Fix to fine-tune URI link for Ruby QUO,QUOxN

Author: idodeclare

src/org/opensolaris/opengrok/analysis/ruby/RubyLexHelper.java

@@ -32,6 +32,7 @@
 import org.opensolaris.opengrok.util.StringUtils;
 import org.opensolaris.opengrok.web.HtmlConsts;
 import org.opensolaris.opengrok.analysis.JFlexJointLexer;
+import org.opensolaris.opengrok.util.RegexUtils;
 
 /**
  * Represents an API for object's using {@link RubyLexHelper}
@@ -97,6 +98,12 @@ class RubyLexHelper implements Resettable {
      */
     private int nendbrace;
 
+    /**
+     * When matching a quoting construct, a Pattern to identify collateral
+     * capture characters is stored.
+     */
+    private Pattern collateralCapture;
+
     public RubyLexHelper(int qUO, int qUOxN, int qUOxL, int qUOxLxN,
 	    RubyLexer lexer,
         int hERE, int hERExN, int hEREin, int hEREinxN) {
@@ -119,6 +126,7 @@ public RubyLexHelper(int qUO, int qUOxN, int qUOxL, int qUOxLxN,
      */
     @Override
     public void reset() {
+        collateralCapture = null;
         endqchar = '\0';
         if (hereSettings != null) hereSettings.clear();
         nendbrace = 0;
@@ -191,6 +199,7 @@ public void qop(boolean doWrite, String capture, int namelength,
             postop = capture.substring(qopname.length());
         }
         nendqchar = 1;
+        collateralCapture = null;
 
         char opc = postop.charAt(0);
         setEndQuoteChar(opc);
@@ -212,9 +221,9 @@ public void setState(String postop, boolean nointerp) {
         boolean nolink = false;
 
         // "no link" for values in the rules for "string links" if `postop'
-        // is file- or URI-like.
-        if (StringUtils.startsWithFnameChars(postop) ||
-            StringUtils.startsWithURIChars(postop)) {
+        // starts path-like or with the e-mail delimiter.
+        if (StringUtils.startsWithFpathChar(postop) ||
+            postop.startsWith("@")) {
             nolink = true;
         }
 
@@ -491,6 +500,26 @@ public int nameLength(String capture) {
         return len;
     }
 
+    /**
+     * Gets a pattern to match the collateral capture for the current quoting
+     * state or null if there is no active quoting state.
+     * @return a defined pattern or null
+     */
+    public Pattern getCollateralCapturePattern() {
+        if (endqchar == '\0') return null;
+        if (collateralCapture != null) return collateralCapture;
+
+        StringBuilder patb = new StringBuilder("[");
+        patb.append(Pattern.quote(String.valueOf(endqchar)));
+        if (nestqchar != '\0') {
+            patb.append(Pattern.quote(String.valueOf(nestqchar)));
+        }
+        patb.append("]");
+        patb.append(RegexUtils.getNotFollowingEscapePattern());
+        collateralCapture = Pattern.compile(patb.toString());
+        return collateralCapture;
+    }
+
     private class HereDocSettings {
         private final String terminator;
         private final int state;

src/org/opensolaris/opengrok/analysis/ruby/RubyProductions.lexh

@@ -504,15 +504,30 @@ Here_EOF3 = [\`][^\r\n\`]*[\`]
         }
     }
 
+    {FNameChar}+ "@" {FNameChar}+ "." {FNameChar}+    {
+        if (takeAllContent()) {
+            writeEMailAddress(yytext());
+        }
+    }
+}
+
+<SCOMMENT, POD, HERE, HERExN, HEREin, HEREinxN> {
     {BrowseableURI}    {
         if (takeAllContent()) {
-            appendLink(yytext(), true);
+            appendLink(yytext(), true, null);
         }
+        // no skipLink() needed except in QUO* states
     }
+}
 
-    {FNameChar}+ "@" {FNameChar}+ "." {FNameChar}+    {
+<QUO, QUOxN> {
+    {BrowseableURI}    {
+        String capture = yytext();
+        Pattern collateralCapture = h.getCollateralCapturePattern();
         if (takeAllContent()) {
-            writeEMailAddress(yytext());
+            appendLink(capture, true, collateralCapture);
+        } else {
+            skipLink(capture, collateralCapture);
         }
     }
 }

src/org/opensolaris/opengrok/analysis/ruby/RubySymbolTokenizer.lex

@@ -30,7 +30,9 @@ package org.opensolaris.opengrok.analysis.ruby;
 
 import java.io.IOException;
 import java.util.Stack;
+import java.util.regex.Pattern;
 import org.opensolaris.opengrok.analysis.JFlexTokenizer;
+import org.opensolaris.opengrok.util.StringUtils;
 import org.opensolaris.opengrok.web.HtmlConsts;
 import org.opensolaris.opengrok.web.Util;
 
@@ -152,9 +154,14 @@ import org.opensolaris.opengrok.web.Util;
 
     protected void appendProject() { /* noop */ }
 
-    protected void appendLink(String s, boolean b) { /* noop */ }
+    protected void appendLink(String s, boolean b, Pattern p) { /* noop */ }
 
     protected void writeEMailAddress(String s) { /* noop */ }
+
+    protected void skipLink(String url, Pattern p) {
+        int n = StringUtils.countPushback(url, p);
+        if (n > 0) yypushback(n);
+    }
 %}
 
 %include Common.lexh

src/org/opensolaris/opengrok/analysis/ruby/RubyXref.lex

@@ -31,6 +31,7 @@ package org.opensolaris.opengrok.analysis.ruby;
 import java.io.IOException;
 import java.io.Reader;
 import java.util.Stack;
+import java.util.regex.Pattern;
 import org.opensolaris.opengrok.analysis.JFlexXref;
 import org.opensolaris.opengrok.web.HtmlConsts;
 import org.opensolaris.opengrok.web.Util;
@@ -138,6 +139,8 @@ import org.opensolaris.opengrok.web.Util;
     }
 
     protected String getUrlPrefix() { return urlPrefix; }
+
+    protected void skipLink(String s, Pattern p) { /* noop */ }
 %}
 
 %include Common.lexh

test/org/opensolaris/opengrok/analysis/ruby/ruby_xrefres.html

@@ -182,5 +182,6 @@
 <a class="l" name="174" href="#174">174</a>  <b>system</b>(<span class="s">&quot;stty -raw echo&quot;</span>)
 <a class="l" name="175" href="#175">175</a><b>end</b>
 <a class="l" name="176" href="#176">176</a><b>print</b> <span class="s">&quot;\n&quot;</span>
-<a class="l" name="177" href="#177">177</a></body>
+<a class="l" name="177" href="#177">177</a><b>print</b> <span class="s">&apos;<a href="http://example.com">http://example.com</a>&apos;</span>
+<a class="l" name="178" href="#178">178</a></body>
 </html>

test/org/opensolaris/opengrok/analysis/ruby/sample.rb

@@ -174,3 +174,4 @@ def right
   system("stty -raw echo")
 end
 print "\n"
+print 'http://example.com'