Now function parse_googlemap behaves like parse_wrmap.

[philipp/winterrodeln/wrpylib.git] / wrpylib / mwmarkup.py

diff --git a/wrpylib/mwmarkup.py b/wrpylib/mwmarkup.py
index f377c822287a227da9b27e22fd82c12615bc03b7..735610c8ba6cc8d9c3fcf8e2167b50c000545c79 100644 (file)
--- a/wrpylib/mwmarkup.py
+++ b/wrpylib/mwmarkup.py
@@ -13,6 +13,11 @@ import re
 import xml.etree.ElementTree
 
 
+class ParseError(RuntimeError):
+    """Exception used by some of the functions"""
+    pass
+
+
 def find_template(wikitext, template_title):
     """Returns the tuple (start, end) of the first occurence of the template '{{template ...}} within wikitext'.
     (None, None) is returned if the template is not found.
@@ -148,14 +153,14 @@ def find_tag(wikitext, tagname, pos=0):
     return match_starttag.start(), match_starttag.end(), match_endtag.start(), match_endtag.end()
 
 
-def parse_googlemap(wikitext, detail=False):
-    """Parses the (unicode) u'<googlemap ...>content</googlemap>' of the googlemap extension
-    out of a page. If wikitext does not contain the googlemap extension text None is returned.
-    If the googlemap contains invalid formatted lines, a RuntimeError is raised.
+def parse_googlemap(wikitext):
+    """Parses the (unicode) u'<googlemap ...>content</googlemap>' of the googlemap extension.
+    If wikitext does not contain the <googlemap> tag or if the <googlemap> tag contains
+    invalid formatted lines, a ParseError is raised.
+    Use find_tag(wikitext, 'googlemap') to find the googlemap tag within an arbitrary
+    wikitext before using this function.
 
     :param wikitext: wikitext containing the template. Example:
-    :param detail: bool. If True, start and end position of <googlemap>...</googlemap> is
-        returned additionally.
 
     wikitext = '''
     <googlemap version="0.9" lat="47.113291" lon="11.272337" zoom="15">
@@ -176,8 +181,7 @@ def parse_googlemap(wikitext, detail=False):
         zoom is the google zoom level as integer or None if not provided
         coords is a list of (lon, lat, symbol, title) tuples.
         paths is a list of (style, coords) tuples.
-        coords is again a list of (lot, lat, symbol, title) tuples.
-        If detail is True, (center, zoom, coords, paths, start, end) is returned."""
+        coords is again a list of (lot, lat, symbol, title) tuples."""
 
     def is_coord(line):
         """Returns True if the line contains a coordinate."""
@@ -199,15 +203,21 @@ def parse_googlemap(wikitext, detail=False):
         if not match is None: return (float(match.group(2)), float(match.group(1)), None, match.group(3))
         match = re.match(u'([0-9]{1,2}\.[0-9]+), ?([0-9]{1,2}\.[0-9]+)', line)
         if not match is None: return (float(match.group(2)), float(match.group(1)), None, None)
-        return RuntimeError(u'Could not parse line ' + line)
-
-    regexp = re.compile(u"(<googlemap[^>]*>)(.*?)(</googlemap>)", re.DOTALL)
-    match = regexp.search(wikitext)
-    if match is None: return None
-    start = match.start()
-    end = match.end()
-    content = match.group(2)
-    gm = xml.etree.ElementTree.XML((match.group(1)+match.group(3)).encode('UTF8'))
+        return ParseError(u'Could not parse line ' + line)
+
+    start, content, endtag, end = find_tag(wikitext, 'googlemap')
+    if start is None:
+        raise ParseError('<googlemap> tag not found.')
+    if content is None:
+        xml_only = wikitext[start:endtag]
+    else:
+        xml_only = wikitext[start:content]+wikitext[endtag:end]
+        
+    try:
+        gm = xml.etree.ElementTree.XML(xml_only.encode('UTF8'))
+    except xml.etree.ElementTree.ParseError as e:
+        row, column = e.position
+        raise ParseError("XML parse error in <googlemap ...>.")
     zoom = gm.get('zoom')
     lon = gm.get('lon')
     lat = gm.get('lat')
@@ -218,7 +228,7 @@ def parse_googlemap(wikitext, detail=False):
 
     coords = []
     paths = []
-    lines = content.split("\n")
+    lines = wikitext[content:endtag].split("\n")
     i = 0
     while i < len(lines):
         line = lines[i].strip()
@@ -257,8 +267,6 @@ def parse_googlemap(wikitext, detail=False):
             coords.append((lon, lat, symbol, title))
             continue
 
-        raise RuntimeError(u'Unknown line syntax: ' + line)
-    if detail:
-        return (center, zoom, coords, paths, start, end)
+        raise ParseError(u'Unknown line syntax: ' + line)
     return (center, zoom, coords, paths)