Developer CD Series 1997 March: Reference Library/Dev.CD Mar 97 RL.toast/mac/Technical Documentation/develop/develop Issue 27/develop Issue 27 code/Internet Config Assistant/InternetAssistant/HTMLParser.cp

Text File | 1996-05-05 | 9.2 KB | 431 lines | [TEXT/CWIE]

/*
    File:        HTMLParser.cp

    Contains:    A simplistic parse of a sub-set of HTML.
                Used to allow styles in dialogs.

    Written by:    Arno Gourdol

    Copyright:    © 1996 by Apple Computer, Inc., all rights reserved.

*/

#include "HTMLParser.h"
#include "assert.h"
#include "macros.h"

#include <TextUtils.h>



// --------------------------------------------------------------------
// HTMLParser
// --------------------------------------------------------------------

HTMLParser::HTMLParser(Handle text, SInt32 fromIndex, SInt32 toIndex) :
    fTextHandle(text),
    fCursor(fromIndex - 1),
    fFinalCursor(toIndex),
    fContinue(true)
{
    assert(text != NULL);
    assert(fromIndex >= 0);

    if (toIndex < 0)
        fFinalCursor = ::GetHandleSize(text) - 1;

    fSavedHandleState = HGetState(fTextHandle);
    HLock(fTextHandle);
    fText = *fTextHandle;
}



// --------------------------------------------------------------------
// ~HTMLParser
// --------------------------------------------------------------------

HTMLParser::~HTMLParser()
{
    HSetState(fTextHandle, fSavedHandleState);
}



// --------------------------------------------------------------------
// IsTag
// --------------------------------------------------------------------

Boolean HTMLParser::IsTag(SInt32 tagStart, SInt32 tagEnd,
                        ConstStr255Param tag)
{
#pragma unused (tagStart)
#pragma unused (tagEnd)
#pragma unused (tag)
    SInt16 result = IdenticalText(&fText[tagStart], &tag[1],
                                        tag[0] - 1, tag[0] - 1, NULL);

    // Furthermore, the next character must be white space or
    // end of tag ???

    return (result == 0);
}



// --------------------------------------------------------------------
// Tag
// --------------------------------------------------------------------

void HTMLParser::Tag(SInt32 tagStart, SInt32 tagEnd)
{
    if (IsTag(tagStart, tagEnd, "\p<B>"))
    {
        TagBoldStart(tagStart, tagEnd);
    }
    else if (IsTag(tagStart, tagEnd, "\p</B>"))
    {
        TagBoldEnd(tagStart, tagEnd);
    }
    else if (IsTag(tagStart, tagEnd, "\p<I>"))
    {
        TagItalicStart(tagStart, tagEnd);
    }
    else if (IsTag(tagStart, tagEnd, "\p</I>"))
    {
        TagItalicEnd(tagStart, tagEnd);
    }
}



// --------------------------------------------------------------------
// TagBoldStart
// --------------------------------------------------------------------

void HTMLParser::TagBoldStart(SInt32 tagStart, SInt32 tagEnd)
{
#pragma unused (tagStart)
#pragma unused (tagEnd)
}



// --------------------------------------------------------------------
// TagBoldEnd
// --------------------------------------------------------------------

void HTMLParser::TagBoldEnd(SInt32 tagStart, SInt32 tagEnd)
{
#pragma unused (tagStart)
#pragma unused (tagEnd)
}



// --------------------------------------------------------------------
// TagItalicStart
// --------------------------------------------------------------------

void HTMLParser::TagItalicStart(SInt32 tagStart, SInt32 tagEnd)
{
#pragma unused (tagStart)
#pragma unused (tagEnd)
}



// --------------------------------------------------------------------
// TagItalicEnd
// --------------------------------------------------------------------

void HTMLParser::TagItalicEnd(SInt32 tagStart, SInt32 tagEnd)
{
#pragma unused (tagStart)
#pragma unused (tagEnd)
}



// --------------------------------------------------------------------
// EndOfText
// --------------------------------------------------------------------

void HTMLParser::EndOfText(SInt32 endOfText)
{
#pragma unused (endOfText)
}



// --------------------------------------------------------------------
// Parse
// --------------------------------------------------------------------

void HTMLParser::Parse(void)
{
    UInt16 c;
    while (fContinue)
    {
        c = GetChar();
        if (c == '<')
        {
            // Potential start of a tag
            UInt32 startTag = GetCursor();

            UInt32 endTag = FindChar('>');

            if (endTag >= 0)
            {
                Tag(startTag, endTag);
                SetCursor(endTag);
            }
        }
    }
    EndOfText(max(GetCursor(), GetFinalCursor()));
}



// --------------------------------------------------------------------
// SkipWhiteSpaces
// --------------------------------------------------------------------

void HTMLParser::SkipWhiteSpaces(void)
{
    UInt16 c;

    do
    {
        c = GetChar();
    } while (c == ' ');
    // ??? Use Character type to catch all the white spaces

    // Roll back one if we have found white space
    SetCursor(GetCursor() - 1);
}



// --------------------------------------------------------------------
// GetChar
// --------------------------------------------------------------------

UInt16 HTMLParser::GetChar(void)
{
    if (GetCursor() >= GetFinalCursor())
    {
        Abort();
        return 0;
    }
    else
    {
        Byte c;
        // ??? Should use FillParseTable to take into account
        // 2-bytes characters
        c = fText[++fCursor];
        return c;
    }
}



// --------------------------------------------------------------------
// FindChar
// --------------------------------------------------------------------
//
// Find a character in the text stream, return its index
// -1 if not found
// Do not change the cursor

SInt32 HTMLParser::FindChar(UInt16 c)
{
    SInt32 result;
    SInt32 saveCursor = GetCursor();

    Byte currentChar;

    do
    {
        currentChar = GetChar();
    } while (fContinue && currentChar != c);

    if (!fContinue)
        result = -1;
    else
        result = GetCursor();

    SetCursor(saveCursor);

    return result;
}



// --------------------------------------------------------------------
// HTMLToStyledText
// --------------------------------------------------------------------

HTMLToStyledText::HTMLToStyledText(CRect frame, Handle text) :
    HTMLParser(text),
    fFrame(frame),
    fStyledText(NULL),
    fLastTextStyleChange(0),
    fBold(0),
    fItalic(0)
{
//    fStyledText = TEStyleNew(fFrame, fFrame);
}



// --------------------------------------------------------------------
// GetStyledText
// --------------------------------------------------------------------

TEHandle HTMLToStyledText::GetStyledText(void)
{
    return fStyledText;
}



// --------------------------------------------------------------------
// ApplyTextStyles
// --------------------------------------------------------------------

void HTMLToStyledText::ApplyTextStyles(SInt32 upToIndex)
{
    assert(upToIndex >= -1 && upToIndex <= GetFinalCursor());

    if (fStyledText == NULL)
    {
        fStyledText = TEStyleNew(fFrame, fFrame);
    }

    {
        {
            SInt32 endOfText = GetHandleSize((**fStyledText).hText);

            OSErr err = PtrAndHand(&fText[fLastTextStyleChange],
                        (**fStyledText).hText,
                        (upToIndex - fLastTextStyleChange) + 1);
            assert(err == noErr);

            TESetSelect(endOfText, 32000, fStyledText);
        }
        {
            TextStyle newStyle;
            newStyle.tsFace = normal;
            newStyle.filler = 0;        // Workaround for style
                                        // TextEdit bug see TE.16
            if (fBold > 0)
                newStyle.tsFace |= bold;
            if (fItalic > 0)
                newStyle.tsFace |= italic;

            TESetStyle(doFace, &newStyle, false, fStyledText);
        }
    }
}



// --------------------------------------------------------------------
// TagBoldStart
// --------------------------------------------------------------------

void HTMLToStyledText::TagBoldStart(SInt32 tagStart, SInt32 tagEnd)
{
    ApplyTextStyles(tagStart - 1);
    fLastTextStyleChange = tagEnd + 1;
    fBold++;
}



// --------------------------------------------------------------------
// TagBoldEnd
// --------------------------------------------------------------------

void HTMLToStyledText::TagBoldEnd(SInt32 tagStart, SInt32 tagEnd)
{
    ApplyTextStyles(tagStart - 1);
    fLastTextStyleChange = tagEnd + 1;
    fBold--;
}



// --------------------------------------------------------------------
// TagItalicStart
// --------------------------------------------------------------------

void HTMLToStyledText::TagItalicStart(SInt32 tagStart, SInt32 tagEnd)
{
    ApplyTextStyles(tagStart - 1);
    fLastTextStyleChange = tagEnd + 1;
    fItalic++;
}



// --------------------------------------------------------------------
// TagItalicEnd
// --------------------------------------------------------------------

void HTMLToStyledText::TagItalicEnd(SInt32 tagStart, SInt32 tagEnd)
{
    ApplyTextStyles(tagStart - 1);
    fLastTextStyleChange = tagEnd + 1;
    fItalic--;
}



// --------------------------------------------------------------------
// EndOfText
// --------------------------------------------------------------------
//
// endOfText is the index to the last character

void HTMLToStyledText::EndOfText(SInt32 endOfText)
{
    if (fStyledText != NULL)
    {
        ApplyTextStyles(endOfText);
        TESetSelect(GetHandleSize((**fStyledText).hText) - 1,
                    GetHandleSize((**fStyledText).hText), fStyledText);
        {
            TextStyle newStyle;
            newStyle.tsFace = bold;
            newStyle.filler = 0;        // Workaround for style
                                        // TextEdit bug see TE.16
            TESetStyle(doFace, &newStyle, false, fStyledText);
        }
        TECalText(fStyledText);
    }
}



// --------------------------------------------------------------------
// ConvertHTMLToStyledText
// --------------------------------------------------------------------
// Parses "text". If it contains HTML tags, create a new TextEdit
// record with all the styles set appropriately.
// Returns NULL otherwise.

void ConvertHTMLToStyledText(Handle text, CRect frame,
                                                    TEHandle* textEdit)
{
    HTMLToStyledText parser(frame, text);

    parser.Parse();

    *textEdit = parser.GetStyledText();
}