From 72bcabf9e138f1e90dc80507a991c2c68270145d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Rimas=20Misevi=C4=8Dius?= <rmisev3@gmail.com>
Date: Wed, 11 Oct 2023 00:13:06 +0300
Subject: [PATCH] Add tests for the IDNToUnicode function

---
 src/url_idna.cpp       | 24 ++++++++++++++++++++----
 test/test-url_host.cpp | 42 +++++++++++++++++++++++++++++++++++-------
 2 files changed, 55 insertions(+), 11 deletions(-)

diff --git a/src/url_idna.cpp b/src/url_idna.cpp
index cc16ec8..5d711f6 100644
--- a/src/url_idna.cpp
+++ b/src/url_idna.cpp
@@ -35,7 +35,7 @@ const UIDNA* getUIDNA() {
             UErrorCode err = U_ZERO_ERROR;
             // https://url.spec.whatwg.org/#idna
             // UseSTD3ASCIIRules = false
-            // Nontransitional_Processing
+            // Transitional_Processing = false
             // CheckBidi = true
             // CheckJoiners = true
             uidna_ptr = uidna_openUTS46(
@@ -93,7 +93,7 @@ namespace icu {
 validation_errc IDNToASCII(const char16_t* src, std::size_t src_len, simple_buffer<char16_t>& output) {
     // https://url.spec.whatwg.org/#concept-domain-to-ascii
     // https://www.unicode.org/reports/tr46/#ToASCII
-    static const uint32_t UIDNA_ERR_MASK = ~static_cast<uint32_t>(
+    static constexpr uint32_t UIDNA_ERR_MASK = ~static_cast<uint32_t>(
         // VerifyDnsLength = false
         UIDNA_ERROR_EMPTY_LABEL
         | UIDNA_ERROR_LABEL_TOO_LONG
@@ -152,6 +152,21 @@ validation_errc IDNToASCII(const char16_t* src, std::size_t src_len, simple_buff
 // TODO: common function template for IDNToASCII and IDNToUnicode
 
 validation_errc IDNToUnicode(const char* src, std::size_t src_len, simple_buffer<char>& output) {
+#if 0
+    // https://url.spec.whatwg.org/#concept-domain-to-unicode
+    // https://www.unicode.org/reports/tr46/#ToUnicode
+    static constexpr uint32_t UIDNA_ERR_MASK = ~static_cast<uint32_t>(
+        // VerifyDnsLength = false
+        UIDNA_ERROR_EMPTY_LABEL
+        | UIDNA_ERROR_LABEL_TOO_LONG
+        | UIDNA_ERROR_DOMAIN_NAME_TOO_LONG
+        // CheckHyphens = false
+        | UIDNA_ERROR_LEADING_HYPHEN
+        | UIDNA_ERROR_TRAILING_HYPHEN
+        | UIDNA_ERROR_HYPHEN_3_4
+        );
+#endif
+
     // uidna_nameToUnicodeUTF8 uses int32_t length
     // https://unicode-org.github.io/icu-docs/apidoc/dev/icu4c/uidna_8h.html#afd9ae1e0ae5318e20c87bcb0149c3ada
     if (src_len > util::unsigned_limit<int32_t>::max())
@@ -171,11 +186,12 @@ validation_errc IDNToUnicode(const char* src, std::size_t src_len, simple_buffer
             &info, &err);
         if (U_SUCCESS(err)) {
             output.resize(output_length);
+            // https://url.spec.whatwg.org/#concept-domain-to-unicode
+            // TODO: Signify domain-to-Unicode validation errors for any returned errors (i.e.
+            // if (info.errors & UIDNA_ERR_MASK) != 0), and then, return result.
             return validation_errc::ok;
         }
 
-        // https://url.spec.whatwg.org/#concept-domain-to-unicode
-        // TODO: Signify domain-to-Unicode validation errors for any returned errors, and then, return result.
         if (err != U_BUFFER_OVERFLOW_ERROR)
             return validation_errc::domain_to_unicode;
 
diff --git a/test/test-url_host.cpp b/test/test-url_host.cpp
index 2bd809d..b0bb120 100644
--- a/test/test-url_host.cpp
+++ b/test/test-url_host.cpp
@@ -24,7 +24,16 @@ class host_out : public upa::host_output {
     }
 };
 
-TEST_SUITE("host_parser::parse_host (isNotSpecial = true)") {
+static std::string long_host() {
+    // Host length = 10 + 102 * 10 = 1030 > 1024 (simple_buffer's fixed buffer length)
+    // Use "xn--" label to avoid ASCII fast path
+    std::string strHost = "xn--2da.90";
+    for (int i = 0; i < 102; ++i)
+        strHost.append(".bcde12345");
+    return strHost;
+}
+
+TEST_SUITE("host_parser::parse_host (is_opaque = true)") {
     TEST_CASE("HostType::Empty") {
         const std::string strHost = "";
         host_out out;
@@ -53,7 +62,7 @@ TEST_SUITE("host_parser::parse_host (isNotSpecial = true)") {
     }
 }
 
-TEST_SUITE("host_parser::parse_host (isNotSpecial = false)") {
+TEST_SUITE("host_parser::parse_host (is_opaque = false)") {
     TEST_CASE("HostType::Empty") {
         const std::string strHost = "";
         host_out out;
@@ -73,11 +82,7 @@ TEST_SUITE("host_parser::parse_host (isNotSpecial = false)") {
     }
 
     TEST_CASE("HostType::Domain with long host") {
-        // Host length = 10 + 102 * 10 = 1030 > 1024 (simple_buffer's fixed buffer length)
-        // Use "xn--" label to avoid ASCII fast path
-        std::string strHost = "xn--2da.90";
-        for (int i = 0; i < 102; ++i)
-            strHost.append(".bcde12345");
+        std::string strHost = long_host();
         host_out out;
 
         REQUIRE(upa::host_parser::parse_host(strHost.data(), strHost.data() + strHost.length(), false, out) == upa::validation_errc::ok);
@@ -241,3 +246,26 @@ TEST_SUITE("url_host") {
         CHECK(hm.type() == upa::HostType::IPv4);
     }
 }
+
+
+// Test upa::IDNToUnicode function
+
+TEST_SUITE("IDNToUnicode") {
+    TEST_CASE("Valid input") {
+        upa::simple_buffer<char> output;
+        CHECK(upa::IDNToUnicode("abc", 3, output) == upa::validation_errc::ok);
+        CHECK(upa::string_view(output.data(), output.size()) == "abc");
+    }
+
+    TEST_CASE("Valid long input") {
+        const std::string input = long_host();
+        upa::simple_buffer<char> output;
+        CHECK(upa::IDNToUnicode(input.data(), input.length(), output) == upa::validation_errc::ok);
+    }
+
+    TEST_CASE("Invalid input") {
+        upa::simple_buffer<char> output;
+        // IDNA errors are not failures for this function, so it returns `ok`
+        CHECK(upa::IDNToUnicode("xn--a.op", 8, output) == upa::validation_errc::ok);
+    }
+}