v2 of drop HTML tags when indexing

v2 of drop HTML tags when indexing

[David Bremner]

This obsoletes id:20170322112306.12060-1-david@tethera.net

Compared to the previous version:

- drop test already pushed upstream

- add diagram of new state machine

- add longer, and hopefully more illuminating comment about how the
  (new) state machines are encoded

- don't create the variable 'current', stick with the existing 'next'
  as a state index/pointer
  

[PATCH 1/6] lib: add content type argument to uuencode filter.

[David Bremner]

The idea is to support more general types of filtering, based on
content type.
---
 lib/index.cc | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/lib/index.cc b/lib/index.cc
index 8c145540..1c04cc3d 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -56,6 +56,7 @@ typedef struct _NotmuchFilterDiscardUuencodeClass NotmuchFilterDiscardUuencodeCl
  **/
 struct _NotmuchFilterDiscardUuencode {
     GMimeFilter parent_object;
+    GMimeContentType *content_type;
     int state;
 };
 
@@ -63,7 +64,7 @@ struct _NotmuchFilterDiscardUuencodeClass {
     GMimeFilterClass parent_class;
 };
 
-static GMimeFilter *notmuch_filter_discard_uuencode_new (void);
+static GMimeFilter *notmuch_filter_discard_uuencode_new (GMimeContentType *content);
 
 static void notmuch_filter_discard_uuencode_finalize (GObject *object);
 
@@ -102,8 +103,9 @@ notmuch_filter_discard_uuencode_finalize (GObject *object)
 static GMimeFilter *
 filter_copy (GMimeFilter *gmime_filter)
 {
-    (void) gmime_filter;
-    return notmuch_filter_discard_uuencode_new ();
+    NotmuchFilterDiscardUuencode *filter = (NotmuchFilterDiscardUuencode *) gmime_filter;
+
+    return notmuch_filter_discard_uuencode_new (filter->content_type);
 }
 
 static void
@@ -196,7 +198,7 @@ filter_reset (GMimeFilter *gmime_filter)
  * Returns: a new #NotmuchFilterDiscardUuencode filter.
  **/
 static GMimeFilter *
-notmuch_filter_discard_uuencode_new (void)
+notmuch_filter_discard_uuencode_new (GMimeContentType *content_type)
 {
     static GType type = 0;
     NotmuchFilterDiscardUuencode *filter;
@@ -220,6 +222,7 @@ notmuch_filter_discard_uuencode_new (void)
 
     filter = (NotmuchFilterDiscardUuencode *) g_object_newv (type, 0, NULL);
     filter->state = 0;
+    filter->content_type = content_type;
 
     return (GMimeFilter *) filter;
 }
@@ -396,7 +399,7 @@ _index_mime_part (notmuch_message_t *message,
     g_mime_stream_mem_set_owner (GMIME_STREAM_MEM (stream), FALSE);
 
     filter = g_mime_stream_filter_new (stream);
-    discard_uuencode_filter = notmuch_filter_discard_uuencode_new ();
+    discard_uuencode_filter = notmuch_filter_discard_uuencode_new (content_type);
 
     g_mime_stream_filter_add (GMIME_STREAM_FILTER (filter),
 			      discard_uuencode_filter);
-- 
2.11.0

[PATCH 2/6] lib/index: Add another layer of indirection in filtering

[David Bremner]

We could add a second gmime filter subclass, but prefer to avoid
duplicating the boilerplate.
---
 lib/index.cc | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/lib/index.cc b/lib/index.cc
index 1c04cc3d..74a750b9 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -29,6 +29,8 @@
 typedef struct _NotmuchFilterDiscardUuencode NotmuchFilterDiscardUuencode;
 typedef struct _NotmuchFilterDiscardUuencodeClass NotmuchFilterDiscardUuencodeClass;
 
+typedef void (*filter_fun) (GMimeFilter *filter, char *in, size_t len, size_t prespace,
+			    char **out, size_t *outlen, size_t *outprespace);
 /**
  * NotmuchFilterDiscardUuencode:
  *
@@ -57,6 +59,7 @@ typedef struct _NotmuchFilterDiscardUuencodeClass NotmuchFilterDiscardUuencodeCl
 struct _NotmuchFilterDiscardUuencode {
     GMimeFilter parent_object;
     GMimeContentType *content_type;
+    filter_fun real_filter;
     int state;
 };
 
@@ -110,7 +113,14 @@ filter_copy (GMimeFilter *gmime_filter)
 
 static void
 filter_filter (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t prespace,
-	       char **outbuf, size_t *outlen, size_t *outprespace)
+	       char **outbuf, size_t *outlen, size_t *outprespace) {
+    NotmuchFilterDiscardUuencode *filter = (NotmuchFilterDiscardUuencode *) gmime_filter;
+    (*filter->real_filter)(gmime_filter, inbuf, inlen, prespace, outbuf, outlen, outprespace);
+}
+
+static void
+filter_filter_uuencode (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t prespace,
+			char **outbuf, size_t *outlen, size_t *outprespace)
 {
     NotmuchFilterDiscardUuencode *filter = (NotmuchFilterDiscardUuencode *) gmime_filter;
     register const char *inptr = inbuf;
@@ -223,7 +233,7 @@ notmuch_filter_discard_uuencode_new (GMimeContentType *content_type)
     filter = (NotmuchFilterDiscardUuencode *) g_object_newv (type, 0, NULL);
     filter->state = 0;
     filter->content_type = content_type;
-
+    filter->real_filter = filter_filter_uuencode;
     return (GMimeFilter *) filter;
 }
 
-- 
2.11.0

[PATCH 3/6] lib/index: separate state table definition from scanner.

[David Bremner]

We want to reuse the scanner definition with a different table
---
 lib/index.cc | 81 +++++++++++++++++++++++++++++++++++-------------------------
 1 file changed, 47 insertions(+), 34 deletions(-)

diff --git a/lib/index.cc b/lib/index.cc
index 74a750b9..02b35b81 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -31,6 +31,15 @@ typedef struct _NotmuchFilterDiscardUuencodeClass NotmuchFilterDiscardUuencodeCl
 
 typedef void (*filter_fun) (GMimeFilter *filter, char *in, size_t len, size_t prespace,
 			    char **out, size_t *outlen, size_t *outprespace);
+
+typedef struct {
+    int state;
+    int a;
+    int b;
+    int next_if_match;
+    int next_if_not_match;
+} scanner_state_t;
+
 /**
  * NotmuchFilterDiscardUuencode:
  *
@@ -119,46 +128,18 @@ filter_filter (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t pres
 }
 
 static void
-filter_filter_uuencode (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t prespace,
-			char **outbuf, size_t *outlen, size_t *outprespace)
+do_filter (const scanner_state_t states[],
+	   int first_skipping_state,
+	   GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t prespace,
+	   char **outbuf, size_t *outlen, size_t *outprespace)
 {
     NotmuchFilterDiscardUuencode *filter = (NotmuchFilterDiscardUuencode *) gmime_filter;
     register const char *inptr = inbuf;
     const char *inend = inbuf + inlen;
     char *outptr;
-
+    int next;
     (void) prespace;
 
-    /* Simple, linear state-transition diagram for our filter.
-     *
-     * If the character being processed is within the range of [a, b]
-     * for the current state then we transition next_if_match
-     * state. If not, we transition to the next_if_not_match state.
-     *
-     * The final two states are special in that they are the states in
-     * which we discard data. */
-    static const struct {
-	int state;
-	int a;
-	int b;
-	int next_if_match;
-	int next_if_not_match;
-    } states[] = {
-	{0,  'b',  'b',  1,  0},
-	{1,  'e',  'e',  2,  0},
-	{2,  'g',  'g',  3,  0},
-	{3,  'i',  'i',  4,  0},
-	{4,  'n',  'n',  5,  0},
-	{5,  ' ',  ' ',  6,  0},
-	{6,  '0',  '7',  7,  0},
-	{7,  '0',  '7',  8,  0},
-	{8,  '0',  '7',  9,  0},
-	{9,  ' ',  ' ',  10, 0},
-	{10, '\n', '\n', 11, 10},
-	{11, 'M',  'M',  12, 0},
-	{12, ' ',  '`',  12, 11}
-    };
-    int next;
 
     g_mime_filter_set_size (gmime_filter, inlen, FALSE);
     outptr = gmime_filter->outbuf;
@@ -174,7 +155,7 @@ filter_filter_uuencode (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, si
 	    next = states[filter->state].next_if_not_match;
 	}
 
-	if (filter->state < 11)
+	if (filter->state < first_skipping_state)
 	    *outptr++ = *inptr;
 
 	filter->state = next;
@@ -187,6 +168,38 @@ filter_filter_uuencode (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, si
 }
 
 static void
+filter_filter_uuencode (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t prespace,
+			char **outbuf, size_t *outlen, size_t *outprespace)
+{
+    /* Simple, linear state-transition diagram for our filter.
+     *
+     * If the character being processed is within the range of [a, b]
+     * for the current state then we transition next_if_match
+     * state. If not, we transition to the next_if_not_match state.
+     *
+     * The final two states are special in that they are the states in
+     * which we discard data. */
+    static const scanner_state_t states[] = {
+	{0,  'b',  'b',  1,  0},
+	{1,  'e',  'e',  2,  0},
+	{2,  'g',  'g',  3,  0},
+	{3,  'i',  'i',  4,  0},
+	{4,  'n',  'n',  5,  0},
+	{5,  ' ',  ' ',  6,  0},
+	{6,  '0',  '7',  7,  0},
+	{7,  '0',  '7',  8,  0},
+	{8,  '0',  '7',  9,  0},
+	{9,  ' ',  ' ',  10, 0},
+	{10, '\n', '\n', 11, 10},
+	{11, 'M',  'M',  12, 0},
+	{12, ' ',  '`',  12, 11}
+    };
+
+    do_filter(states, 11,
+	      gmime_filter, inbuf, inlen, prespace, outbuf, outlen, outprespace);
+}
+
+static void
 filter_complete (GMimeFilter *filter, char *inbuf, size_t inlen, size_t prespace,
 		 char **outbuf, size_t *outlen, size_t *outprespace)
 {
-- 
2.11.0

[PATCH 4/6] lib/index: generalize filter name

[David Bremner]

We can't very well call it uuencode if it is going to filter other
things as well.
---
 lib/index.cc | 92 +++++++++++++++++++++++++++++++-----------------------------
 1 file changed, 48 insertions(+), 44 deletions(-)

diff --git a/lib/index.cc b/lib/index.cc
index 02b35b81..3bb1ac1c 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -26,8 +26,8 @@
 
 /* Oh, how I wish that gobject didn't require so much noisy boilerplate!
  * (Though I have at least eliminated some of the stock set...) */
-typedef struct _NotmuchFilterDiscardUuencode NotmuchFilterDiscardUuencode;
-typedef struct _NotmuchFilterDiscardUuencodeClass NotmuchFilterDiscardUuencodeClass;
+typedef struct _NotmuchFilterDiscardNonTerms NotmuchFilterDiscardNonTerms;
+typedef struct _NotmuchFilterDiscardNonTermsClass NotmuchFilterDiscardNonTermsClass;
 
 typedef void (*filter_fun) (GMimeFilter *filter, char *in, size_t len, size_t prespace,
 			    char **out, size_t *outlen, size_t *outprespace);
@@ -41,44 +41,29 @@ typedef struct {
 } scanner_state_t;
 
 /**
- * NotmuchFilterDiscardUuencode:
+ * NotmuchFilterDiscardNonTerms:
  *
  * @parent_object: parent #GMimeFilter
  * @encode: encoding vs decoding
  * @state: State of the parser
  *
- * A filter to discard uuencoded portions of an email.
- *
- * A uuencoded portion is identified as beginning with a line
- * matching:
- *
- *	begin [0-7][0-7][0-7] .*
- *
- * After that detection, and beginning with the following line,
- * characters will be discarded as long as the first character of each
- * line begins with M and subsequent characters on the line are within
- * the range of ASCII characters from ' ' to '`'.
- *
- * This is not a perfect UUencode filter. It's possible to have a
- * message that will legitimately match that pattern, (so that some
- * legitimate content is discarded). And for most UUencoded files, the
- * final line of encoded data (the line not starting with M) will be
- * indexed.
+ * A filter to discard non terms portions of an email, i.e. stuff not
+ * worth indexing.
  **/
-struct _NotmuchFilterDiscardUuencode {
+struct _NotmuchFilterDiscardNonTerms {
     GMimeFilter parent_object;
     GMimeContentType *content_type;
     filter_fun real_filter;
     int state;
 };
 
-struct _NotmuchFilterDiscardUuencodeClass {
+struct _NotmuchFilterDiscardNonTermsClass {
     GMimeFilterClass parent_class;
 };
 
-static GMimeFilter *notmuch_filter_discard_uuencode_new (GMimeContentType *content);
+static GMimeFilter *notmuch_filter_discard_non_terms_new (GMimeContentType *content);
 
-static void notmuch_filter_discard_uuencode_finalize (GObject *object);
+static void notmuch_filter_discard_non_terms_finalize (GObject *object);
 
 static GMimeFilter *filter_copy (GMimeFilter *filter);
 static void filter_filter (GMimeFilter *filter, char *in, size_t len, size_t prespace,
@@ -91,14 +76,14 @@ static void filter_reset (GMimeFilter *filter);
 static GMimeFilterClass *parent_class = NULL;
 
 static void
-notmuch_filter_discard_uuencode_class_init (NotmuchFilterDiscardUuencodeClass *klass)
+notmuch_filter_discard_non_terms_class_init (NotmuchFilterDiscardNonTermsClass *klass)
 {
     GObjectClass *object_class = G_OBJECT_CLASS (klass);
     GMimeFilterClass *filter_class = GMIME_FILTER_CLASS (klass);
 
     parent_class = (GMimeFilterClass *) g_type_class_ref (GMIME_TYPE_FILTER);
 
-    object_class->finalize = notmuch_filter_discard_uuencode_finalize;
+    object_class->finalize = notmuch_filter_discard_non_terms_finalize;
 
     filter_class->copy = filter_copy;
     filter_class->filter = filter_filter;
@@ -107,7 +92,7 @@ notmuch_filter_discard_uuencode_class_init (NotmuchFilterDiscardUuencodeClass *k
 }
 
 static void
-notmuch_filter_discard_uuencode_finalize (GObject *object)
+notmuch_filter_discard_non_terms_finalize (GObject *object)
 {
     G_OBJECT_CLASS (parent_class)->finalize (object);
 }
@@ -115,15 +100,15 @@ notmuch_filter_discard_uuencode_finalize (GObject *object)
 static GMimeFilter *
 filter_copy (GMimeFilter *gmime_filter)
 {
-    NotmuchFilterDiscardUuencode *filter = (NotmuchFilterDiscardUuencode *) gmime_filter;
+    NotmuchFilterDiscardNonTerms *filter = (NotmuchFilterDiscardNonTerms *) gmime_filter;
 
-    return notmuch_filter_discard_uuencode_new (filter->content_type);
+    return notmuch_filter_discard_non_terms_new (filter->content_type);
 }
 
 static void
 filter_filter (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t prespace,
 	       char **outbuf, size_t *outlen, size_t *outprespace) {
-    NotmuchFilterDiscardUuencode *filter = (NotmuchFilterDiscardUuencode *) gmime_filter;
+    NotmuchFilterDiscardNonTerms *filter = (NotmuchFilterDiscardNonTerms *) gmime_filter;
     (*filter->real_filter)(gmime_filter, inbuf, inlen, prespace, outbuf, outlen, outprespace);
 }
 
@@ -133,7 +118,7 @@ do_filter (const scanner_state_t states[],
 	   GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t prespace,
 	   char **outbuf, size_t *outlen, size_t *outprespace)
 {
-    NotmuchFilterDiscardUuencode *filter = (NotmuchFilterDiscardUuencode *) gmime_filter;
+    NotmuchFilterDiscardNonTerms *filter = (NotmuchFilterDiscardNonTerms *) gmime_filter;
     register const char *inptr = inbuf;
     const char *inend = inbuf + inlen;
     char *outptr;
@@ -167,6 +152,25 @@ do_filter (const scanner_state_t states[],
     *outbuf = gmime_filter->outbuf;
 }
 
+/*
+ *
+ * A uuencoded portion is identified as beginning with a line
+ * matching:
+ *
+ *	begin [0-7][0-7][0-7] .*
+ *
+ * After that detection, and beginning with the following line,
+ * characters will be discarded as long as the first character of each
+ * line begins with M and subsequent characters on the line are within
+ * the range of ASCII characters from ' ' to '`'.
+ *
+ * This is not a perfect UUencode filter. It's possible to have a
+ * message that will legitimately match that pattern, (so that some
+ * legitimate content is discarded). And for most UUencoded files, the
+ * final line of encoded data (the line not starting with M) will be
+ * indexed.
+ */
+
 static void
 filter_filter_uuencode (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t prespace,
 			char **outbuf, size_t *outlen, size_t *outprespace)
@@ -210,7 +214,7 @@ filter_complete (GMimeFilter *filter, char *inbuf, size_t inlen, size_t prespace
 static void
 filter_reset (GMimeFilter *gmime_filter)
 {
-    NotmuchFilterDiscardUuencode *filter = (NotmuchFilterDiscardUuencode *) gmime_filter;
+    NotmuchFilterDiscardNonTerms *filter = (NotmuchFilterDiscardNonTerms *) gmime_filter;
 
     filter->state = 0;
 }
@@ -218,32 +222,32 @@ filter_reset (GMimeFilter *gmime_filter)
 /**
  * notmuch_filter_discard_uuencode_new:
  *
- * Returns: a new #NotmuchFilterDiscardUuencode filter.
+ * Returns: a new #NotmuchFilterDiscardNonTerms filter.
  **/
 static GMimeFilter *
-notmuch_filter_discard_uuencode_new (GMimeContentType *content_type)
+notmuch_filter_discard_non_terms_new (GMimeContentType *content_type)
 {
     static GType type = 0;
-    NotmuchFilterDiscardUuencode *filter;
+    NotmuchFilterDiscardNonTerms *filter;
 
     if (!type) {
 	static const GTypeInfo info = {
-	    sizeof (NotmuchFilterDiscardUuencodeClass),
+	    sizeof (NotmuchFilterDiscardNonTermsClass),
 	    NULL, /* base_class_init */
 	    NULL, /* base_class_finalize */
-	    (GClassInitFunc) notmuch_filter_discard_uuencode_class_init,
+	    (GClassInitFunc) notmuch_filter_discard_non_terms_class_init,
 	    NULL, /* class_finalize */
 	    NULL, /* class_data */
-	    sizeof (NotmuchFilterDiscardUuencode),
+	    sizeof (NotmuchFilterDiscardNonTerms),
 	    0,    /* n_preallocs */
 	    NULL, /* instance_init */
 	    NULL  /* value_table */
 	};
 
-	type = g_type_register_static (GMIME_TYPE_FILTER, "NotmuchFilterDiscardUuencode", &info, (GTypeFlags) 0);
+	type = g_type_register_static (GMIME_TYPE_FILTER, "NotmuchFilterDiscardNonTerms", &info, (GTypeFlags) 0);
     }
 
-    filter = (NotmuchFilterDiscardUuencode *) g_object_newv (type, 0, NULL);
+    filter = (NotmuchFilterDiscardNonTerms *) g_object_newv (type, 0, NULL);
     filter->state = 0;
     filter->content_type = content_type;
     filter->real_filter = filter_filter_uuencode;
@@ -332,7 +336,7 @@ _index_mime_part (notmuch_message_t *message,
 		  GMimeObject *part)
 {
     GMimeStream *stream, *filter;
-    GMimeFilter *discard_uuencode_filter;
+    GMimeFilter *discard_non_terms_filter;
     GMimeDataWrapper *wrapper;
     GByteArray *byte_array;
     GMimeContentDisposition *disposition;
@@ -422,10 +426,10 @@ _index_mime_part (notmuch_message_t *message,
     g_mime_stream_mem_set_owner (GMIME_STREAM_MEM (stream), FALSE);
 
     filter = g_mime_stream_filter_new (stream);
-    discard_uuencode_filter = notmuch_filter_discard_uuencode_new (content_type);
+    discard_non_terms_filter = notmuch_filter_discard_non_terms_new (content_type);
 
     g_mime_stream_filter_add (GMIME_STREAM_FILTER (filter),
-			      discard_uuencode_filter);
+			      discard_non_terms_filter);
 
     charset = g_mime_object_get_content_type_parameter (part, "charset");
     if (charset) {
@@ -447,7 +451,7 @@ _index_mime_part (notmuch_message_t *message,
 
     g_object_unref (stream);
     g_object_unref (filter);
-    g_object_unref (discard_uuencode_filter);
+    g_object_unref (discard_non_terms_filter);
 
     g_byte_array_append (byte_array, (guint8 *) "\0", 1);
     body = (char *) g_byte_array_free (byte_array, FALSE);
-- 
2.11.0

[PATCH 5/6] lib/index.cc: generalize filter state machine

[David Bremner]

To match things more complicated than fixed strings, we need states
with multiple out arrows.
---
 lib/index.cc | 25 ++++++++++++++++---------
 1 file changed, 16 insertions(+), 9 deletions(-)

diff --git a/lib/index.cc b/lib/index.cc
index 3bb1ac1c..1b420b75 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -129,16 +129,23 @@ do_filter (const scanner_state_t states[],
     g_mime_filter_set_size (gmime_filter, inlen, FALSE);
     outptr = gmime_filter->outbuf;
 
+    next = filter->state;
     while (inptr < inend) {
-	if (*inptr >= states[filter->state].a &&
-	    *inptr <= states[filter->state].b)
-	{
-	    next = states[filter->state].next_if_match;
-	}
-	else
-	{
-	    next = states[filter->state].next_if_not_match;
-	}
+	 /* Each state is defined by a contiguous set of rows of the
+	 * state table marked by a common value for '.state'. The
+	 * state numbers must be equal to the index of the first row
+	 * in a given state; thus the loop condition here looks for a
+	 * jump to a first row of a state, which is a real transition
+	 * in the underlying DFA.
+	 */
+	do {
+	    if (*inptr >= states[next].a && *inptr <= states[next].b)  {
+		next = states[next].next_if_match;
+	    } else  {
+		next = states[next].next_if_not_match;
+	    }
+
+	} while (next != states[next].state);
 
 	if (filter->state < first_skipping_state)
 	    *outptr++ = *inptr;
-- 
2.11.0

[PATCH 6/6] lib/index: add simple html filter

[David Bremner]

Just drop all (HTML) tags
---
 lib/index.cc               | 38 +++++++++++++++++++++++++++++++++++++-
 test/T680-html-indexing.sh |  5 ++++-
 2 files changed, 41 insertions(+), 2 deletions(-)

diff --git a/lib/index.cc b/lib/index.cc
index 1b420b75..217b58c0 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -211,6 +211,39 @@ filter_filter_uuencode (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, si
 }
 
 static void
+filter_filter_html (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t prespace,
+		    char **outbuf, size_t *outlen, size_t *outprespace)
+{
+/* The following table is intended to implement this DFA (in 'dot'
+   format). Note that 2 and 3 are "hidden" states used to step through
+   the possible out edges of state 1.
+
+digraph html_filter {
+	0 -> 1  [label="<"];
+	0 -> 0;
+	1 -> 4 [label="'"];
+	1 -> 5 [label="\""];
+	1 -> 0 [label=">"];
+	1 -> 1;
+	4 -> 1 [label="'"];
+	4 -> 4;
+	5 -> 1 [label="\""];
+	5 -> 5;
+}
+*/
+    static const scanner_state_t states[] = {
+	{0,  '<',  '<',  1,  0},
+	{1,  '\'', '\'', 4,  2},  /* scanning for quote or > */
+	{1,  '"',  '"',  5,  3},
+	{1,  '>',  '>',  0,  1},
+	{4,  '\'', '\'', 1,  4},  /* inside single quotes */
+	{5,  '"', '"',   1,  5},  /* inside double quotes */
+    };
+    do_filter(states, 1,
+	      gmime_filter, inbuf, inlen, prespace, outbuf, outlen, outprespace);
+}
+
+static void
 filter_complete (GMimeFilter *filter, char *inbuf, size_t inlen, size_t prespace,
 		 char **outbuf, size_t *outlen, size_t *outprespace)
 {
@@ -257,7 +290,10 @@ notmuch_filter_discard_non_terms_new (GMimeContentType *content_type)
     filter = (NotmuchFilterDiscardNonTerms *) g_object_newv (type, 0, NULL);
     filter->state = 0;
     filter->content_type = content_type;
-    filter->real_filter = filter_filter_uuencode;
+    if (g_mime_content_type_is_type (content_type, "text", "html"))
+	filter->real_filter = filter_filter_html;
+    else
+	filter->real_filter = filter_filter_uuencode;
     return (GMimeFilter *) filter;
 }
 
diff --git a/test/T680-html-indexing.sh b/test/T680-html-indexing.sh
index 5e9cc4cb..74f33708 100755
--- a/test/T680-html-indexing.sh
+++ b/test/T680-html-indexing.sh
@@ -5,10 +5,13 @@ test_description="indexing of html parts"
 add_email_corpus html
 
 test_begin_subtest 'embedded images should not be indexed'
-test_subtest_known_broken
 notmuch search kwpza7svrgjzqwi8fhb2msggwtxtwgqcxp4wbqr4wjddstqmeqa7 > OUTPUT
 test_expect_equal_file /dev/null OUTPUT
 
+test_begin_subtest 'ignore > in attribute text'
+notmuch search swordfish | notmuch_search_sanitize > OUTPUT
+test_expect_equal_file /dev/null OUTPUT
+
 test_begin_subtest 'non tag text should be indexed'
 notmuch search hunter2 | notmuch_search_sanitize > OUTPUT
 cat <<EOF > EXPECTED
-- 
2.11.0