highlight.js

展示数据结构的一些实用技巧. 包含: 1.运用kmp算法计算无穷概率 2.矩阵乘法的十种经典运算技巧 3.位运算的实用技巧(1) (2) (3)

var IDENT_RE = '[a-zA-Z][a-zA-Z0-9_]*';
var UNDERSCORE_IDENT_RE = '[a-zA-Z_][a-zA-Z0-9_]*';
var NUMBER_RE = '\\b\\d+(\\.\\d+)?';
var C_NUMBER_RE = '\\b(0x[A-Za-z0-9]+|\\d+(\\.\\d+)?)';

// Common modes
var APOS_STRING_MODE = {
  className: 'string',
  begin: '\'', end: '(^|[^\\\\])\'',
  relevance: 0
}
var QUOTE_STRING_MODE = {
  className: 'string',
  begin: '"', end: '(^|[^\\\\])"',
  relevance: 0
}
var C_LINE_COMMENT_MODE = {
  className: 'comments',
  begin: '//', end: '$',
  relevance: 0
}
var C_BLOCK_COMMENT_MODE = {
  className: 'comments',
  begin: '/\\*', end: '\\*/'
}
var HASH_COMMENT_MODE = {
  className: 'comments',
  begin: '#', end: '$'
}
var C_NUMBER_MODE = {
  className: 'number',
  begin: C_NUMBER_RE, end: '^',
  relevance: 0
}

var LANGUAGES = {}

var PASCAL_KEYWORDS = {'abs': 1,'addr': 1,'and': 1,'ansichar': 1,'ansistring': 1,'array': 1,'as': 1,'asm': 1,'begin': 1,'boolean': 1,'byte': 1,'cardinal': 1,'case': 1,'char': 1,'class': 1,'comp': 1,'const': 1,'constructor': 1,'currency': 1,'destructor': 1,'div': 1,'do': 1,'double': 1,'downto': 1,'else': 1,'end': 1,'except': 1,'exports': 1,'extended': 1,'false': 1,'file': 1,'finalization': 1,'finally': 1,'for': 1,'function': 1,'goto': 1,'if': 1,'implementation': 1,'in': 1,'inherited': 1,'int64': 1,'initialization': 1,'integer': 1,'interface': 1,'is': 1,'label': 1,'library': 1,'longint': 1,'longword': 1,'mod': 1,'nil': 1,'not': 1,'object': 1,'of': 1,'on': 1,'operator': 1,'or': 1,'packed': 1,'pansichar': 1,'pansistring': 1,'pchar': 1,'pcurrency': 1,'pdatetime': 1,'pextended': 1,'pint64': 1,'pointer': 1,'private': 1,'procedure': 1,'program': 1,'property': 1,'pshortstring': 1,'pstring': 1,'pvariant': 1,'pwidechar': 1,'pwidestring': 1,'protected': 1,'public': 1,'published': 1,'raise': 1,'real': 1,'real48': 1,'record': 1,'repeat': 1,'set': 1,'shl': 1,'shortint': 1,'shortstring': 1,'shr': 1,'single': 1,'smallint': 1,'string': 1,'then': 1,'threadvar': 1,'to': 1,'true': 1,'type': 1,'unit': 1,'until': 1,'uses': 1,'val': 1,'var': 1,'varirnt': 1,'while': 1,'widechar': 1,'widestring': 1,'with': 1,'word': 1,'write': 1,'writeln': 1,'xor': 1,'assign': 1,'reset': 1,'rewrite': 1,'exit': 1,'halt': 1,'break': 1,'read': 1,'readln': 1,'new': 1,'length': 1,'append': 1,'close': 1};

LANGUAGES.pascal = {
  defaultMode: {
    lexems: [IDENT_RE],
    illegal: '("|\\$[G-Zg-z]|\\/\\*|</)',
    contains: ['comments', 'string', 'number', 'function', 'class'],
    keywords: PASCAL_KEYWORDS
  },
  case_insensitive: true,
  modes: [
    {
      className: 'comments',
      begin: '{', end: '}'
    },
    {
      className: 'comments',
      begin: '\\(\\*', end: '\\*\\)',
      relevance: 10
    },
    C_LINE_COMMENT_MODE,
    {
      className: 'number',
      begin: NUMBER_RE, end: '^',
      relevance: 0
    },
    {
      className: 'string',
      begin: '\'', end: '\'',
      contains: ['quote'],
      relevance: 0
    },
    {
      className: 'quote',
      begin: '\'\'', end: '^'
    },
    {
      className: 'function',
      begin: 'function', end: '[:;]',
      lexems: [IDENT_RE],
      keywords: {'function': 1},
      contains: ['title', 'params', 'comments'],
      relevance: 0
    },
    {
      className: 'function',
      begin: '(procedure|constructor|destructor)', end: ';',
      lexems: [IDENT_RE],
      keywords: {'constructor': 1, 'destructor': 1, 'procedure': 1},
      contains: ['title', 'params', 'comments'],
      relevance: 10
    },
    {
      className: 'title',
      begin: IDENT_RE, end: '^'
    },
    {
      className: 'params',
      begin: '\\(', end: '\\)',
      lexems: [IDENT_RE],
      keywords: PASCAL_KEYWORDS,
      contains: ['string']
    }
  ]
};//pascal, written by matrix67.com

LANGUAGES.cpp = {
  defaultMode: {
    lexems: [UNDERSCORE_IDENT_RE],
    illegal: '</',
    contains: ['comments', 'string', 'number', 'preprocessor'],
    keywords: {'false': 1, 'int': 1, 'float': 1, 'while': 1, 'private': 1, 'char': 1, 'catch': 1, 'export': 1, 'virtual': 1, 'operator': 1, 'sizeof': 1, 'dynamic_cast': 1, 'typedef': 1, 'const_cast': 1, 'const': 1, 'struct': 1, 'for': 1, 'static_cast': 1, 'union': 1, 'namespace': 1, 'unsigned': 1, 'long': 1, 'throw': 1, 'volatile': 1, 'static': 1, 'protected': 1, 'bool': 1, 'template': 1, 'mutable': 1, 'if': 1, 'public': 1, 'friend': 1, 'do': 1, 'return': 1, 'goto': 1, 'auto': 1, 'void': 1, 'enum': 1, 'else': 1, 'break': 1, 'new': 1, 'extern': 1, 'using': 1, 'true': 1, 'class': 1, 'asm': 1, 'case': 1, 'typeid': 1, 'short': 1, 'reinterpret_cast': 1, 'default': 1, 'double': 1, 'register': 1, 'explicit': 1, 'signed': 1, 'typename': 1, 'try': 1, 'this': 1, 'switch': 1, 'continue': 1, 'wchar_t': 1, 'inline': 1, 'delete': 1}
  },
  modes: [
    C_LINE_COMMENT_MODE,
    C_BLOCK_COMMENT_MODE,
    C_NUMBER_MODE,
    QUOTE_STRING_MODE,
    {
      className: 'string',
      begin: '\'', end: '[^\\\\]\'',
      illegal: '[^\\\\][^\']'
    },
    {
      className: 'preprocessor',
      begin: '#', end: '$'
    }
  ]
};//cpp

LANGUAGES.javascript = {
  defaultMode: {
    lexems: [UNDERSCORE_IDENT_RE],
    contains: ['string', 'comments', 'number', 'regexp', 'function', 'literal'],
    keywords: {'in': 1, 'if': 1, 'for': 1, 'while': 1, 'finally': 1, 'var': 1, 'new': 1, 'function': 1, 'do': 1, 'return': 1, 'void': 1, 'else': 1, 'break': 1, 'catch': 1, 'instanceof': 1, 'with': 1, 'throw': 1, 'case': 1, 'default': 1, 'try': 1, 'this': 1, 'switch': 1, 'continue': 1, 'typeof': 1, 'delete': 1}
  },
  modes: [
    C_LINE_COMMENT_MODE,
    C_BLOCK_COMMENT_MODE,
    C_NUMBER_MODE,
    {
      className: 'literal',
      begin: '\\b(true|false|null)', end: '^'
    },
    APOS_STRING_MODE,
    QUOTE_STRING_MODE,
    {
      className: 'regexp',
      begin: '\\/[^\\/]', end: '(^|[^\\\\])\\/[gim]*'
    },
    {
      className: 'function',
      begin: 'function ', end: '{',
      lexems: [UNDERSCORE_IDENT_RE],
      keywords: {'function': 1},
      contains: ['title', 'params']
    },
    {
      className: 'title',
      begin: UNDERSCORE_IDENT_RE, end: '^'
    },
    {
      className: 'params',
      begin: '\\(', end: '\\)',
      contains: ['string', 'comments']
    }
  ]
};//javascript

function langRe(language, value, global) {
  var mode =  'm' + (language.case_insensitive ? 'i' : '') + (global ? 'g' : '');
  return new RegExp(value, mode);
}//re

for (var i in LANGUAGES) {
  var language = LANGUAGES[i];
  for (var key in language.modes) {
    if (language.modes[key].begin)
      language.modes[key].beginRe = langRe(language, '^' + language.modes[key].begin);
    if (language.modes[key].end)
      language.modes[key].endRe = langRe(language, '^' + language.modes[key].end);
    if (language.modes[key].illegal)
      language.modes[key].illegalRe = langRe(language, '^(?:' + language.modes[key].illegal + ')');
    language.defaultMode.illegalRe = langRe(language, '^(?:' + language.defaultMode.illegal + ')');
  }//for
}//for

var selected_languages = {};

function Highlighter(language_name, value) {
  currentMode = function(){
    return modes[modes.length - 1];
  }//currentMode
  
  function subMode(lexem) {
    if (!currentMode().contains)
      return null;
    for (var key in language.modes)
      if (contains(currentMode().contains, language.modes[key].className) && language.modes[key].beginRe.test(lexem))
        return language.modes[key];
    return null;
  }//subMode

  function endOfMode(mode_index, lexem) {
    if (modes[mode_index].end && modes[mode_index].endRe.test(lexem))
      return 1;
    if (modes[mode_index].endsWithParent) {
      var level = endOfMode(mode_index - 1, lexem);
      return level ? level + 1 : 0;
    }//if
    return 0;
  }//endOfMode
  
  function isIllegal(lexem) {
    if (!currentMode().illegalRe)
      return false;
    return currentMode().illegalRe.test(lexem);
  }//isIllegal

  function eatModeChunk(value, index) {
    if (!currentMode().terminators) {
      var terminators = [];
      
      if (currentMode().contains)
        for (var key in language.modes) {
          if (contains(currentMode().contains, language.modes[key].className) &&
              !contains(terminators, language.modes[key].begin))
            terminators[terminators.length] = language.modes[key].begin;
        }//for
      
      var mode_index = modes.length - 1;
      do {
        if (modes[mode_index].end && !contains(terminators, modes[mode_index].end))
          terminators[terminators.length] = modes[mode_index].end;
        mode_index--;
      } while (modes[mode_index + 1].endsWithParent);
      
      if (currentMode().illegal)
        if (!contains(terminators, currentMode().illegal))
          terminators[terminators.length] = currentMode().illegal;
      
      var terminator_re = '(' + terminators[0];
      for (var i = 0; i < terminators.length; i++)
        terminator_re += '|' + terminators[i];
      terminator_re += ')';
      currentMode().terminators = langRe(language, terminator_re);
    }//if
    value = value.substr(index);
    var match = currentMode().terminators.exec(value);
    if (!match) 
      return [value, '', true];
    if (match.index == 0)
      return ['', match[0], false];
    else
      return [value.substr(0, match.index), match[0], false];
  }//eatModeChunk
  
  function escape(value) {
    return value.replace(/&/gm, '&amp;').replace(/</gm, '&lt;').replace(/>/gm, '&gt;');
  }//escape
  
  function processKeywords(buffer) {
    var mode = currentMode();
    if (!mode.keywords || !mode.lexems)
      return escape(buffer);
    if (!mode.lexemsRe) {
      var lexems = [];
      for (var key in mode.lexems)
        if (!contains(lexems, mode.lexems[key]))
          lexems[lexems.length] = mode.lexems[key];
      var lexems_re = '(' + lexems[0];
      for (var i = 1; i < lexems.length; i++)
        lexems_re += '|' + lexems[i];
      lexems_re += ')';
      mode.lexemsRe = langRe(language, lexems_re, true);
    }//if
    var result = '';
    var last_index = 0;
    mode.lexemsRe.lastIndex = 0;
    var match = mode.lexemsRe.exec(buffer);
    while (match) {
      result += escape(buffer.substr(last_index, match.index - last_index));
      if (mode.keywords[language.case_insensitive ? match[0].toLowerCase() : match[0]]) {
        keyword_count++;
        result += '<span class="keyword">' + escape(match[0]) + '</span>';
      } else {
        result += escape(match[0]);
      }//if
      last_index = mode.lexemsRe.lastIndex;
      match = mode.lexemsRe.exec(buffer);
    }//while
    result += escape(buffer.substr(last_index, buffer.length - last_index));
    return result;
  }//processKeywords
  
  function processModeInfo(buffer, lexem, end) {
    if (end) {
      result += processKeywords(currentMode().buffer + buffer);
      return;
    }//if
    if (isIllegal(lexem))
      throw 'Illegal';
    var new_mode = subMode(lexem);
    if (new_mode) {
      currentMode().buffer += buffer;
      result += processKeywords(currentMode().buffer);
      if (new_mode.excludeBegin) {
        result += lexem + '<span class="' + new_mode.className + '">';
        new_mode.buffer = '';
      } else {
        result += '<span class="' + new_mode.className + '">';
        new_mode.buffer = lexem;
      }//if
      modes[modes.length] = new_mode;
      relevance += currentMode().relevance != undefined ? currentMode().relevance : 1;
      return;
    }//if
    var end_level = endOfMode(modes.length - 1, lexem);
    if (end_level) {
      currentMode().buffer += buffer;
      if (currentMode().excludeEnd) {
        result += processKeywords(currentMode().buffer) + '</span>' + lexem;
      } else {
        result += processKeywords(currentMode().buffer + lexem) + '</span>';
      }
      while (end_level > 1) {
        result += '</span>';
        end_level--;
        modes.length--;
      }//while
      modes.length--;
      currentMode().buffer = '';
      return;
    }//if
  }//processModeInfo

  function highlight(value) {
    var index = 0;
    language.defaultMode.buffer = '';
    for (var mode_info = eatModeChunk(value, index); index < value.length; mode_info = eatModeChunk(value, index)) {
      processModeInfo(mode_info[0], mode_info[1], mode_info[2]);
      index += mode_info[0].length + mode_info[1].length;
    }//for
    if(modes.length > 1)
      throw 'Illegal';
  }//highlight
  
  this.language_name = language_name;
  var language = LANGUAGES[language_name];
  var modes = [language.defaultMode];
  var relevance = 0;
  var keyword_count = 0;
  var result = '';
  try {
    highlight(value);
    this.relevance = relevance;
    this.keyword_count = keyword_count;
    this.result = result;
  } catch (e) {
    if (e == 'Illegal') {
      this.relevance = 0;
      this.keyword_count = 0;
      this.result = escape(value);
    } else {
      throw e;
    }//if
  }//try
}//Highlighter

function contains(array, item) {
  if (!array)
    return false;
  for (var key in array)
    if (array[key] == item)
      return true;
  return false;
}//contains

function blockText(block) {
  var result = '';
  for (var i = 0; i < block.childNodes.length; i++)
    if (block.childNodes[i].nodeType == 3)
      result += block.childNodes[i].nodeValue;
    else if (block.childNodes[i].nodeName == 'BR')
      result += '\n';
    else
      throw 'Complex markup';
  return result;
}//blockText

function initHighlight(block) {
  if (block.className.search(/\bno\-highlight\b/) != -1)
    return;
  try {
    blockText(block);
  } catch (e) {
    if (e == 'Complex markup')
      return;
  }//try
  var classes = block.className.split(/\s+/);
  for (var i = 0; i < classes.length; i++) {
    if (LANGUAGES[classes[i]]) {
      highlightLanguage(block, classes[i]);
      return;
    }//if
  }//for
}//initHighlight

function highlightLanguage(block, language) {
  var highlight = new Highlighter(language, blockText(block));
  // See these 4 lines? This is IE's notion of "block.innerHTML = result". Love this browser :-/
  var container = document.createElement('div');
  container.innerHTML = '<pre><code class="' + block.className + '">' + highlight.result + '</code></pre>';
  var environment = block.parentNode.parentNode;
  environment.replaceChild(container.firstChild, block.parentNode);
}//highlightLanguage

function initHighlighting() {
  if (initHighlighting.called)
    return;
  initHighlighting.called = true;
  if (arguments.length) {
    for (var i = 0; i < arguments.length; i++) {
      if (LANGUAGES[arguments[i]]) {
        selected_languages[arguments[i]] = LANGUAGES[arguments[i]];
      }//if
    }//for
  } else
    selected_languages = LANGUAGES;
  var pres = document.getElementsByTagName('pre');
  for (var i = 0; i < pres.length; i++) {
    if (pres[i].firstChild && pres[i].firstChild.nodeName == 'CODE')
      initHighlight(pres[i].firstChild);
  }//for
}//initHighlighting

function initHighlightingOnLoad() {
  var original_arguments = arguments;
  var handler = function(){initHighlighting.apply(null, original_arguments)};
  if (window.addEventListener) {
    window.addEventListener('DOMContentLoaded', handler, false);
    window.addEventListener('load', handler, false);
  } else if (window.attachEvent)
    window.attachEvent('onload', handler);
  else
    window.onload = handler;
}//initHighlightingOnLoad