File indexing completed on 2024-04-28 11:35:09

0001 /*
0002     SPDX-FileCopyrightText: 2005 Ingo Kloecker <kloecker@kde.org>
0003     SPDX-FileCopyrightText: 2007 Allen Winter <winter@kde.org>
0004 
0005     SPDX-License-Identifier: LGPL-2.0-only
0006 */
0007 
0008 #include "ktexttohtmltest.h"
0009 #include "kcoreaddons_debug.h"
0010 
0011 #include "../src/lib/text/ktexttohtml.h"
0012 #include "../src/lib/text/ktexttohtml_p.h"
0013 
0014 #include <QDebug>
0015 #include <QTest>
0016 #include <QUrl>
0017 
0018 QTEST_MAIN(KTextToHTMLTest)
0019 
0020 Q_DECLARE_METATYPE(KTextToHTML::Options)
0021 
0022 #ifndef Q_OS_WIN
0023 void initLocale()
0024 {
0025     setenv("LC_ALL", "en_US.utf-8", 1);
0026 }
0027 Q_CONSTRUCTOR_FUNCTION(initLocale)
0028 #endif
0029 
0030 void KTextToHTMLTest::testGetEmailAddress()
0031 {
0032     // empty input
0033     const QString emptyQString;
0034     KTextToHTMLHelper ll1(emptyQString, 0);
0035     QVERIFY(ll1.getEmailAddress().isEmpty());
0036 
0037     // no '@' at scan position
0038     KTextToHTMLHelper ll2(QStringLiteral("foo@bar.baz"), 0);
0039     QVERIFY(ll2.getEmailAddress().isEmpty());
0040 
0041     // '@' in local part
0042     KTextToHTMLHelper ll3(QStringLiteral("foo@bar@bar.baz"), 7);
0043     QVERIFY(ll3.getEmailAddress().isEmpty());
0044 
0045     // empty local part
0046     KTextToHTMLHelper ll4(QStringLiteral("@bar.baz"), 0);
0047     QVERIFY(ll4.getEmailAddress().isEmpty());
0048     KTextToHTMLHelper ll5(QStringLiteral(".@bar.baz"), 1);
0049     QVERIFY(ll5.getEmailAddress().isEmpty());
0050     KTextToHTMLHelper ll6(QStringLiteral(" @bar.baz"), 1);
0051     QVERIFY(ll6.getEmailAddress().isEmpty());
0052     KTextToHTMLHelper ll7(QStringLiteral(".!#$%&'*+-/=?^_`{|}~@bar.baz"), qstrlen(".!#$%&'*+-/=?^_`{|}~"));
0053     QVERIFY(ll7.getEmailAddress().isEmpty());
0054 
0055     // allowed special chars in local part of address
0056     KTextToHTMLHelper ll8(QStringLiteral("a.!#$%&'*+-/=?^_`{|}~@bar.baz"), qstrlen("a.!#$%&'*+-/=?^_`{|}~"));
0057     QCOMPARE(ll8.getEmailAddress(), QStringLiteral("a.!#$%&'*+-/=?^_`{|}~@bar.baz"));
0058 
0059     // '@' in domain part
0060     KTextToHTMLHelper ll9(QStringLiteral("foo@bar@bar.baz"), 3);
0061     QVERIFY(ll9.getEmailAddress().isEmpty());
0062 
0063     // domain part without dot
0064     KTextToHTMLHelper lla(QStringLiteral("foo@bar"), 3);
0065     QVERIFY(lla.getEmailAddress().isEmpty());
0066     KTextToHTMLHelper llb(QStringLiteral("foo@bar."), 3);
0067     QVERIFY(llb.getEmailAddress().isEmpty());
0068     KTextToHTMLHelper llc(QStringLiteral(".foo@bar"), 4);
0069     QVERIFY(llc.getEmailAddress().isEmpty());
0070     KTextToHTMLHelper lld(QStringLiteral("foo@bar "), 3);
0071     QVERIFY(lld.getEmailAddress().isEmpty());
0072     KTextToHTMLHelper lle(QStringLiteral(" foo@bar"), 4);
0073     QVERIFY(lle.getEmailAddress().isEmpty());
0074     KTextToHTMLHelper llf(QStringLiteral("foo@bar-bar"), 3);
0075     QVERIFY(llf.getEmailAddress().isEmpty());
0076 
0077     // empty domain part
0078     KTextToHTMLHelper llg(QStringLiteral("foo@"), 3);
0079     QVERIFY(llg.getEmailAddress().isEmpty());
0080     KTextToHTMLHelper llh(QStringLiteral("foo@."), 3);
0081     QVERIFY(llh.getEmailAddress().isEmpty());
0082     KTextToHTMLHelper lli(QStringLiteral("foo@-"), 3);
0083     QVERIFY(lli.getEmailAddress().isEmpty());
0084 
0085     // simple address
0086     KTextToHTMLHelper llj(QStringLiteral("foo@bar.baz"), 3);
0087     QCOMPARE(llj.getEmailAddress(), QStringLiteral("foo@bar.baz"));
0088     KTextToHTMLHelper llk(QStringLiteral("foo@bar.baz."), 3);
0089     QCOMPARE(llk.getEmailAddress(), QStringLiteral("foo@bar.baz"));
0090     KTextToHTMLHelper lll(QStringLiteral(".foo@bar.baz"), 4);
0091     QCOMPARE(lll.getEmailAddress(), QStringLiteral("foo@bar.baz"));
0092     KTextToHTMLHelper llm(QStringLiteral("foo@bar.baz-"), 3);
0093     QCOMPARE(llm.getEmailAddress(), QStringLiteral("foo@bar.baz"));
0094     KTextToHTMLHelper lln(QStringLiteral("-foo@bar.baz"), 4);
0095     QCOMPARE(lln.getEmailAddress(), QStringLiteral("foo@bar.baz"));
0096     KTextToHTMLHelper llo(QStringLiteral("foo@bar.baz "), 3);
0097     QCOMPARE(llo.getEmailAddress(), QStringLiteral("foo@bar.baz"));
0098     KTextToHTMLHelper llp(QStringLiteral(" foo@bar.baz"), 4);
0099     QCOMPARE(llp.getEmailAddress(), QStringLiteral("foo@bar.baz"));
0100     KTextToHTMLHelper llq(QStringLiteral("foo@bar-bar.baz"), 3);
0101     QCOMPARE(llq.getEmailAddress(), QStringLiteral("foo@bar-bar.baz"));
0102 }
0103 
0104 void KTextToHTMLTest::testGetUrl()
0105 {
0106     QStringList brackets;
0107     brackets << QString() << QString(); // no brackets
0108     brackets << QStringLiteral("<") << QStringLiteral(">");
0109     brackets << QStringLiteral("[") << QStringLiteral("]");
0110     brackets << QStringLiteral("\"") << QStringLiteral("\"");
0111     brackets << QStringLiteral("<link>") << QStringLiteral("</link>");
0112 
0113     for (int i = 0; i < brackets.count(); i += 2) {
0114         testGetUrl2(brackets[i], brackets[i + 1]);
0115     }
0116 }
0117 
0118 void KTextToHTMLTest::testGetUrl2(const QString &left, const QString &right)
0119 {
0120     QStringList schemas;
0121     schemas << QStringLiteral("http://");
0122     schemas << QStringLiteral("https://");
0123     schemas << QStringLiteral("vnc://");
0124     schemas << QStringLiteral("fish://");
0125     schemas << QStringLiteral("ftp://");
0126     schemas << QStringLiteral("ftps://");
0127     schemas << QStringLiteral("sftp://");
0128     schemas << QStringLiteral("smb://");
0129     schemas << QStringLiteral("file://");
0130     schemas << QStringLiteral("irc://");
0131     schemas << QStringLiteral("ircs://");
0132 
0133     QStringList urls;
0134     urls << QStringLiteral("www.kde.org");
0135     urls << QStringLiteral("user@www.kde.org");
0136     urls << QStringLiteral("user:pass@www.kde.org");
0137     urls << QStringLiteral("user:pass@www.kde.org:1234");
0138     urls << QStringLiteral("user:pass@www.kde.org:1234/sub/path");
0139     urls << QStringLiteral("user:pass@www.kde.org:1234/sub/path?a=1");
0140     urls << QStringLiteral("user:pass@www.kde.org:1234/sub/path?a=1#anchor");
0141     urls << QStringLiteral("user:pass@www.kde.org:1234/sub/\npath  \n /long/  path \t  ?a=1#anchor");
0142     urls << QStringLiteral("user:pass@www.kde.org:1234/sub/path/special(123)?a=1#anchor");
0143     urls << QStringLiteral("user:pass@www.kde.org:1234/sub/path:with:colon/special(123)?a=1#anchor");
0144     urls << QStringLiteral("user:pass@www.kde.org:1234/sub/path:with:colon/special(123)?a=1#anchor[bla");
0145     urls << QStringLiteral("user:pass@www.kde.org:1234/sub/path:with:colon/special(123)?a=1#anchor[bla]");
0146     urls << QStringLiteral("user:pass@www.kde.org:1234/\nsub/path:with:colon/\nspecial(123)?\na=1#anchor[bla]");
0147     urls << QStringLiteral("user:pass@www.kde.org:1234/  \n  sub/path:with:colon/  \n\t   \t   special(123)?") + QStringLiteral("\n\t  \n\t   a=1#anchor[bla]");
0148 
0149     for (const QString &schema : std::as_const(schemas)) {
0150         for (QString url : std::as_const(urls)) {
0151             // by definition: if the URL is enclosed in brackets, the URL itself is not allowed
0152             // to contain the closing bracket, as this would be detected as the end of the URL
0153             if ((left.length() == 1) && (url.contains(right[0]))) {
0154                 continue;
0155             }
0156 
0157             // if the url contains a whitespace, it must be enclosed with brackets
0158             if ((url.contains(QLatin1Char('\n')) || url.contains(QLatin1Char('\t')) || url.contains(QLatin1Char(' '))) && left.isEmpty()) {
0159                 continue;
0160             }
0161 
0162             QString test(left + schema + url + right);
0163             KTextToHTMLHelper ll(test, left.length());
0164             QString gotUrl = ll.getUrl();
0165 
0166             // we want to have the url without whitespace
0167             url.remove(QLatin1Char(' '));
0168             url.remove(QLatin1Char('\n'));
0169             url.remove(QLatin1Char('\t'));
0170 
0171             bool ok = (gotUrl == (schema + url));
0172             if (!ok) {
0173                 qCDebug(KCOREADDONS_DEBUG) << "got:" << gotUrl;
0174             }
0175             QVERIFY2(ok, qPrintable(test));
0176         }
0177     }
0178 
0179     QStringList urlsWithoutSchema;
0180     urlsWithoutSchema << QStringLiteral(".kde.org");
0181     urlsWithoutSchema << QStringLiteral(".kde.org:1234/sub/path");
0182     urlsWithoutSchema << QStringLiteral(".kde.org:1234/sub/path?a=1");
0183     urlsWithoutSchema << QStringLiteral(".kde.org:1234/sub/path?a=1#anchor");
0184     urlsWithoutSchema << QStringLiteral(".kde.org:1234/sub/path/special(123)?a=1#anchor");
0185     urlsWithoutSchema << QStringLiteral(".kde.org:1234/sub/path:with:colon/special(123)?a=1#anchor");
0186     urlsWithoutSchema << QStringLiteral(".kde.org:1234/sub/path:with:colon/special(123)?a=1#anchor[bla");
0187     urlsWithoutSchema << QStringLiteral(".kde.org:1234/sub/path:with:colon/special(123)?a=1#anchor[bla]");
0188     urlsWithoutSchema << QStringLiteral(".kde.org:1234/\nsub/path:with:colon/\nspecial(123)?\na=1#anchor[bla]");
0189     urlsWithoutSchema << QStringLiteral(".kde.org:1234/  \n  sub/path:with:colon/  \n\t   \t   special(123)?") + QStringLiteral("\n\t  \n\t   a=1#anchor[bla]");
0190 
0191     QStringList starts;
0192     starts << QStringLiteral("www") << QStringLiteral("ftp") << QStringLiteral("news:www");
0193 
0194     for (const QString &start : std::as_const(starts)) {
0195         for (QString url : std::as_const(urlsWithoutSchema)) {
0196             // by definition: if the URL is enclosed in brackets, the URL itself is not allowed
0197             // to contain the closing bracket, as this would be detected as the end of the URL
0198             if ((left.length() == 1) && (url.contains(right[0]))) {
0199                 continue;
0200             }
0201 
0202             // if the url contains a whitespace, it must be enclosed with brackets
0203             if ((url.contains(QLatin1Char('\n')) || url.contains(QLatin1Char('\t')) || url.contains(QLatin1Char(' '))) && left.isEmpty()) {
0204                 continue;
0205             }
0206 
0207             QString test(left + start + url + right);
0208             KTextToHTMLHelper ll(test, left.length());
0209             QString gotUrl = ll.getUrl();
0210 
0211             // we want to have the url without whitespace
0212             url.remove(QLatin1Char(' '));
0213             url.remove(QLatin1Char('\n'));
0214             url.remove(QLatin1Char('\t'));
0215 
0216             bool ok = (gotUrl == (start + url));
0217             if (!ok) {
0218                 qCDebug(KCOREADDONS_DEBUG) << "got:" << gotUrl;
0219             }
0220             QVERIFY2(ok, qPrintable(gotUrl));
0221         }
0222     }
0223 
0224     // test max url length
0225     QString url = QStringLiteral("https://www.kde.org/this/is/a_very_loooooong_url/test/test/test");
0226     {
0227         KTextToHTMLHelper ll(url, 0, 10);
0228         QVERIFY(ll.getUrl().isEmpty()); // url too long
0229     }
0230     {
0231         KTextToHTMLHelper ll(url, 0, url.length() - 1);
0232         QVERIFY(ll.getUrl().isEmpty()); // url too long
0233     }
0234     {
0235         KTextToHTMLHelper ll(url, 0, url.length());
0236         QCOMPARE(ll.getUrl(), url);
0237     }
0238     {
0239         KTextToHTMLHelper ll(url, 0, url.length() + 1);
0240         QCOMPARE(ll.getUrl(), url);
0241     }
0242 
0243     // mailto
0244     {
0245         QString addr = QStringLiteral("mailto:test@kde.org");
0246         QString test(left + addr + right);
0247         KTextToHTMLHelper ll(test, left.length());
0248 
0249         QString gotUrl = ll.getUrl();
0250 
0251         bool ok = (gotUrl == addr);
0252         if (!ok) {
0253             qCDebug(KCOREADDONS_DEBUG) << "got:" << gotUrl;
0254         }
0255         QVERIFY2(ok, qPrintable(gotUrl));
0256     }
0257 }
0258 
0259 void KTextToHTMLTest::testHtmlConvert_data()
0260 {
0261     QTest::addColumn<QString>("plainText");
0262     QTest::addColumn<KTextToHTML::Options>("flags");
0263     QTest::addColumn<QString>("htmlText");
0264 
0265     // Linker error when using PreserveSpaces, therefore the hardcoded 0x01 or 0x09
0266 
0267     // Test preserving whitespace correctly
0268     QTest::newRow("") << " foo" << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "&nbsp;foo";
0269     QTest::newRow("") << "  foo" << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "&nbsp;&nbsp;foo";
0270     QTest::newRow("") << "  foo  " << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "&nbsp;&nbsp;foo&nbsp;&nbsp;";
0271     QTest::newRow("") << "  foo " << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "&nbsp;&nbsp;foo&nbsp;";
0272     QTest::newRow("") << "bla bla bla bla bla" << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "bla bla bla bla bla";
0273     QTest::newRow("") << "bla bla bla \n  bla bla bla " << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0274                       << "bla bla bla&nbsp;<br />\n&nbsp;&nbsp;bla bla bla&nbsp;";
0275     QTest::newRow("") << "bla bla  bla" << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "bla bla&nbsp;&nbsp;bla";
0276     QTest::newRow("") << " bla bla \n bla bla a\n  bla bla " << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0277                       << "&nbsp;bla bla&nbsp;<br />\n&nbsp;bla bla a<br />\n"
0278                          "&nbsp;&nbsp;bla bla&nbsp;";
0279 
0280     // Test highlighting with *, / and _
0281     QTest::newRow("") << "Ce paragraphe _contient_ des mots ou des _groupes de mots_ à mettre en"
0282                          " forme…"
0283                       << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0284                       << "Ce paragraphe <u>_contient_</u> des mots ou des"
0285                          " <u>_groupes de mots_</u> à mettre en forme…";
0286     QTest::newRow("punctation-bug") << "Ce texte *a l'air* de _fonctionner_, à condition"
0287                                        " d’utiliser le guillemet ASCII."
0288                                     << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0289                                     << "Ce texte <b>*a l'air*</b> de <u>_fonctionner_</u>, à"
0290                                        " condition d’utiliser le guillemet ASCII.";
0291     QTest::newRow("punctation-bug") << "Un répertoire /est/ un *dossier* où on peut mettre des"
0292                                        " *fichiers*."
0293                                     << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0294                                     << "Un répertoire <i>/est/</i> un"
0295                                        " <b>*dossier*</b> où on peut mettre des <b>*fichiers*</b>.";
0296     QTest::newRow("punctation-bug") << "*BLA BLA BLA BLA*." << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0297                                     << "<b>BLA BLA BLA BLA</b>.";
0298     QTest::newRow("") << "Je vais tenter de repérer des faux positif*" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0299                       << "Je vais tenter de repérer des faux positif*";
0300     QTest::newRow("") << "*Ouais !* *Yes!*" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0301                       << "<b>*Ouais !*</b> <b>*Yes!*</b>";
0302 
0303     QTest::newRow("multispace") << "*Ouais     foo*" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0304                                 << "<b>*Ouais     foo*</b>";
0305 
0306     QTest::newRow("multispace3") << "*Ouais:     foo*" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0307                                  << "<b>*Ouais:     foo*</b>";
0308 
0309     QTest::newRow("multi-") << "** Ouais:  foo **" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0310                             << "** Ouais:&nbsp;&nbsp;foo **";
0311 
0312     QTest::newRow("multi-") << "*** Ouais:  foo ***" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0313                             << "*** Ouais:&nbsp;&nbsp;foo ***";
0314 
0315     QTest::newRow("nohtmlversion") << "* Ouais:     foo *" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0316                                    << "* Ouais:&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;foo *";
0317 
0318     QTest::newRow("nohtmlversion2") << "*Ouais:     foo *" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0319                                     << "*Ouais:&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;foo *";
0320 
0321     QTest::newRow("nohtmlversion3") << "* Ouais:     foo*" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0322                                     << "* Ouais:&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;foo*";
0323 
0324     QTest::newRow("nohtmlversion3") << "* Ouais: *ff sfsdf* foo *" << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0325                                     << "* Ouais: <b>*ff sfsdf*</b> foo *";
0326 
0327     QTest::newRow("") << "the /etc/{rsyslog.d,syslog-ng.d}/package.rpmnew file"
0328                       << KTextToHTML::Options(KTextToHTML::PreserveSpaces | KTextToHTML::HighlightText)
0329                       << "the /etc/{rsyslog.d,syslog-ng.d}/package.rpmnew file";
0330 
0331     // This test has problems with the encoding, apparently.
0332     // QTest::newRow( "" ) << "*Ça fait plaisir de pouvoir utiliser des lettres accentuées dans du"
0333     //                       " texte mis en forme*." << 0x09 << "<b>Ça fait plaisir de pouvoir"
0334     //                       " utiliser des lettres accentuées dans du texte mis en forme</b>.";
0335 
0336     // Bug reported by dfaure, the <hostname> would get lost
0337     QTest::newRow("") << "QUrl url(\"http://strange<hostname>/\");" << KTextToHTML::Options(KTextToHTML::ReplaceSmileys | KTextToHTML::HighlightText)
0338                       << "QUrl url(&quot;<a href=\"http://strange<hostname>/\">"
0339                          "http://strange&lt;hostname&gt;/</a>&quot;);";
0340 
0341     // Bug: 211128 - plain text emails should not replace ampersand & with &amp;
0342     QTest::newRow("bug211128") << "https://green-site/?Ticket=85&Page=next" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0343                                << "<a href=\"https://green-site/?Ticket=85&Page=next\">"
0344                                   "https://green-site/?Ticket=85&amp;Page=next</a>";
0345 
0346     QTest::newRow("dotBeforeEnd") << "Look at this file: www.example.com/example.h" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0347                                   << "Look at this file: <a href=\"http://www.example.com/example.h\">"
0348                                      "www.example.com/example.h</a>";
0349     QTest::newRow("dotInMiddle") << "Look at this file: www.example.com/.bashrc" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0350                                  << "Look at this file: <a href=\"http://www.example.com/.bashrc\">"
0351                                     "www.example.com/.bashrc</a>";
0352 
0353     // A dot at the end of an URL is explicitly ignored
0354     QTest::newRow("dotAtEnd") << "Look at this file: www.example.com/test.cpp." << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0355                               << "Look at this file: <a href=\"http://www.example.com/test.cpp\">"
0356                                  "www.example.com/test.cpp</a>.";
0357 
0358     // Bug 313719 - URL in parenthesis
0359     QTest::newRow("url-in-parenthesis-1") << "KDE (website https://www.kde.org)" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0360                                           << "KDE (website <a href=\"https://www.kde.org\">https://www.kde.org</a>)";
0361     QTest::newRow("url-in-parenthesis-2") << "KDE website (https://www.kde.org)" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0362                                           << "KDE website (<a href=\"https://www.kde.org\">https://www.kde.org</a>)";
0363     QTest::newRow("url-in-parenthesis-3") << "bla (https://www.kde.org - section 5.2)" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0364                                           << "bla (<a href=\"https://www.kde.org\">https://www.kde.org</a> - section 5.2)";
0365 
0366     // Fix url as foo <<url> <url>> when we concatened them.
0367     QTest::newRow("url-with-url")
0368         << "foo <https://www.kde.org/ <https://www.kde.org/>>" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0369         << "foo &lt;<a href=\"https://www.kde.org/ \">https://www.kde.org/ </a>&lt;<a href=\"https://www.kde.org/\">https://www.kde.org/</a>&gt;&gt;";
0370 
0371     // Fix url exploit
0372     QTest::newRow("url-exec-html") << "https://\"><!--" << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "https://&quot;&gt;&lt;!--";
0373 
0374     QTest::newRow("url-exec-html-2") << "https://192.168.1.1:\"><!--" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0375                                      << "https://192.168.1.1:&quot;&gt;&lt;!--";
0376 
0377     QTest::newRow("url-exec-html-3") << "https://<IP>:\"><!--" << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "https://&lt;IP&gt;:&quot;&gt;&lt;!--";
0378 
0379     QTest::newRow("url-exec-html-4") << "https://<IP>:/\"><!--" << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "https://&lt;IP&gt;:/&quot;&gt;&lt;!--";
0380 
0381     QTest::newRow("url-exec-html-5") << "https://<IP>:/\"><script>alert(1);</script><!--" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0382                                      << "https://&lt;IP&gt;:/&quot;&gt;&lt;script&gt;alert(1);&lt;/script&gt;&lt;!--";
0383 
0384     QTest::newRow("url-exec-html-6") << "https://<IP>:/\"><script>alert(1);</script><!--\nTest2" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0385                                      << "https://&lt;IP&gt;:/&quot;&gt;&lt;script&gt;alert(1);&lt;/script&gt;&lt;!--\nTest2";
0386 
0387     QTest::newRow("url-with-ref-in-[") << "https://www.kde.org[1]" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0388                                        << "<a href=\"https://www.kde.org\">https://www.kde.org</a>[1]";
0389 
0390     QTest::newRow("url-with-ref-in-[2") << "[http://www.example.org/][whatever]" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0391                                         << "[<a href=\"http://www.example.org/\">http://www.example.org/</a>][whatever]";
0392     // Bug 346132
0393     QTest::newRow("url-with-ref-in-<") << "http://www.foo.bar<http://foo.bar/>" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0394                                        << "<a href=\"http://www.foo.bar\">http://www.foo.bar</a>&lt;<a href=\"http://foo.bar/\">http://foo.bar/</a>&gt;";
0395 
0396     QTest::newRow("url-with-ref-in-]") << "[Please visit our booth 24-25 http://example.com/]" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0397                                        << "[Please visit our booth 24-25 <a href=\"http://example.com/\">http://example.com/</a>]";
0398 
0399     QTest::newRow("two url with space") << "http://www.kde.org/standards/kcfg/1.0 http://www.kde.org/" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0400                                         << "<a href=\"http://www.kde.org/standards/kcfg/1.0\">http://www.kde.org/standards/kcfg/1.0</a> <a "
0401                                            "href=\"http://www.kde.org/\">http://www.kde.org/</a>";
0402 
0403     // Bug kmail
0404     QTest::newRow("two url with space-2")
0405         << "@@ -55,6 +55,10 @@ xsi:schemaLocation=\"http://www.kde.org/standards/kcfg/1.0 http://www.kde.org/"
0406         << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0407         << "@@ -55,6 +55,10 @@ xsi:schemaLocation=&quot;<a href=\"http://www.kde.org/standards/kcfg/1.0\">http://www.kde.org/standards/kcfg/1.0</a> <a "
0408            "href=\"http://www.kde.org/\">http://www.kde.org/</a>";
0409 
0410     const auto opt = KTextToHTML::PreserveSpaces | KTextToHTML::ConvertPhoneNumbers;
0411     // tel: urls
0412     QTest::newRow("tel url compact") << "bla bla <tel:+491234567890> bla bla" << opt
0413                                      << "bla bla &lt;<a href=\"tel:+491234567890\">tel:+491234567890</a>&gt; bla bla";
0414     QTest::newRow("tel url fancy") << "bla bla tel:+49-321-123456 bla bla" << opt << "bla bla <a href=\"tel:+49-321-123456\">tel:+49-321-123456</a> bla bla";
0415 
0416     // negative tel: url tests
0417     QTest::newRow("empty tel url") << "bla tel: blub" << opt << "bla tel: blub";
0418 
0419     // phone numbers
0420     QTest::newRow("tel compact international") << "call +49123456789, then hang up" << opt
0421                                                << "call <a href=\"tel:+49123456789\">+49123456789</a>, then hang up";
0422     QTest::newRow("tel parenthesis/spaces international")
0423         << "phone:+33 (01) 12 34 56 78 blub" << opt << "phone:<a href=\"tel:+330112345678\">+33 (01) 12 34 56 78</a> blub";
0424     QTest::newRow("tel dashes international") << "bla +44-321-1-234-567" << opt << "bla <a href=\"tel:+443211234567\">+44-321-1-234-567</a>";
0425     QTest::newRow("tel dashes/spaces international") << "+1 123-456-7000 blub" << opt << "<a href=\"tel:+11234567000\">+1 123-456-7000</a> blub";
0426     QTest::newRow("tel spaces international") << "bla +32 1 234 5678 blub" << opt << "bla <a href=\"tel:+3212345678\">+32 1 234 5678</a> blub";
0427     QTest::newRow("tel slash domestic") << "bla 030/12345678 blub" << opt << "bla <a href=\"tel:03012345678\">030/12345678</a> blub";
0428     QTest::newRow("tel slash/space domestic") << "Tel.: 089 / 12 34 56 78" << opt << "Tel.: <a href=\"tel:08912345678\">089 / 12 34 56 78</a>";
0429     QTest::newRow("tel follow by parenthesis") << "Telefon: 0 18 05 / 12 23 46 (14 Cent/Min.*)" << opt
0430                                                << "Telefon: <a href=\"tel:01805122346\">0 18 05 / 12 23 46</a> (14 Cent/Min.*)";
0431     QTest::newRow("tel space single digit at end") << "0123/123 456 7" << opt << "<a href=\"tel:01231234567\">0123/123 456 7</a>";
0432     QTest::newRow("tel space around dash") << "bla +49 (0) 12 23 - 45 6000 blub" << opt
0433                                            << "bla <a href=\"tel:+4901223456000\">+49 (0) 12 23 - 45 6000</a> blub";
0434     QTest::newRow("tel two numbers speparated by dash")
0435         << "bla +49 (0) 12 23 46 78 - +49 0123/123 456 78 blub" << opt
0436         << "bla <a href=\"tel:+49012234678\">+49 (0) 12 23 46 78</a> - <a href=\"tel:+49012312345678\">+49 0123/123 456 78</a> blub";
0437 
0438     // negative tests for phone numbers
0439     QTest::newRow("non-tel number") << "please send 1200 cakes" << opt << "please send 1200 cakes";
0440     QTest::newRow("non-tel alpha-numeric") << "bla 1-123-456-ABCD blub" << opt << "bla 1-123-456-ABCD blub";
0441     QTest::newRow("non-tel alpha prefix") << "ABCD0123-456-789" << opt << "ABCD0123-456-789";
0442     QTest::newRow("non-tel date") << "bla 02/03/2019 blub" << opt << "bla 02/03/2019 blub";
0443     QTest::newRow("non-tel too long") << "bla +012-4567890123456 blub" << opt << "bla +012-4567890123456 blub";
0444     QTest::newRow("non-tel unbalanced") << "bla +012-456789(01 blub" << opt << "bla +012-456789(01 blub";
0445     QTest::newRow("non-tel nested") << "bla +012-4(56(78)90)1 blub" << opt << "bla +012-4(56(78)90)1 blub";
0446     QTest::newRow("tel extraction disabled") << "call +49123456789 now" << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "call +49123456789 now";
0447 
0448     QTest::newRow("bug-414360")
0449         << "https://www.openstreetmap.org/directions?engine=graphhopper_foot&route=44.85765%2C-0.55931%3B44.85713%2C-0.56117#map=18/44.85756/-0.56094"
0450         << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0451         << "<a "
0452            "href=\"https://www.openstreetmap.org/directions?engine=graphhopper_foot&route=44.85765%2C-0.55931%3B44.85713%2C-0.56117#map=18/44.85756/"
0453            "-0.56094\">https://www.openstreetmap.org/directions?engine=graphhopper_foot&amp;route=44.85765%2C-0.55931%3B44.85713%2C-0.56117#map=18/44.85756/"
0454            "-0.56094</a>";
0455 
0456     // xmpp bug 422291
0457     QTest::newRow("xmpp1") << "xmpp:username@server.tld" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0458                            << "<a href=\"xmpp:username@server.tld\">xmpp:username@server.tld</a>";
0459     QTest::newRow("xmpp2") << "xmpp:conversations@conference.siacs.eu" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0460                            << "<a href=\"xmpp:conversations@conference.siacs.eu\">xmpp:conversations@conference.siacs.eu</a>";
0461     QTest::newRow("xmpp3") << "xmpp:conversations@conference.siacs.eu?join" << KTextToHTML::Options(KTextToHTML::PreserveSpaces)
0462                            << "<a href=\"xmpp:conversations@conference.siacs.eu?join\">xmpp:conversations@conference.siacs.eu?join</a>";
0463 
0464     // Test news: only
0465     QTest::newRow("news") << "news: " << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "news:&nbsp;";
0466 
0467     QTest::newRow("ftp") << "ftp: " << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "ftp:&nbsp;";
0468     QTest::newRow("mailto") << "mailto: " << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "mailto:&nbsp;";
0469     QTest::newRow("empty") << "" << KTextToHTML::Options(KTextToHTML::PreserveSpaces) << "";
0470 }
0471 
0472 void KTextToHTMLTest::testHtmlConvert()
0473 {
0474     QFETCH(QString, plainText);
0475     QFETCH(KTextToHTML::Options, flags);
0476     QFETCH(QString, htmlText);
0477 
0478     QEXPECT_FAIL("punctation-bug", "Linklocator does not properly detect punctation as boundaries", Continue);
0479 
0480     const QString actualHtml = KTextToHTML::convertToHtml(plainText, flags);
0481     QCOMPARE(actualHtml, htmlText);
0482 }
0483 
0484 #include "moc_ktexttohtmltest.cpp"