cpptohtm.cpp

把C++源代码转换成HTML

//  FILE: CPPtoHTML.cpp
//    
//  DESCRIPTION: 
//  Create an HTML page from C or C++ code.
//
//  Note that this program was written using Rogue Wave's Tools.h++ 7.0.3
//  with Standard Template Library support.  Converting this code to use
//  the STL and String classes should not be difficult.
//
//  REV: 1.0
//  CREATED: 09/29/96   11:19:46
//  AUTHOR:  Mike Benzinger

// Disable warning about signed/unsigned mismatch.
#pragma warning( disable : 4018 )

// Disable warning about unary minus operator applied to unsigned type.
#pragma warning( disable : 4146 )

// Include the necessary header files.
#include <iostream>
#include <fstream>

#include <ctype.h>

#include <rw/cstring.h>
#include <rw/re.h>
#include <rw/tvordvec.h>

typedef RWTValOrderedVector<CString> StringVector;


//  createKeywordList
// 
//  DESCRIPTION:
//    Create a list of C and C++ keywords.
// 
//  PARAMETERS:
//    keywordList - List of keywords to be populated
// 
//  RETURNS:
//    Nothing

void
createKeywordList(StringVector& keywordList)
{
  keywordList.insert("asm");
  keywordList.insert("auto");
  keywordList.insert("bool");
  keywordList.insert("break");
  keywordList.insert("case");
  keywordList.insert("catch");
  keywordList.insert("char");
  keywordList.insert("class");
  keywordList.insert("const");
  keywordList.insert("const_cast");
  keywordList.insert("continue");
  keywordList.insert("default");
  keywordList.insert("define");
  keywordList.insert("delete");
  keywordList.insert("disable");
  keywordList.insert("do");
  keywordList.insert("double");
  keywordList.insert("dynamic_cast");
  keywordList.insert("else");
  keywordList.insert("enum");
  keywordList.insert("error");
  keywordList.insert("explicit");
  keywordList.insert("extern");
  keywordList.insert("false");
  keywordList.insert("float");
  keywordList.insert("for");
  keywordList.insert("friend");
  keywordList.insert("goto");
  keywordList.insert("if");
  keywordList.insert("ifdef");
  keywordList.insert("ifndef");
  keywordList.insert("include");
  keywordList.insert("inline");
  keywordList.insert("int");
  keywordList.insert("long");
  keywordList.insert("mutable");
  keywordList.insert("namespace");
  keywordList.insert("new");
  keywordList.insert("NULL");
  keywordList.insert("operator");
  keywordList.insert("pragma");
  keywordList.insert("private");
  keywordList.insert("protected");
  keywordList.insert("public");
  keywordList.insert("register");
  keywordList.insert("reinterpret_cast");
  keywordList.insert("return");
  keywordList.insert("short");
  keywordList.insert("signed");
  keywordList.insert("sizeof");
  keywordList.insert("static");
  keywordList.insert("static_cast");
  keywordList.insert("struct");
  keywordList.insert("switch");
  keywordList.insert("template");
  keywordList.insert("this");
  keywordList.insert("throw");
  keywordList.insert("true");
  keywordList.insert("try");
  keywordList.insert("typedef");
  keywordList.insert("typeid");
  keywordList.insert("uchar");
  keywordList.insert("uint");
  keywordList.insert("ulong");
  keywordList.insert("union");
  keywordList.insert("unsigned");
  keywordList.insert("ushort");
  keywordList.insert("using");
  keywordList.insert("virtual");
  keywordList.insert("void");
  keywordList.insert("volatile");
  keywordList.insert("warning");
  keywordList.insert("wchar_t");
  keywordList.insert("while");
}

// Define constants for the HTML colors.
const CString black       = "\"#000000\"";
const CString blue        = "\"#0000FF\"";
const CString cyan        = "\"#00FFFF\"";
const CString green       = "\"#00FF00\"";
const CString magenta     = "\"#FF00FF\"";
const CString red         = "\"#FF0000\"";
const CString yellow      = "\"#FFFF00\"";
const CString white       = "\"#FFFFFF\"";
const CString darkBlue    = "\"#000080\"";
const CString darkCyan    = "\"#008080\"";
const CString darkGreen   = "\"#008000\"";
const CString darkMagenta = "\"#800080\"";
const CString darkRed     = "\"#800000\"";
const CString darkYellow  = "\"#808000\"";
const CString darkGray    = "\"#808080\"";
const CString lightGray   = "\"#C0C0C0\"";

//  startColor
// 
//  DESCRIPTION:
//    Compose an HTML tag to change the font color.
// 
//  PARAMETERS:
//    color - Desired font color
// 
//  RETURNS:
//    String defining the font color

CString
startColor(CString color)
{
  return( CString("<FONT COLOR=" + color + ">") );
}


//  endColor
// 
//  DESCRIPTION:
//    Compose an HTML tag to revert back to the previous color.
// 
//  PARAMETERS:
//    None
// 
//  RETURNS:
//    String defining the end font HTML tag.

CString
endColor()
{
  return( CString("</FONT>") );
}


//  processToken
// 
//  DESCRIPTION:
//    If a token is a keyword, bold it and change its color to blue.
// 
//  PARAMETERS:
//    token       - Token to be analyzed
//    htmlLine    - HTML output line
//    keywordList - List of keywords
// 
//  RETURNS:
//    Nothing

CString
processToken(CString& token, bool& tokenFound, StringVector& keywordList)
{
  // If the keyword is a keyword, then bold it and change its color to blue.
  // Otherwise, simply add the token to the HTML output line with no changes.
  CString htmlString;

  if ( keywordList.contains(token) )
    htmlString = "<B>" + startColor(blue) + token + endColor() + "</B>";
  else
    htmlString = token;  

  // Reset the token to blank and indicate that no token has been found.
  token      = "";
  tokenFound = false;

  // Return the HTML string.
  return( htmlString );
}


//  colorInclude
// 
//  DESCRIPTION:
//    Change the color of a system include file in the format <include.h> so
//    that it is colored the same as a string.
// 
//  PARAMETERS:
//    htmlLine - The HTML output line to be modified.
// 
//  RETURNS:
//    Nothing

void
colorInclude(CString& htmlLine, StringVector& headerNamesList)
{
  // Extract the header name from the HTML line with the format <headerFile.hpp>.
  // Note that since the "<" symbol is a special character for HTML, they are all
  // converted to "&lt".
  CString headerName;

  if ( htmlLine.contains("&lt") )
  {
    // Extract the header name to determine if it's in the list of
    // headers to have hypertext links.
    headerName = htmlLine;

    headerName.replace(RWCRExpr(".*&lt;</FONT>"), "");
    headerName.replace(RWCRExpr("<FONT.*&gt;.*"), "");
  }

  // Otherwise check for a quoted include file.  Note, again, that since the
  // '"' symbol is a special character for HTML, they are all converted to
  // "&quot".

  else if (htmlLine.contains("&quot") )
  {
    // Extract the header name to determine if it's in the list of
    // headers to have hypertext links.
    headerName = htmlLine;

    headerName.replace(RWCRExpr(".*>&quot;"), "");
    headerName.replace(RWCRExpr("&quot;<.*"), "");
  }

  // Force the header name to lower case to ensure that the compare will be
  // done properly.
  headerName.toLower();

  // If the name is in the list, then create a hypertext link.
  if ( headerNamesList.contains(headerName) )
  {
    // Create the hypertext reference name.
    CString hyperTextHRef;

    hyperTextHRef = headerName;
    hyperTextHRef.replace(RWCRExpr("\\."), "_");
    hyperTextHRef += ".htm";

    // Create the hypertext link.
    CString hyperTextLink;

    hyperTextLink = "<A HREF=\"" + hyperTextHRef + "\">" + headerName + "</A>";

    htmlLine.replace(RWCRExpr(headerName), hyperTextLink);
  }

  // Otherwise, if this is a include file in the form <headerFile.hpp> then
  // change the color of the header name to red.
  else if ( htmlLine.contains("&lt") )
  {
    // Convert the dark red color used for symbols and operators to red.
    htmlLine.replace(RWCRExpr(darkRed), red, CString::all);

    // Since by default, the less than symbol is colored and the coloring is
    // turned off immediately afterwards, remove the </FONT> which halts the
    // red coloring so that it will continue coloring the name of the file.    
    htmlLine.replace(RWCRExpr("lt;</FONT>"), "lt;");

    // If there are slashes in the header name, then remove all coloring
    // HTML flags.  An example is #include <rw/cstring.h>.
    RWCRExpr removeSlashColor("<FONT COLOR=" + red + ">/</FONT>");
  
    htmlLine.replace(removeSlashColor, "/", CString::all);

    // Remove the font coloring from before the greater than symbol.
    htmlLine.replace(RWCRExpr("<FONT COLOR=" + red + ">&gt"), "&gt");
  }
}


//  main
// 
//  DESCRIPTION:
//    Main routine for converting C or C++ code to HTML.
//
//  PARAMETERS:
//    argc - Number of input arguments.
// 
//  RETURNS:
//    0

int
main(int argc, char *argv[])
{
  // Check to be sure that the input file name has been passed in.
  if ( argc < 2 )
  {
    cout << "Usage: " << argv[0] << " sourceFile [headerNamesFile] [background]" << endl;
    
    exit( 0 );
  }
  
  // Open the input file.
  ifstream cfile(argv[1]);

  if ( ! cfile )
  {
    cout << "Unable to open input file '" << argv[1] << "'" << endl;

    exit( 1 );
  }

  // If a header names file is present, load the list of header files for
  // which a hyper text link will be created.  Note that header names should
  // be specified one per line.
  StringVector headerNamesList;

  if ( argc >= 3 )
  {
    // Open the header names list file.
    ifstream hdrFile(argv[2]);

    if ( ! hdrFile )
    {
      cout << "Unable to open header names file '" << argv[2] << "'" << endl;

      exit( 1 );
    }

    // Input all header names, convert them to lower case and add them to
    // the list.
    CString headerFileName;

    while ( ! hdrFile.eof() )
    {
      hdrFile >> headerFileName;

      headerFileName.toLower();

      headerNamesList.insert(headerFileName);
    }
  }

  // If the background name exists, then read it for output in the body statement.
  CString background;

  if ( argc >= 4 )
  {
    background = argv[3];
  }
  
  // Create an order vector of CStrings and add all C and C++ keywords to it.
  StringVector keywordList;

  createKeywordList(keywordList);

  // Output the header HTML.
  cout << "<HTML>" << endl;

  // Strip the drive and directory off the file name.  In other words, the
  // following regular expression and replace function will transform the file
  // name "C:\ADIR\BDIR\CDIR\CODEFILE.CPP" to "CODEFILE.CPP".  Note that a
  // back slash is used as an escape character in both C++ strings and regular
  // expressions.  Therefore, in order to have a back slash interpreted as a
  // regular character, two back slashes are needed for the C++ string as well
  // as the regular expression.
  CString fileName = argv[1];
  
  fileName.replace(RWCRExpr(".:"), "");
  fileName.replace(RWCRExpr("\\\\(.+\\\\)*"), "");

  // Create the title for the HTML document.
  cout << "<HEAD>" << endl;
  cout << "<TITLE>" << fileName << "</TITLE>" << endl;
  cout << "</HEAD>" << endl;

  // Output the body statement and the background if any.
  cout << "<BODY ";

  if ( background.length() > 0 )
  {
    cout << "BACKGROUND=\"" << background << "\">" << endl;
  }

  else
  {