Newer
Older
# -*- coding: utf-8 -*-
# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# #
# MapServer REST API is a python wrapper around MapServer which #
# allows to manipulate a mapfile in a RESTFul way. It has been #
# developped to match as close as possible the way the GeoServer #
# REST API acts. #
# #
# Copyright (C) 2011-2013 Neogeo Technologies. #
# #
# This file is part of MapServer Rest API. #
# #
# MapServer Rest API is free software: you can redistribute it #
# and/or modify it under the terms of the GNU General Public License #
# as published by the Free Software Foundation, either version 3 of #
# the License, or (at your option) any later version. #
# #
# MapServer Rest API is distributed in the hope that it will be #
# useful, but WITHOUT ANY WARRANTY; without even the implied warranty #
# of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the #
# GNU General Public License for more details. #
# #
# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
import pyxml
import StringIO
import urlparse
from xml.etree import ElementTree as etree
from cgi import escape
from xml.sax.saxutils import unescape
def should_be_url(s):
"""This is used to find out if it might be a good idea to
consider a string is a URL.
"""
parsed = urlparse.urlparse(s)
return parsed.scheme and parsed.netloc
__dump_xml_max_element_id = 0
def dump_xml(xml, fp, indent=0, indent_depth=2, reinit=True):
"""Recursive function that transforms an ElementTree to html
written to the file like stream fp.
indent can be used to specify the amount of indentation wanted.
"""
def new_id():
"""Gets a new unique element ID."""
global __dump_xml_max_element_id
__dump_xml_max_element_id += 1
return __dump_xml_max_element_id
def line(fmt, *args):
"""Writes a line to fp with corect indentation."""
fp.write(('%s' + fmt + '\n') % tuple([indent*' ']+list(args)))
if reinit:
# We need to reset element ID if we where nto called recursivly.
global __dump_xml_max_element_id
__dump_xml_max_element_id = 0
if xml.text:
if should_be_url(xml.text):
line('<a href="%s">%s</a>', escape(unescape(xml.text), quote=True), escape(unescape(xml.text)))
else:
line('<pre>%s</pre>', escape(unescape(xml.text)))
elif 'href' in xml.attrib:
line('<a href="%s">%s</a>', escape(xml.attrib['href'], quote=True), escape(xml.attrib['href']))
else:
line('<table class="table table-condensed table-bordered table-hover table-striped" style="margin:0 0 0.2em 0">')
indent += indent_depth
for child in xml:
id = new_id()
line('<tr>')
indent += indent_depth
if 'href' in child.attrib:
line('<td class="key" id="value_%d" onClick="toggle(\'entry_%d\')">href</td>', id, id)
else:
line('<td class="key" id="value_%d" onClick="toggle(\'entry_%d\')">%s%s</td>', id, id,
escape(child.tag), '' if not child.attrib else escape(" %s" % (child.attrib)))
line('<td><div id="entry_%d">', id)
dump_xml(child, fp, indent=indent+indent_depth, reinit=False)
line('</div></td>')
indent -= indent_depth
line('</tr>')
indent -= indent_depth
line('</table>')
def dump(obj, fp, indent=0, *args, **kwargs):
"""Writes the html represention of obj to the file-like object fp.
This uses pyxml to first transform the object into xml.
*args and **kwargs are forwarded to pyxml.xml()
"""
xml = pyxml.xml(obj, *args, **kwargs)
dump_xml(xml, fp, indent)
def dumps(obj, *args, **kwargs):
"""Returns the html representation of obj as a string."""
stream = StringIO.StringIO()
dump(obj, stream, *args, **kwargs)
stream.flush()
return stream.getvalue()