3
Google Summer of Code is finished, spent a couple of weeks away from
4
the project to think about it and also to take a break. Working on it
5
now I discovered bugs in the query parser (now fixed), and I also
6
un-convoluted the querylib logic (well slighlty).
8
Updated ZFS files to latest SVN.
12
Fixed index document count, and created new config variable to store
13
the size. (Search now has 3 global vars in $CFG, date, size and complete,
14
see indexer.php for var names). Index size is cached to provide an always
15
current value for the index - this is to take into account the fact that
16
deleted documents are in fact not removed from the index, but instead just
17
marked as deleted and not returned in search results. The actual document
18
still features in the index, and skews sizes. When the index optimiser is
19
completed in ZFS, then these deleted documents will be pruned, thus
20
correctly modifying the index size.
22
Additional commenting added.
24
Query page logic very slightly modified to clean up GET string a bit (removed
27
Add/delete functions added to other document types.
29
A few TODO fields added to source, indicating changes still to come (or at
30
least to be considered).
34
Add/delete/update cron functions finished - can be called seperately
35
or all at once via cron.php.
37
Document date field added to index and database summary.
39
Some index db functionality abstracted out to indexlib.php - can
40
use IndexDBControl class to add/del documents from database, and
41
to make sure the db table is functioning.
43
DB sql files changed to add some extra fields.
45
Default 'simple' query modified to search title and author, as well
46
as contents of document, to provide better results for users.
50
First revision of the advanced search page completed. Functional,
51
but needs a date search field still.
55
Added resource search type, and the ability to specify custom 'virtual'
56
models to search - allowing for non-module specific information to be
57
indexed. Specify the extra search types to use in lib.php.
61
Added delete logic to documents; the moodle database log is checked
62
and any found delete events are used to remove the referenced documents
63
from the database table and search index.
65
Added database table name constant to lib.php, must change files using
66
the static table name.
68
Changed documents to use 'docid' instead of 'id' to reference the moodle
69
instance id, since Zend Search adds it's own internal 'id' field. Noticed
70
this whilst working on deletions.
72
Added some additional fields to the permissions checking method, must still
77
Query logic moved into the SearchQuery class in querylib.php. Should be able
78
to include this file in any page and run a query against the index (PHP 5
79
checks must be added to those pages then, though).
81
Index info can be retrieved using IndexInfo class in indexlib.php.
83
Abstracted some stuff away, to reduce rendundancy and decrease the
84
likelihood of errors. Improved the stats.php page to include some
85
diagnostics for adminstrators.
87
delete.php skeleton created for removing deleted documents from the
88
index. cron.php will contain the logic for running delete.php,
89
update.php and eventually add.php.
93
(Warning: It took me 1900 seconds to index the forum, go make coffee
94
whilst you wait.) [Moodle.org forum data]
96
Forum search functions changed to use 'get_recordset' instead of
97
'get_records', for speed reasons. This provides a significant improvement,
98
but indexing is still slow - getting data from the database and Zend's
99
tokenising _seem_ to be the prime suspects at the moment.
101
/search/tests/ added - index.php can be used to see which modules are
102
ready to be included in the search index, and it informs you of any
103
errors - should be a prerequisite for indexing.
105
Search result pagination added to query.php, will default to 20 until
106
an admin page for the search module is written.
110
Search-enabling functions moved out've the mod's lib.php files and into
111
/search/documents/mod_document.php - this requires the search module to
112
operate without requiring modification of lib files.
114
SearchDocument base class improved, and the way module documents extend
115
it. A custom-data field has been added to allow modules to add any custom
116
data they wish to be stored in the index - this field is serialised into
117
the index as a binary field.
119
Database field 'type' renamed to 'doctype' to match the renaming in the
120
index, 'type' seems to be a reserved word in Lucene. Several index field
121
names change to be more descriptive (cid -> course_id). URLs are now
122
stored in the index, and don't have to be generated on the fly during
123
display of query results.
127
Started cleaning and standardising things.
131
This is the initial release (prototype) of Moodle's new search module -
132
so basically watch out for sharp edges.
134
The structure has not been finalised, but this is what is working at the
135
moment, when I start looking at other content to index, it will most likely
136
change. I don't recommend trying to make your own content modules indexable,
137
at least not until the whole flow is finalised. I will be implementing the
138
functions needed to index all of the default content modules on Moodle, so
139
expect that around mid-August.
141
Wiki pages were my goal for this release, they can be indexed and searched,
142
but not updated or deleted at this stage (was waiting for ZF 0.14 actually).
144
I need to check the PostgreSQL sql file, I don't have a PG7 install lying
145
around to test on, so the script is untested.
147
To index for the first time, login as an admin user and browse to /search/index.php
148
or /search/stats.php - there will be a message and a link telling you to go index.
150
email: cynnical@gmail.com
b'\\ No newline at end of file'
1
This directoery contains the central implementation of
2
Moodle's Global Search Engine.
4
The Global Search Engine stores indexes about a huge quantity
5
of information from within modules, block or resources stored
6
by Moodle either in the database or the file system.
8
The administrator initialy indexes the existing content. Once this
9
first initialization performed, the search engine maintains indexes
10
regularily, adding new entries, deleting obsolete one or updating
11
some that have changed.
13
Search will produce links for acceding the information in a similar
14
context as usually accessed, from the current user point of view.
15
Results filtering removes from results any link to information the
16
current user would not be allowed to acces on a straight situation.
21
The search engine is now part of Moodle core distribution.
23
Some extra libraries might be added for converting physical documents to text
24
so it can be indexed. Moodle CVS (entry contrib/patches/global_search_libraries)
25
provides packs for antiword and xpdf GPL libraries the search engine is ready for
26
shockwave indexing, but will not provide Adobe Search converters that should be
27
obtained at http://www.adobe.com/licensing/developer/
29
1. Go to the block administration panel and setup once the Global Search
30
block. This will initialize useful parameters for the global search engine.
32
2. Insert a new Global Search block somewhere in a course or top-level screen.
34
3. Launch an empty search (you must be administrator).
36
4. Go to the statistics screen.
38
5. Activate indexation (indexersplash.php). Beware, if your Moodle has
39
a large amount of content, indexing process may be VERY LONG.
41
To search, go back to the search block and try a query.
43
Handled information for indexing
44
################################
46
In the actual state, the engine indexes the following information:
48
- assignment descriptions
50
- database records (using textual fields only)
53
- glossary comments on entries
54
- Moodle native resources
55
- physical MSWord files as resources (.doc)
56
- physical Powerpoint files as resources (.ppt)
57
- physical PDF files as resources
58
- physical text files as resources (.txt)
59
- physical html files as resources (.htm and .html)
60
- physical xml files as resources (.xml)
65
Some third party plugins are also searchable using the new Search API implementation
72
The reviewed search engine API allows:
74
- indexing of blocks contents
75
- indexation of modules or blocks containing a complex information model
76
- securing the access to the results
77
- adding indexing handling for additional modules and plugins adding a php calibrated script
78
- adding physical filetype handling adding a php calibrated script
80
Global Search on NFS Mounted clusters
81
#####################################
83
This version contains a patched Lucene Zend implementation that allows using the Global Search engine in an NFS mounted shared volume for Web clustering. This implementation
84
remains highly experimental and not all tests have been processed. Some changes may
85
occur in the SoftLockManager that was added to the Lucene engine.
90
- Should be added more information to index such as forum and glossary attachements,
91
so will other standard module contents.
92
- extending the search capability to a mnet network information space by aggregating remote search responses.
b'\\ No newline at end of file'