Mercurial > p > roundup > code
annotate roundup/backends/indexer_dbm.py @ 2623:4e1030d49cea
fix: Option defaults were applied as strings...
...not converted to internal representation;
add Option methods str2value and value2str for value conversion
from and to the strings kept in .ini file;
options MAIL_USERNAME, MAIL_TLS_KEYFILE and MAIL_TLS_CERTFILE
default to empty strings: NODEFAULT value is a configuration
error, and for these options it is ok to keep them unset.
| author | Alexander Smishlajev <a1s@users.sourceforge.net> |
|---|---|
| date | Sun, 25 Jul 2004 14:36:50 +0000 |
| parents | 93f03c6714d8 |
| children | d530b68e4b42 49d1fd44881a |
| 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 # |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
17 #$Id: indexer_dbm.py,v 1.1 2004-03-19 04:47:59 richard Exp $ |
|
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 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
27 class Indexer: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
28 '''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
|
29 searching. |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
30 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
31 Three structures are created by the indexer:: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
32 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
33 files {identifier: (fileid, wordcount)} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
34 words {word: {fileid: count}} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
35 fileids {fileid: identifier} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
36 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
37 where identifier is (classname, nodeid, propertyname) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
38 ''' |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
39 def __init__(self, db_path): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
40 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
|
41 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
|
42 self.reindex = 0 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
43 self.quiet = 9 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
44 self.changed = 0 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
45 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
46 # 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
|
47 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
|
48 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
|
49 not os.path.exists(version)): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
50 # 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
|
51 self.force_reindex() |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
52 elif os.path.exists(version): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
53 version = open(version).read() |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
54 # 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
|
55 if version.strip() != '1': |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
56 self.force_reindex() |
|
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 def force_reindex(self): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
59 '''Force a reindex condition |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
60 ''' |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
61 if os.path.exists(self.indexdb_path): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
62 shutil.rmtree(self.indexdb_path) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
63 os.makedirs(self.indexdb_path) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
64 os.chmod(self.indexdb_path, 0775) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
65 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
|
66 self.reindex = 1 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
67 self.changed = 1 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
68 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
69 def should_reindex(self): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
70 '''Should we reindex? |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
71 ''' |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
72 return self.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 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
|
75 '''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
|
76 identifier. |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
77 ''' |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
78 # make sure the index is loaded |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
79 self.load_index() |
|
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 # remove old entries for this identifier |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
82 if self.files.has_key(identifier): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
83 self.purge_entry(identifier) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
84 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
85 # split into words |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
86 words = self.splitter(text, mime_type) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
87 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
88 # 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
|
89 # (_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
|
90 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
|
91 file_index = abs(self.files['_TOP'][0]) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
92 self.files[identifier] = (file_index, len(words)) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
93 self.fileids[file_index] = identifier |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
94 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
95 # find the unique words |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
96 filedict = {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
97 for word in words: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
98 if filedict.has_key(word): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
99 filedict[word] = filedict[word]+1 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
100 else: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
101 filedict[word] = 1 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
102 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
103 # now add to the totals |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
104 for word in filedict.keys(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
105 # 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
|
106 if self.words.has_key(word): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
107 entry = self.words[word] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
108 else: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
109 # new word |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
110 entry = {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
111 self.words[word] = entry |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
112 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
113 # 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
|
114 entry[file_index] = filedict[word] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
115 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
116 # save needed |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
117 self.changed = 1 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
118 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
119 def splitter(self, text, ftype): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
120 '''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
|
121 ''' |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
122 if ftype == 'text/plain': |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
123 words = self.text_splitter(text) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
124 else: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
125 return [] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
126 return words |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
127 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
128 def text_splitter(self, text): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
129 """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
|
130 """ |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
131 # case insensitive |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
132 text = str(text).upper() |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
133 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
134 # 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
|
135 # 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
|
136 # 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
|
137 # place |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
138 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
|
139 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
140 def search(self, search_terms, klass, ignore={}, |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
141 dre=re.compile(r'([^\d]+)(\d+)')): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
142 '''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
|
143 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
|
144 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
145 "dre" is a helper, not an argument. |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
146 ''' |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
147 # do the index lookup |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
148 hits = self.find(search_terms) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
149 if not hits: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
150 return {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
151 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
152 designator_propname = {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
153 for nm, propclass in klass.getprops().items(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
154 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
|
155 designator_propname[propclass.classname] = nm |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
156 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
157 # 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
|
158 # and files |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
159 nodeids = {} # this is the answer |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
160 propspec = {} # used to do the klass.find |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
161 for propname in designator_propname.values(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
162 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
|
163 for classname, nodeid, property in hits.values(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
164 # 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
|
165 if ignore.has_key((classname, property)): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
166 continue |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
167 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
168 # 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
|
169 if classname == klass.classname: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
170 if not nodeids.has_key(nodeid): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
171 nodeids[nodeid] = {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
172 continue |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
173 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
174 # 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
|
175 if not designator_propname.has_key(classname): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
176 continue |
|
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 # 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
|
179 linkprop = designator_propname[classname] # eg, msg -> messages |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
180 propspec[linkprop][nodeid] = 1 |
|
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 # retain only the meaningful entries |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
183 for propname, idset in propspec.items(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
184 if not idset: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
185 del propspec[propname] |
|
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 # 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
|
188 for resid in klass.find(**propspec): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
189 resid = str(resid) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
190 if not nodeids.has_key(id): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
191 nodeids[resid] = {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
192 node_dict = nodeids[resid] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
193 # now figure out where it came from |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
194 for linkprop in propspec.keys(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
195 for nodeid in klass.get(resid, linkprop): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
196 if propspec[linkprop].has_key(nodeid): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
197 # OK, this node[propname] has a winner |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
198 if not node_dict.has_key(linkprop): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
199 node_dict[linkprop] = [nodeid] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
200 else: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
201 node_dict[linkprop].append(nodeid) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
202 return nodeids |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
203 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
204 # 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
|
205 # the (fail) case. |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
206 def find(self, wordlist): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
207 '''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
|
208 ''' |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
209 if not hasattr(self, 'words'): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
210 self.load_index() |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
211 self.load_index(wordlist=wordlist) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
212 entries = {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
213 hits = None |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
214 for word in wordlist: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
215 if not 2 < len(word) < 25: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
216 # 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
|
217 continue |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
218 word = word.upper() |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
219 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
|
220 entries[word] = entry # of matching files |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
221 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
|
222 return {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
223 if hits is None: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
224 hits = {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
225 for k in entry.keys(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
226 if not self.fileids.has_key(k): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
227 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
|
228 hits[k] = self.fileids[k] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
229 else: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
230 # Eliminate hits for every non-match |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
231 for fileid in hits.keys(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
232 if not entry.has_key(fileid): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
233 del hits[fileid] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
234 if hits is None: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
235 return {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
236 return hits |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
237 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
238 segments = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ#_-!" |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
239 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
|
240 # 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
|
241 if self.index_loaded() and not reload: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
242 return 0 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
243 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
244 # Ok, now let's actually load it |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
245 db = {'WORDS': {}, 'FILES': {'_TOP':(0,None)}, 'FILEIDS': {}} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
246 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
247 # Identify the relevant word-dictionary segments |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
248 if not wordlist: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
249 segments = self.segments |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
250 else: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
251 segments = ['-','#'] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
252 for word in wordlist: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
253 segments.append(word[0].upper()) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
254 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
255 # Load the segments |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
256 for segment in segments: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
257 try: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
258 f = open(self.indexdb + segment, 'rb') |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
259 except IOError, error: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
260 # probably just nonexistent segment index file |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
261 if error.errno != errno.ENOENT: raise |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
262 else: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
263 pickle_str = zlib.decompress(f.read()) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
264 f.close() |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
265 dbslice = marshal.loads(pickle_str) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
266 if dbslice.get('WORDS'): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
267 # if it has some words, add them |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
268 for word, entry in dbslice['WORDS'].items(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
269 db['WORDS'][word] = entry |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
270 if dbslice.get('FILES'): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
271 # if it has some files, add them |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
272 db['FILES'] = dbslice['FILES'] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
273 if dbslice.get('FILEIDS'): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
274 # if it has fileids, add them |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
275 db['FILEIDS'] = dbslice['FILEIDS'] |
|
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 self.words = db['WORDS'] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
278 self.files = db['FILES'] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
279 self.fileids = db['FILEIDS'] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
280 self.changed = 0 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
281 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
282 def save_index(self): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
283 # 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
|
284 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
|
285 return |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
286 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
287 # 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
|
288 for segment in self.segments: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
289 try: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
290 os.remove(self.indexdb + segment) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
291 except OSError, error: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
292 # probably just nonexistent segment index file |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
293 if error.errno != errno.ENOENT: raise |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
294 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
295 # 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
|
296 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
|
297 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
|
298 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
299 # 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
|
300 letters = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ#_" |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
301 segdicts = {} # Need batch of empty dicts |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
302 for segment in letters: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
303 segdicts[segment] = {} |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
304 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
|
305 initchar = word[0].upper() |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
306 segdicts[initchar][word] = entry |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
307 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
308 # save |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
309 for initchar in letters: |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
310 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
|
311 pickle_str = marshal.dumps(db) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
312 filename = self.indexdb + initchar |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
313 pickle_fh = open(filename, 'wb') |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
314 pickle_fh.write(zlib.compress(pickle_str)) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
315 os.chmod(filename, 0664) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
316 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
317 # save done |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
318 self.changed = 0 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
319 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
320 def purge_entry(self, identifier): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
321 '''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
|
322 ''' |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
323 self.load_index() |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
324 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
325 if not self.files.has_key(identifier): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
326 return |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
327 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
328 file_index = self.files[identifier][0] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
329 del self.files[identifier] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
330 del self.fileids[file_index] |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
331 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
332 # 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
|
333 for key, occurs in self.words.items(): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
334 if occurs.has_key(file_index): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
335 del occurs[file_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 # save needed |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
338 self.changed = 1 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
339 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
340 def index_loaded(self): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
341 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
|
342 hasattr(self,'words')) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
343 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
344 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
345 def rollback(self): |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
346 ''' load last saved index info. ''' |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
347 self.load_index(reload=1) |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
348 |
|
93f03c6714d8
A few big changes in this commit:
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
349 # vim: set filetype=python ts=4 sw=4 et si |
