svnno****@sourc*****
svnno****@sourc*****
Fri Mar 7 04:31:04 JST 2008
Revision: 3450
http://svn.sourceforge.jp/cgi-bin/viewcvs.cgi?root=kazehakase&view=rev&rev=3450
Author: pal_gene
Date: 2008-03-07 04:31:04 +0900 (Fri, 07 Mar 2008)
Log Message:
-----------
Refactor. separate function.
Modified Paths:
--------------
kazehakase/trunk/module/search/kz-hyper-estraier-search.c
Modified: kazehakase/trunk/module/search/kz-hyper-estraier-search.c
===================================================================
--- kazehakase/trunk/module/search/kz-hyper-estraier-search.c 2008-03-06 19:31:00 UTC (rev 3449)
+++ kazehakase/trunk/module/search/kz-hyper-estraier-search.c 2008-03-06 19:31:04 UTC (rev 3450)
@@ -592,6 +592,51 @@
return g_strndup(start, strstr(start, "<") - start);
}
+/**
+ * get utf-8 contents string for given html file
+ * @param filepath file name to get document contents.
+ * @return gchar* newly allocated utf-8 string, or NULL failed.
+ */
+static gchar *
+get_utf8_contents(const char *filepath)
+{
+ gchar *contents = NULL;
+
+ if (g_file_get_contents(filepath, &contents, NULL, NULL))
+ {
+ gchar *encoding = get_document_encoding(contents);
+ if (!encoding)
+ encoding = g_strdup(est_enc_name(contents,
+ strlen(contents),
+ ESTLANGJA));
+
+ if (g_ascii_strcasecmp(encoding, "UTF-8") != 0)
+ {
+ gchar *utf8_contents, *upper_encoding;
+ upper_encoding = g_ascii_strup(encoding, -1);
+ utf8_contents = g_convert(contents, -1,
+ "UTF-8", upper_encoding,
+ NULL, NULL,
+ NULL);
+ g_free(upper_encoding);
+ g_free(contents);
+ if (utf8_contents)
+ {
+ contents = utf8_contents;
+ }
+ else
+ {
+ g_warning("failed convert encoding. [enc:file] [%s:%s]\n",
+ encoding, filepath);
+ contents = NULL;
+ }
+ }
+ g_free(encoding);
+ }
+
+ return contents;
+}
+
gboolean
register_document (KzSearch *search, const gchar *uri, const gchar *title, const gchar *contents, GTime mtime)
{
@@ -730,7 +775,7 @@
}
else
{
- gchar *uri, *title, *contents, *encoding;
+ gchar *uri, *title, *contents;
GTime mtime;
struct stat st;
@@ -743,46 +788,12 @@
continue;
}
- g_file_get_contents(new_path, &contents, NULL, NULL);
- encoding = get_document_encoding(contents);
- if (!encoding)
- encoding = g_strdup(est_enc_name(contents,
- strlen(contents),
- ESTLANGJA));
- if (encoding)
- {
- if (g_ascii_strcasecmp(encoding, "UTF-8") != 0)
- {
- gchar *utf8_contents, *upper_encoding;
- upper_encoding = g_ascii_strup(encoding, -1);
- utf8_contents = g_convert(contents, -1,
- "UTF-8", upper_encoding,
- NULL, NULL,
- NULL);
- g_free(upper_encoding);
- g_free(contents);
- if (utf8_contents)
- {
- contents = utf8_contents;
- }
- else
- {
- g_warning("failed convert encoding. [enc:file] [%s:%s]\n",
- encoding, new_path);
- g_free(encoding);
- g_free(new_path);
- continue;
- }
- }
- g_free(encoding);
- }
- else
- {
- g_warning("unknown encoding. [%s]\n", new_path);
- g_free(contents);
+ contents = get_utf8_contents(new_path);
+ if (!contents) {
g_free(new_path);
continue;
}
+
g_stat(new_path, &st);
mtime = st.st_mtime;
title = get_document_title(contents);