/*++
Copyright (c) 1996 Microsoft Corporation
Module Name :
linkpars.cpp
Abstract:
Link parser class implementation. This class responsible for
parsing the html file for hyperlink.
Author:
Michael Cheuk (mcheuk)
Project:
Link Checker
Revision History:
--*/
#include "stdafx.h"
#include "LinkPars.h"
#include "link.h"
#include "lcmgr.h"
#ifdef _DEBUG
#define new DEBUG_NEW
#undef THIS_FILE
static char THIS_FILE[] = __FILE__;
#endif
// Constants
const CString strLocalHost_c(_T("localhost"));
void
CLinkParser::Parse(
const CString& strData,
const CString& strBaseURL,
CLinkPtrList& rLinkPtrList
)
/*++
Routine Description:
Parse a page of html data
Arguments:
strData - page of html
strBaseURL - base URL
rLinkPtrList - reference to links list. The new links will
will be added to this list.
Return Value:
N/A
--*/
{
// Look for the first '<'
LPCTSTR lpszOpen = _tcschr(strData, _TUCHAR('<'));
while(lpszOpen != NULL)
{
// Look for the '>'
LPCTSTR lpszClose = _tcschr(lpszOpen, _TUCHAR('>'));
if(lpszClose)
{
// The possible tag must be longer than 7 bytes (a href=)
int iCount = (int)(lpszClose - lpszOpen) - 1; // skip the '<'
if( iCount > 7 )
{
int iIndex = lpszOpen - ((LPCTSTR)strData) + 1; // skip the '<'
CString strPossibleURL(strData.Mid(iIndex, iCount));
// Parse the possible tag
if(ParsePossibleTag(strPossibleURL))
{
CString strURL;
BOOL fLocalLink;
// We found a valid tag. Time to create new link.
if( CreateURL(strPossibleURL, strBaseURL, strURL, fLocalLink) )
{
rLinkPtrList.AddLink(strURL, strBaseURL, strPossibleURL, fLocalLink);
}
}
}
}
// Look for the next '<'
lpszOpen = _tcschr(++lpszOpen, _TUCHAR('<'));
}
} // CLinkParser::Parse
BOOL
CLinkParser::ParsePossibleTag(
CString& strTag
)
/*++
Routine Description:
Parse a single "<.....>" for possible hyperlink
Arguments:
strTag - value inside a "<.....>" excluding '<' & '>'
If this is a hyperlink tag, the hyperlink URL
will be put in strTag.
Return Value:
BOOL - TRUE if hyperlink tag. FALSE otherwise.
--*/
{
// Make a working copy
CString strWorkCopy(strTag);
// Let's work with lower case
strWorkCopy.MakeLower();
//
// Check for,
//
// HyperLink:
//
//
//
//
// CGI
//
//
// Style Sheet
//
//
if( strWorkCopy[0] == _T('a') ||
strWorkCopy.Find(_T("link")) == 0 )
{
return GetTagValue(strTag, CString(_T("href")));
}
//
// Check for,
//
//
//
// Table:
//