comparison libpurple/util.c @ 17895:1b74553c5c08

When converting HTML to XHTML and plain text on message send, links in the HTML are converted to 'text' <'url'> in plain text instead of just 'text'.

17894:bbd92ee894f2

purple_markup_html_to_xhtml(const char *html, char **xhtml_out,
						  char **plain_out)
{
	GString *xhtml = g_string_new("");
	GString *plain = g_string_new("");
	GList *tags = NULL, *tag;
	const char *c = html;

	while(c && *c) {
		if(*c == '<') {

				}
				if(tag) {
					while(tags) {
						struct purple_parse_tag *pt = tags->data;
						g_string_append_printf(xhtml, "</%s>", pt->dest_tag);
						if(tags == tag)
							break;
						tags = g_list_remove(tags, pt);
						g_free(pt);
					}

						plain = g_string_append_c(plain, '<');
						c++;
					}
				}
			} else { /* opening tag */
				ALLOW_TAG("a");
				ALLOW_TAG("blockquote");
				ALLOW_TAG("cite");
				ALLOW_TAG("div");
				ALLOW_TAG("em");
				ALLOW_TAG("h1");

					pt->src_tag = "sup";
					pt->dest_tag = "span";
					tags = g_list_prepend(tags, pt);
					c = strchr(c, '>') + 1;
					xhtml = g_string_append(xhtml, "<span style='vertical-align:super;'>");
					continue;
				}
				if(!g_ascii_strncasecmp(c, "<font", 5) && (*(c+5) == '>' || *(c+5) == ' ')) {
					const char *p = c;
					GString *style = g_string_new("");

		*xhtml_out = g_strdup(xhtml->str);
	if(plain_out)
		*plain_out = g_strdup(plain->str);
	g_string_free(xhtml, TRUE);
	g_string_free(plain, TRUE);
}

/* The following are probably reasonable changes:
 * - \n should be converted to a normal space
 * - in addition to <br>, <p> and <div> etc. should also be converted into \n

17895:1b74553c5c08

purple_markup_html_to_xhtml(const char *html, char **xhtml_out,
						  char **plain_out)
{
	GString *xhtml = g_string_new("");
	GString *plain = g_string_new("");
	GString *url = NULL;
	GList *tags = NULL, *tag;
	const char *c = html;

	while(c && *c) {
		if(*c == '<') {

				}
				if(tag) {
					while(tags) {
						struct purple_parse_tag *pt = tags->data;
						g_string_append_printf(xhtml, "</%s>", pt->dest_tag);
						if(!strcmp(pt->src_tag, "a")) {
							/* if this is a link, we have to add the url to the plaintext, too */
							g_string_append_printf(plain, " <%s>", g_strstrip(url->str));
						}
						if(tags == tag)
							break;
						tags = g_list_remove(tags, pt);
						g_free(pt);
					}

						plain = g_string_append_c(plain, '<');
						c++;
					}
				}
			} else { /* opening tag */
				/*ALLOW_TAG("a");*/
				ALLOW_TAG("blockquote");
				ALLOW_TAG("cite");
				ALLOW_TAG("div");
				ALLOW_TAG("em");
				ALLOW_TAG("h1");

					pt->src_tag = "sup";
					pt->dest_tag = "span";
					tags = g_list_prepend(tags, pt);
					c = strchr(c, '>') + 1;
					xhtml = g_string_append(xhtml, "<span style='vertical-align:super;'>");
					continue;
				}
				if(!g_ascii_strncasecmp(c, "<a", 2) && (*(c+2) == '>' || *(c+2) == ' ')) {
					const char *p = c;
					struct purple_parse_tag *pt;
					while(*p && *p != '>') {
						if(!g_ascii_strncasecmp(p, "href=", strlen("href="))) {
							const char *q = p + strlen("href=");
							g_string_free(url, TRUE);
							url = g_string_new("");
							if(*q == '\'' || *q == '\"')
								q++;
							while(*q && *q != '\"' && *q != '\'' && *q != ' ') {
								url = g_string_append_c(url, *q);
								q++;
							}
							p = q;
						}
						p++;
					}
					if ((c = strchr(c, '>')) != NULL)
						c++;
					else
						c = p;
					pt = g_new0(struct purple_parse_tag, 1);
					pt->src_tag = "a";
					pt->dest_tag = "a";
					tags = g_list_prepend(tags, pt);
					g_string_append_printf(xhtml, "<a href='%s'>", g_strstrip(url->str));
					continue;
				}
				if(!g_ascii_strncasecmp(c, "<font", 5) && (*(c+5) == '>' || *(c+5) == ' ')) {
					const char *p = c;
					GString *style = g_string_new("");

		*xhtml_out = g_strdup(xhtml->str);
	if(plain_out)
		*plain_out = g_strdup(plain->str);
	g_string_free(xhtml, TRUE);
	g_string_free(plain, TRUE);
	g_string_free(url, TRUE);
}

/* The following are probably reasonable changes:
 * - \n should be converted to a normal space
 * - in addition to <br>, <p> and <div> etc. should also be converted into \n