X-Git-Url: https://git.toastfreeware.priv.at/philipp/winterrodeln/wrpylib.git/blobdiff_plain/bc655ea731d3c9d3f283ac2dd047f832c507acb5..f6c4fcdf278c631e8900c06845eebfe36aa5dc51:/wrpylib/wrmwmarkup.py diff --git a/wrpylib/wrmwmarkup.py b/wrpylib/wrmwmarkup.py index bc72a39..264c7e2 100644 --- a/wrpylib/wrmwmarkup.py +++ b/wrpylib/wrmwmarkup.py @@ -1,5 +1,4 @@ #!/usr/bin/python3.4 -# -*- coding: iso-8859-15 -*- # $Id$ # $HeadURL$ """This module contains winterrodeln specific functions that are processing the MediaWiki markup. @@ -7,12 +6,133 @@ import re import xml.etree.ElementTree import collections -import formencode +import mwparserfromhell import wrpylib.wrvalidators import wrpylib.mwmarkup +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 -WRMAP_POINT_TYPES = ['gasthaus', 'haltestelle', 'parkplatz', 'achtung', 'foto', 'verleih', 'punkt'] -WRMAP_LINE_TYPES = ['rodelbahn', 'gehweg', 'alternative', 'lift', 'anfahrt', 'linie'] + +def inn_to_gasthausbox(inn): + """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): @@ -20,228 +140,6 @@ class ParseError(RuntimeError): pass -class RodelbahnboxDictConverter(formencode.Validator): - """Converts a dict with Rodelbahnbox properties to a Sledrun class. Does no validation.""" - - def to_python(self, value, state=None): - """value is a dict of properties. If state is an object with the attribute sledrun, this sledrun class will be populated or updated.""" - props = value - if isinstance(state, object) and hasattr(state, 'sledrun'): - sledrun = state.sledrun - else: - class Sledrun(object): - pass - sledrun = Sledrun() - for k, v in props.items(): - if k == 'Position': sledrun.position_latitude, sledrun.position_longitude = v - elif k == 'Position oben': sledrun.top_latitude, sledrun.top_longitude = v - elif k == 'Höhe oben': sledrun.top_elevation = v - elif k == 'Position unten': sledrun.bottom_latitude, sledrun.bottom_longitude = v - elif k == 'Höhe unten': sledrun.bottom_elevation = v - elif k == 'Länge': sledrun.length = v - elif k == 'Schwierigkeit': sledrun.difficulty = v - elif k == 'Lawinen': sledrun.avalanches = v - elif k == 'Betreiber': sledrun.operator = v - elif k == 'Öffentliche Anreise': sledrun.public_transport = v - elif k == 'Aufstieg möglich': sledrun.walkup_possible = v - elif k == 'Aufstieg getrennt': sledrun.walkup_separate, sledrun.walkup_separate_comment = v - elif k == 'Gehzeit': sledrun.walkup_time = v - elif k == 'Aufstiegshilfe': sledrun.lift, sledrun.lift_details = v - elif k == 'Beleuchtungsanlage': sledrun.night_light, sledrun.night_light_comment = v - elif k == 'Beleuchtungstage': sledrun.night_light_days, sledrun.night_light_days_comment = v - elif k == 'Rodelverleih': sledrun.sled_rental, sledrun.sled_rental_comment = v - elif k == 'Gütesiegel': sledrun.cachet = v - elif k == 'Webauskunft': sledrun.information_web = v - elif k == 'Telefonauskunft': sledrun.information_phone = v - elif k == 'Bild': sledrun.image = v - elif k == 'In Übersichtskarte': sledrun.show_in_overview = v - elif k == 'Forumid': sledrun.forum_id = v - return sledrun - - def from_python(self, value, state=None): - """Converts a sledrun class to a dict of Rodelbahnbox properties. value is a sledrun instance.""" - sledrun = value - r = collections.OrderedDict() - r['Position'] = (sledrun.position_latitude, sledrun.position_longitude) - r['Position oben'] = (sledrun.top_latitude, sledrun.top_longitude) - r['Höhe oben'] = sledrun.top_elevation - r['Position unten'] = (sledrun.bottom_latitude, sledrun.bottom_longitude) - r['Höhe unten'] = sledrun.bottom_elevation - r['Länge'] = sledrun.length - r['Schwierigkeit'] = sledrun.difficulty - r['Lawinen'] = sledrun.avalanches - r['Betreiber'] = sledrun.operator - r['Öffentliche Anreise'] = sledrun.public_transport - r['Aufstieg möglich'] = sledrun.walkup_possible - r['Aufstieg getrennt'] = (sledrun.walkup_separate, sledrun.walkup_separate_comment) - r['Gehzeit'] = sledrun.walkup_time - r['Aufstiegshilfe'] = (sledrun.lift, sledrun.lift_details) - r['Beleuchtungsanlage'] = (sledrun.night_light, sledrun.night_light_comment) - r['Beleuchtungstage'] = (sledrun.night_light_days, sledrun.night_light_days_comment) - r['Rodelverleih'] = (sledrun.sled_rental, sledrun.sled_rental_comment) - r['Gütesiegel'] = sledrun.cachet - r['Webauskunft'] = sledrun.information_web - r['Telefonauskunft'] = sledrun.information_phone - r['Bild'] = sledrun.image - r['In Übersichtskarte'] = sledrun.show_in_overview - r['Forumid'] = sledrun.forum_id - return r - - -class WinterrodelnTemplateDict(formencode.Validator): - """Private helper class for RodelbahnboxValidator or GasthausboxValidator""" - def __init__(self, template_title): - self.template_title = template_title - - def to_python(self, value, state): - title, anonym_params, named_params = value - if title != self.template_title: - raise formencode.Invalid('Template title has to be "{}".'.format(self.template_title), value, state) - if len(anonym_params) > 0: - raise formencode.Invalid('No anonymous parameters are allowed in "{}".'.format(self.template_title), value, state) - return named_params - - def from_python(self, value, state): - return self.template_title, [], value - - -class RodelbahnboxValidator(wrpylib.wrvalidators.RodelbahnboxDictValidator): - def __init__(self): - wrpylib.wrvalidators.RodelbahnboxDictValidator.__init__(self) - self.pre_validators=[wrpylib.mwmarkup.TemplateValidator(as_table=True, as_table_keylen=20), WinterrodelnTemplateDict('Rodelbahnbox')] - self.chained_validators = [RodelbahnboxDictConverter()] - - -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.""" - # find Rodelbahnbox - start, end = wrpylib.mwmarkup.find_template(wikitext, 'Rodelbahnbox') - if start is None: raise formencode.Invalid("Rodelbahnbox nicht gefunden", wikitext, None) - - # convert to sledrun - if sledrun is None: - state = None - else: - class State(object): - pass - state = State() - state.sledrun = sledrun - return start, end, RodelbahnboxValidator().to_python(wikitext[start:end], state) - - -def sledrun_to_rodelbahnbox(sledrun, version=None): - """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 '1.4' is supported.""" - assert version in [None, '1.4'] - return RodelbahnboxValidator().from_python(sledrun) - - -class GasthausboxDictConverter(formencode.Validator): - """Converts a dict with Gasthausbox properties to a Inn class. Does no validation.""" - - def to_python(self, value, state=None): - """value is a dict of properties. If state is an object with the attribute inn, this inn class will be populated or updated.""" - props = value - if isinstance(state, object) and hasattr(state, 'inn'): - inn = state.inn - else: - class Inn(object): - pass - inn = Inn() - for k, v in props.items(): - if k == 'Position': inn.position_latitude, inn.position_longitude = v - elif k == 'Höhe': inn.position_elevation = v - elif k == 'Betreiber': inn.operator = v - elif k == 'Sitzplätze': inn.seats = v - elif k == 'Übernachtung': inn.overnight, inn.overnight_comment = v - elif k == 'Rauchfrei': inn.nonsmoker_area, inn.smoker_area = v - elif k == 'Rodelverleih': inn.sled_rental, inn.sled_rental_comment = v - elif k == 'Handyempfang': inn.mobile_provider = v - elif k == 'Homepage': inn.homepage = v - elif k == 'E-Mail': inn.email_list = v - elif k == 'Telefon': inn.phone_list = v - elif k == 'Bild': inn.image = v - elif k == 'Rodelbahnen': inn.sledding_list = v - return inn - - def from_python(self, value, state=None): - """Converts an inn class to a dict of Gasthausbox properties. value is an Inn instance.""" - inn = value - r = collections.OrderedDict() - r['Position'] = (inn.position_latitude, inn.position_longitude) - r['Höhe'] = inn.position_elevation - r['Betreiber'] = inn.operator - r['Sitzplätze'] = inn.seats - r['Übernachtung'] = (inn.overnight, inn.overnight_comment) - r['Rauchfrei'] = (inn.nonsmoker_area, inn.smoker_area) - r['Rodelverleih'] = (inn.sled_rental, inn.sled_rental_comment) - r['Handyempfang'] = inn.mobile_provider - r['Homepage'] = inn.homepage - r['E-Mail'] = inn.email_list - r['Telefon'] = inn.phone_list - r['Bild'] = inn.image - r['Rodelbahnen'] = inn.sledding_list - return r - - -class GasthausboxValidator(wrpylib.wrvalidators.GasthausboxDictValidator): - def __init__(self): - wrpylib.wrvalidators.GasthausboxDictValidator.__init__(self) - self.pre_validators=[wrpylib.mwmarkup.TemplateValidator(as_table=True, as_table_keylen=17), WinterrodelnTemplateDict('Gasthausbox')] - self.chained_validators = [GasthausboxDictConverter()] - - -def gasthausbox_to_inn(wikitext, inn=None): - """Converts a inn wiki page containing a {{Gasthausbox}} to an inn. - inn may be an instance of WrInnCache or an "empty" class (default). - raises a formencode.Invalid exception if the format is not OK or the Gasthausbox is not found. - :return: (start, end, inn) tuple.""" - # find Gasthausbox - start, end = wrpylib.mwmarkup.find_template(wikitext, 'Gasthausbox') - if start is None: raise formencode.Invalid("No 'Gasthausbox' found", wikitext, None) - - # convert to inn - if inn is None: - state = None - else: - class State(object): - pass - state = State() - state.inn = inn - return start, end, GasthausboxValidator().to_python(wikitext[start:end], state) - - -def inn_to_gasthausbox(inn): - """Converts the inn class to the {{Gasthausbox}} representation.""" - return GasthausboxValidator().from_python(inn) - - -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['1'].strip()) - ele = wrpylib.wrvalidators.UnsignedNone().to_python(params['2'].strip()) - return start, end, lat, lon, ele - - -def create_template_latlon_ele(template_title, lat, lon, ele): - geo = wrpylib.wrvalidators.GeoNone().from_python((lat, lon)) - if len(geo) == 0: geo = ' ' - ele = wrpylib.wrvalidators.UnsignedNone().from_python(ele) - if len(ele) == 0: ele = ' ' - 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}}'""" @@ -275,21 +173,21 @@ def create_template_unsigned(template_title, unsigned): def find_template_Hoehenunterschied(wikitext): - """Same as find_template_unsigned with template '{{Höhenunterschied|350}}'""" - return find_template_unsigned(wikitext, '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('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, '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('Bahnlänge', length) + return create_template_unsigned('Bahnlänge', length) def find_template_Gehzeit(wikitext): @@ -307,7 +205,7 @@ def find_template_Forumlink(wikitext): if start is None: return (None,) * 3 title, params = wrpylib.mwmarkup.split_template(wikitext[start:end]) forumid = params['1'].strip() - if forumid == '': unsigned_value = None + 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') @@ -332,68 +230,6 @@ def find_template_Haltestelle(wikitext): 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_templates(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 googlemap_to_wrmap(attributes, coords, paths): - """Converts the output of parse_googlemap to the GeoJSON format wrmap uses. - :returns: (GeoJSON as nested Python datatypes) - """ - json_features = [] - - # point - for point in coords: - lon, lat, symbol, title = point - properties = {'type': 'punkt' if symbol is None else symbol.lower()} - if title is not None: properties['name'] = title - json_features.append({ - 'type': 'Feature', - 'geometry': {'type': 'Point', 'coordinates': [lon, lat]}, - 'properties': properties}) - - # path - for path in paths: - style, entries = path - style = style.lower() - PATH_TYPES = {'6#ff014e9a': 'rodelbahn', '6#ffe98401': 'gehweg', '6#ff7f7fff': 'alternative', '3#ff000000': 'lift', '3#ffe1e100': 'anfahrt'} - if style in PATH_TYPES: - properties = {'type': PATH_TYPES[style]} - else: - properties = {'type': 'line'} - properties['dicke'] = style[0] - properties['farbe'] = style[4:] - 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, - 'properties': attributes} - return 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 @@ -413,6 +249,10 @@ 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. If wikitext does not contain the tag or if the tag contains @@ -424,9 +264,9 @@ def parse_wrmap(wikitext): 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 @@ -457,7 +297,7 @@ def parse_wrmap(wikitext): # point if is_point: properties = {'type': feature.tag} - allowed_properties = set(['name', 'wiki']) + allowed_properties = {'name', 'wiki'} wrong_properties = set(feature.attrib.keys()) - allowed_properties if len(wrong_properties) > 0: raise ParseError("The attribute '{}' is not allowed at <{}>.".format(list(wrong_properties)[0], feature.tag)) @@ -473,7 +313,7 @@ def parse_wrmap(wikitext): # line if is_line: properties = {'type': feature.tag} - allowed_properties = set(['farbe', 'dicke']) + allowed_properties = {'farbe', 'dicke'} wrong_properties = set(feature.attrib.keys()) - allowed_properties if len(wrong_properties) > 0: raise ParseError("The attribute '{}' is not allowed at <{}>.".format(list(wrong_properties)[0], feature.tag)) @@ -553,4 +393,3 @@ def create_wrmap(geojson): 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') -