Mercurial > p > roundup > code

annotate roundup/backends/indexer_rdbms.py @ 2929:7a8a02646d4e

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
backend is an attribute of tracker instances
author Alexander Smishlajev <a1s@users.sourceforge.net>
date Thu, 18 Nov 2004 16:23:58 +0000
parents d530b68e4b42
children f8d0fd056ac0
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
1 ''' This implements the full-text indexer over two RDBMS tables. The first
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
2 is a mapping of words to occurance IDs. The second maps the IDs to (Class,
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
3 propname, itemid) instances.
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
4 '''
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
5 import re
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
6
2872
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2098
diff changeset
7 from indexer_dbm import Indexer, is_stopword
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
8
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
9 class Indexer(Indexer):
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
10 def __init__(self, db):
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
11 self.db = db
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
12 self.reindex = 0
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
13
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
14 def close(self):
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
15 '''close the indexing database'''
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
16 # just nuke the circular reference
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
17 self.db = None
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
18
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
19 def force_reindex(self):
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
20 '''Force a reindexing of the database. This essentially
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
21 empties the tables ids and index and sets a flag so
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
22 that the databases are reindexed'''
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
23 self.reindex = 1
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
24
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
25 def should_reindex(self):
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
26 '''returns True if the indexes need to be rebuilt'''
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
27 return self.reindex
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
28
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
29 def add_text(self, identifier, text, mime_type='text/plain'):
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
30 ''' "identifier" is (classname, itemid, property) '''
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
31 if mime_type != 'text/plain':
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
32 return
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
33
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
34 # first, find the id of the (classname, itemid, property)
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
35 a = self.db.arg
2098
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
36 sql = 'select _textid from __textids where _class=%s and '\
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
37 '_itemid=%s and _prop=%s'%(a, a, a)
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
38 self.db.cursor.execute(sql, identifier)
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
39 r = self.db.cursor.fetchone()
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
40 if not r:
2098
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
41 id = self.db.newid('__textids')
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
42 sql = 'insert into __textids (_textid, _class, _itemid, _prop)'\
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
43 ' values (%s, %s, %s, %s)'%(a, a, a, a)
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
44 self.db.cursor.execute(sql, (id, ) + identifier)
2098
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
45 self.db.cursor.execute('select max(_textid) from __textids')
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
46 id = self.db.cursor.fetchone()[0]
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
47 else:
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
48 id = int(r[0])
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
49 # clear out any existing indexed values
2098
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
50 sql = 'delete from __words where _textid=%s'%a
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
51 self.db.cursor.execute(sql, (id, ))
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
52
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
53 # ok, find all the words in the text
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
54 wordlist = re.findall(r'\b\w{2,25}\b', str(text).upper())
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
55 words = {}
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
56 for word in wordlist:
2872
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2098
diff changeset
57 if is_stopword(word):
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2098
diff changeset
58 continue
d530b68e4b42 don't index common words [SF#1046612]
Richard Jones <richard@users.sourceforge.net>
parents: 2098
diff changeset
59 words[word] = 1
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
60 words = words.keys()
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
61
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
62 # for each word, add an entry in the db
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
63 for word in words:
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
64 # don't dupe
2098
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
65 sql = 'select * from __words where _word=%s and _textid=%s'%(a, a)
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
66 self.db.cursor.execute(sql, (word, id))
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
67 if self.db.cursor.fetchall():
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
68 continue
2098
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
69 sql = 'insert into __words (_word, _textid) values (%s, %s)'%(a, a)
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
70 self.db.cursor.execute(sql, (word, id))
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
71
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
72 def find(self, wordlist):
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
73 '''look up all the words in the wordlist.
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
74 If none are found return an empty dictionary
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
75 * more rules here
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
76 '''
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
77 l = [word.upper() for word in wordlist if 26 > len(word) > 2]
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
78
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
79 a = ','.join([self.db.arg] * len(l))
2098
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
80 sql = 'select distinct(_textid) from __words where _word in (%s)'%a
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
81 self.db.cursor.execute(sql, tuple(l))
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
82 r = self.db.cursor.fetchall()
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
83 if not r:
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
84 return {}
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
85 a = ','.join([self.db.arg] * len(r))
2098
18addf2a8596 Implemented proper datatypes in mysql and postgresql backends...
Richard Jones <richard@users.sourceforge.net>
parents: 2093
diff changeset
86 sql = 'select _class, _itemid, _prop from __textids '\
2093
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
87 'where _textid in (%s)'%a
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
88 self.db.cursor.execute(sql, tuple([int(id) for (id,) in r]))
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
89 # self.search_index has the results as {some id: identifier} ...
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
90 # sigh
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
91 r = {}
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
92 k = 0
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
93 for c,n,p in self.db.cursor.fetchall():
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
94 key = (str(c), str(n), str(p))
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
95 r[k] = key
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
96 k += 1
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
97 return r
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
98
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
99 def save_index(self):
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
100 # the normal RDBMS backend transaction mechanisms will handle this
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
101 pass
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
102
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
103 def rollback(self):
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
104 # the normal RDBMS backend transaction mechanisms will handle this
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
105 pass
3f6024ab2c7a That's the last of the RDBMS migration steps done! Yay!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
106
Roundup Issue Tracker: http://roundup-tracker.org/