v3 of drop HTML tags when indexing

v3 of drop HTML tags when indexing

[David Bremner]

this obsoletes id:20170510113910.28444-1-david@tethera.net

This is a simplification of the previous version, rather than swapping
function pointers, we swap scanner tables. Unforunately because of the
reorganization, the interdiff is unhelpfully large.

The series shrunk from 6 to 4 patches; that's partly a cheat as the
old 2/6 is now part of 4/4.

[patch v3 1/4] lib/index: generalize name of indexing filter

[David Bremner]

In followup commits we will generalize the functionality of this
filter to deal with other types of non-indexable content.
---
 lib/index.cc | 52 ++++++++++++++++++++++++++--------------------------
 1 file changed, 26 insertions(+), 26 deletions(-)

diff --git a/lib/index.cc b/lib/index.cc
index 8c145540..d9612e03 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -26,11 +26,11 @@
 
 /* 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 _NotmuchFilterDiscardNonTerm NotmuchFilterDiscardNonTerm;
+typedef struct _NotmuchFilterDiscardNonTermClass NotmuchFilterDiscardNonTermClass;
 
 /**
- * NotmuchFilterDiscardUuencode:
+ * NotmuchFilterDiscardNonTerm:
  *
  * @parent_object: parent #GMimeFilter
  * @encode: encoding vs decoding
@@ -54,18 +54,18 @@ typedef struct _NotmuchFilterDiscardUuencodeClass NotmuchFilterDiscardUuencodeCl
  * final line of encoded data (the line not starting with M) will be
  * indexed.
  **/
-struct _NotmuchFilterDiscardUuencode {
+struct _NotmuchFilterDiscardNonTerm {
     GMimeFilter parent_object;
     int state;
 };
 
-struct _NotmuchFilterDiscardUuencodeClass {
+struct _NotmuchFilterDiscardNonTermClass {
     GMimeFilterClass parent_class;
 };
 
-static GMimeFilter *notmuch_filter_discard_uuencode_new (void);
+static GMimeFilter *notmuch_filter_discard_non_term_new (void);
 
-static void notmuch_filter_discard_uuencode_finalize (GObject *object);
+static void notmuch_filter_discard_non_term_finalize (GObject *object);
 
 static GMimeFilter *filter_copy (GMimeFilter *filter);
 static void filter_filter (GMimeFilter *filter, char *in, size_t len, size_t prespace,
@@ -78,14 +78,14 @@ 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_term_class_init (NotmuchFilterDiscardNonTermClass *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_term_finalize;
 
     filter_class->copy = filter_copy;
     filter_class->filter = filter_filter;
@@ -94,7 +94,7 @@ notmuch_filter_discard_uuencode_class_init (NotmuchFilterDiscardUuencodeClass *k
 }
 
 static void
-notmuch_filter_discard_uuencode_finalize (GObject *object)
+notmuch_filter_discard_non_term_finalize (GObject *object)
 {
     G_OBJECT_CLASS (parent_class)->finalize (object);
 }
@@ -103,14 +103,14 @@ static GMimeFilter *
 filter_copy (GMimeFilter *gmime_filter)
 {
     (void) gmime_filter;
-    return notmuch_filter_discard_uuencode_new ();
+    return notmuch_filter_discard_non_term_new ();
 }
 
 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;
+    NotmuchFilterDiscardNonTerm *filter = (NotmuchFilterDiscardNonTerm *) gmime_filter;
     register const char *inptr = inbuf;
     const char *inend = inbuf + inlen;
     char *outptr;
@@ -185,40 +185,40 @@ filter_complete (GMimeFilter *filter, char *inbuf, size_t inlen, size_t prespace
 static void
 filter_reset (GMimeFilter *gmime_filter)
 {
-    NotmuchFilterDiscardUuencode *filter = (NotmuchFilterDiscardUuencode *) gmime_filter;
+    NotmuchFilterDiscardNonTerm *filter = (NotmuchFilterDiscardNonTerm *) gmime_filter;
 
     filter->state = 0;
 }
 
 /**
- * notmuch_filter_discard_uuencode_new:
+ * notmuch_filter_discard_non_term_new:
  *
- * Returns: a new #NotmuchFilterDiscardUuencode filter.
+ * Returns: a new #NotmuchFilterDiscardNonTerm filter.
  **/
 static GMimeFilter *
-notmuch_filter_discard_uuencode_new (void)
+notmuch_filter_discard_non_term_new (void)
 {
     static GType type = 0;
-    NotmuchFilterDiscardUuencode *filter;
+    NotmuchFilterDiscardNonTerm *filter;
 
     if (!type) {
 	static const GTypeInfo info = {
-	    sizeof (NotmuchFilterDiscardUuencodeClass),
+	    sizeof (NotmuchFilterDiscardNonTermClass),
 	    NULL, /* base_class_init */
 	    NULL, /* base_class_finalize */
-	    (GClassInitFunc) notmuch_filter_discard_uuencode_class_init,
+	    (GClassInitFunc) notmuch_filter_discard_non_term_class_init,
 	    NULL, /* class_finalize */
 	    NULL, /* class_data */
-	    sizeof (NotmuchFilterDiscardUuencode),
+	    sizeof (NotmuchFilterDiscardNonTerm),
 	    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, "NotmuchFilterDiscardNonTerm", &info, (GTypeFlags) 0);
     }
 
-    filter = (NotmuchFilterDiscardUuencode *) g_object_newv (type, 0, NULL);
+    filter = (NotmuchFilterDiscardNonTerm *) g_object_newv (type, 0, NULL);
     filter->state = 0;
 
     return (GMimeFilter *) filter;
@@ -306,7 +306,7 @@ _index_mime_part (notmuch_message_t *message,
 		  GMimeObject *part)
 {
     GMimeStream *stream, *filter;
-    GMimeFilter *discard_uuencode_filter;
+    GMimeFilter *discard_non_term_filter;
     GMimeDataWrapper *wrapper;
     GByteArray *byte_array;
     GMimeContentDisposition *disposition;
@@ -396,10 +396,10 @@ _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 ();
+    discard_non_term_filter = notmuch_filter_discard_non_term_new ();
 
     g_mime_stream_filter_add (GMIME_STREAM_FILTER (filter),
-			      discard_uuencode_filter);
+			      discard_non_term_filter);
 
     charset = g_mime_object_get_content_type_parameter (part, "charset");
     if (charset) {
@@ -421,7 +421,7 @@ _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_term_filter);
 
     g_byte_array_append (byte_array, (guint8 *) "\0", 1);
     body = (char *) g_byte_array_free (byte_array, FALSE);
-- 
2.11.0

[patch v3 2/4] lib/index: separate state table definition from scanner.

[David Bremner]

We want to reuse the scanner definition with a different table.  This
is mainly code movement, and making the state table part of the filter
struct/class.
---
 lib/index.cc | 70 ++++++++++++++++++++++++++++++++++--------------------------
 1 file changed, 40 insertions(+), 30 deletions(-)

diff --git a/lib/index.cc b/lib/index.cc
index d9612e03..19ddc39c 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -24,6 +24,40 @@
 
 #include <xapian.h>
 
+
+typedef struct {
+    int state;
+    int a;
+    int b;
+    int next_if_match;
+    int next_if_not_match;
+} scanner_state_t;
+
+/* Simple, linear state-transition diagram for the uuencode 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 int first_uuencode_skipping_state = 11;
+static const scanner_state_t uuencode_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}
+};
+
 /* 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 _NotmuchFilterDiscardNonTerm NotmuchFilterDiscardNonTerm;
@@ -57,6 +91,8 @@ typedef struct _NotmuchFilterDiscardNonTermClass NotmuchFilterDiscardNonTermClas
 struct _NotmuchFilterDiscardNonTerm {
     GMimeFilter parent_object;
     int state;
+    int first_skipping_state;
+    const scanner_state_t *states;
 };
 
 struct _NotmuchFilterDiscardNonTermClass {
@@ -111,41 +147,13 @@ filter_filter (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t pres
 	       char **outbuf, size_t *outlen, size_t *outprespace)
 {
     NotmuchFilterDiscardNonTerm *filter = (NotmuchFilterDiscardNonTerm *) gmime_filter;
+    const scanner_state_t *states = filter->states;
     register const char *inptr = inbuf;
     const char *inend = inbuf + inlen;
     char *outptr;
 
     (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);
@@ -162,7 +170,7 @@ filter_filter (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t pres
 	    next = states[filter->state].next_if_not_match;
 	}
 
-	if (filter->state < 11)
+	if (filter->state < filter->first_skipping_state)
 	    *outptr++ = *inptr;
 
 	filter->state = next;
@@ -220,6 +228,8 @@ notmuch_filter_discard_non_term_new (void)
 
     filter = (NotmuchFilterDiscardNonTerm *) g_object_newv (type, 0, NULL);
     filter->state = 0;
+    filter->states = uuencode_states;
+    filter->first_skipping_state = first_uuencode_skipping_state;
 
     return (GMimeFilter *) filter;
 }
-- 
2.11.0

[patch v3 3/4] 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 19ddc39c..8a18abf4 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -159,16 +159,23 @@ filter_filter (GMimeFilter *gmime_filter, char *inbuf, size_t inlen, size_t pres
     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 < filter->first_skipping_state)
 	    *outptr++ = *inptr;
-- 
2.11.0

[patch v3 4/4] lib/index: add simple html filter

[David Bremner]

The filter just drops all (HTML) tags. As an enabling change, pass the
content type to the filter constructor so we can decide which scanner
to user.
---
 lib/index.cc               | 48 +++++++++++++++++++++++++++++++++++++++-------
 test/T680-html-indexing.sh |  5 ++++-
 2 files changed, 45 insertions(+), 8 deletions(-)

diff --git a/lib/index.cc b/lib/index.cc
index 8a18abf4..0c4e2329 100644
--- a/lib/index.cc
+++ b/lib/index.cc
@@ -58,6 +58,33 @@ static const scanner_state_t uuencode_states[] = {
     {12, ' ',  '`',  12, 11}
 };
 
+/* 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 int first_html_skipping_state = 1;
+static const scanner_state_t html_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 */
+};
+
 /* 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 _NotmuchFilterDiscardNonTerm NotmuchFilterDiscardNonTerm;
@@ -90,6 +117,7 @@ typedef struct _NotmuchFilterDiscardNonTermClass NotmuchFilterDiscardNonTermClas
  **/
 struct _NotmuchFilterDiscardNonTerm {
     GMimeFilter parent_object;
+    GMimeContentType *content_type;
     int state;
     int first_skipping_state;
     const scanner_state_t *states;
@@ -99,7 +127,7 @@ struct _NotmuchFilterDiscardNonTermClass {
     GMimeFilterClass parent_class;
 };
 
-static GMimeFilter *notmuch_filter_discard_non_term_new (void);
+static GMimeFilter *notmuch_filter_discard_non_term_new (GMimeContentType *content);
 
 static void notmuch_filter_discard_non_term_finalize (GObject *object);
 
@@ -138,8 +166,8 @@ notmuch_filter_discard_non_term_finalize (GObject *object)
 static GMimeFilter *
 filter_copy (GMimeFilter *gmime_filter)
 {
-    (void) gmime_filter;
-    return notmuch_filter_discard_non_term_new ();
+    NotmuchFilterDiscardNonTerm *filter = (NotmuchFilterDiscardNonTerm *) gmime_filter;
+    return notmuch_filter_discard_non_term_new (filter->content_type);
 }
 
 static void
@@ -211,7 +239,7 @@ filter_reset (GMimeFilter *gmime_filter)
  * Returns: a new #NotmuchFilterDiscardNonTerm filter.
  **/
 static GMimeFilter *
-notmuch_filter_discard_non_term_new (void)
+notmuch_filter_discard_non_term_new (GMimeContentType *content_type)
 {
     static GType type = 0;
     NotmuchFilterDiscardNonTerm *filter;
@@ -234,9 +262,15 @@ notmuch_filter_discard_non_term_new (void)
     }
 
     filter = (NotmuchFilterDiscardNonTerm *) g_object_newv (type, 0, NULL);
+    filter->content_type = content_type;
     filter->state = 0;
-    filter->states = uuencode_states;
-    filter->first_skipping_state = first_uuencode_skipping_state;
+    if (g_mime_content_type_is_type (content_type, "text", "html")) {
+      filter->states = html_states;
+      filter->first_skipping_state = first_html_skipping_state;
+    } else {
+      filter->states = uuencode_states;
+      filter->first_skipping_state = first_uuencode_skipping_state;
+    }
 
     return (GMimeFilter *) filter;
 }
@@ -413,7 +447,7 @@ _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_non_term_filter = notmuch_filter_discard_non_term_new ();
+    discard_non_term_filter = notmuch_filter_discard_non_term_new (content_type);
 
     g_mime_stream_filter_add (GMIME_STREAM_FILTER (filter),
 			      discard_non_term_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

Re: v3 of drop HTML tags when indexing

[David Bremner]

David Bremner <david@tethera.net> writes:

> this obsoletes id:20170510113910.28444-1-david@tethera.net
>
> This is a simplification of the previous version, rather than swapping
> function pointers, we swap scanner tables. Unforunately because of the
> reorganization, the interdiff is unhelpfully large.
>
> The series shrunk from 6 to 4 patches; that's partly a cheat as the
> old 2/6 is now part of 4/4.

Pushed this series to master

d