This PostgreSQL extension provides a shared ispell dictionary, i.e. a dictionary that's stored in shared segment. The traditional ispell implementation means that each session initializes and stores the dictionary on it's own, which means a lot of CPU/RAM is wasted.
This extension allocates an area in shared segment (you have to choose the size in advance) and then loads the dictionary into it when it's used for the first time.
If you need just snowball-type dictionaries, this extension is not really interesting for you. But if you really need an ispell dictionary, this may save you a lot of resources.
Before build and install shared_ispell
you should ensure following:
- PostgreSQL version is 9.6 or later.
Installing the extension is quite simple. In that case all you need to do is this:
$ git clone git@github.com:postgrespro/shared_ispell.git
$ cd shared_ispell
$ make USE_PGXS=1
$ make USE_PGXS=1 install
and then (after connecting to the database)
db=# CREATE EXTENSION shared_ispell;
Important: Don't forget to set the
PG_CONFIG
variable in case you want to testshared_ispell
on a custom build of PostgreSQL. Read more here.
No the functions are created, but you still need to load the shared module. This needs to be done from postgresql.conf, as the module needs to allocate space in the shared memory segment. So add this to the config file (or update the current values)
# libraries to load
shared_preload_libraries = 'shared_ispell'
# config of the shared memory
shared_ispell.max_size = 32MB
Yes, there's a single GUC variable that defines the maximum size of the shared segment. This is a hard limit, the shared segment is not extensible and you need to set it so that all the dictionaries fit into it and not much memory is wasted.
To find out how much memory you actually need, use a large value (e.g. 200MB) and load all the dictionaries you want to use. Then use the shared_ispell_mem_used() function to find out how much memory was actually used (and set the max_size GUC variable accordingly).
Don't set it exactly to that value, leave there some free space, so that you can reload the dictionaries without changing the GUC max_size limit (which requires a restart of the DB). Ssomething like 512kB should be just fine.
The shared segment can contain several dictionaries at the same time, the amount of memory is the only limit. There's no limit on number of dictionaries / words etc. Just the max_size GUC variable.
Technically, the extension defines a 'shared_ispell' template that you may use to define custom dictionaries. E.g. you may do this
CREATE TEXT SEARCH DICTIONARY czech_shared (
TEMPLATE = shared_ispell,
DictFile = czech,
AffFile = czech,
StopWords = czech
);
CREATE TEXT SEARCH CONFIGURATION public.czech_shared
( COPY = pg_catalog.simple );
ALTER TEXT SEARCH CONFIGURATION czech_shared
ALTER MAPPING FOR asciiword, asciihword, hword_asciipart,
word, hword, hword_part
WITH czech_shared;
and then do the usual stuff, e.g.
db=# SELECT ts_lexize('czech_shared', 'automobile');
or whatever you want.
The extension provides five management functions, that allow you to manage and get info about the preloaded dictionaries. The first two functions
shared_ispell_mem_used()
shared_ispell_mem_available()
allow you to get info about the shared segment (used and free memory) e.g. to properly size the segment (max_size). Then there are functions return list of dictionaries / stop lists loaded in the shared segment
shared_ispell_dicts()
shared_ispell_stoplists()
e.g. like this
db=# SELECT * FROM shared_ispell_dicts();
dict_name | affix_name | words | affixes | bytes
-----------+------------+-------+---------+----------
bulgarian | bulgarian | 79267 | 12 | 7622128
czech | czech | 96351 | 2544 | 12715000
(2 rows)
db=# SELECT * FROM shared_ispell_stoplists();
stop_name | words | bytes
-----------+-------+-------
czech | 259 | 4552
(1 row)
The last function allows you to reset the dictionary (e.g. so that you can reload the updated files from disk). The sessions that already use the dictionaries will be forced to reinitialize them (the first one will rebuild and copy them in the shared segment, the other ones will use this prepared data).
db=# SELECT shared_ispell_reset();
That's all for now ...
The original version of this module located in the Tomas Vondra's GitHub. That version does not handle affixes that require full regular expressions (regex_t, implemented in regex.h).
This version of the module can handle that affixes with full regular exressions. To handle it the module loads and stores affix files in each sessions. The affix list is tiny and takes a little time and memory to parse. Actually this is Tomas idea, but there is not related code in the GitHub.
Tomas Vondra GitHub