coordinates(geo['coordinates']) + '\n' + last_json_feature = feature_xml + feature_xml.attrib = json_feature['properties'] + del feature_xml.attrib['type'] + + if last_json_feature is not None: + last_json_feature.tail = '\n\n' + return xml.etree.ElementTree.tostring(wrmap_xml, encoding='utf-8').decode('utf-8')