Merge branch 'master' into attrdict_phase1

Bump version to 5.1.38

CMakeLists.txt

@@ -336,6 +336,9 @@ set(CPACK_DEBIAN_PACKAGE_HOMEPAGE "http://www.html-tidy.org/")
 #set(CPACK_DEBIAN_PACKAGE_DEPENDS "libc")
 set(CPACK_DEBIAN_PACKAGE_SECTION "Libraries")
 
+## RPM config
+set(CPACK_RPM_EXCLUDE_FROM_AUTO_FILELIST_ADDITION "/usr/share/man" "/usr/share/man/man1")
+
 set(CPACK_SOURCE_IGNORE_FILES "${CMAKE_CURRENT_SOURCE_DIR}/test/;${CMAKE_CURRENT_SOURCE_DIR}/build/;${CMAKE_CURRENT_SOURCE_DIR}/.git/")
 
 if (NOT WIN32 AND NOT APPLE)

include/tidy.h

@@ -744,9 +744,13 @@ TIDY_EXPORT int TIDY_CALL         tidySaveStdout( TidyDoc tdoc );
 /** Save to given TidyBuffer object */
 TIDY_EXPORT int TIDY_CALL         tidySaveBuffer( TidyDoc tdoc, TidyBuffer* buf );
 
-/** Save document to application buffer.  If buffer is not big enough,
+/** Save document to application buffer.  If TidyShowMarkup and
-**  ENOMEM will be returned and the necessary buffer size will be placed
+**  the document has no errors, or TidyForceOutput, the current 
-**  in *buflen.
+**  document, per the current configuration, will be Pretty Printed
+**  to the application buffer.  The document byte length,
+**  not character length, will be placed in *buflen. The document 
+**  will not be null terminated. If the buffer is not big enough,
+**  ENOMEM will be returned, else the actual document status.
 */
 TIDY_EXPORT int TIDY_CALL         tidySaveString( TidyDoc tdoc,
                                                  tmbstr buffer, uint* buflen );

src/istack.c

@@ -10,6 +10,9 @@
 #include "attrs.h"
 #include "streamio.h"
 #include "tmbstr.h"
+#if !defined(NDEBUG) && defined(_MSC_VER)
+#include "sprtf.h"
+#endif
 
 /* duplicate attributes */
 AttVal *TY_(DupAttrs)( TidyDocImpl* doc, AttVal *attrs)
@@ -115,6 +118,7 @@ static void PopIStack( TidyDocImpl* doc )
         TY_(FreeAttribute)( doc, av );
     }
     TidyDocFree(doc, istack->element);
+    istack->element = NULL; /* remove the freed element */
 }
 
 static void PopIStackUntil( TidyDocImpl* doc, TidyTagId tid )
@@ -267,9 +271,12 @@ Node *TY_(InsertedToken)( TidyDocImpl* doc )
     node->end = lexer->txtend; /* was : lexer->txtstart; */
     istack = lexer->insert;
 
-#if 0 && defined(_DEBUG)
+/* #if 0 && defined(_DEBUG) */
+#if !defined(NDEBUG) && defined(_MSC_VER)
     if ( lexer->istacksize == 0 )
-        fprintf( stderr, "0-size istack!\n" );
+    {
+        SPRTF( "WARNING: ZERO sized istack!\n" );
+    }
 #endif
 
     node->element = TY_(tmbstrdup)(doc->allocator, istack->element);

src/lexer.c

@@ -52,7 +52,7 @@
 static void check_me(char *name);
 static Bool show_attrs = yes;
 #define MX_TXT 8
-static char buffer[MX_TXT+8]; /* NOTE extra for '...'\0 tail */
+static char buffer[(MX_TXT*4)+8]; /* NOTE extra for '...'\0 tail */
 static tmbstr get_text_string(Lexer* lexer, Node *node)
 {
     uint len = node->end - node->start;
@@ -61,13 +61,14 @@ static tmbstr get_text_string(Lexer* lexer, Node *node)
     unsigned char c;
     uint i = 0;
     Bool insp = no;
-    buffer[0] = (char)0;
+    if (len <= ((MX_TXT * 2) + 3)) {
+        buffer[0] = 0;
         while (cp < end) {
             c = *cp;
+            cp++;
             if (c == '\n') {
                 buffer[i++] = '\\';
                 buffer[i++] = 'n';
-            insp = yes;
             } else if ( c == ' ' ) {
                 if (!insp)
                     buffer[i++] = c;
@@ -76,16 +77,58 @@ static tmbstr get_text_string(Lexer* lexer, Node *node)
                 buffer[i++] = c;
                 insp = no;
             }
+        }
+    } else {
+        char *end1 = cp + MX_TXT;
+        char *bgn = cp + (len - MX_TXT);
+        buffer[0] = 0;
+        if (bgn < end1)
+            bgn = end1;
+        while (cp < end1) {
+            c = *cp;
             cp++;
+            if (c == '\n') {
+                buffer[i++] = '\\';
+                buffer[i++] = 'n';
+            } else if ( c == ' ' ) {
+                if (!insp)
+                    buffer[i++] = c;
+                insp = yes;
+            } else {
+                buffer[i++] = c;
+                insp = no;
+            }
             if (i >= MX_TXT)
                 break;
         }
-    if (i < len) {
+        c = '.';
-        buffer[i++] = '.';
+        if ((i < len)&&(cp < bgn)) {
-        if (i < len) {
+            buffer[i++] = c;
-            buffer[i++] = '.';
+            cp++;
-            if (i < len) {
+            if ((i < len)&&(cp < bgn)) {
-                buffer[i++] = '.';
+                buffer[i++] = c;
+                cp++;
+                if ((i < len)&&(cp < bgn)) {
+                    buffer[i++] = c;
+                    cp++;
+                }
+            }
+        }
+        cp = bgn;
+        insp = no;
+        while (cp < end) {
+            c = *cp;
+            cp++;
+            if (c == '\n') {
+                buffer[i++] = '\\';
+                buffer[i++] = 'n';
+            } else if ( c == ' ' ) {
+                if (!insp)
+                    buffer[i++] = c;
+                insp = yes;
+            } else {
+                buffer[i++] = c;
+                insp = no;
             }
         }
     }
@@ -101,7 +144,7 @@ static void Show_Node( TidyDocImpl* doc, const char *msg, Node *node )
     tmbstr src = lex ? "lexer" : "stream";
     SPRTF("R=%d C=%d: ", line, col );
     // DEBUG: Be able to set a TRAP on a SPECIFIC row,col
-    if ((line == 8) && (col == 36)) {
+    if ((line == 67) && (col == 95)) {
         check_me("Show_Node"); // just a debug trap
     }
     if (lexer && lexer->token && 
@@ -2265,8 +2308,11 @@ Node* TY_(GetToken)( TidyDocImpl* doc, GetTokenMode mode )
     assert( !(lexer->pushed || lexer->itoken) );
 
     /* at start of block elements, unclosed inline
-       elements are inserted into the token stream */
+       elements are inserted into the token stream 
-    if (lexer->insert || lexer->inode) {
+       Issue #341 - Can NOT insert a token if NO istacksize  
+     */
+    if ((lexer->insert || lexer->inode) && lexer->istacksize)
+    {
         /*\ Issue #92: could fix by the following, but instead chose not to stack these 2
          *  if ( !(lexer->insert && (nodeIsINS(lexer->insert) || nodeIsDEL(lexer->insert))) ) {
         \*/