dovecot-2.2: lib-mail, fts: Put application/xhtml+xml MIME parts...

Tue Oct 27 21:57:14 UTC 2015

details:   http://hg.dovecot.org/dovecot-2.2/rev/c02969e65b64
changeset: 19334:c02969e65b64
user:      Timo Sirainen <tss at iki.fi>
date:      Tue Oct 27 23:56:48 2015 +0200
description:
lib-mail, fts: Put application/xhtml+xml MIME parts through html parser as well.

diffstat:

 src/lib-mail/mail-html2text.h       |   7 +++++++
 src/lib-mail/message-snippet.c      |   2 +-
 src/lib-mail/test-message-snippet.c |  14 ++++++++++++++
 src/plugins/fts/fts-parser-html.c   |   2 +-
 4 files changed, 23 insertions(+), 2 deletions(-)

diffs (63 lines):

diff -r c7d384bc3964 -r c02969e65b64 src/lib-mail/mail-html2text.h

--- a/src/lib-mail/mail-html2text.h	Mon Oct 26 17:20:49 2015 +0200
+++ b/src/lib-mail/mail-html2text.h	Tue Oct 27 23:56:48 2015 +0200
@@ -12,4 +12,11 @@
 			 buffer_t *output);
 void mail_html2text_deinit(struct mail_html2text **ht);
 
+static inline bool
+mail_html2text_content_type_match(const char *content_type)
+{
+	return strcasecmp(content_type, "text/html") == 0 ||
+		strcasecmp(content_type, "application/xhtml+xml") == 0;
+}
+
 #endif
diff -r c7d384bc3964 -r c02969e65b64 src/lib-mail/message-snippet.c
--- a/src/lib-mail/message-snippet.c	Mon Oct 26 17:20:49 2015 +0200
+++ b/src/lib-mail/message-snippet.c	Tue Oct 27 23:56:48 2015 +0200
@@ -115,7 +115,7 @@
 			ct = message_decoder_current_content_type(decoder);
 			if (ct == NULL)
 				/* text/plain */ ;
-			else if (strcasecmp(ct, "text/html") == 0) {
+			else if (mail_html2text_content_type_match(ct)) {
 				ctx.html2text = mail_html2text_init(MAIL_HTML2TEXT_FLAG_SKIP_QUOTED);
 				ctx.plain_output = buffer_create_dynamic(pool, 1024);
 			} else if (strncasecmp(ct, "text/", 5) != 0)
diff -r c7d384bc3964 -r c02969e65b64 src/lib-mail/test-message-snippet.c
--- a/src/lib-mail/test-message-snippet.c	Mon Oct 26 17:20:49 2015 +0200
+++ b/src/lib-mail/test-message-snippet.c	Tue Oct 27 23:56:48 2015 +0200
@@ -51,6 +51,20 @@
 	  "</div><br =class=3D\"\"></body></html>=\n",
 	  100,
 	  "Hi, How is it going? > -foo" },
+
+	{ "Content-Transfer-Encoding: quoted-printable\n"
+	  "Content-Type: application/xhtml+xml;\n"
+	  "      charset=utf-8\n"
+	  "\n"
+	  "<html><head><meta http-equiv=3D\"Content-Type\" content=3D\"text/html =\n"
+	  "charset=3Dutf-8\"></head><body style=3D\"word-wrap: break-word; =\n"
+	  "-webkit-nbsp-mode: space; -webkit-line-break: after-white-space;\" =\n"
+	  "class=3D\"\">Hi,<div class=3D\"\"><br class=3D\"\"></div><div class=3D\"\">How =\n"
+	  "is it going? <blockquote>quoted text is ignored</blockquote>\n"
+	  "> -foo\n"
+	  "</div><br =class=3D\"\"></body></html>=\n",
+	  100,
+	  "Hi, How is it going? > -foo" },
 };
 
 static void test_message_snippet(void)
diff -r c7d384bc3964 -r c02969e65b64 src/plugins/fts/fts-parser-html.c
--- a/src/plugins/fts/fts-parser-html.c	Mon Oct 26 17:20:49 2015 +0200
+++ b/src/plugins/fts/fts-parser-html.c	Tue Oct 27 23:56:48 2015 +0200
@@ -19,7 +19,7 @@
 {
 	struct html_fts_parser *parser;
 
-	if (strcasecmp(content_type, "text/html") != 0)
+	if (!mail_html2text_content_type_match(content_type))
 		return NULL;
 
 	parser = i_new(struct html_fts_parser, 1);
