annotate piecrust/uriutil.py @ 550:6f216c1ab6b1

bake: Add a flag to know which record entries got collapsed from last run. This makes it possible to find entries for things that were actually baked during the current run, as opposed to skipped because they were "clean".
author Ludovic Chabant <ludovic@chabant.com>
date Tue, 04 Aug 2015 21:22:30 -0700
parents 4284c673bb91
children e85f29b28b84
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
1 import re
33
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
2 import os.path
3
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
3 import string
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
4 import logging
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
5 import functools
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
6
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
7
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
8 logger = logging.getLogger(__name__)
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
9
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
10
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
11 class UriError(Exception):
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
12 def __init__(self, uri):
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
13 super(UriError, self).__init__("Invalid URI: %s" % uri)
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
14
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
15
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
16 @functools.total_ordering
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
17 class UriInfo(object):
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
18 def __init__(self, uri, source, args, taxonomy=None, page_num=1):
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
19 self.uri = uri
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
20 self.source = source
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
21 self.args = args
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
22 self.taxonomy = taxonomy
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
23 self.page_num = page_num
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
24
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
25 def __eq__(self, other):
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
26 return ((self.uri, self.source, self.args, self.taxonomy,
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
27 self.page_num) ==
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
28 (other.uri, other.source, other.args, other.taxonomy,
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
29 other.page_num))
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
30
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
31 def __lt__(self, other):
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
32 return ((self.uri, self.source, self.args, self.taxonomy,
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
33 self.page_num) <
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
34 (other.uri, other.source, other.args, other.taxonomy,
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
35 other.page_num))
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
36
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
37
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
38 pagenum_pattern = re.compile(r'/(\d+)/?$')
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
39
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
40
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
41 def parse_uri(routes, uri):
6
f5ca5c5bed85 More Python 3 fixes, modularization, and new unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 5
diff changeset
42 if uri.find('..') >= 0:
3
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
43 raise UriError(uri)
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
44
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
45 page_num = 1
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
46 match = pagenum_pattern.search(uri)
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
47 if match is not None:
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
48 uri = uri[:match.start()]
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
49 page_num = int(match.group(1))
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
50
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
51 uri = '/' + uri.strip('/')
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
52
5
474c9882decf Upgrade to Python 3.
Ludovic Chabant <ludovic@chabant.com>
parents: 3
diff changeset
53 for rn, rc in routes.items():
3
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
54 pattern = route_to_pattern(rn)
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
55 m = re.match(pattern, uri)
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
56 if m is not None:
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
57 args = m.groupdict()
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
58 return UriInfo(uri, rc['source'], args, rc.get('taxonomy'),
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
59 page_num)
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
60
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
61 return None
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
62
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
63
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
64 r2p_pattern = re.compile(r'%(\w+)%')
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
65
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
66
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
67 def route_to_pattern(route):
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
68 return r2p_pattern.sub(r'(?P<\1>[\w\-]+)', route)
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
69
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
70
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
71 def multi_replace(text, replacements):
5
474c9882decf Upgrade to Python 3.
Ludovic Chabant <ludovic@chabant.com>
parents: 3
diff changeset
72 reps = dict((re.escape(k), v) for k, v in replacements.items())
474c9882decf Upgrade to Python 3.
Ludovic Chabant <ludovic@chabant.com>
parents: 3
diff changeset
73 pattern = re.compile("|".join(list(reps.keys())))
3
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
74 return pattern.sub(lambda m: reps[re.escape(m.group(0))], text)
f485ba500df3 Gigantic change to basically make PieCrust 2 vaguely functional.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
75
12
30a42341cfa8 Define page slugs properly, avoid recursions with debug data.
Ludovic Chabant <ludovic@chabant.com>
parents: 6
diff changeset
76
329
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
77 def split_uri(app, uri):
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
78 root = app.config.get('site/root')
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
79 uri_root = uri[:len(root)]
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
80 if uri_root != root:
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
81 raise Exception("URI '%s' is not a full URI." % uri)
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
82 uri = uri[len(root):]
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
83 return uri_root, uri
12
30a42341cfa8 Define page slugs properly, avoid recursions with debug data.
Ludovic Chabant <ludovic@chabant.com>
parents: 6
diff changeset
84
33
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
85
298
b7ab1b503510 data: Fix incorrect next/previous page URLs in pagination data.
Ludovic Chabant <ludovic@chabant.com>
parents: 33
diff changeset
86 def split_sub_uri(app, uri):
329
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
87 root = app.config.get('site/root')
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
88 if not uri.startswith(root):
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
89 raise Exception("URI '%s' is not a full URI." % uri)
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
90
33
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
91 pretty_urls = app.config.get('site/pretty_urls')
484
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
92 trailing_slash = app.config.get('site/trailing_slash')
33
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
93 if not pretty_urls:
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
94 uri, ext = os.path.splitext(uri)
495
4284c673bb91 internal: Fix some edge-cases for splitting sub-URIs.
Ludovic Chabant <ludovic@chabant.com>
parents: 484
diff changeset
95 else:
484
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
96 uri = uri.rstrip('/')
33
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
97
298
b7ab1b503510 data: Fix incorrect next/previous page URLs in pagination data.
Ludovic Chabant <ludovic@chabant.com>
parents: 33
diff changeset
98 page_num = 1
33
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
99 pgn_suffix_re = app.config.get('__cache/pagination_suffix_re')
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
100 m = re.search(pgn_suffix_re, uri)
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
101 if m:
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
102 uri = uri[:m.start()]
298
b7ab1b503510 data: Fix incorrect next/previous page URLs in pagination data.
Ludovic Chabant <ludovic@chabant.com>
parents: 33
diff changeset
103 page_num = int(m.group('num'))
33
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
104
484
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
105 if len(uri) < len(root):
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
106 # The only reasons the URI could have gotten shorter are:
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
107 # - if the regexp "ate" the trailing slash of the root.
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
108 # - if we stripped the trailing slash on a root URL.
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
109 uri += '/'
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
110
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
111 if len(uri) > len(root):
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
112 # Now if we don't have a root URI, make it conform to the rules
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
113 # (re-add the extension, or re-add the trailing slash).
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
114 if not pretty_urls:
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
115 uri += ext
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
116 elif trailing_slash:
d4321317beae internal: Correctly split sub URIs. Add unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 329
diff changeset
117 uri += '/'
33
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
118
329
422052d2e978 internal: Try handling URLs in a consistent way.
Ludovic Chabant <ludovic@chabant.com>
parents: 298
diff changeset
119 return uri, page_num
33
62c7a97c8340 Get the un-paginated URL of a page early and pass that around.
Ludovic Chabant <ludovic@chabant.com>
parents: 12
diff changeset
120