annotate wikked/fs.py @ 124:6c261cb94631

Fixed wiki history page and API endpoint.
author Ludovic Chabant <ludovic@chabant.com>
date Sat, 23 Nov 2013 13:35:37 -0800
parents 827e236aa7c6
children 72e5f588f989
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
1 import os
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
2 import os.path
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
3 import re
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
4 import string
24
8a83b0e91633 The wiki `fs` only scans known file extensions.
Ludovic Chabant <ludovic@chabant.com>
parents: 18
diff changeset
5 import codecs
49
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
6 import logging
124
6c261cb94631 Fixed wiki history page and API endpoint.
Ludovic Chabant <ludovic@chabant.com>
parents: 110
diff changeset
7 from utils import PageNotFoundError, title_to_url, path_to_url
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
8
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
9
82
9afe4a1dbd1e Refactoring of core wiki classes:
Ludovic Chabant <ludovic@chabant.com>
parents: 65
diff changeset
10 class PageInfo(object):
101
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
11 def __init__(self, url, path):
82
9afe4a1dbd1e Refactoring of core wiki classes:
Ludovic Chabant <ludovic@chabant.com>
parents: 65
diff changeset
12 self.url = url
9afe4a1dbd1e Refactoring of core wiki classes:
Ludovic Chabant <ludovic@chabant.com>
parents: 65
diff changeset
13 self.path = path
101
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
14 self._content = None
82
9afe4a1dbd1e Refactoring of core wiki classes:
Ludovic Chabant <ludovic@chabant.com>
parents: 65
diff changeset
15
9afe4a1dbd1e Refactoring of core wiki classes:
Ludovic Chabant <ludovic@chabant.com>
parents: 65
diff changeset
16 @property
101
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
17 def content(self):
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
18 if self._content is None:
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
19 with codecs.open(self.path, 'r', encoding='utf-8') as f:
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
20 self._content = f.read()
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
21 return self._content
82
9afe4a1dbd1e Refactoring of core wiki classes:
Ludovic Chabant <ludovic@chabant.com>
parents: 65
diff changeset
22
9afe4a1dbd1e Refactoring of core wiki classes:
Ludovic Chabant <ludovic@chabant.com>
parents: 65
diff changeset
23
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
24 class FileSystem(object):
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
25 """ A class responsible for mapping page URLs to
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
26 file-system paths, and for scanning the file-system
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
27 to list existing pages.
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
28 """
100
fd6eccb24882 Refactoring to get rid of `slugify` callbacks.
Ludovic Chabant <ludovic@chabant.com>
parents: 82
diff changeset
29 def __init__(self, root, logger=None):
35
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
30 self.root = unicode(root)
49
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
31
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
32 if logger is None:
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
33 logger = logging.getLogger('wikked.fs')
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
34 self.logger = logger
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
35
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
36 self.excluded = []
24
8a83b0e91633 The wiki `fs` only scans known file extensions.
Ludovic Chabant <ludovic@chabant.com>
parents: 18
diff changeset
37 self.page_extensions = None
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
38
3
59cad6ce1a1c Added support for history and diffing.
Ludovic Chabant <ludovic@chabant.com>
parents: 0
diff changeset
39 def getPageInfos(self, subdir=None):
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
40 basepath = self.root
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
41 if subdir is not None:
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
42 basepath = self.getPhysicalNamespacePath(subdir)
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
43
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
44 for dirpath, dirnames, filenames in os.walk(basepath):
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
45 dirnames[:] = [d for d in dirnames if os.path.join(dirpath, d) not in self.excluded]
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
46 for filename in filenames:
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
47 path = os.path.join(dirpath, filename)
8
793234411100 Fixed Mercurial files incorrectly included as pages.
Ludovic Chabant <ludovic@chabant.com>
parents: 3
diff changeset
48 if path in self.excluded:
793234411100 Fixed Mercurial files incorrectly included as pages.
Ludovic Chabant <ludovic@chabant.com>
parents: 3
diff changeset
49 continue
18
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
50 page_info = self.getPageInfo(path)
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
51 if page_info is not None:
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
52 yield page_info
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
53
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
54 def getPageInfo(self, path):
35
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
55 if not isinstance(path, unicode):
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
56 path = unicode(path)
18
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
57 for e in self.excluded:
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
58 if path.startswith(e):
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
59 return None
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
60 return self._getPageInfo(path)
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
61
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
62 def getPage(self, url):
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
63 path = self.getPhysicalPagePath(url)
101
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
64 return PageInfo(url, path)
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
65
101
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
66 def setPage(self, url, content):
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
67 path = self.getPhysicalPagePath(url)
49
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
68 with codecs.open(path, 'w', encoding='utf-8') as f:
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
69 f.write(content)
101
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
70 return PageInfo(url, path)
49
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
71
3
59cad6ce1a1c Added support for history and diffing.
Ludovic Chabant <ludovic@chabant.com>
parents: 0
diff changeset
72 def pageExists(self, url):
59cad6ce1a1c Added support for history and diffing.
Ludovic Chabant <ludovic@chabant.com>
parents: 0
diff changeset
73 try:
59cad6ce1a1c Added support for history and diffing.
Ludovic Chabant <ludovic@chabant.com>
parents: 0
diff changeset
74 self.getPhysicalPagePath(url)
59cad6ce1a1c Added support for history and diffing.
Ludovic Chabant <ludovic@chabant.com>
parents: 0
diff changeset
75 return True
59cad6ce1a1c Added support for history and diffing.
Ludovic Chabant <ludovic@chabant.com>
parents: 0
diff changeset
76 except PageNotFoundError:
59cad6ce1a1c Added support for history and diffing.
Ludovic Chabant <ludovic@chabant.com>
parents: 0
diff changeset
77 return False
59cad6ce1a1c Added support for history and diffing.
Ludovic Chabant <ludovic@chabant.com>
parents: 0
diff changeset
78
49
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
79 def getPhysicalPagePath(self, url):
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
80 return self._getPhysicalPath(url, True)
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
81
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
82 def getPhysicalNamespacePath(self, url):
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
83 return self._getPhysicalPath(url, False)
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
84
18
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
85 def _getPageInfo(self, path):
35
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
86 rel_path = os.path.relpath(path, self.root)
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
87 rel_path_split = os.path.splitext(rel_path)
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
88 ext = rel_path_split[1].lstrip('.')
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
89 name = rel_path_split[0]
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
90 if len(ext) == 0:
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
91 return None
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
92 if self.page_extensions is not None and ext not in self.page_extensions:
2b35d719f342 Handle wiki and page permissions for read/write access.
Ludovic Chabant <ludovic@chabant.com>
parents: 24
diff changeset
93 return None
65
c6dcf6716d26 Fixed slugification bug on Windows.
Ludovic Chabant <ludovic@chabant.com>
parents: 49
diff changeset
94
124
6c261cb94631 Fixed wiki history page and API endpoint.
Ludovic Chabant <ludovic@chabant.com>
parents: 110
diff changeset
95 url = path_to_url(unicode(name), strip_ext=True)
101
13249e5ca51c Big refactor for better database caching:
Ludovic Chabant <ludovic@chabant.com>
parents: 100
diff changeset
96 return PageInfo(url, path)
18
67c150d5ed53 Added ability to get a single page's info from the file-system.
Ludovic Chabant <ludovic@chabant.com>
parents: 8
diff changeset
97
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
98 def _getPhysicalPath(self, url, is_file):
110
827e236aa7c6 Various front-end fixes:
Ludovic Chabant <ludovic@chabant.com>
parents: 101
diff changeset
99 if url[0] != '/':
827e236aa7c6 Various front-end fixes:
Ludovic Chabant <ludovic@chabant.com>
parents: 101
diff changeset
100 raise ValueError("Page URLs need to be absolute: " + url)
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
101 if string.find(url, '..') >= 0:
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
102 raise ValueError("Page URLs can't contain '..': " + url)
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
103
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
104 # For each "part" in the given URL, find the first
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
105 # file-system entry that would get slugified to an
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
106 # equal string.
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
107 current = self.root
110
827e236aa7c6 Various front-end fixes:
Ludovic Chabant <ludovic@chabant.com>
parents: 101
diff changeset
108 parts = unicode(url[1:]).lower().split('/')
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
109 for i, part in enumerate(parts):
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
110 names = os.listdir(current)
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
111 for name in names:
100
fd6eccb24882 Refactoring to get rid of `slugify` callbacks.
Ludovic Chabant <ludovic@chabant.com>
parents: 82
diff changeset
112 name_formatted = title_to_url(name)
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
113 if is_file and i == len(parts) - 1:
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
114 # If we're looking for a file and this is the last part,
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
115 # look for something similar but with an extension.
49
fb6ae96756c1 Added unit tests.
Ludovic Chabant <ludovic@chabant.com>
parents: 47
diff changeset
116 if re.match(r"%s\.[a-z]+" % re.escape(part), name_formatted):
0
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
117 current = os.path.join(current, name)
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
118 break
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
119 else:
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
120 if name_formatted == part:
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
121 current = os.path.join(current, name)
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
122 break
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
123 else:
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
124 # Failed to find a part of the URL.
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
125 raise PageNotFoundError("No such page: " + url)
c946f4facfa2 Initial commit.
Ludovic Chabant <ludovic@chabant.com>
parents:
diff changeset
126 return current
110
827e236aa7c6 Various front-end fixes:
Ludovic Chabant <ludovic@chabant.com>
parents: 101
diff changeset
127