Repository: lucenenet Updated Branches: refs/heads/api-work 54bad2c2d -> 548e768cc
Lucene.Net.Tests.Analysis.Common.Analysis.CharFilter.HTMLStripCharFilterTest: Added note about version compatibility level 4.8.1 and reformatted tests for easier reading Project: http://git-wip-us.apache.org/repos/asf/lucenenet/repo Commit: http://git-wip-us.apache.org/repos/asf/lucenenet/commit/31ceeb20 Tree: http://git-wip-us.apache.org/repos/asf/lucenenet/tree/31ceeb20 Diff: http://git-wip-us.apache.org/repos/asf/lucenenet/diff/31ceeb20 Branch: refs/heads/api-work Commit: 31ceeb20653ba84a26606f28c7f51e7baebe361b Parents: 54bad2c Author: Shad Storhaug <[email protected]> Authored: Sun Mar 26 04:19:18 2017 +0700 Committer: Shad Storhaug <[email protected]> Committed: Sun Mar 26 04:19:18 2017 +0700 ---------------------------------------------------------------------- .../Analysis/CharFilter/HTMLStripCharFilter.cs | 1 + .../CharFilters/HTMLStripCharFilterTest.cs | 228 ++++++++++++++++++- 2 files changed, 217 insertions(+), 12 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/lucenenet/blob/31ceeb20/src/Lucene.Net.Analysis.Common/Analysis/CharFilter/HTMLStripCharFilter.cs ---------------------------------------------------------------------- diff --git a/src/Lucene.Net.Analysis.Common/Analysis/CharFilter/HTMLStripCharFilter.cs b/src/Lucene.Net.Analysis.Common/Analysis/CharFilter/HTMLStripCharFilter.cs index 7dba4f6..7184212 100644 --- a/src/Lucene.Net.Analysis.Common/Analysis/CharFilter/HTMLStripCharFilter.cs +++ b/src/Lucene.Net.Analysis.Common/Analysis/CharFilter/HTMLStripCharFilter.cs @@ -29,6 +29,7 @@ namespace Lucene.Net.Analysis.CharFilters /// <summary> /// A <see cref="CharFilter"/> that wraps another <see cref="TextReader"/> and attempts to strip out HTML constructs. /// </summary> + // LUCENENET NOTE: Version compatibility level 4.8.1 (added fix for SOLR-5983: HTMLStripCharFilter is treating CDATA sections incorrectly) public sealed class HTMLStripCharFilter : BaseCharFilter { /// <summary>This character denotes the end of file</summary> http://git-wip-us.apache.org/repos/asf/lucenenet/blob/31ceeb20/src/Lucene.Net.Tests.Analysis.Common/Analysis/CharFilters/HTMLStripCharFilterTest.cs ---------------------------------------------------------------------- diff --git a/src/Lucene.Net.Tests.Analysis.Common/Analysis/CharFilters/HTMLStripCharFilterTest.cs b/src/Lucene.Net.Tests.Analysis.Common/Analysis/CharFilters/HTMLStripCharFilterTest.cs index 2b07c0d..0db491d 100644 --- a/src/Lucene.Net.Tests.Analysis.Common/Analysis/CharFilters/HTMLStripCharFilterTest.cs +++ b/src/Lucene.Net.Tests.Analysis.Common/Analysis/CharFilters/HTMLStripCharFilterTest.cs @@ -25,6 +25,7 @@ namespace Lucene.Net.Analysis.CharFilters * limitations under the License. */ + // LUCENENET NOTE: Version compatibility level 4.8.1 (added fix for SOLR-5983: HTMLStripCharFilter is treating CDATA sections incorrectly) public class HTMLStripCharFilterTest : BaseTokenStreamTestCase { @@ -56,8 +57,12 @@ namespace Lucene.Net.Analysis.CharFilters [Test] public virtual void Test() { - string html = "<div class=\"foo\">this is some text</div> here is a <a href=\"#bar\">link</a> and " + "another <a href=\"http://lucene.apache.org/\";>link</a>. " + "This is an entity: & plus a <. Here is an &. <!-- is a comment -->"; - string gold = "\nthis is some text\n here is a link and " + "another link. " + "This is an entity: & plus a <. Here is an &. "; + string html = "<div class=\"foo\">this is some text</div> here is a <a href=\"#bar\">link</a> and " + + "another <a href=\"http://lucene.apache.org/\";>link</a>. " + + "This is an entity: & plus a <. Here is an &. <!-- is a comment -->"; + string gold = "\nthis is some text\n here is a link and " + + "another link. " + + "This is an entity: & plus a <. Here is an &. "; AssertHTMLStripsTo(html, gold, null); } @@ -95,7 +100,8 @@ namespace Lucene.Net.Analysis.CharFilters builder.Append((char)ch); } // Compare trim()'d output to gold - assertEquals("'" + builder.ToString().Trim() + "' is not equal to '" + gold + "'", gold, builder.ToString().Trim()); + assertEquals("'" + builder.ToString().Trim() + "' is not equal to '" + gold + "'", + gold, builder.ToString().Trim()); } [Test] @@ -144,7 +150,169 @@ namespace Lucene.Net.Analysis.CharFilters [Test] public virtual void TestMalformedHTML() { - string[] testGold = new string[] { "a <a hr<ef=aa<a>> </close</a>", "a <a hr<ef=aa> </close", "<a href=http://dmoz.org/cgi-bin/add.cgi?where=/arts/\"; class=lu style=\"font-size: 9px\" target=dmoz>Submit a Site</a>", "Submit a Site", "<a href=javascript:ioSwitch('p8','http://www.csmonitor.com/') title=expand id=e8 class=expanded rel=http://www.csmonitor.com/>Christian Science", "Christian Science", "<link rel=\"alternate\" type=\"application/rss+xml\" title=\"San Francisco \" 2008 RSS Feed\" href=\"http://2008.sf.wordcamp.org/feed/\"; />", "\n", "<a href=\" http://www.surgery4was.happyhost.org/video-of-arthroscopic-knee-surgery symptoms.html, heat congestive heart failure <a href=\" http://www.symptoms1bad.happyhost.org/canine";, "<a href=\" http://www.surgery4was.happyhost.org/video-of-arthroscopic-knee-surgery symptoms.html, heat congestive heart failure <a href=\" http://www.symptoms1bad.happyhost.org/canine";, "<a href=\"http://ucblibraries.colorado.edu/how/index.htm\"cl ass=\"pageNavAreaText\">", "", "<link title=\"^\\\" 21Sta's Blog\" rel=\"search\" type=\"application/opensearchdescription+xml\" href=\"http://21sta.com/blog/inc/opensearch.php\"; />", "\n", "<a href=\"#postcomment\" title=\"\"Leave a comment\";\">?", "?", "<a href='/modern-furniture' ' id='21txt' class='offtab' onMouseout=\"this.className='offtab'; return true;\" onMouseover=\"this.className='ontab'; return true;\">", "", "<a href='http://alievi.wordpress.com/category/01-todos-posts/' style='font-size: 275%; padding: 1px; margin: 1px;' title='01 - Todos Post's (83)'>", "", "The <a href=<a href=\"http://www.advancedmd.com>medical\">http://www.advancedmd.com>medical</a> practice software</a>", "The <a href=medical\">http://www.advancedmd.com>medical practice software", "<a href=\"node/21426\" class=\"clipTitle2\" title=\"Levi.com/BMX 2008 Clip of the Week 29 \"Morgan Wade Leftover Clips\"\">Levi.com/BMX 2008 Clip of the Week 29...", "Levi.com/BMX 2008 Clip of the Week 29...", "<a href=\"printer_friendly.php?branch=&year=&submit=go&screen=\";\">Printer Friendly", "Printer Friendly", "<a href=#\" ondragstart=\"return false\" onclick=\"window.external.AddFavorite('http://www.amazingtextures.com', 'Amazing Textures');return false\" onmouseover=\"window.status='Add to Favorites';return true\">Add to Favorites", "Add to Favorites", "<a href=\"../at_home/at_home_search.html\"../_home/at_home_search.html\">At", "At", "E-mail: <a href=\"\"mailto:[email protected]\"; \">[email protected] </a>", "E-mail: [email protected] ", "<li class=\"farsi\"><a title=\"A'13?\" alt=\"A'13?\" href=\"http://www.america.gov/persian\"; alt=\"\" name=\"A'13?\"A'13? title=\"A'13?\">A'13?</a></li>", "\nA'13?\n", "<li><a href=\"#28\" title=\"Hubert \"Geese\" Ausby\">Hubert \"Geese\" Ausby</a></li>", "\nHubert \"Geese\" Ausby\n", "<href=\"http://anbportal.com/mms/login.asp\";>", "\n", "<a href=\"", "<a href=\"", "<a href=\">", "", "<a rel=\"nofollow\" href=\"http://anissanina31.skyrock.c om/1895039493-Hi-tout-le-monde.html\" title=\" Hi, tout le monde !>#</a>", "#", "<a href=\"http://annunciharleydavidsonusate.myblog.it/\"; title=\"Annunci Moto e Accessori Harley Davidson\" target=\"_blank\"><img src=\"http://annunciharleydavidsonusate.myblog.it/images/Antipixel.gif\"; /></a>", "", "<a href=\"video/addvideo&v=120838887181\" onClick=\"return confirm('Are you sure you want add this video to your profile? If it exists some video in your profile will be overlapped by this video!!')\" \" onmouseover=\"this.className='border2'\" onmouseout=\"this.className=''\">", "", "<a href=#Services & Support>", "", "<input type=\"image\" src=\"http://apologyindex.com/ThemeFiles/83401-72905/images/btn_search.gif\"value=\"Search\"; name=\"Search\" alt=\"Search\" class=\"searchimage\" onclick=\"incom ='&sc=' + document.getElementById('sel').value ; var dt ='&dt=' + document.getElementById('dt').value; var searchKeyword = document.getElementById('q').value ; searchKeyword = searchKeyword.r eplace(/\\s/g,''); if (searchKeyword.length < 3){alert('Nothing to search. Search keyword should contain atleast 3 chars.'); return false; } var al='&al=' + document.getElementById('advancedlink').style.display ; document.location.href='http://apologyindex.com/search.aspx?q=' + document.getElementById('q').value + incom + dt + al;\" />", "", "<input type=\"image\" src=\"images/afbe.gif\" width=\"22\" height=\"22\" hspace=\"4\" title=\"Add to Favorite\" alt=\"Add to Favorite\"onClick=\" if(window.sidebar){ window.sidebar.addPanel(document.title,location.href,''); }else if(window.external){ window.external.AddFavorite(location.href,document.title); }else if(window.opera&&window.print) { return true; }\">", "", "<area shape=\"rect\" coords=\"12,153,115,305\" href=\"http://statenislandtalk.com/v-web/gallery/Osmundsen-family\"Art's Norwegian Roots in Rogaland\">", "\n", "<a rel=\"nofollow\" href=\"http://arth26.skyrock.com/660188240-bonzai.html\"; title=\"bonza>#", "#", "<a href= >", "", "<ahref=http:..", "<ahref=http:..", "<ahref=http:..>", "\n", "<ahref=\"http://aseigo.bddf.ca/cms/1025\";>A", "\nA", "<a href=\"javascript:calendar_window=window.open('/calendar.aspx?formname=frmCalendar.txtDate','calendar_window','width=154,height=188');calendar_window.focus()\">", "", "<a href=\"/applications/defenseaerospace/19+rackmounts\" title=\"19\" Rackmounts\">", "", "<a href=http://www.azimprimerie.fr/flash/backup/lewes-zip-code/savage-model-110-manual.html title=savage model 110 manual rel=dofollow>", "", "<a class=\"at\" name=\"Lamborghini href=\"http://lamborghini.coolbegin.com\";>Lamborghini /a>", "Lamborghini /a>", "<A href='newslink.php?news_link=http%3A%2F%2Fwww.worldnetdaily.com%2Findex.php%3Ffa%3DPAGE.view%26pageId%3D85729&news_title=Florida QB makes 'John 3:16' hottest Google search Tebow inscribed Bible reference on eye black for championship game' TARGET=_blank>", "", "<a href=/myspace !style='color:#993333'>", "", "<meta name=3DProgId content=3DExcel.Sheet>" , "\n", "<link id=3D\"shLink\" href=3D\"PSABrKelly-BADMINTONCupResults08FINAL2008_09_19=_files/sheet004.htm\">", "\n", "<td bgcolor=3D\"#FFFFFF\" nowrap>", "\n", "<a href=\"http://basnect.info/usersearch/\"predicciones-mundiales-2009\".html\";>\"predicciones mundiales 2009\"</a>", "\"predicciones mundiales 2009\"", "<a class=\"comment-link\" href=\"https://www.blogger.com/comment.g?blogID=19402125&postID=114070605958684588\"location.href=https://www.blogger.com/comment.g?blogID=19402125&postID=114070605958684588;>", "", "<a href = \"/videos/Bishop\"/\" title = \"click to see more Bishop\" videos\">Bishop\"</a>", "Bishop\"", "<a href=\"http://bhaa.ie/calendar/event.php?eid=20081203150127531\"\";>BHAA Eircom 2 & 5 miles CC combined start</a>", "BHAA Eircom 2 & 5 miles CC combined start", "<a href=\"http://people.tribe.net/wolfmana\"; onClick='setClick(\"Application[tribe].Person[bb7df210-9dc0-478c-917f-436b896bcb79]\")'\" title=\"Mana\">", "", "<a href=\"http://blog.edu-cyberpg.com/ ct.ashx?id=6143c528-080c-4bb2-b765-5ec56c8256d3&url=http%3a%2f%2fwww.gsa.ac.uk%2fmackintoshsketchbook%2f\"\" eudora=\"autourl\">", "", "<input type=\"text\" value=\"<search here>\">", "<input type=\"text\" value=\"\n\">", "<input type=\"text\" value=\"<search here\">", "<input type=\"text\" value=\"\n", "<input type=\"text\" value=\"search here>\">", "\">", "<input type=\"text\" value=\"<search here>\" onFocus=\"this.value='<search here>'\">", "", "<![if ! IE]>\n<link href=\"http://i.deviantart.com/icons/favicon.png\"; rel=\"shortcut icon\"/>\n<![endif]>", "\n\n\n", "<![if supportMisalignedColumns]>\n<tr height=0 style='display:none'>\n<td width=64 style='width:48pt'></td>\n</tr>\n<![endif]>", "\n\n\n\n\n\n\n\n" }; + string[] testGold = { + "a <a hr<ef=aa<a>> </close</a>", + "a <a hr<ef=aa> </close", + + "<a href=http://dmoz.org/cgi-bin/add.cgi?where=/arts/\"; class=lu style=\"font-size: 9px\" target=dmoz>Submit a Site</a>", + "Submit a Site", + + "<a href=javascript:ioSwitch('p8','http://www.csmonitor.com/') title=expand id=e8 class=expanded rel=http://www.csmonitor.com/>Christian Science", + "Christian Science", + + "<link rel=\"alternate\" type=\"application/rss+xml\" title=\"San Francisco \" 2008 RSS Feed\" href=\"http://2008.sf.wordcamp.org/feed/\"; />", + "\n", + + "<a href=\" http://www.surgery4was.happyhost.org/video-of-arthroscopic-knee-surgery symptoms.html, heat congestive heart failure <a href=\" http://www.symptoms1bad.happyhost.org/canine";, + "<a href=\" http://www.surgery4was.happyhost.org/video-of-arthroscopic-knee-surgery symptoms.html, heat congestive heart failure <a href=\" http://www.symptoms1bad.happyhost.org/canine";, + + "<a href=\"http://ucblibraries.colorado.edu/how/index.htm\"class=\"pageNavAreaText\";>", + "", + + "<link title=\"^\\\" 21Sta's Blog\" rel=\"search\" type=\"application/opensearchdescription+xml\" href=\"http://21sta.com/blog/inc/opensearch.php\"; />", + "\n", + + "<a href=\"#postcomment\" title=\"\"Leave a comment\";\">?", + "?", + + "<a href='/modern-furniture' ' id='21txt' class='offtab' onMouseout=\"this.className='offtab'; return true;\" onMouseover=\"this.className='ontab'; return true;\">", + "", + + "<a href='http://alievi.wordpress.com/category/01-todos-posts/' style='font-size: 275%; padding: 1px; margin: 1px;' title='01 - Todos Post's (83)'>", + "", + + "The <a href=<a href=\"http://www.advancedmd.com>medical\">http://www.advancedmd.com>medical</a> practice software</a>", + "The <a href=medical\">http://www.advancedmd.com>medical practice software", + + "<a href=\"node/21426\" class=\"clipTitle2\" title=\"Levi.com/BMX 2008 Clip of the Week 29 \"Morgan Wade Leftover Clips\"\">Levi.com/BMX 2008 Clip of the Week 29...", + "Levi.com/BMX 2008 Clip of the Week 29...", + + "<a href=\"printer_friendly.php?branch=&year=&submit=go&screen=\";\">Printer Friendly", + "Printer Friendly", + + "<a href=#\" ondragstart=\"return false\" onclick=\"window.external.AddFavorite('http://www.amazingtextures.com', 'Amazing Textures');return false\" onmouseover=\"window.status='Add to Favorites';return true\">Add to Favorites", + "Add to Favorites", + + "<a href=\"../at_home/at_home_search.html\"../_home/at_home_search.html\">At", + "At", + + "E-mail: <a href=\"\"mailto:[email protected]\"; \">[email protected] </a>", + "E-mail: [email protected] ", + + "<li class=\"farsi\"><a title=\"A'13?\" alt=\"A'13?\" href=\"http://www.america.gov/persian\"; alt=\"\" name=\"A'13?\"A'13? title=\"A'13?\">A'13?</a></li>", + "\nA'13?\n", + + "<li><a href=\"#28\" title=\"Hubert \"Geese\" Ausby\">Hubert \"Geese\" Ausby</a></li>", + "\nHubert \"Geese\" Ausby\n", + + "<href=\"http://anbportal.com/mms/login.asp\";>", + "\n", + + "<a href=\"", + "<a href=\"", + + "<a href=\">", + "", + + "<a rel=\"nofollow\" href=\"http://anissanina31.skyrock.com/1895039493-Hi-tout-le-monde.html\"; title=\" Hi, tout le monde !>#</a>", + "#", + + "<a href=\"http://annunciharleydavidsonusate.myblog.it/\"; title=\"Annunci Moto e Accessori Harley Davidson\" target=\"_blank\"><img src=\"http://annunciharleydavidsonusate.myblog.it/images/Antipixel.gif\"; /></a>", + "", + + "<a href=\"video/addvideo&v=120838887181\" onClick=\"return confirm('Are you sure you want add this video to your profile? If it exists some video in your profile will be overlapped by this video!!')\" \" onmouseover=\"this.className='border2'\" onmouseout=\"this.className=''\">", + "", + + "<a href=#Services & Support>", + "", + + "<input type=\"image\" src=\"http://apologyindex.com/ThemeFiles/83401-72905/images/btn_search.gif\"value=\"Search\"; name=\"Search\" alt=\"Search\" class=\"searchimage\" onclick=\"incom ='&sc=' + document.getElementById('sel').value ; var dt ='&dt=' + document.getElementById('dt').value; var searchKeyword = document.getElementById('q').value ; searchKeyword = searchKeyword.replace(/\\s/g,''); if (searchKeyword.length < 3){alert('Nothing to search. Search keyword should contain atleast 3 chars.'); return false; } var al='&al=' + document.getElementById('advancedlink').style.display ; document.location.href='http://apologyindex.com/search.aspx?q=' + document.getElementById('q').value + incom + dt + al;\" />", + "", + + "<input type=\"image\" src=\"images/afbe.gif\" width=\"22\" height=\"22\" hspace=\"4\" title=\"Add to Favorite\" alt=\"Add to Favorite\"onClick=\" if(window.sidebar){ window.sidebar.addPanel(document.title,location.href,''); }else if(window.external){ window.external.AddFavorite(location.href,document.title); }else if(window.opera&&window.print) { return true; }\">", + "", + + "<area shape=\"rect\" coords=\"12,153,115,305\" href=\"http://statenislandtalk.com/v-web/gallery/Osmundsen-family\"Art's Norwegian Roots in Rogaland\">", + "\n", + + "<a rel=\"nofollow\" href=\"http://arth26.skyrock.com/660188240-bonzai.html\"; title=\"bonza>#", + "#", + + "<a href= >", + "", + + "<ahref=http:..", + "<ahref=http:..", + + "<ahref=http:..>", + "\n", + + "<ahref=\"http://aseigo.bddf.ca/cms/1025\";>A", + "\nA", + + "<a href=\"javascript:calendar_window=window.open('/calendar.aspx?formname=frmCalendar.txtDate','calendar_window','width=154,height=188');calendar_window.focus()\">", + "", + + "<a href=\"/applications/defenseaerospace/19+rackmounts\" title=\"19\" Rackmounts\">", + "", + + "<a href=http://www.azimprimerie.fr/flash/backup/lewes-zip-code/savage-model-110-manual.html title=savage model 110 manual rel=dofollow>", + "", + + "<a class=\"at\" name=\"Lamborghini href=\"http://lamborghini.coolbegin.com\";>Lamborghini /a>", + "Lamborghini /a>", + + "<A href='newslink.php?news_link=http%3A%2F%2Fwww.worldnetdaily.com%2Findex.php%3Ffa%3DPAGE.view%26pageId%3D85729&news_title=Florida QB makes 'John 3:16' hottest Google search Tebow inscribed Bible reference on eye black for championship game' TARGET=_blank>", + "", + + "<a href=/myspace !style='color:#993333'>", + "", + + "<meta name=3DProgId content=3DExcel.Sheet>", + "\n", + + "<link id=3D\"shLink\" href=3D\"PSABrKelly-BADMINTONCupResults08FINAL2008_09_19=_files/sheet004.htm\">", + "\n", + + "<td bgcolor=3D\"#FFFFFF\" nowrap>", + "\n", + + "<a href=\"http://basnect.info/usersearch/\"predicciones-mundiales-2009\".html\";>\"predicciones mundiales 2009\"</a>", + "\"predicciones mundiales 2009\"", + + "<a class=\"comment-link\" href=\"https://www.blogger.com/comment.g?blogID=19402125&postID=114070605958684588\"location.href=https://www.blogger.com/comment.g?blogID=19402125&postID=114070605958684588;>", + "", + + "<a href = \"/videos/Bishop\"/\" title = \"click to see more Bishop\" videos\">Bishop\"</a>", + "Bishop\"", + + "<a href=\"http://bhaa.ie/calendar/event.php?eid=20081203150127531\"\";>BHAA Eircom 2 & 5 miles CC combined start</a>", + "BHAA Eircom 2 & 5 miles CC combined start", + + "<a href=\"http://people.tribe.net/wolfmana\"; onClick='setClick(\"Application[tribe].Person[bb7df210-9dc0-478c-917f-436b896bcb79]\")'\" title=\"Mana\">", + "", + + "<a href=\"http://blog.edu-cyberpg.com/ct.ashx?id=6143c528-080c-4bb2-b765-5ec56c8256d3&url=http%3a%2f%2fwww.gsa.ac.uk%2fmackintoshsketchbook%2f\"\"; eudora=\"autourl\">", + "", + + "<input type=\"text\" value=\"<search here>\">", + "<input type=\"text\" value=\"\n\">", + + "<input type=\"text\" value=\"<search here\">", + "<input type=\"text\" value=\"\n", + + "<input type=\"text\" value=\"search here>\">", + "\">", + + "<input type=\"text\" value=\"<search here>\" onFocus=\"this.value='<search here>'\">", + "", + + "<![if ! IE]>\n<link href=\"http://i.deviantart.com/icons/favicon.png\"; rel=\"shortcut icon\"/>\n<![endif]>", + "\n\n\n", + + "<![if supportMisalignedColumns]>\n<tr height=0 style='display:none'>\n<td width=64 style='width:48pt'></td>\n</tr>\n<![endif]>", + "\n\n\n\n\n\n\n\n" + }; for (int i = 0; i < testGold.Length; i += 2) { AssertHTMLStripsTo(testGold[i], testGold[i + 1], null); @@ -249,7 +417,8 @@ namespace Lucene.Net.Analysis.CharFilters while ((ch = reader.Read()) > 0) { int correction = reader.CorrectOffset(off); - assertTrue("invalid offset correction: " + off + "->" + correction + " for doc of length: " + length, correction <= length); + assertTrue("invalid offset correction: " + off + "->" + correction + " for doc of length: " + length, + correction <= length); off++; } } @@ -284,7 +453,9 @@ namespace Lucene.Net.Analysis.CharFilters [Test] public virtual void TestServerSideIncludes() { - string test = "one<img src=\"image.png\"\n" + " alt = \"Alt: <!--#echo var='${IMAGE_CAPTION:<!--comment-->\\'Comment\\'}' -->\"\n\n" + " title=\"Title: <!--#echo var=\"IMAGE_CAPTION\"-->\">two"; + string test = "one<img src=\"image.png\"\n" + + " alt = \"Alt: <!--#echo var='${IMAGE_CAPTION:<!--comment-->\\'Comment\\'}' -->\"\n\n" + + " title=\"Title: <!--#echo var=\"IMAGE_CAPTION\"-->\">two"; string gold = "onetwo"; AssertHTMLStripsTo(test, gold, null); @@ -317,7 +488,11 @@ namespace Lucene.Net.Analysis.CharFilters [Test] public virtual void TestStyle() { - string test = "one<style type=\"text/css\">\n" + "<!--\n" + "@import url('http://www.lasletrasdecanciones.com/css.css');\n" + "-->\n" + "</style>two"; + string test = "one<style type=\"text/css\">\n" + + "<!--\n" + + "@import url('http://www.lasletrasdecanciones.com/css.css');\n" + + "-->\n" + + "</style>two"; string gold = "one\ntwo"; AssertHTMLStripsTo(test, gold, null); } @@ -334,7 +509,13 @@ namespace Lucene.Net.Analysis.CharFilters [Test] public virtual void TestBR() { - string[] testGold = new string[] { "one<BR />two<br>three", "one\ntwo\nthree", "one<BR some stuff here too>two</BR>", "one\ntwo\n" }; + string[] testGold = { + "one<BR />two<br>three", + "one\ntwo\nthree", + + "one<BR some stuff here too>two</BR>", + "one\ntwo\n" + }; for (int i = 0; i < testGold.Length; i += 2) { AssertHTMLStripsTo(testGold[i], testGold[i + 1], null); @@ -361,13 +542,37 @@ namespace Lucene.Net.Analysis.CharFilters public virtual void TestCDATA() { int maxNumElems = 100; - string randomHtmlishString1 = TestUtil.RandomHtmlishString(Random(), maxNumElems).Replace(">", " ").replaceFirst("^--", "__"); // Don't create a comment (disallow "<!--") and don't include a closing ">" + string randomHtmlishString1 // Don't create a comment (disallow "<!--") and don't include a closing ">" + = TestUtil.RandomHtmlishString(Random(), maxNumElems).Replace(">", " ").replaceFirst("^--", "__"); string closedAngleBangNonCDATA = "<!" + randomHtmlishString1 + "-[CDATA[&]]>"; - string randomHtmlishString2 = TestUtil.RandomHtmlishString(Random(), maxNumElems).Replace(">", " ").replaceFirst("^--", "__"); // Don't create a comment (disallow "<!--") and don't include a closing ">" + string randomHtmlishString2 // Don't create a comment (disallow "<!--") and don't include a closing ">" + = TestUtil.RandomHtmlishString(Random(), maxNumElems).Replace(">", " ").replaceFirst("^--", "__"); string unclosedAngleBangNonCDATA = "<!" + randomHtmlishString1 + "-[CDATA["; - string[] testGold = new string[] { "one<![CDATA[<one><two>three<four></four></two></one>]]>two", "one<one><two>three<four></four></two></one>two", "one<![CDATA[two<![CDATA[three]]]]><![CDATA[>four]]>five", "onetwo<![CDATA[three]]>fourfive", "<! [CDATA[&]]>", "", "<! [CDATA[&] ] >", "", "<! [CDATA[&]]", "<! [CDATA[&]]", "<!\u2009[CDATA[&]]>", "", "<!\u2009[CDATA[&]\u2009]\u2009>", "", "<!\u2009[CDATA[&]\u2009]\u2009", "<!\u2009[CDATA[&]\u2009]\u2009", closedAngleBangNonCDATA, "", "<![CDATA[", "", "<![CDATA[<br>", "<br>", "<![CDATA[<br>]]", "<br>]]", "<![CDATA[<br>]]>", "<br>", "<![CDATA[<br>] ] >", "<br>] ] >", "<![CDATA[<br>]\u2009]\u2009>", "<br>]\u2009]\u2009>", "<!\u2009[CDATA[", "<!\u2009[CDATA[", unclosedAngleBangNonCDATA, unclosedAngleBangNonCDATA }; + string[] testGold = { + "one<![CDATA[<one><two>three<four></four></two></one>]]>two", + "one<one><two>three<four></four></two></one>two", + + "one<![CDATA[two<![CDATA[three]]]]><![CDATA[>four]]>five", + "onetwo<![CDATA[three]]>fourfive", + + "<! [CDATA[&]]>", "", + "<! [CDATA[&] ] >", "", + "<! [CDATA[&]]", "<! [CDATA[&]]", // unclosed angle bang - all input is output + "<!\u2009[CDATA[&]]>", "", + "<!\u2009[CDATA[&]\u2009]\u2009>", "", + "<!\u2009[CDATA[&]\u2009]\u2009", "<!\u2009[CDATA[&]\u2009]\u2009", // unclosed angle bang - all input is output + closedAngleBangNonCDATA, "", + "<![CDATA[", "", + "<![CDATA[<br>", "<br>", + "<![CDATA[<br>]]", "<br>]]", + "<![CDATA[<br>]]>", "<br>", + "<![CDATA[<br>] ] >", "<br>] ] >", + "<![CDATA[<br>]\u2009]\u2009>", "<br>]\u2009]\u2009>", + "<!\u2009[CDATA[", "<!\u2009[CDATA[", + unclosedAngleBangNonCDATA, unclosedAngleBangNonCDATA + }; for (int i = 0; i < testGold.Length; i += 2) { AssertHTMLStripsTo(testGold[i], testGold[i + 1], null); @@ -525,5 +730,4 @@ namespace Lucene.Net.Analysis.CharFilters assertEquals("'" + builder.ToString() + "' is not equal to '" + gold + "'", gold, builder.ToString()); } } - } \ No newline at end of file
![http://annunciharleydavidsonusate.myblog.it/images/Antipixel.gif\"](http://annunciharleydavidsonusate.myblog.it/images/Antipixel.gif\"){kind=link}
![http://apologyindex.com/ThemeFiles/83401-72905/images/btn_search.gif\"value=\"Search\"](http://apologyindex.com/ThemeFiles/83401-72905/images/btn_search.gif\"value=\"Search\"){kind=link}
![http://i.deviantart.com/icons/favicon.png\"](http://i.deviantart.com/icons/favicon.png\"){kind=link}