message/mime-object: cap references

Cap the references, since the threading algo is recursive.

lib/message/mu-mime-object.cc

@@ -26,6 +26,7 @@
 #include <mutex>
 #include <ranges>
 #include <regex>
+#include <unordered_set>
 #include <fcntl.h>
 #include <sys/stat.h>
 #include <errno.h>
@@ -476,16 +477,16 @@ MimeMessage::contacts(Contact::Type ctype) const noexcept
  * message-ids (in that order). Duplicates are removed.
  *
  * The _first_ one in the list determines the thread-id for the message.
+ *
+ * The number of references is capped (max_references), since the thread algo is
+ * recursive, and should still be plenty big.
  */
 std::vector<std::string>
 MimeMessage::references() const noexcept
 {
-	// is ref already in the list? O(n) but with small n.
+	constexpr size_t max_references = 256;
-	auto is_dup = [](auto&& seq, const std::string& ref) {
-		return std::ranges::any_of(seq, [&](auto&& str) { return ref == str; });
-	};
 
-	auto on_blacklist = [](auto&& msgid) {
+	const auto on_blacklist = [](const char* msgid) {
 		// don't include empty message-ids
 		if (!*msgid)
 			return true;
@@ -498,6 +499,7 @@ MimeMessage::references() const noexcept
 	};
 
 	std::vector<std::string> refs;
+	std::unordered_set<std::string> seen; // O(1) dup-check, keeps this O(n)
 	for (auto&& ref_header: { "References", "In-reply-to" }) {
 
 		auto hdr{header(ref_header)};
@@ -508,16 +510,28 @@ MimeMessage::references() const noexcept
 		if (!mime_refs)
 			continue; /* try the next header */
 
-		refs.reserve(refs.size() + g_mime_references_length(mime_refs));
+		const auto n_refs{g_mime_references_length(mime_refs)};
+		refs.reserve(refs.size() + n_refs);
 
-		for (auto i = 0; i != g_mime_references_length(mime_refs); ++i) {
+		for (auto i = 0; i != n_refs; ++i) {
 			const auto msgid{g_mime_references_get_message_id(mime_refs, i)};
-			if (msgid && !is_dup(refs, msgid) && !on_blacklist(msgid))
+			if (msgid && !on_blacklist(msgid) && seen.emplace(msgid).second)
 				refs.emplace_back(msgid);
 		}
 		g_mime_references_free(mime_refs);
 	}
 
+	// cap hostile/pathological headers: keep the first reference (thread-id)
+	// plus the most-recent ancestors (the tail).
+	if (refs.size() > max_references) {
+		std::vector<std::string> capped;
+		capped.reserve(max_references);
+		capped.emplace_back(std::move(refs.front()));
+		for (auto it = refs.end() - (max_references - 1); it != refs.end(); ++it)
+			capped.emplace_back(std::move(*it));
+		refs = std::move(capped);
+	}
+
 	return refs;
 }
 

lib/message/test-mu-message.cc

@@ -866,6 +866,44 @@ On Thu, Aug 04, 2022 at 05:31:39PM +0100, Robin Murphy wrote:
 }
 
 
+static void
+test_message_references_capped()
+{
+	// a too-big References header must not produce an
+	// unbounded reference list (see MimeMessage::references): keep
+	// first (thread-id) + most-recent ancestors.
+	constexpr size_t n_refs = 1000;
+	constexpr size_t cap	= 256; // must match max_references
+
+	std::string hdr{"References:"};
+	for (size_t i = 0; i != n_refs; ++i)
+		hdr += mu_format(" <r{:04}@x>", i);
+
+	const auto msgtext =
+		hdr + "\n" +
+		"To: \"Robin Murphy\" <robin.murphy@arm.com>\n"
+		"From: \"Dan Carpenter\" <dan.carpenter@oracle.com>\n"
+		"Subject: capped\n"
+		"Date: Fri, 5 Aug 2022 09:37:02 +0300\n"
+		"Message-Id: <capped@kadam>\n"
+		"\n"
+		"body\n";
+
+	auto message{Message::make_from_text(
+			msgtext,
+			"/home/test/Maildir/inbox/cur/162342449279256.88888_1.evergrey:2,S")};
+	g_assert_true(!!message);
+
+	const auto refs{message->references()};
+	g_assert_cmpuint(refs.size(), ==, cap);
+	// thread-id (first) preserved, and the immediate parent (last) kept.
+	assert_equal(refs.front(), "r0000@x");
+	assert_equal(refs.back(), mu_format("r{:04}@x", n_refs - 1));
+	// the middle is dropped: second entry is the start of the kept tail.
+	assert_equal(refs.at(1), mu_format("r{:04}@x", n_refs - (cap - 1)));
+}
+
+
 static void
 test_message_outlook_body()
 {
@@ -1109,6 +1147,8 @@ main(int argc, char* argv[])
 			test_message_calendar);
 	g_test_add_func("/message/message/references",
 			test_message_references);
+	g_test_add_func("/message/message/references-capped",
+			test_message_references_capped);
 	g_test_add_func("/message/message/outlook-body",
 			test_message_outlook_body);
 	g_test_add_func("/message/message/message-id",