/*
  pprint.c -- pretty print parse tree  
  
  (c) 1998-2007 (W3C) MIT, ERCIM, Keio University
  See tidy.h for the copyright notice.
  
  CVS Info :

    $Author: arnaud02 $ 
    $Date: 2007/02/11 17:14:27 $ 
    $Revision: 1.118 $ 

*/

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include "pprint.h"
#include "tidy-int.h"
#include "parser.h"
#include "entities.h"
#include "tmbstr.h"
#include "utf8.h"

/*
  Block-level and unknown elements are printed on
  new lines and their contents indented 2 spaces

  Inline elements are printed inline.

  Inline content is wrapped on spaces (except in
  attribute values or preformatted text, after
  start tags and before end tags
*/

static void PPrintAsp( TidyDocImpl* doc, uint indent, Node* node );
static void PPrintJste( TidyDocImpl* doc, uint indent, Node* node );
static void PPrintPhp( TidyDocImpl* doc, uint indent, Node* node );
static int  TextEndsWithNewline( Lexer *lexer, Node *node, uint mode );
static int  TextStartsWithWhitespace( Lexer *lexer, Node *node, uint start, uint mode );
static Bool InsideHead( TidyDocImpl* doc, Node *node );
static Bool ShouldIndent( TidyDocImpl* doc, Node *node );

#if SUPPORT_ASIAN_ENCODINGS
/* #431953 - start RJ Wraplen adjusted for smooth international ride */

#if 0
uint CWrapLen( TidyDocImpl* doc, uint ind )
{
    ctmbstr lang = cfgStr( doc, TidyLanguage );
    uint wraplen = cfg( doc, TidyWrapLen );

    if ( !TY_(tmbstrcasecmp)(lang, "zh") )
        /* Chinese characters take two positions on a fixed-width screen */ 
        /* It would be more accurate to keep a parallel linelen and wraphere
           incremented by 2 for Chinese characters and 1 otherwise, but this
           is way simpler.
        */
        return (ind + (( wraplen - ind ) / 2)) ; 
    
    if ( !TY_(tmbstrcasecmp)(lang, "ja") )
        /* average Japanese text is 30% kanji */
        return (ind + ((( wraplen - ind ) * 7) / 10)) ; 
    
    return wraplen;
}
#endif

typedef enum
{
  UC00, /* None                       */
  UCPC, /* Punctuation, Connector     */
  UCPD, /* Punctuation, Dash          */
  UCPE, /* Punctuation, Close         */
  UCPS, /* Punctuation, Open          */
  UCPI, /* Punctuation, Initial quote */
  UCPF, /* Punctuation, Final quote   */
  UCPO, /* Punctuation, Other         */
  UCZS, /* Separator, Space           */
  UCZL, /* Separator, Line            */
  UCZP  /* Separator, Paragraph       */
} UnicodeCategory;

/*
  From the original code, the following characters are removed:

    U+2011 (non-breaking hyphen)
    U+202F (narrow non-break space)
    U+2044 (fraction slash) 
    U+200B (zero width space)
    ...... (bidi formatting control characters)

  U+2011 and U+202F are non-breaking, U+2044 is a Sm character,
  U+200B is a non-visible space, wrapping after it would make
  this space visible, bidi should be done using HTML features
  and the characters are neither Px or Zx.

  The following Unicode 3.0 punctuation characters are added:

    U+2048 (question exclamation mark)
    U+2049 (exclamation question mark)
    U+204A (tironian sign et)
    U+204B (reversed pilcrow sign)
    U+204C (black leftwards bullet)
    U+204D (black rightwards bullet)
    U+3030 (wavy dash)
    U+30FB (katakana middle dot)
    U+FE63 (small hyphen-minus)
    U+FE68 (small reverse solidus)
    U+FF3F (fullwidth low line)
    U+FF5B (fullwidth left curly bracket)
    U+FF5D (fullwidth right curly bracket)

  Other additional characters were not included in Unicode 3.0.
  The table is based on Unicode 4.0. It must include only those
  characters marking a wrapping point, "before" if the general
  category is UCPS or UCPI, otherwise "after".
*/
static struct _unicode4cat
{
  unsigned long code;
  UnicodeCategory category;
} const unicode4cat[] =
{
#if 0
  { 0x037E, UCPO }, { 0x0387, UCPO }, { 0x055A, UCPO }, { 0x055B, UCPO },
  { 0x055C, UCPO }, { 0x055D, UCPO }, { 0x055E, UCPO }, { 0x055F, UCPO },
  { 0x0589, UCPO }, { 0x058A, UCPD }, { 0x05BE, UCPO }, { 0x05C0, UCPO },
  { 0x05C3, UCPO }, { 0x05F3, UCPO }, { 0x05F4, UCPO }, { 0x060C, UCPO },
  { 0x060D, UCPO }, { 0x061B, UCPO }, { 0x061F, UCPO }, { 0x066A, UCPO },
  { 0x066B, UCPO }, { 0x066C, UCPO }, { 0x066D, UCPO }, { 0x06D4, UCPO },
  { 0x0700, UCPO }, { 0x0701, UCPO }, { 0x0702, UCPO }, { 0x0703, UCPO },
  { 0x0704, UCPO }, { 0x0705, UCPO }, { 0x0706, UCPO }, { 0x0707, UCPO },
  { 0x0708, UCPO }, { 0x0709, UCPO }, { 0x070A, UCPO }, { 0x070B, UCPO },
  { 0x070C, UCPO }, { 0x070D, UCPO }, { 0x0964, UCPO }, { 0x0965, UCPO },
  { 0x0970, UCPO }, { 0x0DF4, UCPO }, { 0x0E4F, UCPO }, { 0x0E5A, UCPO },
  { 0x0E5B, UCPO }, { 0x0F04, UCPO }, { 0x0F05, UCPO }, { 0x0F06, UCPO },
  { 0x0F07, UCPO }, { 0x0F08, UCPO }, { 0x0F09, UCPO }, { 0x0F0A, UCPO },
  { 0x0F0B, UCPO }, { 0x0F0D, UCPO }, { 0x0F0E, UCPO }, { 0x0F0F, UCPO },
  { 0x0F10, UCPO }, { 0x0F11, UCPO }, { 0x0F12, UCPO }, { 0x0F3A, UCPS },
  { 0x0F3B, UCPE }, { 0x0F3C, UCPS }, { 0x0F3D, UCPE }, { 0x0F85, UCPO },
  { 0x104A, UCPO }, { 0x104B, UCPO }, { 0x104C, UCPO }, { 0x104D, UCPO },
  { 0x104E, UCPO }, { 0x104F, UCPO }, { 0x10FB, UCPO }, { 0x1361, UCPO },
  { 0x1362, UCPO }, { 0x1363, UCPO }, { 0x1364, UCPO }, { 0x1365, UCPO },
  { 0x1366, UCPO }, { 0x1367, UCPO }, { 0x1368, UCPO }, { 0x166D, UCPO },
  { 0x166E, UCPO }, { 0x1680, UCZS }, { 0x169B, UCPS }, { 0x169C, UCPE },
  { 0x16EB, UCPO }, { 0x16EC, UCPO }, { 0x16ED, UCPO }, { 0x1735, UCPO },
  { 0x1736, UCPO }, { 0x17D4, UCPO }, { 0x17D5, UCPO }, { 0x17D6, UCPO },
  { 0x17D8, UCPO }, { 0x17D9, UCPO }, { 0x17DA, UCPO }, { 0x1800, UCPO },
  { 0x1801, UCPO }, { 0x1802, UCPO }, { 0x1803, UCPO }, { 0x1804, UCPO },
  { 0x1805, UCPO }, { 0x1806, UCPD }, { 0x1807, UCPO }, { 0x1808, UCPO },
  { 0x1809, UCPO }, { 0x180A, UCPO }, { 0x180E, UCZS }, { 0x1944, UCPO },
  { 0x1945, UCPO }, 
#endif
  { 0x2000, UCZS }, { 0x2001, UCZS }, { 0x2002, UCZS }, { 0x2003, UCZS },
  { 0x2004, UCZS }, { 0x2005, UCZS }, { 0x2006, UCZS }, { 0x2008, UCZS },
  { 0x2009, UCZS }, { 0x200A, UCZS }, { 0x2010, UCPD }, { 0x2012, UCPD },
  { 0x2013, UCPD }, { 0x2014, UCPD }, { 0x2015, UCPD }, { 0x2016, UCPO },
  { 0x2017, UCPO }, { 0x2018, UCPI }, { 0x2019, UCPF }, { 0x201A, UCPS },
  { 0x201B, UCPI }, { 0x201C, UCPI }, { 0x201D, UCPF }, { 0x201E, UCPS },
  { 0x201F, UCPI }, { 0x2020, UCPO }, { 0x2021, UCPO }, { 0x2022, UCPO },
  { 0x2023, UCPO }, { 0x2024, UCPO }, { 0x2025, UCPO }, { 0x2026, UCPO },
  { 0x2027, UCPO }, { 0x2028, UCZL }, { 0x2029, UCZP }, { 0x2030, UCPO },
  { 0x2031, UCPO }, { 0x2032, UCPO }, { 0x2033, UCPO }, { 0x2034, UCPO },
  { 0x2035, UCPO }, { 0x2036, UCPO }, { 0x2037, UCPO }, { 0x2038, UCPO },
  { 0x2039, UCPI }, { 0x203A, UCPF }, { 0x203B, UCPO }, { 0x203C, UCPO },
  { 0x203D, UCPO }, { 0x203E, UCPO }, { 0x203F, UCPC }, { 0x2040, UCPC },
  { 0x2041, UCPO }, { 0x2042, UCPO }, { 0x2043, UCPO }, { 0x2045, UCPS },
  { 0x2046, UCPE }, { 0x2047, UCPO }, { 0x2048, UCPO }, { 0x2049, UCPO },
  { 0x204A, UCPO }, { 0x204B, UCPO }, { 0x204C, UCPO }, { 0x204D, UCPO },
  { 0x204E, UCPO }, { 0x204F, UCPO }, { 0x2050, UCPO }, { 0x2051, UCPO },
  { 0x2053, UCPO }, { 0x2054, UCPC }, { 0x2057, UCPO }, { 0x205F, UCZS },
  { 0x207D, UCPS }, { 0x207E, UCPE }, { 0x208D, UCPS }, { 0x208E, UCPE },
  { 0x2329, UCPS }, { 0x232A, UCPE }, { 0x23B4, UCPS }, { 0x23B5, UCPE },
  { 0x23B6, UCPO }, { 0x2768, UCPS }, { 0x2769, UCPE }, { 0x276A, UCPS },
  { 0x276B, UCPE }, { 0x276C, UCPS }, { 0x276D, UCPE }, { 0x276E, UCPS },
  { 0x276F, UCPE }, { 0x2770, UCPS }, { 0x2771, UCPE }, { 0x2772, UCPS },
  { 0x2773, UCPE }, { 0x2774, UCPS }, { 0x2775, UCPE }, { 0x27E6, UCPS },
  { 0x27E7, UCPE }, { 0x27E8, UCPS }, { 0x27E9, UCPE }, { 0x27EA, UCPS },
  { 0x27EB, UCPE }, { 0x2983, UCPS }, { 0x2984, UCPE }, { 0x2985, UCPS },
  { 0x2986, UCPE }, { 0x2987, UCPS }, { 0x2988, UCPE }, { 0x2989, UCPS },
  { 0x298A, UCPE }, { 0x298B, UCPS }, { 0x298C, UCPE }, { 0x298D, UCPS },
  { 0x298E, UCPE }, { 0x298F, UCPS }, { 0x2990, UCPE }, { 0x2991, UCPS },
  { 0x2992, UCPE }, { 0x2993, UCPS }, { 0x2994, UCPE }, { 0x2995, UCPS },
  { 0x2996, UCPE }, { 0x2997, UCPS }, { 0x2998, UCPE }, { 0x29D8, UCPS },
  { 0x29D9, UCPE }, { 0x29DA, UCPS }, { 0x29DB, UCPE }, { 0x29FC, UCPS },
  { 0x29FD, UCPE }, { 0x3001, UCPO }, { 0x3002, UCPO }, { 0x3003, UCPO },
  { 0x3008, UCPS }, { 0x3009, UCPE }, { 0x300A, UCPS }, { 0x300B, UCPE },
  { 0x300C, UCPS }, { 0x300D, UCPE }, { 0x300E, UCPS }, { 0x300F, UCPE },
  { 0x3010, UCPS }, { 0x3011, UCPE }, { 0x3014, UCPS }, { 0x3015, UCPE },
  { 0x3016, UCPS }, { 0x3017, UCPE }, { 0x3018, UCPS }, { 0x3019, UCPE },
  { 0x301A, UCPS }, { 0x301B, UCPE }, { 0x301C, UCPD }, { 0x301D, UCPS },
  { 0x301E, UCPE }, { 0x301F, UCPE }, { 0x3030, UCPD }, { 0x303D, UCPO },
  { 0x30A0, UCPD }, { 0x30FB, UCPC }, { 0xFD3E, UCPS }, { 0xFD3F, UCPE },
  { 0xFE30, UCPO }, { 0xFE31, UCPD }, { 0xFE32, UCPD }, { 0xFE33, UCPC },
  { 0xFE34, UCPC }, { 0xFE35, UCPS }, { 0xFE36, UCPE }, { 0xFE37, UCPS },
  { 0xFE38, UCPE }, { 0xFE39, UCPS }, { 0xFE3A, UCPE }, { 0xFE3B, UCPS },
  { 0xFE3C, UCPE }, { 0xFE3D, UCPS }, { 0xFE3E, UCPE }, { 0xFE3F, UCPS },
  { 0xFE40, UCPE }, { 0xFE41, UCPS }, { 0xFE42, UCPE }, { 0xFE43, UCPS },
  { 0xFE44, UCPE }, { 0xFE45, UCPO }, { 0xFE46, UCPO }, { 0xFE47, UCPS },
  { 0xFE48, UCPE }, { 0xFE49, UCPO }, { 0xFE4A, UCPO }, { 0xFE4B, UCPO },
  { 0xFE4C, UCPO }, { 0xFE4D, UCPC }, { 0xFE4E, UCPC }, { 0xFE4F, UCPC },
  { 0xFE50, UCPO }, { 0xFE51, UCPO }, { 0xFE52, UCPO }, { 0xFE54, UCPO },
  { 0xFE55, UCPO }, { 0xFE56, UCPO }, { 0xFE57, UCPO }, { 0xFE58, UCPD },
  { 0xFE59, UCPS }, { 0xFE5A, UCPE }, { 0xFE5B, UCPS }, { 0xFE5C, UCPE },
  { 0xFE5D, UCPS }, { 0xFE5E, UCPE }, { 0xFE5F, UCPO }, { 0xFE60, UCPO },
  { 0xFE61, UCPO }, { 0xFE63, UCPD }, { 0xFE68, UCPO }, { 0xFE6A, UCPO },
  { 0xFE6B, UCPO }, { 0xFF01, UCPO }, { 0xFF02, UCPO }, { 0xFF03, UCPO },
  { 0xFF05, UCPO }, { 0xFF06, UCPO }, { 0xFF07, UCPO }, { 0xFF08, UCPS },
  { 0xFF09, UCPE }, { 0xFF0A, UCPO }, { 0xFF0C, UCPO }, { 0xFF0D, UCPD },
  { 0xFF0E, UCPO }, { 0xFF0F, UCPO }, { 0xFF1A, UCPO }, { 0xFF1B, UCPO },
  { 0xFF1F