Functionality improvement and bugfix of mwmarkup.py and adding lots of template parsi...

git-svn-id: http://www.winterrodeln.org/svn/servermediawiki/trunk/wrpylib@1011 7aebc617-e5e2-0310-91dc-80fb5f6d2477

diff --git a/wrpylib/mwmarkup.py b/wrpylib/mwmarkup.py
index d0543e0ea7c82a6fabc7c33dddefe327a05f8cdd..b43050253a8c8cfc9cd2092e75da9348f89f738d 100644 (file)
--- a/wrpylib/mwmarkup.py
+++ b/wrpylib/mwmarkup.py
@@ -28,7 +28,8 @@ def find_template(wikitext, template_title):
     >>> print wikitext.__getslice__(*find_template(wikitext, u'Color'))
     {{Color|red|red text}}
 
-    The search is done with regular expression.
+    The search is done with regular expression. It gives wrong results when parsing a template
+    containing the characters "}}"
 
     :param wikitext: The text (preferalbe unicode) that has the template in it.
     :param template_title: The page title of the template with or without namespace (but as in the wikitext).
@@ -36,15 +37,14 @@ def find_template(wikitext, template_title):
         (start, end) of the first occurence with start >= 0 and end > start.
         (None, None) if the template is not found.
     """ 
-    match = re.search(u"\{\{" + template_title + "[^\}]*\}\}", wikitext,  re.DOTALL)
+    match = re.search(u"\{\{" + template_title + "\s*(\|[^\}]*)?\}\}", wikitext,  re.DOTALL)
     if match is None: return None, None
     return match.start(), match.end()
 
 
-
 def split_template(template):
     """Takes a template, like u'{{Color|red|text=Any text}}' and translates it to a Python tuple
-    (template_title, parameters) where parameters is a Python dictionary {1: u'red', u'text'=u'Any text'}.
+    (template_title, parameters) where parameters is a Python dictionary {u'1': u'red', u'text'=u'Any text'}.
     Anonymous parameters get integer keys (converted to unicode) starting with 1 
     like in MediaWiki, named parameters are unicode strings.
     Whitespace is stripped.
@@ -98,8 +98,11 @@ def create_template(template_title, anonym_params=[], named_param_keys=[], named
         as_table_keylen = max([len(k) for k in named_param_keys])
     for i in xrange(len(named_param_keys)):
         key = named_param_keys[i]
-        if as_table: key = key.ljust(as_table_keylen)
-        parts.append(key + equal_char + named_param_values[i])
+        if as_table: 
+            key = key.ljust(as_table_keylen)
+            parts.append((key + equal_char + named_param_values[i]).rstrip())
+        else:
+            parts.append(key + equal_char + named_param_values[i])
     return pipe_char.join(parts) + end_char
 
 

diff --git a/wrpylib/wrmwmarkup.py b/wrpylib/wrmwmarkup.py
index 990082b281298994cdb362d75fc90ab89f824f15..17f9f2080a6d2c7b988ab0300c68f74d2d7a6a27 100644 (file)
--- a/wrpylib/wrmwmarkup.py
+++ b/wrpylib/wrmwmarkup.py
@@ -29,7 +29,7 @@ def rodelbahnbox_to_sledrun(wikitext, sledrun=None):
     # 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_name, properties = wrpylib.mwmarkup.split_template(wikitext[start:end])
+    template_title, properties = wrpylib.mwmarkup.split_template(wikitext[start:end])
     
     # process properties
     for key, value in properties.iteritems():
@@ -43,9 +43,9 @@ def rodelbahnbox_to_sledrun(wikitext, sledrun=None):
         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'Gehzeit': sledrun.walkup_time = _conv(wrpylib.wrvalidators.UnsignedNone().to_python, value, key) # 90
         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)'
@@ -92,10 +92,10 @@ def sledrun_to_rodelbahnbox(sledrun, version):
     if version == '1.4':
         keys.append(u'Aufstieg möglich')
         values.append(wrpylib.wrvalidators.GermanBoolNone().from_python(sledrun.walkup_possible))
-    keys.append(u'Gehzeit')
-    values.append(wrpylib.wrvalidators.UnsignedNone().from_python(sledrun.walkup_time))
     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')
@@ -130,7 +130,7 @@ def gasthausbox_to_inn(wikitext, inn=None):
     # 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_name, properties = wrpylib.mwmarkup.split_template(wikitext[start:end])
+    template_title, properties = wrpylib.mwmarkup.split_template(wikitext[start:end])
 
     # Process properties
     for key, value in properties.iteritems():
@@ -185,3 +185,114 @@ def inn_to_gasthausbox(inn):
     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
+
+
+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')
+
+
+def create_template_PositionOben(lat, lon, ele):
+    return create_template_latlon_ele(u'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')
+
+
+def find_template_unsigned(wikitext, template_title):
+    """Finds the first occurance of the '{{template_title|1890}}' template
+    and returns the tuple (start, end, unsigned_value) or (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,) * 3
+    title, params = wrpylib.mwmarkup.split_template(wikitext[start:end])
+    unsigned_value = wrpylib.wrvalidators.UnsignedNone().to_python(params[u'1'].strip())
+    return start, end, unsigned_value
+
+
+def find_template_Hoehenunterschied(wikitext):
+    """Same as find_template_unsigned with template '{{Höhenunterschied|350}}'"""
+    return find_template_unsigned(wikitext, u'Höhenunterschied')
+
+
+def find_template_Bahnlaenge(wikitext):
+    """Same as find_template_unsigned with template '{{Bahnlänge|4500}}'"""
+    return find_template_unsigned(wikitext, u'Bahnlänge')
+
+
+def find_template_Gehzeit(wikitext):
+    """Same as find_template_unsigned with template '{{Gehzeit|60}}'"""
+    return find_template_unsigned(wikitext, u'Gehzeit')
+
+
+def find_template_Forumlink(wikitext):
+    """Same as find_template_unsigned with template '{{Forumlink|26}}'"""
+    start, end = wrpylib.mwmarkup.find_template(wikitext, u'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'<nummer einfügen>': unsigned_value = None
+    else: unsigned_value = wrpylib.wrvalidators.UnsignedNone().to_python(forumid)
+    return start, end, unsigned_value
+    # return find_template_unsigned(wikitext, u'Forumlink')
+
+
+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')
+
+
+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')
+    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())
+    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
+