Fix dangling Erlang span by subclassing JFlexXrefSimple

idodeclare · idodeclare · commit 2b0f4fdf48dc · 2017-12-04T20:00:08.000-06:00
Also:
- Make ErlangXref case-sensitive, revising {File} to match
  any case.
- Escape all HTML special characters in ErlangXref.
- Use Common.xref in ErlangSymbolTokenizer.
diff --git a/src/org/opensolaris/opengrok/analysis/erlang/ErlangSymbolTokenizer.lex b/src/org/opensolaris/opengrok/analysis/erlang/ErlangSymbolTokenizer.lex
@@ -45,6 +45,7 @@ super(in);
 
 %state STRING COMMENT QATOM
 
+%include Common.lexh
 %include Erlang.lexh
 %%
 
@@ -80,19 +81,21 @@ super(in);
  }
 
 <STRING> {
+ \\[\"\\]    {}
  \"     { yybegin(YYINITIAL); }
- \\\\ | \\\"   {}
 }
 
 <QATOM> {
+ \\[\'\\]    {}
  \'     { yybegin(YYINITIAL); }
- \\\\ | \\\'   {}
 }
 
 <COMMENT> {
- \n    { yybegin(YYINITIAL);}
+ {EOL}    { yybegin(YYINITIAL);}
 }
 
 <YYINITIAL, STRING, QATOM, COMMENT> {
+{ErlangWhiteSpace}    {}
+
 [^]    {}
 }
diff --git a/src/org/opensolaris/opengrok/analysis/erlang/ErlangXref.lex b/src/org/opensolaris/opengrok/analysis/erlang/ErlangXref.lex
@@ -27,18 +27,18 @@
  */
 
 package org.opensolaris.opengrok.analysis.erlang;
-import org.opensolaris.opengrok.analysis.JFlexXref;
+import org.opensolaris.opengrok.analysis.JFlexXrefSimple;
 import java.io.IOException;
 import java.io.Writer;
 import java.io.Reader;
+import org.opensolaris.opengrok.web.HtmlConsts;
 import org.opensolaris.opengrok.web.Util;
 
 %%
 %public
 %class ErlangXref
-%extends JFlexXref
+%extends JFlexXrefSimple
 %unicode
-%ignorecase
 %int
 %include CommonXref.lexh
 %{
@@ -51,7 +51,9 @@ import org.opensolaris.opengrok.web.Util;
 
 IncludeDirective = (include|include_lib)
 
-File = [a-zA-Z]{FNameChar}* "." ("erl"|"hrl"|"app"|"asn"|"yrl"|"asn1"|"xml"|"html")
+File = [a-zA-Z]{FNameChar}* "." ([Ee][Rr][Ll] | [Hh][Rr][Ll] | [Aa][Pp][Pp] |
+    [Aa][Ss][Nn] | [Yy][Rr][Ll] | [Aa][Ss][Nn][1] | [Xx][Mm][Ll] |
+    [Hh][Tt][Mm][Ll]?)
 
 %state  STRING COMMENT QATOM
 
@@ -63,7 +65,9 @@ File = [a-zA-Z]{FNameChar}* "." ("erl"|"hrl"|"app"|"asn"|"yrl"|"asn1"|"xml"|"htm
 <YYINITIAL>{
 
 "?" {Identifier} {  // Macros
-    out.write("<span class=\"xm\">"); out.write(yytext()); out.write("</span>");
+    disjointSpan(HtmlConsts.MACRO_CLASS);
+    out.write(yytext());
+    disjointSpan(null);
 }
 
 {Identifier} {
@@ -98,41 +102,59 @@ File = [a-zA-Z]{FNameChar}* "." ("erl"|"hrl"|"app"|"asn"|"yrl"|"asn1"|"xml"|"htm
     writeSymbol(id, Consts.modules_kwd, yyline);
 }
 
-{ErlInt}        { out.write("<span class=\"n\">"); out.write(yytext()); out.write("</span>"); }
-{Number}        { out.write("<span class=\"n\">"); out.write(yytext()); out.write("</span>"); }
+{ErlInt} |
+    {Number}    {
+    disjointSpan(HtmlConsts.NUMBER_CLASS);
+    out.write(yytext());
+    disjointSpan(null);
+}
+
+ \"     {
+    pushSpan(STRING, HtmlConsts.STRING_CLASS);
+    out.write(htmlize(yytext()));
+ }
 
- \"     { yybegin(STRING);out.write("<span class=\"s\">\"");}
- \'     { yybegin(QATOM);out.write("<span class=\"s\">\'");}
- "%"   { yybegin(COMMENT);out.write("<span class=\"c\">%");}
+ \'     {
+    pushSpan(QATOM, HtmlConsts.STRING_CLASS);
+    out.write(htmlize(yytext()));
+ }
+
+ "%"    {
+    pushSpan(COMMENT, HtmlConsts.COMMENT_CLASS);
+    out.write(yytext());
+ }
 }
 
 <STRING> {
- \"     { yybegin(YYINITIAL); out.write("\"</span>"); }
- \\\\   { out.write("\\\\"); }
- \\\"   { out.write("\\\""); }
+ \\[\"\\]    { out.write(htmlize(yytext())); }
+ \"    {
+    out.write(htmlize(yytext()));
+    yypop();
+ }
 }
 
 <QATOM> {
- \'     { yybegin(YYINITIAL); out.write("\'</span>"); }
- \\\\   { out.write("\\\\"); }
- \\\'   { out.write("\\\'"); }
+ \\[\'\\]    { out.write(htmlize(yytext())); }
+ \'    {
+    out.write(htmlize(yytext()));
+    yypop();
+ }
 }
 
 <COMMENT> {
   {ErlangWhspChar}*{EOL} {
-    yybegin(YYINITIAL); out.write("</span>");
+    yypop();
     startNewLine();
   }
 }
 
 <YYINITIAL, STRING, COMMENT, QATOM> {
-"&"     {out.write( "&amp;");}
-"<"     {out.write( "&lt;");}
-">"     {out.write( "&gt;");}
+[&<>\'\"]    { out.write(htmlize(yytext())); }
 {ErlangWhspChar}*{EOL}      { startNewLine(); }
  {ErlangWhiteSpace}   { out.write(yytext()); }
  [!-~]  { out.write(yycharat(0)); }
- .      { writeUnicodeChar(yycharat(0)); }
+
+ [^]    { writeUnicodeChar(yycharat(0)); }
 }
 
 <STRING, COMMENT, STRING, QATOM> {
diff --git a/src/org/opensolaris/opengrok/web/HtmlConsts.java b/src/org/opensolaris/opengrok/web/HtmlConsts.java
@@ -39,6 +39,7 @@ public class HtmlConsts {
     public static final String AUTHOR_CLASS = "a";
     public static final String COMMENT_CLASS = "c";
     public static final String DELETED_CLASS = "d";
+    public static final String MACRO_CLASS = "xm";
     public static final String NUMBER_CLASS = "n";
     public static final String STRING_CLASS = "s";
 
diff --git a/test/org/opensolaris/opengrok/analysis/erlang/sample_xref.html b/test/org/opensolaris/opengrok/analysis/erlang/sample_xref.html
@@ -12,7 +12,7 @@
 <a class="l" name="4" href="#4">4</a><span class="c">%% purpose with or without fee is hereby granted, provided that the above</span>
 <a class="l" name="5" href="#5">5</a><span class="c">%% copyright notice and this permission notice appear in all copies.</span>
 <a class="l" name="6" href="#6">6</a><span class="c">%%</span>
-<a class="l" name="7" href="#7">7</a><span class="c">%% THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES</span>
+<a class="l" name="7" href="#7">7</a><span class="c">%% THE SOFTWARE IS PROVIDED &quot;AS IS&quot; AND THE AUTHOR DISCLAIMS ALL WARRANTIES</span>
 <a class="l" name="8" href="#8">8</a><span class="c">%% WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF</span>
 <a class="l" name="9" href="#9">9</a><span class="c">%% MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR</span>
 <a class="hl" name="10" href="#10">10</a><span class="c">%% ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES</span>
@@ -44,10 +44,10 @@
 <a class="l" name="36" href="#36">36</a>					<a href="/source/s?defs=Transport" class="intelliWindow-symbol" data-definition-place="undefined-in-file">Transport</a>:<a href="/source/s?defs=close" class="intelliWindow-symbol" data-definition-place="undefined-in-file">close</a>(<a href="/source/s?defs=CSocket" class="intelliWindow-symbol" data-definition-place="undefined-in-file">CSocket</a>)
 <a class="l" name="37" href="#37">37</a>			<b>end</b>;
 <a class="l" name="38" href="#38">38</a>		<span class="c">%% Reduce the accept rate if we run out of file descriptors.</span>
-<a class="l" name="39" href="#39">39</a>		<span class="c">%% We can't accept anymore anyway, so we might as well wait</span>
+<a class="l" name="39" href="#39">39</a>		<span class="c">%% We can&apos;t accept anymore anyway, so we might as well wait</span>
 <a class="hl" name="40" href="#40">40</a>		<span class="c">%% a little for the situation to resolve itself.</span>
 <a class="l" name="41" href="#41">41</a>		{<a href="/source/s?defs=error" class="intelliWindow-symbol" data-definition-place="undefined-in-file">error</a>, <a href="/source/s?defs=emfile" class="intelliWindow-symbol" data-definition-place="undefined-in-file">emfile</a>} -&gt;
-<a class="l" name="42" href="#42">42</a>			<a href="/source/s?defs=error_logger" class="intelliWindow-symbol" data-definition-place="undefined-in-file">error_logger</a>:<a href="/source/s?defs=warning_msg" class="intelliWindow-symbol" data-definition-place="undefined-in-file">warning_msg</a>(<span class="s">"Ranch acceptor reducing accept rate: out of file descriptors~n"</span>),
+<a class="l" name="42" href="#42">42</a>			<a href="/source/s?defs=error_logger" class="intelliWindow-symbol" data-definition-place="undefined-in-file">error_logger</a>:<a href="/source/s?defs=warning_msg" class="intelliWindow-symbol" data-definition-place="undefined-in-file">warning_msg</a>(<span class="s">&quot;Ranch acceptor reducing accept rate: out of file descriptors~n&quot;</span>),
 <a class="l" name="43" href="#43">43</a>			<b>receive</b> <b>after</b> <span class="n">0xFF</span> -&gt; <a href="/source/s?defs=ok" class="intelliWindow-symbol" data-definition-place="undefined-in-file">ok</a> <b>end</b>;
 <a class="l" name="44" href="#44">44</a>		<span class="c">%% We want to crash if the listening socket got closed.</span>
 <a class="l" name="45" href="#45">45</a>		{<a href="/source/s?defs=error" class="intelliWindow-symbol" data-definition-place="undefined-in-file">error</a>, <a href="/source/s?defs=Reason" class="intelliWindow-symbol" data-definition-place="undefined-in-file">Reason</a>} <b>when</b> <a href="/source/s?defs=Reason" class="intelliWindow-symbol" data-definition-place="undefined-in-file">Reason</a> =/= <a href="/source/s?defs=closed" class="intelliWindow-symbol" data-definition-place="undefined-in-file">closed</a> -&gt;
@@ -59,7 +59,7 @@
 <a class="l" name="51" href="#51">51</a><a class="xf" name="flush"/><a href="/source/s?refs=flush" class="xf intelliWindow-symbol" data-definition-place="def">flush</a>() -&gt;
 <a class="l" name="52" href="#52">52</a>	<b>receive</b> <a href="/source/s?defs=Msg" class="intelliWindow-symbol" data-definition-place="undefined-in-file">Msg</a> -&gt;
 <a class="l" name="53" href="#53">53</a>		<a href="/source/s?defs=error_logger" class="intelliWindow-symbol" data-definition-place="undefined-in-file">error_logger</a>:<a href="/source/s?defs=error_msg" class="intelliWindow-symbol" data-definition-place="undefined-in-file">error_msg</a>(
-<a class="l" name="54" href="#54">54</a>			<span class="s">"Ranch acceptor received unexpected message: ~p~n"</span>,
+<a class="l" name="54" href="#54">54</a>			<span class="s">&quot;Ranch acceptor received unexpected message: ~p~n&quot;</span>,
 <a class="l" name="55" href="#55">55</a>			[<a href="/source/s?defs=Msg" class="intelliWindow-symbol" data-definition-place="undefined-in-file">Msg</a>]),
 <a class="l" name="56" href="#56">56</a>		<a class="d intelliWindow-symbol" href="#flush" data-definition-place="defined-in-file">flush</a>()
 <a class="l" name="57" href="#57">57</a>	<b>after</b> <span class="n">0</span> -&gt;
diff --git a/test/org/opensolaris/opengrok/analysis/erlang/truncated_xref.html b/test/org/opensolaris/opengrok/analysis/erlang/truncated_xref.html
@@ -5,5 +5,5 @@
       class="xref">
 <head>
 <title>sampleFile - OpenGrok cross reference for /sampleFile</title></head><body>
-<a class="l" name="1" href="#1">1</a>  <span class="s">"Oops this is trun</span></body>
+<a class="l" name="1" href="#1">1</a>  <span class="s">&quot;Oops this is trun</span></body>
 </html>

Original file line number	Diff line number	Diff line change
`@@ -45,6 +45,7 @@ super(in);`
`45`	`45`
`46`	`46`	`%state STRING COMMENT QATOM`
`47`	`47`
	`48`	`+%include Common.lexh`
`48`	`49`	`%include Erlang.lexh`
`49`	`50`	`%%`
`50`	`51`
`@@ -80,19 +81,21 @@ super(in);`
`80`	`81`	`}`
`81`	`82`
`82`	`83`	`<STRING> {`
	`84`	`+ \\[\"\\] {}`
`83`	`85`	`\" { yybegin(YYINITIAL); }`
`84`		`- \\\\ \| \\\" {}`
`85`	`86`	`}`
`86`	`87`
`87`	`88`	`<QATOM> {`
	`89`	`+ \\[\'\\] {}`
`88`	`90`	`\' { yybegin(YYINITIAL); }`
`89`		`- \\\\ \| \\\' {}`
`90`	`91`	`}`
`91`	`92`
`92`	`93`	`<COMMENT> {`
`93`		`- \n { yybegin(YYINITIAL);}`
	`94`	`+ {EOL} { yybegin(YYINITIAL);}`
`94`	`95`	`}`
`95`	`96`
`96`	`97`	`<YYINITIAL, STRING, QATOM, COMMENT> {`
	`98`	`+{ErlangWhiteSpace} {}`
	`99`	`+`
`97`	`100`	`[^] {}`
`98`	`101`	`}`