svnno****@sourc*****
svnno****@sourc*****
Sun Mar 2 03:51:37 JST 2008
Revision: 3431
http://svn.sourceforge.jp/cgi-bin/viewcvs.cgi?root=kazehakase&view=rev&rev=3431
Author: pal_gene
Date: 2008-03-02 03:51:36 +0900 (Sun, 02 Mar 2008)
Log Message:
-----------
*Generalize search expression from tool-bar
*Zenkaku white space convert (using UTF-8 Normalize and Composite)
*Support 'site:' expression
usage:
* Only show the google.com URI from history that contain word "kazehakase"
history-search:kazehakase site:google.com
* Drop google.com URI from history that contain word "kazehakase"
history-search:kazehakase -site:google.com
Modified Paths:
--------------
kazehakase/trunk/module/search/kz-hyper-estraier-search.c
kazehakase/trunk/module/search/kz-search-common.h
Modified: kazehakase/trunk/module/search/kz-hyper-estraier-search.c
===================================================================
--- kazehakase/trunk/module/search/kz-hyper-estraier-search.c 2008-03-01 10:12:11 UTC (rev 3430)
+++ kazehakase/trunk/module/search/kz-hyper-estraier-search.c 2008-03-01 18:51:36 UTC (rev 3431)
@@ -258,15 +258,16 @@
G_OBJECT_CLASS(parent_class)->dispose(object);
}
+
static gchar *
create_search_result_html (KzSearch *search, const gchar *text)
{
ESTCOND *cond;
CBLIST *highlights;
int *results, n_results, i;
- gchar *except_word, *tmp;
+ gchar *except_word, *tmp, *utf8;
gchar **texts;
- GString *html, *phrase, *desc_str;
+ GString *html, *phrase, *attr_uri_phrase, *desc_str;
gint num_summary = 128, max_results = 20, half_of_summary;
KzHyperEstraierSearch *he_search;
@@ -279,54 +280,88 @@
cond = est_cond_new();
highlights = cblistopen();
- tmp = g_utf8_casefold(text, -1);
- texts = g_strsplit(tmp, " ", -1);
+ /* convert from kz_conf except keyword */
+ phrase = g_string_new(text);
+ except_word = KZ_CONF_GET_STR("History", "except_keyword");
+ if (except_word && *except_word)
+ {
+ texts = g_strsplit(except_word, ",", -1);
+ g_free(except_word);
+ tmp = g_strjoinv(" -", texts);
+ g_strfreev(texts);
+ g_string_append(phrase, " -");
+ g_string_append(phrase, tmp);
+ g_free(tmp);
+ }
+ utf8 = g_locale_to_utf8(phrase->str, -1, NULL, NULL, NULL);
+ g_string_free(phrase, TRUE);
+ tmp = g_utf8_normalize(utf8, -1, G_NORMALIZE_ALL_COMPOSE);
+ g_free(utf8);
+ utf8 = g_utf8_strdown(tmp, -1);
g_free(tmp);
+
+ /* start word split */
+ texts = g_strsplit(utf8, " ", -1);
+ g_free(utf8);
phrase = g_string_sized_new(0);
for (i = 0; texts[i]; i++)
{
- if(*texts[i] == '-')
+ tmp = texts[i];
+ guint flag = 0;
+ gint cond_i;
+ if(!tmp || !*tmp || g_unichar_isspace(*tmp)) continue;
+ /* extract supported condition [ALL KZ_HISTORY_SEARCH CODE] */
+ for(cond_i = 0; cond_i < KZ_SEARCH_FLAG_SIZE; cond_i++)
{
- g_string_append(phrase, " " ESTOPDIFF " ");
- g_string_append(phrase, texts[i]+1);
+ if(g_str_has_prefix(tmp, KZ_SEARCH_FLAGS[cond_i].exp))
+ {
+ flag |= KZ_SEARCH_FLAGS[cond_i].mask;
+ tmp += strlen(KZ_SEARCH_FLAGS[cond_i].exp);
+ }
}
- else{
- cblistpush(highlights, texts[i], -1);
- g_string_append(phrase, " " ESTOPISECT " ");
- g_string_append(phrase, texts[i]);
+
+ /* write out for search engine expression [ENGINE SPECIFIC CODE] */
+ switch(flag & KZ_SEARCH_FLAG_GROUP_OPTION) {
+ case KZ_SEARCH_FLAG_MASK_SITE:
+ attr_uri_phrase = g_string_sized_new(0);
+ g_string_printf(attr_uri_phrase,
+ "%s %s %s",
+ ESTDATTRURI,
+ (flag & KZ_SEARCH_FLAG_MASK_NOT)?"!ISTRINC":"ISTRINC",
+ tmp);
+ est_cond_add_attr(cond, attr_uri_phrase->str);
+ g_string_free(attr_uri_phrase, TRUE);
+ break;
+ default:
+ if(flag & KZ_SEARCH_FLAG_MASK_NOT)
+ {
+ g_string_append(phrase, " " ESTOPDIFF " ");
+ }
+ else
+ {
+ g_string_append(phrase, " " ESTOPISECT " ");
+ cblistpush(highlights, tmp, -1);
+ }
+ g_string_append(phrase, tmp);
}
}
g_strfreev(texts);
- except_word = KZ_CONF_GET_STR("History", "except_keyword");
- if (except_word && *except_word)
- {
- tmp = g_utf8_casefold(except_word, -1);
- g_free(except_word);
- texts = g_strsplit(tmp, ",", -1);
- g_free(tmp);
- i = 0;
- while (texts[i])
- {
- g_string_append(phrase, " " ESTOPDIFF " ");
- g_string_append(phrase, texts[i]);
- i++;
- }
- g_strfreev(texts);
- }
-
est_cond_set_phrase(cond, phrase->str);
g_string_free(phrase, TRUE);
+ /* start actual search */
KZ_CONF_GET("History", "num_summary", num_summary, INT);
KZ_CONF_GET("History", "max_results", max_results, INT);
half_of_summary = num_summary / 2;
results = est_mtdb_search(he_search->db, cond, &n_results, NULL);
- desc_str = g_string_sized_new(num_summary*2 - 1); /* almost typical in num_summary*2 */
- html = g_string_sized_new((num_summary*2+512)*max_results); /* typical in num_summary*2*max_results */
+ /* almost in typical num_summary*2 */
+ desc_str = g_string_sized_new(num_summary*2 - 1);
+ /* in typical (num_summary*2+html_tags)*max_results */
+ html = g_string_sized_new((num_summary*2+512)*max_results);
g_string_append(html, DTD"\n");
g_string_append(html, "<html>\n");
Modified: kazehakase/trunk/module/search/kz-search-common.h
===================================================================
--- kazehakase/trunk/module/search/kz-search-common.h 2008-03-01 10:12:11 UTC (rev 3430)
+++ kazehakase/trunk/module/search/kz-search-common.h 2008-03-01 18:51:36 UTC (rev 3431)
@@ -21,6 +21,8 @@
#ifndef __KZ_SEARCH_COMMON_H__
#define __KZ_SEARCH_COMMON_H__
+#include <glib.h>
+
G_BEGIN_DECLS
#define DTD "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\" \"http://www.w3.org/TR/html4/strict.dtd\">"
@@ -47,4 +49,30 @@
G_END_DECLS
+
+typedef struct {
+ unsigned int mask;
+ char *exp;
+} KZ_SEARCH_FLAG;
+
+#define KZ_SEARCH_FLAG_GROUP_OPERATOR 0x0000000f
+#define KZ_SEARCH_FLAG_MASK_NOT 0x00000001
+#define KZ_SEARCH_FLAG_GROUP_OPTION 0x000000f0 /* exclusive */
+#define KZ_SEARCH_FLAG_MASK_SITE 0x00000010
+
+//NOTE: synchronize FLAG_NUMBER and FLAGS order
+enum KZ_SEARCH_FLAG_NUMBER {
+ KZ_SEARCH_FLAG_NOT,
+ KZ_SEARCH_FLAG_SITE
+};
+
+KZ_SEARCH_FLAG KZ_SEARCH_FLAGS[] = {
+ { KZ_SEARCH_FLAG_MASK_NOT, "-"},
+ { KZ_SEARCH_FLAG_MASK_SITE, "site:"}
+};
+
+#define KZ_SEARCH_FLAG_SIZE (sizeof(KZ_SEARCH_FLAGS)/sizeof(KZ_SEARCH_FLAG))
+
+
+
#endif /* __KZ_SEARCH_COMMON_H__ */