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

Author: idodeclare

src/org/opensolaris/opengrok/analysis/perl/PerlLexHelper.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 PerlLexHelper}
@@ -113,6 +114,12 @@ class PerlLexHelper implements Resettable {
      */
     private boolean waitq;
 
+    /**
+     * When matching a quoting construct, a Pattern to identify collateral
+     * capture characters is stored.
+     */
+    private Pattern collateralCapture;
+
     public PerlLexHelper(int qUO, int qUOxN, int qUOxL, int qUOxLxN,
         PerlLexer lexer,
         int hERE, int hERExN, int hEREin, int hEREinxN) {
@@ -135,6 +142,7 @@ public PerlLexHelper(int qUO, int qUOxN, int qUOxL, int qUOxLxN,
      */
     @Override
     public void reset() {
+        collateralCapture = null;
         endqchar = '\0';
         if (hereSettings != null) hereSettings.clear();
         nendqchar = 0;
@@ -227,6 +235,7 @@ public void qop(boolean doWrite, String capture, int namelength,
         }
         waitq = false;
         nendqchar = 1;
+        collateralCapture = null;
 
         switch (qopname) {
             case "tr":
@@ -265,9 +274,9 @@ public void setState(String ltpostop, boolean nointerp) {
         boolean nolink = false;
 
         // "no link" for values in the rules for "string links" if `ltpostop'
-        // is file- or URI-like.
-        if (StringUtils.startsWithFnameChars(ltpostop) ||
-            StringUtils.startsWithURIChars(ltpostop)) {
+        // starts path-like or with the e-mail delimiter.
+        if (StringUtils.startsWithFpathChar(ltpostop) ||
+            ltpostop.startsWith("@")) {
             nolink = true;
         }
 
@@ -604,6 +613,26 @@ public void specialID(String capture) throws IOException {
         }
     }
 
+    /**
+     * 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;
+    }
+
     class HereDocSettings {
         private final String terminator;
         private final int state;

src/org/opensolaris/opengrok/analysis/perl/PerlProductions.lexh

@@ -525,15 +525,30 @@ Mpunc2IN = ([!=]"~" | [\:\?\=\+\-\<\>] | "=="|"!="|"<="|">="|"<=>"|"=>")
         }
     }
 
+    {FNameChar}+ "@" {FNameChar}+ "." {FNameChar}+    {
+        if (takeAllContent()) {
+            writeEMailAddress(yytext());
+        }
+    }
+}
+
+<SCOMMENT, POD, FMT, 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(yytext(), collateralCapture);
         }
     }
 }

src/org/opensolaris/opengrok/analysis/perl/PerlSymbolTokenizer.lex

@@ -29,7 +29,9 @@
 package org.opensolaris.opengrok.analysis.perl;
 
 import java.io.IOException;
+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;
 
@@ -137,9 +139,14 @@ 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/perl/PerlXref.lex

@@ -30,6 +30,7 @@ package org.opensolaris.opengrok.analysis.perl;
 
 import java.io.IOException;
 import java.io.Reader;
+import java.util.regex.Pattern;
 import org.opensolaris.opengrok.analysis.JFlexXref;
 import org.opensolaris.opengrok.web.HtmlConsts;
 import org.opensolaris.opengrok.web.Util;
@@ -127,6 +128,8 @@ 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/perl/sample.pl

@@ -382,3 +382,4 @@ ($)
 "left",  "middle", "right"
 .
 print "1\n";
+print 'http://example.com';

test/org/opensolaris/opengrok/analysis/perl/samplexrefres.html

@@ -389,5 +389,6 @@
 <a class="l" name="382" href="#382">382</a><span class="s">&quot;left&quot;,  &quot;middle&quot;, &quot;right&quot;</span>
 <a class="l" name="383" href="#383">383</a><span class="s">.</span>
 <a class="l" name="384" href="#384">384</a><b>print</b> <span class="s">&quot;1\n&quot;</span>;
-<a class="l" name="385" href="#385">385</a></body>
+<a class="l" name="385" href="#385">385</a><b>print</b> <span class="s">&apos;<a href="http://example.com">http://example.com</a>&apos;</span>;
+<a class="l" name="386" href="#386">386</a></body>
 </html>