annotate roundup/backends/indexer_dbm.py @ 3992:fe2af84a5ca5

allow binary data for "content" props through rawToHyperdb
author Richard Jones <richard@users.sourceforge.net>
date Mon, 18 Aug 2008 06:21:53 +0000
parents 5f4db2650da3
children 2ff6f39aa391
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2089
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
1 #
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
2 # This module is derived from the module described at:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
3 # http://gnosis.cx/publish/programming/charming_python_15.txt
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
4 #
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
5 # Author: David Mertz (mertz@gnosis.cx)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
6 # Thanks to: Pat Knight (p.knight@ktgroup.co.uk)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
7 # Gregory Popovitch (greg@gpy.com)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
8 #
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
9 # The original module was released under this license, and remains under
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
10 # it:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
11 #
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
12 # This file is released to the public domain. I (dqm) would
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
13 # appreciate it if you choose to keep derived works under terms
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
14 # that promote freedom, but obviously am giving up any rights
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
15 # to compel such.
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
16 #
3613
5f4db2650da3 implement close() on all indexers [SF#1242477]
Richard Jones <richard@users.sourceforge.net>
parents: 3555
diff changeset
17 #$Id: indexer_dbm.py,v 1.9 2006-04-27 05:48:26 richard Exp $
2089
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
18 '''This module provides an indexer class, RoundupIndexer, that stores text
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
19 indices in a roundup instance. This class makes searching the content of
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
20 messages, string properties and text files possible.
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
21 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
22 __docformat__ = 'restructuredtext'
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
23
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
24 import os, shutil, re, mimetypes, marshal, zlib, errno
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
25 from roundup.hyperdb import Link, Multilink
3544
5cd1c83dea50 Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents: 3295
diff changeset
26 from roundup.backends.indexer_common import Indexer as IndexerBase
2872
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
27
3544
5cd1c83dea50 Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents: 3295
diff changeset
28 class Indexer(IndexerBase):
2089
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
29 '''Indexes information from roundup's hyperdb to allow efficient
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
30 searching.
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
31
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
32 Three structures are created by the indexer::
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
33
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
34 files {identifier: (fileid, wordcount)}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
35 words {word: {fileid: count}}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
36 fileids {fileid: identifier}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
37
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
38 where identifier is (classname, nodeid, propertyname)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
39 '''
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents: 3092
diff changeset
40 def __init__(self, db):
3544
5cd1c83dea50 Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents: 3295
diff changeset
41 IndexerBase.__init__(self, db)
3295
a615cc230160 added Xapian indexer; replaces standard indexers if Xapian is available
Richard Jones <richard@users.sourceforge.net>
parents: 3092
diff changeset
42 self.indexdb_path = os.path.join(db.config.DATABASE, 'indexes')
2089
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
43 self.indexdb = os.path.join(self.indexdb_path, 'index.db')
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
44 self.reindex = 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
45 self.quiet = 9
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
46 self.changed = 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
47
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
48 # see if we need to reindex because of a change in code
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
49 version = os.path.join(self.indexdb_path, 'version')
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
50 if (not os.path.exists(self.indexdb_path) or
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
51 not os.path.exists(version)):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
52 # for now the file itself is a flag
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
53 self.force_reindex()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
54 elif os.path.exists(version):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
55 version = open(version).read()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
56 # check the value and reindex if it's not the latest
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
57 if version.strip() != '1':
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
58 self.force_reindex()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
59
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
60 def force_reindex(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
61 '''Force a reindex condition
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
62 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
63 if os.path.exists(self.indexdb_path):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
64 shutil.rmtree(self.indexdb_path)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
65 os.makedirs(self.indexdb_path)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
66 os.chmod(self.indexdb_path, 0775)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
67 open(os.path.join(self.indexdb_path, 'version'), 'w').write('1\n')
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
68 self.reindex = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
69 self.changed = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
70
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
71 def should_reindex(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
72 '''Should we reindex?
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
73 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
74 return self.reindex
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
75
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
76 def add_text(self, identifier, text, mime_type='text/plain'):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
77 '''Add some text associated with the (classname, nodeid, property)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
78 identifier.
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
79 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
80 # make sure the index is loaded
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
81 self.load_index()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
82
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
83 # remove old entries for this identifier
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
84 if self.files.has_key(identifier):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
85 self.purge_entry(identifier)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
86
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
87 # split into words
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
88 words = self.splitter(text, mime_type)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
89
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
90 # Find new file index, and assign it to identifier
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
91 # (_TOP uses trick of negative to avoid conflict with file index)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
92 self.files['_TOP'] = (self.files['_TOP'][0]-1, None)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
93 file_index = abs(self.files['_TOP'][0])
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
94 self.files[identifier] = (file_index, len(words))
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
95 self.fileids[file_index] = identifier
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
96
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
97 # find the unique words
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
98 filedict = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
99 for word in words:
3544
5cd1c83dea50 Features and fixes.
Richard Jones <richard@users.sourceforge.net>
parents: 3295
diff changeset
100 if self.is_stopword(word):
2872
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
101 continue
2089
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
102 if filedict.has_key(word):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
103 filedict[word] = filedict[word]+1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
104 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
105 filedict[word] = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
106
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
107 # now add to the totals
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
108 for word in filedict.keys():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
109 # each word has a dict of {identifier: count}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
110 if self.words.has_key(word):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
111 entry = self.words[word]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
112 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
113 # new word
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
114 entry = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
115 self.words[word] = entry
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
116
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
117 # make a reference to the file for this word
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
118 entry[file_index] = filedict[word]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
119
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
120 # save needed
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
121 self.changed = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
122
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
123 def splitter(self, text, ftype):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
124 '''Split the contents of a text string into a list of 'words'
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
125 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
126 if ftype == 'text/plain':
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
127 words = self.text_splitter(text)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
128 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
129 return []
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
130 return words
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
131
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
132 def text_splitter(self, text):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
133 """Split text/plain string into a list of words
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
134 """
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
135 # case insensitive
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
136 text = str(text).upper()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
137
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
138 # Split the raw text, losing anything longer than 25 characters
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
139 # since that'll be gibberish (encoded text or somesuch) or shorter
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
140 # than 3 characters since those short words appear all over the
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
141 # place
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
142 return re.findall(r'\b\w{2,25}\b', text)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
143
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
144 # we override this to ignore not 2 < word < 25 and also to fix a bug -
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
145 # the (fail) case.
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
146 def find(self, wordlist):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
147 '''Locate files that match ALL the words in wordlist
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
148 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
149 if not hasattr(self, 'words'):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
150 self.load_index()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
151 self.load_index(wordlist=wordlist)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
152 entries = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
153 hits = None
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
154 for word in wordlist:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
155 if not 2 < len(word) < 25:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
156 # word outside the bounds of what we index - ignore
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
157 continue
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
158 word = word.upper()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
159 entry = self.words.get(word) # For each word, get index
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
160 entries[word] = entry # of matching files
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
161 if not entry: # Nothing for this one word (fail)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
162 return {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
163 if hits is None:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
164 hits = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
165 for k in entry.keys():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
166 if not self.fileids.has_key(k):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
167 raise ValueError, 'Index is corrupted: re-generate it'
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
168 hits[k] = self.fileids[k]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
169 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
170 # Eliminate hits for every non-match
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
171 for fileid in hits.keys():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
172 if not entry.has_key(fileid):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
173 del hits[fileid]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
174 if hits is None:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
175 return {}
3076
2817a4db901d Change indexer_common.search() to take a list of nodeids...
Johannes Gijsbers <jlgijsbers@users.sourceforge.net>
parents: 3058
diff changeset
176 return hits.values()
2089
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
177
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
178 segments = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ#_-!"
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
179 def load_index(self, reload=0, wordlist=None):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
180 # Unless reload is indicated, do not load twice
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
181 if self.index_loaded() and not reload:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
182 return 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
183
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
184 # Ok, now let's actually load it
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
185 db = {'WORDS': {}, 'FILES': {'_TOP':(0,None)}, 'FILEIDS': {}}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
186
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
187 # Identify the relevant word-dictionary segments
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
188 if not wordlist:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
189 segments = self.segments
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
190 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
191 segments = ['-','#']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
192 for word in wordlist:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
193 segments.append(word[0].upper())
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
194
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
195 # Load the segments
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
196 for segment in segments:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
197 try:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
198 f = open(self.indexdb + segment, 'rb')
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
199 except IOError, error:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
200 # probably just nonexistent segment index file
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
201 if error.errno != errno.ENOENT: raise
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
202 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
203 pickle_str = zlib.decompress(f.read())
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
204 f.close()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
205 dbslice = marshal.loads(pickle_str)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
206 if dbslice.get('WORDS'):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
207 # if it has some words, add them
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
208 for word, entry in dbslice['WORDS'].items():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
209 db['WORDS'][word] = entry
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
210 if dbslice.get('FILES'):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
211 # if it has some files, add them
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
212 db['FILES'] = dbslice['FILES']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
213 if dbslice.get('FILEIDS'):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
214 # if it has fileids, add them
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
215 db['FILEIDS'] = dbslice['FILEIDS']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
216
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
217 self.words = db['WORDS']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
218 self.files = db['FILES']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
219 self.fileids = db['FILEIDS']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
220 self.changed = 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
221
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
222 def save_index(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
223 # only save if the index is loaded and changed
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
224 if not self.index_loaded() or not self.changed:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
225 return
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
226
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
227 # brutal space saver... delete all the small segments
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
228 for segment in self.segments:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
229 try:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
230 os.remove(self.indexdb + segment)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
231 except OSError, error:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
232 # probably just nonexistent segment index file
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
233 if error.errno != errno.ENOENT: raise
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
234
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
235 # First write the much simpler filename/fileid dictionaries
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
236 dbfil = {'WORDS':None, 'FILES':self.files, 'FILEIDS':self.fileids}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
237 open(self.indexdb+'-','wb').write(zlib.compress(marshal.dumps(dbfil)))
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
238
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
239 # The hard part is splitting the word dictionary up, of course
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
240 letters = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ#_"
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
241 segdicts = {} # Need batch of empty dicts
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
242 for segment in letters:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
243 segdicts[segment] = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
244 for word, entry in self.words.items(): # Split into segment dicts
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
245 initchar = word[0].upper()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
246 segdicts[initchar][word] = entry
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
247
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
248 # save
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
249 for initchar in letters:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
250 db = {'WORDS':segdicts[initchar], 'FILES':None, 'FILEIDS':None}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
251 pickle_str = marshal.dumps(db)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
252 filename = self.indexdb + initchar
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
253 pickle_fh = open(filename, 'wb')
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
254 pickle_fh.write(zlib.compress(pickle_str))
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
255 os.chmod(filename, 0664)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
256
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
257 # save done
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
258 self.changed = 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
259
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
260 def purge_entry(self, identifier):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
261 '''Remove a file from file index and word index
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
262 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
263 self.load_index()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
264
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
265 if not self.files.has_key(identifier):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
266 return
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
267
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
268 file_index = self.files[identifier][0]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
269 del self.files[identifier]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
270 del self.fileids[file_index]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
271
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
272 # The much harder part, cleanup the word index
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
273 for key, occurs in self.words.items():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
274 if occurs.has_key(file_index):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
275 del occurs[file_index]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
276
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
277 # save needed
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
278 self.changed = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
279
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
280 def index_loaded(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
281 return (hasattr(self,'fileids') and hasattr(self,'files') and
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
282 hasattr(self,'words'))
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
283
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
284 def rollback(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
285 ''' load last saved index info. '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
286 self.load_index(reload=1)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
287
3613
5f4db2650da3 implement close() on all indexers [SF#1242477]
Richard Jones <richard@users.sourceforge.net>
parents: 3555
diff changeset
288 def close(self):
5f4db2650da3 implement close() on all indexers [SF#1242477]
Richard Jones <richard@users.sourceforge.net>
parents: 3555
diff changeset
289 pass
5f4db2650da3 implement close() on all indexers [SF#1242477]
Richard Jones <richard@users.sourceforge.net>
parents: 3555
diff changeset
290
5f4db2650da3 implement close() on all indexers [SF#1242477]
Richard Jones <richard@users.sourceforge.net>
parents: 3555
diff changeset
291
2089
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
292 # vim: set filetype=python ts=4 sw=4 et si

Roundup Issue Tracker: http://roundup-tracker.org/