* mu-query.cc, mu-store.cc: use the escaping for some fields; update test case

3 files changed, 37 insertions, 109 deletions

diff --git a/src/mu-query.cc b/src/mu-query.cc
index 08a97af..0640ff1 100644
--- a/src/mu-query.cc
+++ b/src/mu-query.cc
@@ -288,97 +288,22 @@ mu_query_destroy (MuQuery *self)
 	g_free (self);
 }
 
-struct _CheckPrefix {
-	const char		*pfx;
-	guint			 len;
-	gboolean		 match;
-};
-typedef struct _CheckPrefix	 CheckPrefix;
-
-static void
-each_check_prefix (MuMsgFieldId mfid, CheckPrefix *cpfx)
-{
-	const char *field_name;
-	char field_shortcut;
-
-	if (!cpfx || cpfx->match)
-		return;
-	
-	field_shortcut = mu_msg_field_shortcut (mfid);
-	if (field_shortcut == cpfx->pfx[0]) {
-		cpfx->match = TRUE;
-		return;
-	}
-
-	field_name = mu_msg_field_name (mfid);
-	if (field_name &&
-	    strncmp (cpfx->pfx, field_name, cpfx->len) == 0) {
-		cpfx->match = TRUE;
-		return;
-	}
-}
-
 
-/* colon is a position inside q pointing at a ':' character. function
- * determines whether the prefix is a registered prefix (like
- * 'subject' or 'from' or 's') */
-static gboolean
-is_xapian_prefix (const char *q, const char *colon)
-{
-	const char *cur;
-	
-	if (colon == q)
-		return FALSE; /* : at beginning, not a prefix */
-	
-	/* track back from colon until a boundary or beginning of the
-	 * str */
-	for (cur = colon - 1; cur >= q; --cur) {
-
-		if (cur == q || !isalpha (*(cur-1))) {
-
-			CheckPrefix cpfx;
-			memset (&cpfx, 0, sizeof(CheckPrefix));
-
-			cpfx.pfx   = cur;
-			cpfx.len   = (colon - cur);
-			cpfx.match = FALSE;
-			
-			mu_msg_field_foreach ((MuMsgFieldForEachFunc)
-					      each_check_prefix,
-					      &cpfx);
-			
-			return (cpfx.match);
-		}
-	}
-	
-	return FALSE;
-}
-	
 /* preprocess a query to make them a bit more permissive */
 char*
 mu_query_preprocess (const char *query)
 {
 	gchar *my_query;
-	gchar *cur;
 
 	g_return_val_if_fail (query, NULL);
+	my_query = g_strdup (query);
 	
-	/* translate the the searchexpr to all lowercase; this
-	 * will fixes some of the false-negatives. A full fix
-	 * probably requires some custom query parser.
-	 */
-	my_query = mu_str_normalize(query, TRUE);
-	
-	for (cur = my_query; *cur; ++cur) {
-		if (*cur == ':') /* we found a ':' */
-			 /* if there's a registered xapian prefix before the
-			  * ':', don't touch it. Otherwise replace ':' with
-			  * a space'... ugly...
-			  */			 
-			if (!is_xapian_prefix (my_query, cur))
-				*cur = ' ';
-	}
-	
+	/* remove accents and turn to lower-case */
+	mu_str_normalize_in_place (my_query, TRUE);
+	/* escape '@', single '_' and ':' if it's not following a
+	 * xapian-pfx with '_' */
+	mu_str_ascii_xapian_escape_in_place (my_query);
+		
 	return my_query;
 }
 
diff --git a/src/mu-store.cc b/src/mu-store.cc
index 1d56975..376022d 100644
--- a/src/mu-store.cc
+++ b/src/mu-store.cc
@@ -314,36 +314,36 @@ static void
 add_terms_values_string (Xapian::Document& doc, MuMsg *msg,
 			 MuMsgFieldId mfid)
 {
-	const char* str;
-	
-	str = mu_msg_get_field_string (msg, mfid);
-	if (!str)
+	const char *orig;
+	char *val;
+		
+	orig = mu_msg_get_field_string (msg, mfid);
+	if (!orig)
 		return;
-
-	const std::string value  (str);
+	val = g_strdup (orig);
+		
 	const std::string prefix (1, mu_msg_field_xapian_prefix(mfid));
+
+	/* the value is what we'll display; the unchanged original */
+	if (mu_msg_field_xapian_value(mfid)) 			 
+		doc.add_value ((Xapian::valueno)mfid, val);
+
+	/* now, let's create some search terms... */
+	if (mu_msg_field_normalize (mfid))
+		mu_str_normalize_in_place (val, TRUE);
+	if (mu_msg_field_xapian_escape (mfid))
+		mu_str_ascii_xapian_escape_in_place (val);
 	
 	if (mu_msg_field_xapian_index (mfid)) {
 		Xapian::TermGenerator termgen;
-		gchar *norm (mu_str_normalize(str, TRUE));
 		termgen.set_document (doc);
-		termgen.index_text_without_positions (norm, 1, prefix);
-		g_free(norm);
-	}
-
-	if (mu_msg_field_xapian_term(mfid)) {
-		/* add a normalized version (accents removed,
-		 * lowercase) */
-		gchar *norm =  mu_str_normalize(str, TRUE);
-		doc.add_term (std::string (prefix + std::string(norm), 0,
-					   MU_STORE_MAX_TERM_LENGTH));
-		g_free (norm);
+		termgen.index_text_without_positions (val, 1, prefix);
 	}
-
-	/* the value is what we'll display; the unchanged original */
-	if (mu_msg_field_xapian_value(mfid)) 			 
-		doc.add_value ((Xapian::valueno)mfid,
-			       value);
+	
+	if (mu_msg_field_xapian_term(mfid))
+		doc.add_term (prefix + std::string(val, 0, MU_STORE_MAX_TERM_LENGTH));
+	
+	g_free (val);
 }
 
 static void
@@ -443,6 +443,9 @@ each_contact_info (MuMsgContact *contact, MsgDoc *data)
 	/* don't normalize e-mail address, but do lowercase it */
 	if (contact->address && strlen (contact->address)) {
 		char *lower = g_utf8_strdown (contact->address, -1);
+
+		g_strdelimit (lower, "@.", '_'); /* FIXME */
+
 		data->_doc->add_term
 			(std::string (*pfxp + lower, 0,
 				      MU_STORE_MAX_TERM_LENGTH));
diff --git a/src/tests/test-mu-query.c b/src/tests/test-mu-query.c
index a72e301..9b15423 100644
--- a/src/tests/test-mu-query.c
+++ b/src/tests/test-mu-query.c
@@ -66,14 +66,14 @@ run_and_count_matches (const char *xpath, const char *query)
 	
 	mquery = mu_query_new (xpath, NULL);
 	g_assert (query);
+
+	/* g_printerr ("\n=>'%s'\n", query); */
 	
 	iter = mu_query_run (mquery, query, MU_MSG_FIELD_ID_NONE,
 			     FALSE, 1, NULL);
 	mu_query_destroy (mquery);
 	g_assert (iter);
 
-	/* g_printerr ("\n=> %s\n", query); */
-	
 	for (count = 0; !mu_msg_iter_is_done(iter);
 	     mu_msg_iter_next(iter), ++count);
 	
@@ -179,10 +179,10 @@ test_mu_query_04 (void)
 	int i;
 	
 	QResults queries[] = {
-//     	{ "frodo@example.com", 1}, /* does not match: see mu-find (1) */
+		{ "frodo@example.com", 1}, /* does not match: see mu-find (1) */
 		{ "f:frodo@example.com", 1},
 		{ "f:Frodo Baggins", 1},
-//		{ "bilbo@anotherexample.com", 1}, /* same things */
+		{ "bilbo@anotherexample.com", 1}, /* same things */
 		{ "t:bilbo@anotherexample.com", 1},
 		{ "t:bilbo", 1},
 		{ "f:bilbo", 0},