xml.cpp

提供windows和unix下各种级别日志,包括 xml

#include "stdafx.h"
#include "xml.h"

#ifdef _DEBUG
#define new DEBUG_NEW
#undef THIS_FILE
static char THIS_FILE[] = __FILE__;
#endif

#define XML_TOKEN_START    1  // <
#define XML_TOKEN_END      2  // >
#define XML_TOKEN_START2   3  // </
#define XML_TOKEN_END2     4  // />
#define XML_TOKEN_PI_START 5  // <?
#define XML_TOKEN_PI_END   6  // ?>
#define XML_TOKEN_TEXT     7
#define XML_TOKEN_NAME     8
#define XML_TOKEN_STRING   9  // "xxx" or 'xxx'
#define XML_TOKEN_EQUAL    10 // =
#define XML_TOKEN_EOF      11
#define XML_TOKEN_CDATA    12
#define XML_TOKEN_INVALID  13

char *tokenTypeStrings[] = 
{
   "", //base 0
   "<",
   ">",
   "</",
   "/>",
   "<?",
   "?>",
   "text",
   "tag/attrname",
   "string",
   "=",
   "EOF",
   "CDATA",
   "invalid",
};


void ASSERT_TOKEN(int x); 

static int bReadTag = 0;
static int tokenType = XML_TOKEN_EOF;
static char tokenValue[MAX_TOKEN_LEN+1];
static char textValue[MAX_TEXT_LEN+1];
static char *pCurrent = NULL;
static char *pBuffer = NULL;
static int lineNumber = 0;
static int colNumber = 0;

void strncpyx(char *dest, const char *src, int len)
{
    if(src==NULL)
    {
        dest[0]=0;
        return;
    } 
    strncpy(dest, src, len-1);
    dest[len-1] = 0;
}

TXMLAttr::TXMLAttr()
{
   m_strName[0]=0;
   m_strValue=NULL;
   brother = NULL;
   lineNo = 0;
}

TXMLAttr::~TXMLAttr()
{
    if(m_strValue)
        delete m_strValue;
}

void TXMLAttr::setName(const char* src)
{
   if(strlen(src) >ATTRNAME_LENGTH)
   {
      strncpy(m_strName,src,ATTRNAME_LENGTH);
      m_strName[ATTRNAME_LENGTH]=0;
   }
   else
   {
       strcpy(m_strName,src);
   }
}

char* TXMLAttr::getName()
{
   return m_strName;
}

void TXMLAttr::setValue(const char* src)
{
   if(m_strValue)
       delete m_strValue;
   m_strValue=new char[strlen(src)+1];
   strcpy(m_strValue,src);
}

char* TXMLAttr::getValue()
{
   return m_strValue;
}

TXMLAttr* TXMLAttr::dup()
{
   TXMLAttr* attr=new TXMLAttr;
   attr->setName(m_strName);
   attr->setValue(m_strValue);
         
   if(brother!=NULL)
      attr->brother=brother->dup();
   attr->lineNo=lineNo;
   return attr;
}
char* TXMLAttr::toStr()
{
   char* s=new char[strlen(m_strName) + strlen(m_strValue) + 4];
   sprintf(s,"%s=\"%s\"",m_strName,m_strValue);
   return s;
}

TXMLElem::TXMLElem()
{
   child=NULL;
   brother=NULL;
   parent=NULL;
   attr = NULL;
   lineNo = 0;

   m_strTag[0]=0;
   m_strContent=NULL;
}

TXMLElem::~TXMLElem()
{
   TXMLElem* p=child;
   while(p!=NULL)
   {
      TXMLElem *pNext = p->brother;
      delete p;
      p = pNext;
   }

   {
      TXMLAttr* p=attr;
      while(p!=NULL)
      {
         TXMLAttr *pNext = p->brother;
         delete p;
         p = pNext;
      }
   }

   if(m_strContent)
   {
       delete m_strContent;
       m_strContent=NULL;
   }

}

TXMLElem* TXMLElem::dup(TXMLElem* aParent)
{
   TXMLElem* elem=new TXMLElem;
   elem->setTag(m_strTag);
   elem->setContent(m_strContent);
  
   TXMLElem* pChild=child;
   while(pChild!=NULL)
   {
      TXMLElem* pDupChild=pChild->dup(elem);
      elem->addChild(pDupChild);
      pChild=pChild->brother;
   }
   elem->parent=aParent;

   if(attr!=NULL)
      elem->attr=attr->dup();
   elem->lineNo=lineNo;
   return elem;
}

void TXMLElem::setTag(const char* src)
{
   if(strlen(src) >TAG_LENGTH)
   {
      strncpy(m_strTag,src,TAG_LENGTH);
      m_strTag[TAG_LENGTH]=0;
   }
   else
   {
       strcpy(m_strTag,src);
   }
}

void TXMLElem::setContent(const char* src)
{
   if(m_strContent)
       delete m_strContent;

   m_strContent = new char[strlen(src)+1];
   strcpy(m_strContent,src);
}

char* TXMLElem::getTag()
{
   return m_strTag;
}
char* TXMLElem::getContent()
{
   return m_strContent;
}

static char *pLast = NULL;
void GetLocation(char *pCurr)
{
   char *p = pLast;
   while(p != pCurr)
   {
      if(*p == '\n')
      {
         lineNumber ++;
         colNumber = 1;
      }
      colNumber ++;
      p++;
   }
   pLast = pCurr;
}

#define TOKEN_HEAD 10

void GetToken()
{
   tokenValue[0] = 0;
   textValue[0] = 0;
   char *p = pCurrent;
   if(bReadTag)
   {
      //skip blank
      while(*p==' ' || *p=='\t' || *p=='\n' || *p=='\r')
      {
         p++;
      }
      //only handle NAME, STRING, EQUAL, END, END2
      GetLocation(p);
      if(*p==0)
      {
         tokenType = XML_TOKEN_EOF;
      }
      else if(strncmp(p, "/>", 2) == 0)
      {
         tokenType = XML_TOKEN_END2;
         strcpy(tokenValue, "/>");
         p+=2;
         bReadTag = 0;
      }
      else if(*p=='>')
      {
         tokenType = XML_TOKEN_END;
         strcpy(tokenValue, ">");
         p++;
         bReadTag = 0;
      }
      else if(*p=='=')
      {
         tokenType = XML_TOKEN_EQUAL;
         strcpy(tokenValue, "=");
         p++;
      }
      else if(*p=='"' || *p=='\'')
      {
         char beginChar=*p;
         p++;
         char *pBegin = p;
//         while(*p !=0  && *p !='\n' && *p != beginChar)
         while(*p !=0  && *p != beginChar)
            p++;
         if( *p != beginChar || p - pBegin+1 > MAX_TOKEN_LEN )
         {
            //too long string
            strncpyx(tokenValue, pBegin, TOKEN_HEAD);
            strcat(tokenValue, "(Too long)");
            tokenType = XML_TOKEN_INVALID;
         }
         else
         {
            tokenType = XML_TOKEN_STRING;
            strncpyx(tokenValue, pBegin, p - pBegin+1);
            xmlTextDecode(tokenValue);
            p++;
         }
      }
      else if(isalpha(*p))
      {
         char *pBegin = p;
         p++;
         while(isalpha(*p) || *p>='0' && *p<='9' || *p == '_' || *p == ':'||*p == '-')
         {
            p++;
         }
         if(p - pBegin+1 > MAX_TOKEN_LEN )
         {
            //too long name
            strncpyx(tokenValue, pBegin, TOKEN_HEAD);
            strcat(tokenValue, "(Too long)");
            tokenType = XML_TOKEN_INVALID;
         }
         else
         {
            tokenType = XML_TOKEN_NAME;
            strncpyx(tokenValue, pBegin, p - pBegin+1);
            
            //luojianmin add on 2004/05/13            
            char* tok=strchr(tokenValue,':');
            if(tok!=NULL && strchr(tok+1,':')!=NULL)
               tokenType = XML_TOKEN_INVALID;
         }
      }
      else
      {
            //unknown token
         strncpyx(tokenValue, p, TOKEN_HEAD);
         tokenType = XML_TOKEN_INVALID;
      }
   }
   else 
   {
restart:
      //skip blank
      while(*p==' ' || *p=='\t' || *p=='\n' || *p=='\r')
      {
         p++;
      }
      //only handle BEGIN, COMMENT, CDATA, PI, TEXT
      GetLocation(p);
      if(*p==0)
      {
         tokenType = XML_TOKEN_EOF;
      }
      else if(strncmp(p, "<!--", 4) == 0)
      {
         char *pTail = strstr(p, "-->");
         if(!pTail)
         {
            //invalid comment
            strcpy(tokenValue, "<!-- ... (NEED -->)");
            tokenType = XML_TOKEN_INVALID;
         }
         else 
         {
            p = pTail + 3;
            goto restart;
         }
      }
      else if(strncmp(p, "<?", 2) == 0)
      {
         char *pTail = strstr(p, "?>");
         if(!pTail)
         {
            //invalid Process Instrution
            strcpy(tokenValue, "<? ... (NEED ?>)");
            tokenType = XML_TOKEN_INVALID;
         }
         else 
         {
            p = pTail + 2;
            goto restart;
         }
      }
      else if(strncmp(p, "<![CDATA[", 9) == 0)
      {
         char *pTail = strstr(p, "]]>");
         if(!pTail)
         {
            //invalid Process Instrution
            strcpy(tokenValue, "<![CDATA ... (NEED ]]>)");
            tokenType = XML_TOKEN_INVALID;
         }
         else 
         {
            //don't check it's length?
            if(pTail - (p+9) >MAX_TEXT_LEN)
            {
               strncpyx(textValue, p, TOKEN_HEAD);
               strcat(textValue, "(Too long)");
               tokenType = XML_TOKEN_INVALID;
            }
            else
            {
               tokenType = XML_TOKEN_CDATA;
               strncpyx(textValue, p+9, pTail-(p+9)+1);
               p = pTail + 3;
            }
         }
      }
      else if(strncmp(p, "</", 2) == 0)
      {
         tokenType = XML_TOKEN_START2;
         strcpy(tokenValue, "</");
         p+=2;
         bReadTag = 1;
      }
      else if(*p=='<')
      {
         tokenType = XML_TOKEN_START;
         strcpy(tokenValue, "<");
         p++;
         bReadTag = 1;
      }
      else
      {
         //we think this is a text
         char *pBegin = p;
         while(*p !=0  && *p !='<')
            p++;
         if(p - pBegin+1 > MAX_TEXT_LEN )
         {
            //too long text
            strncpyx(textValue, pBegin, TOKEN_HEAD);
            strcat(textValue, "(Too long)");
            tokenType = XML_TOKEN_INVALID;
         }
         else
         {
            tokenType = XML_TOKEN_TEXT;
            strncpyx(textValue, pBegin, p - pBegin+1);
            xmlTextDecode(textValue);
         }
      }
   }
   pCurrent = p;
}

void TXMLElem::addAttr(TXMLAttr* pAttr)
{
   if(attr==NULL)
   {
      attr=pAttr;
      return;
   }
   TXMLAttr* p=attr;
   while(p->brother!=NULL)
      p=p->brother;
   p->brother=pAttr;
}

char* TXMLElem::GetAttrValue(const char *name)
{
   TXMLAttr* p=GetAttr(name);
   if(p)
   {
       return p->getValue();
   }
   else
      return NULL;
}

TXMLAttr *TXMLElem::GetAttr(const char *name)
{
   for(TXMLAttr* p=attr; p!=NULL; p=p->brother)
   {
      if(stricmp(name, p->getName()) == 0)
         return p;
   }
   return NULL;
}

int TXMLElem::GetAttrNum()
{
   int i=0;
   for(TXMLAttr* p=attr; p!=NULL; p=p->brother)
   {
      i++;
   }
   return i;
}


TXMLAttr *TXMLElem::GetAttr(int index)
{
   int i=0;
   for(TXMLAttr* p=attr; p!=NULL; p=p->brother)
   {
      if(i == index)
         return p;
      i++;
   }
   return NULL;
}

TXMLElem *TXMLElem::GetSubElem(const char *tag1)
{
   for(TXMLElem* p=child; p!=NULL; p=p->brother)
   {
      if(stricmp(tag1, p->getTag()) == 0)
         return p;
   }
   return NULL;
}

int TXMLElem::GetSubElemNum()
{
   int i=0;
   for(TXMLElem* p=child; p!=NULL; p=p->brother)
   {
      i++;
   }
   return i;
}

int TXMLElem::GetSubElemNumByTag(char* strTag)
{
   if (strTag==NULL) return 0;
   int i=0;
   for(TXMLElem* p=child; p!=NULL; p=p->brother)
   {
      if(p->getTag()!=NULL && stricmp(p->getTag(),strTag)==0)
         i++;
   }
   return i;
}

int TXMLElem::GetSubElemName(int index,char* name)
{
   TXMLElem* p=GetSubElem(index);
   if(p!=NULL)