xmlwf.c

一个基于SAX分析方法的XML文档解析函数包

//#include "CommonData.h"
#include "StringProcess.h"
#include "CharSet.h"
#include "XMLwf.h"
#include "NetWork.h"
#include <stdio.h>
#include <time.h>
#include <stdlib.h>
#include <string.h>

#define XML_ASSERT(x)  if(x==NULL){((PXML_Parser)parser)->parserErrorHandler(((PXML_Parser)parser)->userData,"Error_in_system:Out_Of_memory");return NULL;}
#define XML_ASSERT_VOID(x)  if(x==NULL){((PXML_Parser)parser)->parserErrorHandler(((PXML_Parser)parser)->userData,"Error_in_system:Out_Of_memory");return;}

/*start_attribute define*/
/*define a element structure of the attibute and the operations about it.*/
typedef struct attElement{
	PXML_String name;
	PXML_String value;
	struct attElement* next;
} AttriPairs;
typedef AttriPairs* PAttriPairs;

void AttriPairsFree(PAttriPairs p)
{
	free(p->name);
	free(p->value);
	free(p);
}

PAttriPairs AttriPairsCreate(const XML_Char* name, const XML_Char* value)
{
	enum XML_STR_PRO_Error t;
	PAttriPairs pAP = malloc(sizeof(AttriPairs));
	if(pAP == NULL) return NULL;
	pAP->name = CreateXML_String();
	if(pAP->name  == NULL) {
		free(pAP);
		return NULL;
	}

	pAP->value = CreateXML_String();
	if(pAP->value == NULL){
		free(pAP);
		return NULL;
	}
	t = XML_StrPCharCat(pAP->name, name);
	if(t != STRING_SUCCESS ){
		free(pAP);
		return NULL;
	}
	t = XML_StrPCharCat(pAP->value, value);
	if(t != STRING_SUCCESS ){
		free(pAP);
		return NULL;
	}

	pAP->next = NULL;
	return pAP;
}

/*define attibutes structures and the operations on it.*/
typedef struct{
	int size;
	PAttriPairs pairsHead;
	PAttriPairs pairsTail;
} Attributes;
typedef Attributes* PAttributes;

PAttributes AsPAttributes(void* p)
{
	return (PAttributes)p;
}

void IniAttributes(PAttributes p)
{
	p->size = 0;
	p->pairsHead = NULL;
	p->pairsTail = NULL;
}

PAttributes AtrributesCreate()
{
	PAttributes pAttri = malloc(sizeof(Attributes));
	if(!pAttri) return NULL;
	IniAttributes(pAttri);
	return pAttri;
}

void AttributesFree(PAttributes pAttri)
{
	int i = 0;
	PAttriPairs current = NULL;
	PAttriPairs next = NULL;
	
	if(!pAttri) return;
	current = pAttri->pairsHead;
	if(current) next = current->next;

	for(;i<pAttri->size; i++){		
		free(current);
		current = next;
		if(next) next = next->next;
	}
	
	free(pAttri);	
}

void addAttribute(Patrributes desti, PAttriPairs  attri)
{
	PAttributes destination = AsPAttributes(desti);

	if(!desti || !attri) return;

	if(!destination->size){
		destination->pairsHead = attri;
		destination->pairsTail = attri;
	}else{
		destination->pairsTail->next = attri;
		destination->pairsTail = attri;
	}	
	destination->size++;
}

enum XML_Error addAttriPairs(PAttributes destination, const XML_Char* name, const XML_Char* value)
{
	PAttriPairs attri = AttriPairsCreate(name, value);
	if(attri == NULL) return ADD_ATTRI_ERROR;
	addAttribute(destination, attri);
	return XML_SUCCESS;
}

PAttriPairs GetAttributeByIndex(const PAttributes attributes, int index)
{
	int i = 0;
	PAttriPairs current = NULL;

	if(!attributes) return NULL;	
	if(index >=attributes->size ) return NULL;	

	current = attributes->pairsHead;	
	for(;i < index;i++)
		current = current->next;
	return current;
}

PAttriPairs GetAttributeByName(const PAttributes attributes, const XML_Char* name)
{
	int i = 0;
	PAttriPairs current = NULL;

	if(!attributes) return NULL;
	if(!attributes->size) return NULL;

	current = attributes->pairsHead;
	for(;i < attributes->size;i++){
		if(!XML_StrPCharCmp(current->name, name)) return current;
		current = current->next;
	}

	return NULL;
}

const XML_Char* __GetNameFromAttriByIndex(void* attri, int index)
{
	PAttriPairs p = NULL;
	PAttributes attributes = AsPAttributes(attri);
	p = GetAttributeByIndex(attributes, index);
	if(p) return p->name->data;
	else return NULL;
}

const XML_Char* __GetValueFromAttriByIndex(void* attri, int index)
{
	PAttriPairs p = NULL;
	const PAttributes attributes = AsPAttributes(attri);
	p = GetAttributeByIndex(attributes, index);
	if(p) return p->value->data;
	else return NULL;
}

const XML_Char* __GetValueByName(void* attri, const XML_Char* name)
{
	const PAttributes attributes = AsPAttributes(attri);
	PAttriPairs attriPairs = NULL;
	attriPairs = GetAttributeByName(attributes, name);
	if(!attriPairs) return NULL;
	return attriPairs->value->data;
}
//end attribute define


_BOOL IsMemberOf(const XML_Char member, const XML_Char* set, int sizeOfSetElement)
{
	int size = sizeOfSetElement;
	int i =0;

	for(;i < size; i++)		
		if(member == set[i]) return TRUE;	
	
	return FALSE;
}

_BOOL IsMemberOfInt(const int member, const int* set, int sizeOfSetElement)
{
	int size = sizeOfSetElement;
	int i =0;

	for(;i < size; i++)
		if(member == set[i]) return TRUE;	

	return FALSE;
}

//begin the functions to process the recusion down analysis.
int row = 1;
int col = 1;

XML_Char buffer[BUF_SIZE + 2];/* The buffer to store the XML_Char from data source.*/
int		 cursor = 2; /*The cursor point the current XML_char in use in 'buffer'.*/
int		 inputType;/*Record the type of input data source.*/
unsigned int sock = -1;/*The SOCKET for data getting from web.*/
FILE* inputStream = NULL;/*The pointer of the stream of file. */
int	     encodingType = ANSI;/*The type of the file*/

void RefreshBuffer()
{
	int				i;
	unsigned short  in;
	unsigned char	tmpChar;
#ifdef XML_UNICODE
	unsigned char	tmpBuf[BUF_SIZE];	
#else
#endif

	buffer[0] = buffer[BUF_SIZE];
	buffer[1] = buffer[BUF_SIZE + 1];
	if(inputType == XML_FILE_NAME || inputType == XML_FILE_STREAM)
		if(encodingType == UNICODE_){
			for(i = 2;i<BUF_SIZE + 2; i++){
				if(feof(inputStream))  break;
				buffer[i] = fgetc(inputStream);
				if(feof(inputStream))  break;
				in = fgetc(inputStream);				
				if(in){
					buffer[i]	= buffer[i]&0x00FF;
					in			= in<<8;									
					buffer[i]	= buffer[i]|in;
				}
			}
		}else if(encodingType == UNICODE_B){
			for(i = 2;i<BUF_SIZE + 2; i++){
				if(feof(inputStream))  break;
				buffer[i] = fgetc(inputStream);				
				if(feof(inputStream))  break;
				in = fgetc(inputStream);
				if(in == 0xFFFF) return;
				if(in){
					buffer[i]	= buffer[i]<<8;
					in			= in&0x00FF;														
					buffer[i]	= buffer[i]|in;
				}		
			}
		}else{
			for(i = 2;i<BUF_SIZE + 2; i++){
				if(feof(inputStream))  break;
				buffer[i]  = fgetc(inputStream);				
			}
		}					
		else if(encodingType == UNICODE_){
			for(i = 2;i<BUF_SIZE + 2; i++){
				if(!HttpStreamGet(sock, &tmpChar, 1))  break;
				buffer[i] = tmpChar;
				if(!HttpStreamGet(sock, &tmpChar, 1))  break;
				in = tmpChar;
				if(in){
					buffer[i]	= buffer[i]&0x00FF;
					in			= in<<8;									
					buffer[i]	= buffer[i]|in;
				}
			}
		}else if(encodingType == UNICODE_B){
			for(i = 2;i<BUF_SIZE + 2; i++){
				if(!HttpStreamGet(sock, &tmpChar, 1))  break;
				buffer[i] = tmpChar;
				if(!HttpStreamGet(sock, &tmpChar, 1))  break;
				in = tmpChar;
				if(in){
					buffer[i]	= buffer[i]<<8;
					in			= in&0x00FF;														
					buffer[i]	= buffer[i]|in;
				}		
			}
		}else{
#ifdef XML_UNICODE
		HttpStreamGet(sock, tmpBuf, BUF_SIZE);
		for(i = 2;i<BUF_SIZE + 2; i++)
			buffer[i] = tmpBuf[i-2];		
#else
		HttpStreamGet(sock, &buffer[2], BUF_SIZE);
#endif			
		}	
	cursor = 2;		
}

XML_Char GetChar()
{
	if(cursor > BUF_SIZE+1) RefreshBuffer();	
	col++;
	return buffer[cursor++];	
}

void GoBack(int step)
{
	if(step == 1) cursor--;
	else cursor = cursor - 2;	
}

void JudgeEncodingType()
{
	int				tmp		= 0;
	unsigned char	nTmp	= 0;

	if(inputType == XML_FILE_NAME || inputType == XML_FILE_STREAM){
		tmp = fgetc(inputStream);
		if(tmp == 0xFF){
			tmp = fgetc(inputStream);
			if(tmp == 0xFE) encodingType = UNICODE_;
			else exit(0);
		}else if(tmp == 0xEF){
			tmp = fgetc(inputStream);
			if(tmp == 0xBB){
				tmp = fgetc(inputStream);
				if(tmp == 0xBF){ encodingType = UTF_8;}
				else exit(0);
			}else exit(0);
		}else if(tmp == 0xFE){
			tmp = fgetc(inputStream);
			if(tmp == 0xFF) encodingType = UNICODE_B;
			else exit(0);
		}else if(tmp == 0xFF){
		}else{
			buffer[1]	= tmp;
			cursor		= 1; 
			encodingType= ANSI;
		}
	}else{
		HttpStreamGet(sock, &nTmp, 1);
		if(nTmp == 0xFF){
			HttpStreamGet(sock, &nTmp, 1);
			if(nTmp == 0xFE) encodingType = UNICODE_;
			else exit(0);
		}else if(nTmp == 0xEF){
			HttpStreamGet(sock, &nTmp, 1);
			if(nTmp == 0xBB){
				HttpStreamGet(sock, &nTmp, 1);
				if(nTmp == 0xBF) encodingType = UTF_8;
				else exit(0);
			}else exit(0);
		}else if(nTmp == 0xFE){
			HttpStreamGet(sock, &nTmp, 1);
			if(nTmp == 0xFF) encodingType = UNICODE_B;
			else exit(0);
		}else encodingType= ANSI;		
	}

	RefreshBuffer();
	if(encodingType == ANSI){
		if(inputType == XML_FILE_NAME || inputType == XML_FILE_STREAM)	buffer[1] = tmp;
		else buffer[1] = nTmp;
		cursor = 1;
	}
}

int initialGlobalVar(unsigned int s, FILE* pf, int type)
{
	if(type<XML_FILE_NAME ||type>XML_URL) return FALSE;
	sock        = s;
	inputStream = pf;
	inputType  = type;
	memset(buffer, '\0', BUF_SIZE+2);
	cursor      = BUF_SIZE+2;
	row         = 1;
	col			= 1;
	encodingType= ANSI;

	JudgeEncodingType();
	return TRUE;
}

void ErrorHappen(Pxml_parser parser,   const char* errorInf)
{
	FILE* log = fopen("c:\\XMLParser.log", "a+");
	if( log != NULL){
		fputs(errorInf, log);
		fputc('\n', log);
	}
	fclose(log);
	
	printf("Error in line %d, colomn %d. %s", row, col, errorInf);

	if(((PXML_Parser)parser)->parserErrorHandler != NULL)    
		((PXML_Parser)parser)->parserErrorHandler( ((PXML_Parser)parser)->userData, errorInf);
	exit(1);
}
/**
Skip the character in 'space'
*/