annotate roundup/backends/indexer_dbm.py @ 3192:eb00a2fa0e0e maint-0.8 0.8.0

pre-release stuff
author Richard Jones <richard@users.sourceforge.net>
date Wed, 16 Feb 2005 00:29:18 +0000
parents d530b68e4b42
children 1c063814d567
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 #
2872
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
17 #$Id: indexer_dbm.py,v 1.2 2004-11-05 05:10:07 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
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
26
2872
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
27 stopwords = [
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
28 "A", "AND", "ARE", "AS", "AT", "BE", "BUT", "BY",
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
29 "FOR", "IF", "IN", "INTO", "IS", "IT",
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
30 "NO", "NOT", "OF", "ON", "OR", "SUCH",
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
31 "THAT", "THE", "THEIR", "THEN", "THERE", "THESE",
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
32 "THEY", "THIS", "TO", "WAS", "WILL", "WITH"
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
33 ]
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
34 is_stopword = {}
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
35 for word in stopwords:
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
36 is_stopword[word] = None
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
37 is_stopword = is_stopword.has_key
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
38
2089
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
39 class Indexer:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
40 '''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
41 searching.
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
42
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
43 Three structures are created by the indexer::
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
44
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
45 files {identifier: (fileid, wordcount)}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
46 words {word: {fileid: count}}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
47 fileids {fileid: identifier}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
48
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
49 where identifier is (classname, nodeid, propertyname)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
50 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
51 def __init__(self, db_path):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
52 self.indexdb_path = os.path.join(db_path, 'indexes')
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
53 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
54 self.reindex = 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
55 self.quiet = 9
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
56 self.changed = 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
57
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
58 # 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
59 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
60 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
61 not os.path.exists(version)):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
62 # 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
63 self.force_reindex()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
64 elif os.path.exists(version):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
65 version = open(version).read()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
66 # 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
67 if version.strip() != '1':
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
68 self.force_reindex()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
69
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
70 def force_reindex(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
71 '''Force a reindex condition
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
72 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
73 if os.path.exists(self.indexdb_path):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
74 shutil.rmtree(self.indexdb_path)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
75 os.makedirs(self.indexdb_path)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
76 os.chmod(self.indexdb_path, 0775)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
77 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
78 self.reindex = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
79 self.changed = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
80
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
81 def should_reindex(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
82 '''Should we reindex?
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
83 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
84 return self.reindex
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
85
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
86 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
87 '''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
88 identifier.
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 # make sure the index is loaded
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
91 self.load_index()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
92
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
93 # remove old entries for this identifier
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
94 if self.files.has_key(identifier):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
95 self.purge_entry(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 # split into words
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
98 words = self.splitter(text, mime_type)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
99
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
100 # 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
101 # (_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
102 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
103 file_index = abs(self.files['_TOP'][0])
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
104 self.files[identifier] = (file_index, len(words))
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
105 self.fileids[file_index] = identifier
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 # find the unique words
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
108 filedict = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
109 for word in words:
2872
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
110 if is_stopword(word):
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2089
diff changeset
111 continue
2089
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
112 if filedict.has_key(word):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
113 filedict[word] = filedict[word]+1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
114 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
115 filedict[word] = 1
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 # now add to the totals
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
118 for word in filedict.keys():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
119 # 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
120 if self.words.has_key(word):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
121 entry = self.words[word]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
122 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
123 # new word
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
124 entry = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
125 self.words[word] = entry
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
126
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
127 # 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
128 entry[file_index] = filedict[word]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
129
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
130 # save needed
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
131 self.changed = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
132
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
133 def splitter(self, text, ftype):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
134 '''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
135 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
136 if ftype == 'text/plain':
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
137 words = self.text_splitter(text)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
138 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
139 return []
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
140 return words
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
141
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
142 def text_splitter(self, text):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
143 """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
144 """
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
145 # case insensitive
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
146 text = str(text).upper()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
147
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
148 # 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
149 # 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
150 # 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
151 # place
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
152 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
153
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
154 def search(self, search_terms, klass, ignore={},
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
155 dre=re.compile(r'([^\d]+)(\d+)')):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
156 '''Display search results looking for [search, terms] associated
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
157 with the hyperdb Class "klass". Ignore hits on {class: property}.
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
158
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
159 "dre" is a helper, not an argument.
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
160 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
161 # do the index lookup
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
162 hits = self.find(search_terms)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
163 if not hits:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
164 return {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
165
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
166 designator_propname = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
167 for nm, propclass in klass.getprops().items():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
168 if isinstance(propclass, Link) or isinstance(propclass, Multilink):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
169 designator_propname[propclass.classname] = nm
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
170
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
171 # build a dictionary of nodes and their associated messages
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
172 # and files
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
173 nodeids = {} # this is the answer
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
174 propspec = {} # used to do the klass.find
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
175 for propname in designator_propname.values():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
176 propspec[propname] = {} # used as a set (value doesn't matter)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
177 for classname, nodeid, property in hits.values():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
178 # skip this result if we don't care about this class/property
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
179 if ignore.has_key((classname, property)):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
180 continue
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
181
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
182 # if it's a property on klass, it's easy
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
183 if classname == klass.classname:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
184 if not nodeids.has_key(nodeid):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
185 nodeids[nodeid] = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
186 continue
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
187
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
188 # make sure the class is a linked one, otherwise ignore
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
189 if not designator_propname.has_key(classname):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
190 continue
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
191
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
192 # it's a linked class - set up to do the klass.find
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
193 linkprop = designator_propname[classname] # eg, msg -> messages
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
194 propspec[linkprop][nodeid] = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
195
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
196 # retain only the meaningful entries
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
197 for propname, idset in propspec.items():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
198 if not idset:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
199 del propspec[propname]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
200
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
201 # klass.find tells me the klass nodeids the linked nodes relate to
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
202 for resid in klass.find(**propspec):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
203 resid = str(resid)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
204 if not nodeids.has_key(id):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
205 nodeids[resid] = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
206 node_dict = nodeids[resid]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
207 # now figure out where it came from
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
208 for linkprop in propspec.keys():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
209 for nodeid in klass.get(resid, linkprop):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
210 if propspec[linkprop].has_key(nodeid):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
211 # OK, this node[propname] has a winner
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
212 if not node_dict.has_key(linkprop):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
213 node_dict[linkprop] = [nodeid]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
214 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
215 node_dict[linkprop].append(nodeid)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
216 return nodeids
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
217
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
218 # 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
219 # the (fail) case.
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
220 def find(self, wordlist):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
221 '''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
222 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
223 if not hasattr(self, 'words'):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
224 self.load_index()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
225 self.load_index(wordlist=wordlist)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
226 entries = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
227 hits = None
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
228 for word in wordlist:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
229 if not 2 < len(word) < 25:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
230 # 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
231 continue
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
232 word = word.upper()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
233 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
234 entries[word] = entry # of matching files
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
235 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
236 return {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
237 if hits is None:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
238 hits = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
239 for k in entry.keys():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
240 if not self.fileids.has_key(k):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
241 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
242 hits[k] = self.fileids[k]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
243 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
244 # Eliminate hits for every non-match
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
245 for fileid in hits.keys():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
246 if not entry.has_key(fileid):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
247 del hits[fileid]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
248 if hits is None:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
249 return {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
250 return hits
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
251
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
252 segments = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ#_-!"
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
253 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
254 # 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
255 if self.index_loaded() and not reload:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
256 return 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
257
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
258 # Ok, now let's actually load it
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
259 db = {'WORDS': {}, 'FILES': {'_TOP':(0,None)}, 'FILEIDS': {}}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
260
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
261 # Identify the relevant word-dictionary segments
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
262 if not wordlist:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
263 segments = self.segments
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
264 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
265 segments = ['-','#']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
266 for word in wordlist:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
267 segments.append(word[0].upper())
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
268
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
269 # Load the segments
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
270 for segment in segments:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
271 try:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
272 f = open(self.indexdb + segment, 'rb')
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
273 except IOError, error:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
274 # probably just nonexistent segment index file
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
275 if error.errno != errno.ENOENT: raise
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
276 else:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
277 pickle_str = zlib.decompress(f.read())
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
278 f.close()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
279 dbslice = marshal.loads(pickle_str)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
280 if dbslice.get('WORDS'):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
281 # if it has some words, add them
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
282 for word, entry in dbslice['WORDS'].items():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
283 db['WORDS'][word] = entry
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
284 if dbslice.get('FILES'):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
285 # if it has some files, add them
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
286 db['FILES'] = dbslice['FILES']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
287 if dbslice.get('FILEIDS'):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
288 # if it has fileids, add them
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
289 db['FILEIDS'] = dbslice['FILEIDS']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
290
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
291 self.words = db['WORDS']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
292 self.files = db['FILES']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
293 self.fileids = db['FILEIDS']
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
294 self.changed = 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
295
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
296 def save_index(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
297 # 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
298 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
299 return
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
300
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
301 # 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
302 for segment in self.segments:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
303 try:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
304 os.remove(self.indexdb + segment)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
305 except OSError, error:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
306 # probably just nonexistent segment index file
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
307 if error.errno != errno.ENOENT: raise
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
308
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
309 # 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
310 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
311 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
312
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
313 # 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
314 letters = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ#_"
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
315 segdicts = {} # Need batch of empty dicts
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
316 for segment in letters:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
317 segdicts[segment] = {}
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
318 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
319 initchar = word[0].upper()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
320 segdicts[initchar][word] = entry
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
321
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
322 # save
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
323 for initchar in letters:
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
324 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
325 pickle_str = marshal.dumps(db)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
326 filename = self.indexdb + initchar
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
327 pickle_fh = open(filename, 'wb')
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
328 pickle_fh.write(zlib.compress(pickle_str))
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
329 os.chmod(filename, 0664)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
330
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
331 # save done
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
332 self.changed = 0
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
333
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
334 def purge_entry(self, identifier):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
335 '''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
336 '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
337 self.load_index()
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
338
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
339 if not self.files.has_key(identifier):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
340 return
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
341
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
342 file_index = self.files[identifier][0]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
343 del self.files[identifier]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
344 del self.fileids[file_index]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
345
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
346 # 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
347 for key, occurs in self.words.items():
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
348 if occurs.has_key(file_index):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
349 del occurs[file_index]
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
350
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
351 # save needed
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
352 self.changed = 1
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
353
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
354 def index_loaded(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
355 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
356 hasattr(self,'words'))
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
357
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
358
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
359 def rollback(self):
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
360 ''' load last saved index info. '''
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
361 self.load_index(reload=1)
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
362
93f03c6714d8 A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
363 # vim: set filetype=python ts=4 sw=4 et si

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