diff options
author | John Mark Bell <jmb@netsurf-browser.org> | 2008-11-30 16:43:37 +0000 |
---|---|---|
committer | John Mark Bell <jmb@netsurf-browser.org> | 2008-11-30 16:43:37 +0000 |
commit | 79a34e5d5a9a93d9232efd595291bc5ebea2f2c5 (patch) | |
tree | c35e7cc49699420e01a973f4bccb639931e9cc1d /src/parse/parse.c | |
parent | 7dad1d392d1c380648f6de13f81ade48ce958be1 (diff) | |
download | libcss-79a34e5d5a9a93d9232efd595291bc5ebea2f2c5.tar.gz libcss-79a34e5d5a9a93d9232efd595291bc5ebea2f2c5.tar.bz2 |
Use parserutils_hash instead of parserutils_dict.
This approximately halves the size of the interned string table.
We now have the following for allzengarden.css:
5507 slots used (of 8192 => 67.224121%)
Data:
3 full blocks: 12288 bytes
10 partial blocks: 38946 bytes (of 40960 => 95.083008%)
Total: 53488 (4112) (32)
Entries:
21 full blocks: 86016 bytes
1 partial blocks: 2096 bytes (of 4096 => 51.171875%)
Total: 90496 (4112) (32)
Hash structures: 65592
Which gives a total dictionary size of 209,576 bytes.
Note that 43% of this is parserutils_hash_entry structures (length-pointer pairs). It would be good, therefore, to be able to purge these.
svn path=/trunk/libcss/; revision=5852
Diffstat (limited to 'src/parse/parse.c')
-rw-r--r-- | src/parse/parse.c | 18 |
1 files changed, 9 insertions, 9 deletions
diff --git a/src/parse/parse.c b/src/parse/parse.c index bfa6c27..d7ad461 100644 --- a/src/parse/parse.c +++ b/src/parse/parse.c @@ -10,7 +10,7 @@ #include <stdbool.h> #include <parserutils/input/inputstream.h> -#include <parserutils/utils/dict.h> +#include <parserutils/utils/hash.h> #include <parserutils/utils/stack.h> #include <parserutils/utils/vector.h> @@ -91,7 +91,7 @@ struct css_parser #define STACK_CHUNK 32 parserutils_stack *states; /**< Stack of states */ - parserutils_dict *dictionary; /**< Dictionary for interned strings */ + parserutils_hash *dictionary; /**< Dictionary for interned strings */ parserutils_vector *tokens; /**< Vector of pending tokens */ @@ -185,7 +185,7 @@ static css_error (*parseFuncs[])(css_parser *parser) = { * CSS_NOMEM on memory exhaustion */ css_error css_parser_create(const char *charset, css_charset_source cs_source, - parserutils_dict *dictionary, css_alloc alloc, void *pw, + parserutils_hash *dictionary, css_alloc alloc, void *pw, css_parser **parser) { css_parser *p; @@ -420,13 +420,13 @@ const char *css_parser_read_charset(css_parser *parser, const uint8_t *css_parser_dict_add(css_parser *parser, const uint8_t *data, size_t len) { - const parserutils_dict_entry *interned; + const parserutils_hash_entry *interned; parserutils_error perror; if (parser == NULL || data == NULL || len == 0) return NULL; - perror = parserutils_dict_insert(parser->dictionary, data, len, + perror = parserutils_hash_insert(parser->dictionary, data, len, &interned); if (perror != PARSERUTILS_OK) return NULL; @@ -584,7 +584,7 @@ css_error getToken(css_parser *parser, const css_token **token) if (t->type != CSS_TOKEN_S && t->data.data != NULL && t->data.len > 0) { /* Insert token text into the dictionary */ - const parserutils_dict_entry *interned; + const parserutils_hash_entry *interned; uint8_t temp[t->data.len]; bool lower = false; @@ -607,7 +607,7 @@ css_error getToken(css_parser *parser, const css_token **token) /* We get to insert it twice - once for the raw * data, and once for a lowercased version that * we need internally. */ - perror = parserutils_dict_insert( + perror = parserutils_hash_insert( parser->dictionary, temp, t->data.len, &interned); @@ -619,13 +619,13 @@ css_error getToken(css_parser *parser, const css_token **token) t->lower.data = interned->data; t->lower.len = interned->len; - perror = parserutils_dict_insert( + perror = parserutils_hash_insert( parser->dictionary, t->data.data, t->data.len, &interned); } else { /* Otherwise, we're not interested in case */ - perror = parserutils_dict_insert( + perror = parserutils_hash_insert( parser->dictionary, t->data.data, t->data.len, &interned); |