diff options
| author | Vsevolod Stakhov <vsevolod@rambler-co.ru> | 2009-07-02 19:41:47 +0400 |
|---|---|---|
| committer | Vsevolod Stakhov <vsevolod@rambler-co.ru> | 2009-07-02 19:41:47 +0400 |
| commit | 7348a381a903eea67611fbce0782cf968b965ebf (patch) | |
| tree | 8c411767a10a26e4d530c0eb897c1b76e23d1b9a /src/classifiers | |
| parent | 59ecc76a83abd9d57e2c808f34d4d27568eeed33 (diff) | |
| download | rspamd-7348a381a903eea67611fbce0782cf968b965ebf.tar.gz rspamd-7348a381a903eea67611fbce0782cf968b965ebf.zip | |
* Rework structure and API of statfiles functions to improve performance and avoid missusage of hash table
* Correct url length calculation in urls command
Diffstat (limited to 'src/classifiers')
| -rw-r--r-- | src/classifiers/winnow.c | 24 |
1 files changed, 11 insertions, 13 deletions
diff --git a/src/classifiers/winnow.c b/src/classifiers/winnow.c index 1b3904e67..edd929af0 100644 --- a/src/classifiers/winnow.c +++ b/src/classifiers/winnow.c @@ -34,7 +34,7 @@ struct winnow_callback_data { statfile_pool_t *pool; struct classifier_ctx *ctx; - char *filename; + stat_file_t *file; double sum; int count; int in_class; @@ -49,7 +49,7 @@ classify_callback (gpointer key, gpointer value, gpointer data) float v; /* Consider that not found blocks have value 1 */ - if ((v = statfile_pool_get_block (cd->pool, cd->filename, node->h1, node->h2, cd->now)) < 0.00001) { + if ((v = statfile_pool_get_block (cd->pool, cd->file, node->h1, node->h2, cd->now)) < 0.00001) { cd->sum += 1; } else { @@ -72,11 +72,11 @@ learn_callback (gpointer key, gpointer value, gpointer data) c = (cd->in_class) ? WINNOW_PROMOTION : WINNOW_DEMOTION; /* Consider that not found blocks have value 1 */ - if ((v = statfile_pool_get_block (cd->pool, cd->filename, node->h1, node->h2, cd->now)) < 0.00001) { - statfile_pool_set_block (cd->pool, cd->filename, node->h1, node->h2, cd->now, c); + if ((v = statfile_pool_get_block (cd->pool, cd->file, node->h1, node->h2, cd->now)) < 0.00001) { + statfile_pool_set_block (cd->pool, cd->file, node->h1, node->h2, cd->now, c); } else { - statfile_pool_set_block (cd->pool, cd->filename, node->h1, node->h2, cd->now, v * c); + statfile_pool_set_block (cd->pool, cd->file, node->h1, node->h2, cd->now, v * c); } cd->count ++; @@ -105,14 +105,13 @@ winnow_classify (struct classifier_ctx *ctx, statfile_pool_t *pool, char *statfi g_assert (ctx != NULL); data.pool = pool; - data.filename = statfile; data.sum = 0; data.count = 0; data.now = time (NULL); data.ctx = ctx; - if (!statfile_pool_is_open (pool, statfile)) { - if (statfile_pool_open (pool, statfile) == -1) { + if ((data.file = statfile_pool_is_open (pool, statfile)) == NULL) { + if ((data.file = statfile_pool_open (pool, statfile)) == NULL) { return; } } @@ -138,22 +137,21 @@ winnow_learn (struct classifier_ctx *ctx, statfile_pool_t *pool, char *statfile, g_assert (ctx != NULL); data.pool = pool; - data.filename = statfile; data.sum = 0; data.count = 0; data.in_class = in_class; data.now = time (NULL); data.ctx = ctx; - if (!statfile_pool_is_open (pool, statfile)) { - if (statfile_pool_open (pool, statfile) == -1) { + if ((data.file = statfile_pool_is_open (pool, statfile)) == NULL) { + if ((data.file = statfile_pool_open (pool, statfile)) == NULL) { return; } } - statfile_pool_lock_file (pool, statfile); + statfile_pool_lock_file (pool, data.file); g_tree_foreach (input, learn_callback, &data); - statfile_pool_unlock_file (pool, statfile); + statfile_pool_unlock_file (pool, data.file); } struct winnow_result_data { |