From c415ad0b6771992e66c70edf373566c6e247089d Mon Sep 17 00:00:00 2001

From: Patrick Griffis <pgriffis@igalia.com>

Date: Tue, 18 Feb 2025 14:29:50 -0600

Subject: [PATCH] sniffer: Add better coverage of skip_insignificant_space()



Conflict: context adaptation and Modify file path adaptation: libsoup/content-sniffer/soup-content-sniffer.c->libsoup/soup-content-sniffer.c

Reference: https://gitlab.gnome.org/GNOME/libsoup/-/commit/c415ad0b6771992e66c70edf373566c6e247089d

---
 libsoup/soup-content-sniffer.c  |  10 +++---
 tests/resources/whitespace.html | Bin 512 -> 0 bytes
 tests/sniffing-test.c           |  53 +++++++++++++++++++++++++++++---
 tests/soup-tests.gresource.xml  |   1 -
 4 files changed, 53 insertions(+), 11 deletions(-)
 delete mode 100644 tests/resources/whitespace.html

diff --git a/libsoup/soup-content-sniffer.c b/libsoup/soup-content-sniffer.c
index c52d2d0..3fb29ad 100644
--- a/libsoup/soup-content-sniffer.c
+++ b/libsoup/soup-content-sniffer.c
@@ -612,8 +612,11 @@ sniff_text_or_binary (SoupContentSniffer *sniffer, SoupBuffer *buffer)
 }
 
 static gboolean
-skip_insignificant_space (const char *resource, int *pos, int resource_length)
+skip_insignificant_space (const char *resource, gsize *pos, gsize resource_length)
 {
+        if (*pos >= resource_length)
+	        return TRUE;
+
 	while ((resource[*pos] == '\x09') ||
 	       (resource[*pos] == '\x20') ||
 	       (resource[*pos] == '\x0A') ||
@@ -632,7 +635,7 @@ sniff_feed_or_html (SoupContentSniffer *sniffer, SoupBuffer *buffer)
 {
 	const char *resource = (const char *)buffer->data;
 	int resource_length = MIN (512, buffer->length);
-	int pos = 0;
+	gsize pos = 0;
 
 	if (resource_length < 3)
 		goto text_html;
@@ -642,9 +645,6 @@ sniff_feed_or_html (SoupContentSniffer *sniffer, SoupBuffer *buffer)
 		pos = 3;
 
  look_for_tag:
-	if (pos >= resource_length)
-		goto text_html;
-
 	if (skip_insignificant_space (resource, &pos, resource_length))
 		goto text_html;
 
diff --git a/tests/resources/whitespace.html b/tests/resources/whitespace.html
deleted file mode 100644
index 7f07a0e639a102284d6f7c0c5d5560170f994553..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 512
TcmcCf)YWAe1tT*Abam|ja4Z2(

diff --git a/tests/sniffing-test.c b/tests/sniffing-test.c
index 0a4569a..e2dfed4 100644
--- a/tests/sniffing-test.c
+++ b/tests/sniffing-test.c
@@ -436,6 +436,52 @@ test_disabled (gconstpointer data)
 	soup_uri_free (uri);
 }
 
+static const gsize MARKUP_LENGTH = strlen ("<!--") + strlen ("-->");
+
+static void
+do_skip_whitespace_test (void)
+{
+        SoupContentSniffer *sniffer = soup_content_sniffer_new ();
+        SoupMessage *msg = soup_message_new (SOUP_METHOD_GET, "http://example.org");
+        const char *test_cases[] = {
+                "",
+                "<rdf:RDF",
+                "<rdf:RDFxmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\"",
+                "<rdf:RDFxmlns=\"http://purl.org/rss/1.0/\"",
+        };
+
+        soup_message_headers_set_content_type (msg->response_headers, "text/html", NULL);
+
+        for (guint i = 0; i < G_N_ELEMENTS (test_cases); i++) {
+                const char *trailing_data = test_cases[i];
+                gsize leading_zeros = 512 - MARKUP_LENGTH - strlen (trailing_data);
+                gsize testsize = MARKUP_LENGTH + leading_zeros + strlen (trailing_data);
+                guint8 *data = g_malloc0 (testsize);
+                guint8 *p = data;
+                char *content_type;
+                GBytes *buffer;
+
+                // Format of <!--[0x00 * $leading_zeros]-->$trailing_data
+                memcpy (p, "<!--", strlen ("<!--"));
+                p += strlen ("<!--");
+                p += leading_zeros;
+                memcpy (p, "-->", strlen ("-->"));
+                p += strlen ("-->");
+                if (strlen (trailing_data))
+                        memcpy (p, trailing_data, strlen (trailing_data));
+                // Purposefully not NUL terminated.
+
+                buffer = g_bytes_new_take (g_steal_pointer (&data), testsize);
+                content_type = soup_content_sniffer_sniff (sniffer, msg, buffer, NULL);
+
+                g_free (content_type);
+                g_bytes_unref (buffer);
+        }
+
+        g_object_unref (msg);
+        g_object_unref (sniffer);
+}
+
 int
 main (int argc, char **argv)
 {
@@ -605,16 +651,13 @@ main (int argc, char **argv)
 			      "type/text_html; charset=UTF-8/test.html => text/html; charset=UTF-8",
 			      do_sniffing_test);
 
-        /* Test hitting skip_insignificant_space() with number of bytes equaling resource_length. */
-	g_test_add_data_func ("/sniffing/whitespace",
-			      "type/text_html/whitespace.html => text/html",
-			      do_sniffing_test);
-
 	/* Test that disabling the sniffer works correctly */
 	g_test_add_data_func ("/sniffing/disabled",
 			      "/text_or_binary/home.gif",
 			      test_disabled);
 
+	g_test_add_func ("/sniffing/whitespace", do_skip_whitespace_test);
+
 	ret = g_test_run ();
 
 	soup_uri_free (base_uri);
diff --git a/tests/soup-tests.gresource.xml b/tests/soup-tests.gresource.xml
index cbef1d4..9c08d17 100644
--- a/tests/soup-tests.gresource.xml
+++ b/tests/soup-tests.gresource.xml
@@ -25,6 +25,5 @@
     <file>resources/text.txt</file>
     <file>resources/text_binary.txt</file>
     <file>resources/tux.webp</file>
-    <file>resources/whitespace.html</file>
   </gresource>
 </gresources>
-- 
2.46.0