[Jtidy-cvs] SF.net SVN: jtidy:[1208] branches/CodeUpdateAndJava5/src/main/java/org/w3c/ tidy

[<ad...@us...>]

Revision: 1208
          http://jtidy.svn.sourceforge.net/jtidy/?rev=1208&view=rev
Author:   aditsu
Date:     2010-11-17 23:53:53 +0000 (Wed, 17 Nov 2010)

Log Message:
-----------
partial fix for test 445557 - modified encodingError to use decimal character values for invalid_char messages, and added special handling for characters between 127 and 160 in readChar

Modified Paths:
--------------
    branches/CodeUpdateAndJava5/src/main/java/org/w3c/tidy/Report.java
    branches/CodeUpdateAndJava5/src/main/java/org/w3c/tidy/StreamInJavaImpl.java

Modified: branches/CodeUpdateAndJava5/src/main/java/org/w3c/tidy/Report.java
===================================================================
--- branches/CodeUpdateAndJava5/src/main/java/org/w3c/tidy/Report.java	2010-11-17 23:46:35 UTC (rev 1207)
+++ branches/CodeUpdateAndJava5/src/main/java/org/w3c/tidy/Report.java	2010-11-17 23:53:53 UTC (rev 1208)
@@ -480,21 +480,11 @@
                 break;
             case VENDOR_SPECIFIC_CHARS:
                 lexer.badChars |= BC_VENDOR_SPECIFIC_CHARS;
-                messageLexer(
-                    code.code(),
-                    lexer,
-                    Level.WARNING,
-                    "invalid_char",
-                    replaceMode, buf);
+                messageLexer(code.code(), lexer, Level.WARNING, "invalid_char", replaceMode, c);
                 break;
             case INVALID_SGML_CHARS:
                 lexer.badChars |= BC_INVALID_SGML_CHARS;
-                messageLexer(
-                    code.code(),
-                    lexer,
-                    Level.WARNING,
-                    "invalid_char",
-                    replaceMode, buf);
+                messageLexer(code.code(), lexer, Level.WARNING, "invalid_char", replaceMode, c);
                 break;
             case INVALID_UTF8:
                 lexer.badChars |= BC_INVALID_UTF8;

Modified: branches/CodeUpdateAndJava5/src/main/java/org/w3c/tidy/StreamInJavaImpl.java
===================================================================
--- branches/CodeUpdateAndJava5/src/main/java/org/w3c/tidy/StreamInJavaImpl.java	2010-11-17 23:46:35 UTC (rev 1207)
+++ branches/CodeUpdateAndJava5/src/main/java/org/w3c/tidy/StreamInJavaImpl.java	2010-11-17 23:53:53 UTC (rev 1208)
@@ -119,6 +119,8 @@
     private int tabsize;
 
     private int tabs;
+    
+    private Lexer lexer;
 
     /**
      * Instantiates a new StreamInJavaImpl.
@@ -259,6 +261,40 @@
             	return c;
             }
         }
+        
+        /* produced e.g. as a side-effect of smart quotes in Word */
+        /* but can't happen if using MACROMAN encoding */
+        if (127 < c && c < 160) {
+        	int c1 = 0;
+        	int replMode = Report.DISCARDED_CHAR;
+        	final String enc = lexer.configuration.getInCharEncodingName();
+        	final String repl = lexer.configuration.getReplacementCharEncoding();
+        	boolean isVendorChar = ("WIN1252".equals(enc) || "MACROMAN".equals(enc));
+        	boolean isWinChar = ("WIN1252".equals(enc) || "WIN1252".equals(repl));
+        	boolean isMacChar = ("MACROMAN".equals(enc) || "MACROMAN".equals(repl));
+        	
+        	/* set error position just before offending character */
+        	lexer.lines = curline;
+        	lexer.columns = curcol;
+        	
+        	if (isWinChar) {
+        		c1 = EncodingUtils.decodeWin1252(c);
+        	}
+        	else if (isMacChar) {
+        		c1 = EncodingUtils.decodeMacRoman(c);
+        	}
+        	if (c1 != 0) {
+        		replMode = Report.REPLACED_CHAR;
+        	}
+        	
+        	if (c1 == 0 && isVendorChar) {
+        		lexer.report.encodingError(lexer, ErrorCode.VENDOR_SPECIFIC_CHARS, c, replMode);
+        	}
+        	else if (!isVendorChar) {
+        		lexer.report.encodingError(lexer, ErrorCode.INVALID_SGML_CHARS, c, replMode);
+        	}
+        	c = c1;
+        }
 
         this.curcol++;
         return c;
@@ -333,9 +369,8 @@
     /**
      * @see org.w3c.tidy.StreamIn#setLexer(org.w3c.tidy.Lexer)
      */
-    public void setLexer(Lexer lexer)
-    {
-        // unused in the java implementation
+    public void setLexer(final Lexer lexer) {
+    	this.lexer = lexer;
     }
 
 }
\ No newline at end of file


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.