tdf#70833: IDNA support when exporing hyperlinks to PDF

Any URLs using non-ASCII IDNA syntax need to be resolved to ASCII-only, as PDF
URI Action's URI needs to be "encoded in 7-bit ASCII."

Introduce URIHelper::resolveIdnaHost (svl/urihelper.hxx), which internally uses
icu::IDNA, which requires to bump the minimal --with-system-icu requirement from
4.2 to 4.6, which means ICU_RECLASSIFIED_CLOSE_PARENTHESIS is always true now.

Change-Id: I0e20d9a20ed2b869fba0cc7c969721411db590b3
Reviewed-on: https://gerrit.libreoffice.org/19669
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Tested-by: Stephan Bergmann <sbergman@redhat.com>

3 files changed, 133 insertions, 0 deletions

diff --git a/svl/Library_svl.mk b/svl/Library_svl.mk
index db08af5f4b3a..3a7b00d1aaa0 100644
--- a/svl/Library_svl.mk
+++ b/svl/Library_svl.mk
@@ -21,6 +21,8 @@ $(eval $(call gb_Library_Library,svl))
 
 $(eval $(call gb_Library_use_externals,svl,\
     boost_headers \
+    icu_headers \
+    icuuc \
     mdds_headers \
     libxml2 \
 ))
diff --git a/svl/qa/unit/test_URIHelper.cxx b/svl/qa/unit/test_URIHelper.cxx
index f27149b57a01..37d53e90ec6c 100644
--- a/svl/qa/unit/test_URIHelper.cxx
+++ b/svl/qa/unit/test_URIHelper.cxx
@@ -198,9 +198,12 @@ public:
 
     void testFindFirstURLInText();
 
+    void testResolveIdnaHost();
+
     CPPUNIT_TEST_SUITE(Test);
     CPPUNIT_TEST(testNormalizedMakeRelative);
     CPPUNIT_TEST(testFindFirstURLInText);
+    CPPUNIT_TEST(testResolveIdnaHost);
     CPPUNIT_TEST(finish);
     CPPUNIT_TEST_SUITE_END();
 
@@ -423,6 +426,66 @@ void Test::testFindFirstURLInText() {
     }
 }
 
+void Test::testResolveIdnaHost() {
+    OUString input;
+
+    input.clear();
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("Foo.M\xC3\xBCnchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://Muenchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://-M\xC3\xBCnchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://M\xC3\xBCnchen-.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://xn--M\xC3\xBCnchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://xy--M\xC3\xBCnchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://.M\xC3\xBCnchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://-bar.M\xC3\xBCnchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://bar-.M\xC3\xBCnchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://xn--bar.M\xC3\xBCnchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    input = OUString::fromUtf8("foo://xy--bar.M\xC3\xBCnchen.de");
+    CPPUNIT_ASSERT_EQUAL(input, URIHelper::resolveIdnaHost(input));
+
+    CPPUNIT_ASSERT_EQUAL(
+        OUString::fromUtf8("foo://M\xC3\xBCnchen@xn--mnchen-3ya.de"),
+        URIHelper::resolveIdnaHost(
+            OUString::fromUtf8("foo://M\xC3\xBCnchen@M\xC3\xBCnchen.de")));
+
+    CPPUNIT_ASSERT_EQUAL(
+        OUString::fromUtf8("foo://xn--mnchen-3ya.de."),
+        URIHelper::resolveIdnaHost(
+            OUString::fromUtf8("foo://M\xC3\xBCnchen.de.")));
+
+    CPPUNIT_ASSERT_EQUAL(
+        OUString::fromUtf8("Foo://bar@xn--mnchen-3ya.de:123/?bar#baz"),
+        URIHelper::resolveIdnaHost(
+            OUString::fromUtf8("Foo://bar@M\xC3\xBCnchen.de:123/?bar#baz")));
+
+    CPPUNIT_ASSERT_EQUAL(
+        OUString::fromUtf8("foo://xn--mnchen-3ya.de"),
+        URIHelper::resolveIdnaHost(
+            OUString::fromUtf8("foo://Mu\xCC\x88nchen.de")));
+}
+
 css::uno::Reference< css::uno::XComponentContext > Test::m_context;
 
 CPPUNIT_TEST_SUITE_REGISTRATION(Test);
diff --git a/svl/source/misc/urihelper.cxx b/svl/source/misc/urihelper.cxx
index ab47bb6de3b6..bb5678a9291f 100644
--- a/svl/source/misc/urihelper.cxx
+++ b/svl/source/misc/urihelper.cxx
@@ -17,6 +17,10 @@
  *   the License at http://www.apache.org/licenses/LICENSE-2.0 .
  */
 
+#include <sal/config.h>
+
+#include <unicode/idna.h>
+
 #include <svl/urihelper.hxx>
 #include <com/sun/star/ucb/Command.hpp>
 #include <com/sun/star/ucb/IllegalIdentifierException.hpp>
@@ -725,4 +729,68 @@ OUString URIHelper::removePassword(OUString const & rURI,
                aObj.GetURLNoPass(eDecodeMechanism, eCharset);
 }
 
+OUString URIHelper::resolveIdnaHost(OUString const & url) {
+    css::uno::Reference<css::uri::XUriReference> uri(
+        css::uri::UriReferenceFactory::create(
+            comphelper::getProcessComponentContext())
+        ->parse(url));
+    if (!(uri.is() && uri->hasAuthority())) {
+        return url;
+    }
+    auto auth(uri->getAuthority());
+    sal_Int32 hostStart = auth.indexOf('@') + 1;
+    sal_Int32 hostEnd = auth.getLength() - 1;
+    while (hostEnd > hostStart && rtl::isAsciiDigit(auth[hostEnd])) {
+        --hostEnd;
+    }
+    if (!(hostEnd > hostStart && auth[hostEnd] == ':')) {
+        hostEnd = auth.getLength() - 1;
+    }
+    auto asciiOnly = true;
+    for (auto i = hostStart; i != hostEnd; ++i) {
+        if (!rtl::isAscii(auth[i])) {
+            asciiOnly = false;
+            break;
+        }
+    }
+    if (asciiOnly) {
+        // Avoid icu::IDNA case normalization in purely non-IDNA domain names:
+        return url;
+    }
+    UErrorCode e = U_ZERO_ERROR;
+    std::unique_ptr<icu::IDNA> idna(
+        icu::IDNA::createUTS46Instance(
+            (UIDNA_USE_STD3_RULES | UIDNA_CHECK_BIDI | UIDNA_CHECK_CONTEXTJ
+             | UIDNA_CHECK_CONTEXTO),
+            e));
+    if (U_FAILURE(e)) {
+        SAL_WARN("vcl.gdi", "icu::IDNA::createUTS46Instance " << e);
+        return url;
+    }
+    icu::UnicodeString ascii;
+    icu::IDNAInfo info;
+    idna->nameToASCII(
+        icu::UnicodeString(
+            reinterpret_cast<UChar const *>(auth.getStr() + hostStart),
+            hostEnd - hostStart),
+        ascii, info, e);
+    if (U_FAILURE(e) || info.hasErrors()) {
+        return url;
+    }
+    OUStringBuffer buf(uri->getScheme());
+    buf.append("://").append(auth.getStr(), hostStart);
+    buf.append(
+        reinterpret_cast<sal_Unicode const *>(ascii.getBuffer()),
+        ascii.length());
+    buf.append(auth.getStr() + hostEnd, auth.getLength() - hostEnd)
+        .append(uri->getPath());
+    if (uri->hasQuery()) {
+        buf.append('?').append(uri->getQuery());
+    }
+    if (uri->hasFragment()) {
+        buf.append('#').append(uri->getFragment());
+    }
+    return buf.makeStringAndClear();
+}
+
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */