htmlout.c

将rft格式文件转换成html格式文件

#include	<stdio.h>
#include	"rtf.h"
#include	"rtftohtml.h"
#include	"string.h"
static void 	AppendText();


static void	PutHex ();
static char	*MapChar();

#define MAXNEST 8

/* Destinations can be files or strings.  */


struct SDest

HName = {
    NULL, 0, 0
},
HHref = {
    NULL, 0, 0
},
Title = {
    NULL, 0, 0
},
SkipCol = {
    NULL, 0, 0
},
BkName = {
    NULL, 0, 0
},
Item = {
    NULL, 0, 0
};

/* SSC codes are used in saving the state */
#define SSCSPECIAL	0
#define SSCSTYLECHG	1
#define SSCNCODES	2	/* Set this to number of SSCodes */

struct SSData {
	int Special;
	int cpos;
	InStateStack *TheState;
};

struct SSStruct {
	struct SDest SD;
	int alloc;
	int used;
	struct SSData *data;
} PNSave,FNSave;

TMatchRec FNMatch;

char * (badstyles[256]);	
int anyfoot;
int anyTOC;
int InBody = 0;
int PNText = 0;
int FNText = 0;
int HToCLev = 0;
int oToCLev = 0;
int ToCidx=0;
int ToCText=0;

struct FDest {
    char filename[256];		/* Name of output file */
    char wbuff[82];		/* Output buffer */
    FILE *outfd;		/* Output file - (file destinations only) */
    int wrap;		    /* should we wrap text or not */
    int HPTags;			/* Highest Paragraph Tag */
    int HTTags;			/* Highest Text Tag */
    int NeedPar;		/* Output Paragraph before next text */
    int InCol2;			/* We are past Column 1 */
    int PTidx;			/* Paragraph Tag Index */
    int TSpecial;		/* Special Tag */
    int cpos;			/* character position */
    int TStack[MAXNEST];	/* Stack of tags */
}
 EmptyFD = {
    "", "",NULL,1, -1, -1, 0, 0, 0,0, 0
},
 File1 = {
    "","", NULL,1, -1, -1, 0, 0, 0,0, 0
},
ToCFile = {
    "","", NULL,1, -1, -1, 0, 0, 0,0, 0
},
Footnote = {
    "","", NULL,1, -1, -1, 0, 0, 0,0, 0
};

#define ConvertCaps	((TStyle&(styleAllCaps|styleSmallCaps))!=0)
int outfuniq = 0;

char *LineBuffer = (char *) NULL;

FILE *PictFile;
int uniqnum = 0;

#define MAXOBUF	8096
int
nextfield(cp, w,  ob,  adj,  EntIsOne)
char *cp;int w; char *ob; int adj; int EntIsOne;
{
	int i,j,br = -1,lp,rp,nb= -1,ow=w;
	for(i=0;i<w&&cp[i]!='\0';i++){
		if(cp[i]==' '||cp[i]=='\t'){
			if(nb==i-1)br=i;
		}	else {
			if(EntIsOne){
				if(cp[i]=='&'){
					while(cp[i]!=';'&&cp[i]!='\0'){ 
						w++;
						i++;
					}
				} else if(cp[i]=='<'){
				    while(cp[i]!='>'&&cp[i]!='\0'){
						w++;
						i++;
					}
				}
			}
			nb=i;
		}
	}
	if(i==0){
		ob[0]='\0';
		return(0);
	}
	if(cp[i]=='\0'&&nb==i-1)br=i;
	if(br==-1)br=w;
	if(w!=ow){ /* now subtract off any entities not used */
		for(i=br,j=0;i<w&&cp[i]!='\0';i++){
			if(cp[i]=='&'){
				while(cp[i]!=';'){
					j++;
					i++;
				}
			}
		}
		w-=j;
	}
	switch(adj){
	case AdjL:			/* left justify - no pad */
		lp=0;
		rp=br;
		break;
	case AdjLPad:			/* left justify */
		lp=0;
		rp=w;
		break;
	case AdjRPad:			/* right justify */
		lp=w-br;
		rp=w;
		break;
	case AdjCpad:			/* center */
		lp=(w-br)/2;
		rp=w;
		break;
	}
	if(rp>=MAXOBUF){
		RTFPanic ("Overflowed MAXOBUF too many Entities in a table cell");
	}
	for(i=0;i<lp;i++)ob[i]=' ';
	for(j=0;j<br;i++,j++)ob[i]=cp[j];
	for(;i<rp;i++)ob[i]=' ';
	ob[i]='\0';
	return(br);
}

void
putwrap( s,  OD, wrap)
char *s;struct FDest *OD;int wrap;
{
	int i;
	int n;
	char obuff[133];
	/* If writing to a table, just save up the text */
	if(!wrap){
		if(OD->cpos>0){
			OD->wbuff[OD->cpos]='\0';
			fputs(OD->wbuff,OD->outfd);
			fputc('\n',OD->outfd);
		}
		fputs(s,OD->outfd);
		OD->cpos= -1;
		return;
	}
	i=OD->cpos;
	while( *s != '\0'){
		if(i==81){
			n=nextfield(OD->wbuff, 80,  obuff, AdjL,0);
			fputs(obuff,OD->outfd);
			if(n<80){	/* only if we have a break */
				fputc('\n',OD->outfd);
			    for(;n<81;n++)
					if(OD->wbuff[n]!=' '&&OD->wbuff[n]!='\t')
						break;
			    for(i=0;n<81;i++,n++)
			    	OD->wbuff[i]=OD->wbuff[n];
			} else  { /* force break at next opportunity */
				i= -1;
			}
		}
		if(i<0){
			if(strchr(" \t\n",*s)!=NULL){ /*found break */
				fputc('\n',OD->outfd);
				i=0;
				s++;
			} else{
				fputc(*(s++),OD->outfd);
			}
		} else if(*s=='\n'){
			OD->wbuff[i]='\0';
			fputs(OD->wbuff,OD->outfd);
			fputc(*(s++),OD->outfd);
			i=0;
			continue;
		} else
			OD->wbuff[i++] = *(s++);
	}
	OD->cpos=i;
}

HTMLInit()
{
	int i;

	if (LineBuffer == (char *) NULL
			&& (LineBuffer = RTFAlloc (512)) == (char *) NULL)
		RTFPanic ("cannot allocate line buffer.");
	for(i=0;i<256;i++)
		badstyles[i]=NULL;
	HName.used=0;
	HHref.used=0;
	Title.used=0;
	Item.used=0;
	SkipCol.used=0;
	File1=EmptyFD;
	Footnote=EmptyFD;
	ToCFile=EmptyFD;
	outfuniq=0;
	uniqnum=0;
	ToCidx=1;
	ToCText=0;
	anyfoot=0;
	anyTOC=0;
	InBody = 0;
	HToCLev=0;
	oToCLev = 0;
	for(i=0;i<PNSave.used;i++){
		if(PNSave.data[i].TheState){
			    RTFFree(PNSave.data[i].TheState);
		}
	}
	PNSave.SD.used=0;
	PNSave.used=0;
	for(i=0;i<FNSave.used;i++){
		if(FNSave.data[i].TheState){
			    RTFFree(FNSave.data[i].TheState);
		}
	}
	FNSave.SD.used=0;
	FNSave.used=0;
	FNMatch.Font= "NotAFont";

	
    strcpy(File1.filename, OutfileName);
    if (OutfileName[0] != '\0') {
	File1.outfd = OpenOutputFile(OutfileName, "w",FTTEXT);
	if (File1.outfd == NULL) {
	    RTFPanic ("Open of %s Failed", OutfileName);
	}
    } else {
	File1.outfd = stdout;
    }
    /* open a file for footnotes */
    sprintf(Footnote.filename, "%s_fn.html", FPrefix);
	Footnote.outfd = OpenOutputFile(Footnote.filename, "w",FTTEXT);
	if (Footnote.outfd == NULL) {
	    RTFPanic ("Open of %s Failed", Footnote.filename);
	}
	if(ToC){
	    /* open a file for Table of Contents */
	    sprintf(ToCFile.filename, "%s_ToC.html", FPrefix);
		ToCFile.outfd = OpenOutputFile(ToCFile.filename, "w",FTTEXT);
		if (ToCFile.outfd == NULL) {
		    RTFPanic ("Open of %s Failed", ToCFile.filename);
		}
		putwrap("<html><head>", &ToCFile,ToCFile.wrap);
	    putwrap("<!-- This document was created from RTF source by rtftohtml version ", &ToCFile,ToCFile.wrap);
	    putwrap(PVers, &ToCFile,ToCFile.wrap);
	    putwrap(" -->", &ToCFile,ToCFile.wrap);
		putwrap("</head><body>", &ToCFile,ToCFile.wrap);
	}
    putwrap("<html><head>", &File1,File1.wrap);
    putwrap("<!-- This document was created from RTF source by rtftohtml version ", &File1,File1.wrap);
    putwrap(PVers, &File1,File1.wrap);
    putwrap(" -->", &File1,File1.wrap);
    putwrap("<html><head>", &Footnote,Footnote.wrap);
    putwrap("<!-- This document was created from RTF source by rtftohtml version ", &Footnote,Footnote.wrap);
    putwrap(PVers, &Footnote,Footnote.wrap);
    putwrap(" -->", &Footnote,Footnote.wrap);
	putwrap("</head><body>", &Footnote,Footnote.wrap);
}
static void
TestStyle();

HTMLCleanup()
{
    char buff[256];
    	
    /* Fake some plain text so that test-style will undo all current markup. */
    TFont="--Not a Font--";
    TStyle=0;
    TSize=9999;
    ParStyle=PMatchArr[0].PStyle;
    TestStyle(&File1);
	if(anyfoot){	/* We have a footnote */
	    putwrap("</body></html>", &Footnote,0);
		fclose(Footnote.outfd);
	} else {
		fclose(Footnote.outfd);
		remove (Footnote.filename); 
	}
	if(ToC){
		if(anyTOC){	/* We have a Table of Contents */
		    while(oToCLev>0){
	    		putwrap("</ol>",&ToCFile,0);
	    		oToCLev--;
	    	}
		putwrap("</body></html>", &ToCFile,0);
			fclose(ToCFile.outfd);
		} else {
			fclose(ToCFile.outfd);
			remove (ToCFile.filename); 
		}
	}
    putwrap("</body></html>", &File1,0);
    if (File1.outfd != stdout)
	fclose(File1.outfd);
}

HTMLStartDivert(){
    struct FDest *OD;
	OD = &File1;
	switch (destination) {
    case rtfObjResult:
    	if(linkself&&Item.used>0&&!inTable){
    		sprintf(LineBuffer,"<a href=\"#%s\">",Item.ptr);
    		putwrap(LineBuffer,OD,0);
	    	Item.used=0;
	    }
		break;
	}
}

int BkNum=0;
HTMLEndDivert()
{
    struct FDest *OD;
	OD = &File1;
    if (ISS->destination != destination) {
		switch (destination) {
		case rtfPict:
		    if(PictFile)fclose(PictFile);
		    RTFSetClassCallback(rtfText, PutHTML);
		    break;
		case rtfFootnote:
		    Footnote.NeedPar = 1;
		    Footnote.InCol2 = 0;
		    break;
		case rtfObjResult:
	    	if(linkself){putwrap("</a>",OD,OD->wrap);}
			break;
	    case rtfBookmarkStart:
	    	if(BkName.used>0){
	    		sprintf(LineBuffer,"<a name=\"%s\">",BkName.ptr);
	    		putwrap(LineBuffer,OD,0);
	    		BkName.used=0;
	    	}else {
	    		sprintf(LineBuffer,"<a name=\"RTFtohtmlxx%d\">",BkNum++);
	    		putwrap(LineBuffer,OD,OD->wrap);
	    	}
			break;
	    case rtfBookmarkEnd:
	    	putwrap("</a>",OD,OD->wrap);
			break;
	    }
	}
	if(ISS->ToCLev != ToCLev&&ToC&&ToCLev>0){
	    	putwrap("</a>",&ToCFile,0);
	    	putwrap("</a>",OD,0);
	    	ToCText=0;
	}

}
int xval=0;
int charnum = 0;
extern int csTop;
void
ReadPict(){
	int mval;
	int level=0;
	int savecsTop;
	if(csTop>0)savecsTop=csTop-1;
	else savecsTop=0;

	while (1)
	{
		if (rtfClass == rtfEOF)
			break;
		if (rtfClass == rtfGroup)
		{
			if (rtfMajor == rtfBeginGroup)
				++level;
			else if (rtfMajor == rtfEndGroup)
			{
				if (--level < 1)
					break;	/* end of initial group */
			}
		} else {
			if(rtfMajor != '\n' && rtfMajor != '\r'
							&& rtfMajor != '\0'){
				if(rtfMajor <= '9'){
					mval=rtfMajor-'0';
				} else if(rtfMajor <= 'F'){
					mval=rtfMajor-'A'+10;
				} else {
					mval=rtfMajor-'a'+10;
				} 
			    if (++charnum == 2) {
			    	xval=xval*16+mval;
					fputc((char) xval, PictFile);
					charnum = 0;
			    } else {