Mercurial > hg > svgsitemap
annotate svgsitemap/middleware.py @ 10:1329127709de
remove orphans in a unique function
| author | Jeff Hammel <jhammel@mozilla.com> |
|---|---|
| date | Sun, 02 Jan 2011 11:43:27 -0800 |
| parents | aa4eab6dc994 |
| children | 1ff7b363e89e |
| rev | line source |
|---|---|
| 0 | 1 """ |
| 2 request dispatcher | |
| 3 """ | |
| 4 | |
| 5 __all__ = ['MapserverMiddleware', 'SVGSiteMap'] | |
| 6 | |
| 7 import os | |
| 1 | 8 import urlparse |
|
6
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
9 from fnmatch import fnmatch |
| 0 | 10 from pygraphviz import AGraph |
| 1 | 11 from webob import Request, Response, exc |
| 0 | 12 |
| 13 class MapserverMiddleware(object): | |
| 14 """silly middleware to serve just the svg""" | |
| 15 def __init__(self, app, svgmap, path='/map'): | |
| 1 | 16 self.app = app |
| 0 | 17 self.svgmap = svgmap |
| 18 self.path = path | |
| 19 | |
| 20 def __call__(self, environ, start_response): | |
| 21 request = Request(environ) | |
| 22 if request.path_info == self.path or not self.path: | |
| 1 | 23 if not os.path.exists(self.svgmap): |
| 24 res = exc.HTTPNotFound() | |
| 25 return res(environ, start_response) | |
| 0 | 26 content = file(self.svgmap).read() |
| 1 | 27 res = Response(content_type='image/svg+xml', body=content) |
| 28 return res(environ, start_response) | |
| 29 return self.app(environ, start_response) | |
| 0 | 30 |
| 31 | |
| 32 class SVGSiteMap(object): | |
| 33 | |
| 34 ### class level variables | |
| 1 | 35 defaults = { 'name': '', |
| 36 'hosts': '', | |
| 37 'external_referers': True, | |
| 2 | 38 'maxwidth': 5, |
| 39 'minwidth': '0.01', | |
|
7
145a4d85b37d
exclude long urls for now ::sigh::
Jeff Hammel <jhammel@mozilla.com>
parents:
6
diff
changeset
|
40 'maxlength': 80, |
|
8
1d16bde33eee
take out some image too ::sigh::
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
41 'excluded': '*.css *.js */static/* /css/* *.ico /backgrounds/* *.png *.jpg', |
| 1 | 42 |
| 43 # input/output | |
|
6
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
44 'file': None, # graphviz file |
| 1 | 45 'output': None, # .svg file |
| 46 | |
| 47 # graph attributes | |
| 0 | 48 'bgcolor': 'black', |
| 49 'fontcolor': 'white', | |
| 50 'fontname': 'Helvetica', | |
| 1 | 51 'fontsize': '10.0', |
| 0 | 52 'nodecolor': 'aqua', |
| 53 'edgecolor': 'lime', | |
| 5 | 54 'shape': 'plaintext', |
| 1 | 55 'len': '1.3', |
| 56 'arrowsize': '0.5', | |
| 0 | 57 } |
| 58 | |
| 59 def __init__(self, app, **kw): | |
| 1 | 60 |
| 61 # boilerplate | |
| 0 | 62 self.app = app |
| 63 self.edges = {} | |
| 1 | 64 self.max = 0 |
| 65 | |
| 66 # set attrs from defaults | |
| 0 | 67 for key in self.defaults: |
| 68 setattr(self, key, kw.get(key, self.defaults[key])) | |
| 1 | 69 |
| 70 # sanity checks + data fixing | |
| 0 | 71 assert self.output, "Please give an output file" |
| 72 assert self.file, "Cannot save file!" | |
| 2 | 73 self.maxwidth = float(self.maxwidth) |
|
6
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
74 if isinstance(self.excluded, basestring): |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
75 self.excluded = self.excluded.split() |
| 1 | 76 if self.hosts: |
| 77 self.hosts = self.hosts.split() | |
| 78 else: | |
| 79 self.hosts = [] | |
| 80 if isinstance(self.external_referers, basestring): | |
| 81 self.external_referers = self.external_referers.lower() == 'true' | |
| 82 | |
| 83 # open the graph | |
| 0 | 84 if os.path.exists(self.file): |
| 1 | 85 self.graph = AGraph(self.file, name=self.name, splines=False, directed=True) |
| 0 | 86 for edge in self.graph.edges(): |
|
6
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
87 if self.exclude(edge[0], edge[1]): |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
88 self.graph.remove_edge(edge[0], edge[1]) |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
89 continue |
|
10
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
90 |
| 1 | 91 count = int(edge.attr['label']) |
| 92 self.edges[(edge[0], edge[1])] = count | |
| 93 if count > self.max: | |
| 94 self.max = count | |
|
10
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
95 self.remove_orphans() |
|
6
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
96 self.set_widths() |
| 0 | 97 else: |
| 1 | 98 self.graph = AGraph(name=self.name, splines=False, directed=True) |
| 99 | |
| 100 # make it pretty | |
| 2 | 101 self.graph.graph_attr['name'] = self.name |
| 0 | 102 self.graph.graph_attr['label'] = self.name |
| 103 self.graph.graph_attr['fontname'] = self.fontname | |
| 104 self.graph.graph_attr['fontcolor'] = self.fontcolor | |
| 105 self.graph.graph_attr['bgcolor'] = self.bgcolor | |
| 106 self.graph.node_attr['color'] = self.nodecolor | |
| 107 self.graph.node_attr['fontcolor'] = self.fontcolor | |
| 1 | 108 self.graph.node_attr['fontname'] = self.fontname |
| 109 self.graph.node_attr['fontsize'] = self.fontsize | |
| 0 | 110 self.graph.node_attr['shape'] = self.shape |
| 111 self.graph.edge_attr['color'] = self.edgecolor | |
| 112 self.graph.edge_attr['fontcolor'] = self.fontcolor | |
| 1 | 113 self.graph.edge_attr['fontname'] = self.fontname |
| 114 self.graph.edge_attr['fontsize'] = self.fontsize | |
| 115 self.graph.edge_attr['len'] = self.len | |
| 116 self.graph.edge_attr['arrowsize'] = self.arrowsize | |
|
9
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
117 |
|
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
118 if self.edges: |
|
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
119 self.save() |
| 0 | 120 |
| 121 def __call__(self, environ, start_response): | |
| 122 request = Request(environ) | |
| 1 | 123 if request.referer: |
| 124 | |
| 125 # parse the URLs | |
| 126 parsed_referer = urlparse.urlsplit(request.referer) | |
| 127 parsed_referee = urlparse.urlsplit(request.url) | |
| 128 islocal = False | |
| 129 | |
| 130 # see if its local or not | |
| 131 localhosts = self.hosts[:] | |
| 132 if parsed_referee.hostname not in localhosts: | |
| 133 localhosts.append(parsed_referee.hostname) | |
| 134 for host in localhosts: | |
| 135 if parsed_referer.hostname == host or parsed_referer.hostname.endswith('.' + host): | |
| 136 islocal = True | |
| 137 break | |
| 138 | |
| 139 # make the connection | |
| 140 if islocal: | |
| 141 self.add(parsed_referer.path, parsed_referee.path) | |
| 142 else: | |
| 143 if self.external_referers: | |
| 144 self.add(request.referer, parsed_referee.path) | |
| 145 | |
| 0 | 146 return self.app(environ, start_response) |
| 147 | |
| 148 def add(self, from_url, to_url): | |
| 1 | 149 """add a conncection in the graph""" |
| 150 | |
| 151 if from_url == to_url: | |
| 152 return # don't do self-references | |
|
6
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
153 if self.exclude(from_url, to_url): |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
154 return # ignore certain urls |
| 1 | 155 |
| 0 | 156 if (from_url, to_url) in self.edges: |
| 157 count = self.edges[(from_url, to_url)] | |
| 158 count += 1 | |
| 1 | 159 if count > self.max: |
| 160 self.max = count | |
| 0 | 161 self.edges[(from_url, to_url)] = count |
| 162 edge = self.graph.get_edge(from_url, to_url) | |
| 163 edge.attr['label'] = str(count) | |
| 164 else: | |
| 165 self.edges[(from_url, to_url)] = 1 | |
| 4 | 166 self.max = 1 |
| 0 | 167 self.graph.add_edge(from_url, to_url, label='1') |
| 168 | |
|
6
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
169 if self.maxwidth: |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
170 self.set_widths() |
| 2 | 171 |
| 0 | 172 for url in from_url, to_url: |
| 173 node = self.graph.get_node(url) | |
| 174 node.attr['label'] = url | |
| 175 node.attr['href'] = url | |
| 176 | |
|
9
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
177 self.save() |
|
6
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
178 |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
179 def exclude(self, *urls): |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
180 """tell whether the edge is excluded""" |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
181 for pattern in self.excluded: |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
182 for url in urls: |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
183 if fnmatch(url, pattern): |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
184 return True |
|
7
145a4d85b37d
exclude long urls for now ::sigh::
Jeff Hammel <jhammel@mozilla.com>
parents:
6
diff
changeset
|
185 for url in urls: |
|
145a4d85b37d
exclude long urls for now ::sigh::
Jeff Hammel <jhammel@mozilla.com>
parents:
6
diff
changeset
|
186 if len(url) > self.maxlength: |
|
145a4d85b37d
exclude long urls for now ::sigh::
Jeff Hammel <jhammel@mozilla.com>
parents:
6
diff
changeset
|
187 return True |
|
6
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
188 return False |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
189 |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
190 def set_widths(self): |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
191 if self.maxwidth: |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
192 for edge in self.graph.edges(): |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
193 count = self.edges[(edge[0], edge[1])] |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
194 width = self.maxwidth * count / self.max |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
195 if not width: |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
196 width = self.minwidth |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
197 edge.attr['style'] = 'setlinewidth(%s)' % width |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
198 |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
199 else: |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
200 for edge in self.graph.edges(): |
|
a88a670c92d0
dont display things that you dont want
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
201 edge.attr['style'] = '' |
|
9
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
202 |
|
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
203 def save(self): |
|
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
204 if self.file: |
|
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
205 self.graph.write(self.file) |
|
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
206 if self.output: |
|
aa4eab6dc994
* dont set node width, height; * move save() to its own function
Jeff Hammel <jhammel@mozilla.com>
parents:
8
diff
changeset
|
207 self.graph.draw(self.output, prog='neato') |
|
10
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
208 |
|
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
209 def remove_orphans(self): |
|
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
210 flag = True |
|
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
211 while flag: |
|
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
212 flag = False |
|
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
213 for node in self.graph.nodes(): |
|
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
214 if not self.graph.neighbors(node) or self.exclude(node): |
|
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
215 flag = True |
|
1329127709de
remove orphans in a unique function
Jeff Hammel <jhammel@mozilla.com>
parents:
9
diff
changeset
|
216 self.graph.remove_node(node) |