/*********************************************************************
* Determines format of input file and calls parse_word_header or
* process_file if
* it is word processor file or copy_out if it is plain text file
* return not 0 when error
********************************************************************/
int analyze_format(FILE *f) {
unsigned char buffer[129];
long offset=0;
FILE *new_file, *ole_file;
int ret_code=69;
if (!signature_check) {
/* forced parsing */
/* no autodetect possible. Assume 8-bit if not overriden on
* command line */
if (!get_unicode_char)
get_unicode_char=get_8bit_char;
return process_file(f,LONG_MAX);
}
catdoc_read(buffer,4,1,f);
buffer[4]=0;
if (strncmp((char *)&buffer,write_sign,2)==0) {
printf("[Windows Write file. Some garbage expected]\n");
get_unicode_char=get_8bit_char;
return process_file(f,LONG_MAX);
} else if (strncmp((char *)&buffer,rtf_sign,4)==0) {
return parse_rtf(f);
} else if (strncmp((char *)&buffer, zip_sign,4) == 0) {
fprintf(stderr,"This file looks like ZIP archive or Office 2007 "
"or later file.\nNot supported by catdoc\n");
exit(1);
} else if (strncmp((char *)&buffer,old_word_sign,2)==0) {
fread(buffer+4,1,124,f);
return parse_word_header(buffer,f,128,0);
}
fread(buffer+4,1,4,f);
if (strncmp((char *)&buffer,ole_sign,8)==0) {
if ((new_file=ole_init(f, buffer, 8)) != NULL) {
set_ole_func();
while((ole_file=ole_readdir(new_file)) != NULL) {
int res=ole_open(ole_file);
if (res >= 0) {
if (strcmp(((oleEntry*)ole_file)->name , "WordDocument") == 0) {
offset=catdoc_read(buffer, 1, 128, ole_file);
ret_code=parse_word_header(buffer,ole_file,-offset,offset);
}
}
ole_close(ole_file);
}
set_std_func();
ole_finish();
} else {
fprintf(stderr,"Broken OLE file. Try using -b switch\n");
exit(1);
}
} else {
copy_out(f,(char *)&buffer);
return 0;
}
return ret_code;
}
/**
*
*
* @param input
* @param filename
*/
void do_ppt(FILE *input,char *filename) {
int itemsread=1;
int rectype;
long reclen;
unsigned char recbuf[8];
while(itemsread) {
itemsread = catdoc_read(recbuf, 1, 8, input);
/* fprintf(stderr,"itemsread=%d: ",itemsread); */
/* for(i=0; i<8; i++) */
/* fprintf(stderr,"%02x ",recbuf[i]); */
/* fprintf(stderr,"\n"); */
if (catdoc_eof(input)) {
process_item(DOCUMENT_END,0,input);
return;
}
if(itemsread < 8)
break;
rectype=getshort(recbuf,2);
reclen=getulong(recbuf,4);
if (reclen < 0) {
return;
}
process_item(rectype,reclen,input);
}
}
/* Reads file from MS-Word 97 and above file. Takes in account strange*
* situation that unicode and non-unicode 256-byte blocks could be *
* intermixed in word file *
* *
* Parameters: *
* *
* f - file to read *
* offset - position of the character inside file (to determine *
* possible block boundaries *
**********************************************************************/
int get_word8_char(FILE *f,long *offset,long fileend) {
int count,i,u;
char c;
if ((i=(*offset)%256) ==0) {
count=catdoc_read(read_buf,1,256,f);
memset(read_buf+count,0,256-count);
buf_is_unicode=0;
if (*offset+(long)count>fileend) {
count=fileend-*offset;
}
while (i<count) {
c=read_buf[i++];
if ((c==0x20|| c==0x0D||ispunct(c))&&i<count&&read_buf[i]==0) {
buf_is_unicode=1;
break;
}
i++;
}
i=0;
}
if (buf_is_unicode) {
u=read_buf[i] | read_buf[i+1]<<8;
(*offset)+=2;
} else {
u=to_unicode(source_charset,read_buf[i]);
(*offset)++;
}
return u;
}
void do_table(FILE *input,char *filename) {
long rectype;
long reclen,build_year=0,build_rel=0,offset=0;
int eof_flag=0;
int itemsread=1;
date_shift=25569.0; /* Windows 1900 date system */
CleanUpFormatIdxUsed();
while (itemsread) {
catdoc_read(rec,2,1,input);
biff_version=getshort(rec,0);
catdoc_read(rec,2,1,input);
reclen=getshort(rec,0);
if ( biff_version == 0x0809 || biff_version == 0x0409 ||
biff_version == 0x0209 || biff_version == 0x0009 ) {
if (reclen==8 || reclen==16) {
if (biff_version == 0x0809 ) {
itemsread=catdoc_read(rec,4,1,input);
if (itemsread == 0)
break;
build_year=getshort(rec+2,0);
build_rel=getshort(rec,0);
(void) build_rel;
if(build_year > 5 ) {
catdoc_read(rec,8,1,input);
biff_version=8;
offset=12;
}
else {
biff_version=7;
offset=4;
}
} else if (biff_version == 0x0209 ) {
biff_version=3;
offset=2;
} else if (biff_version == 0x0409 ) {
offset=2;
biff_version=4;
} else {
biff_version=2;
}
itemsread=catdoc_read(rec,reclen-offset,1,input);
break;
} else {
fprintf(stderr,"%s: Invalid BOF record\n",filename);
return;
}
} else {
itemsread=catdoc_read(rec,126,1,input);
}
}
if (catdoc_eof(input)) {
fprintf(stderr,"%s: No BOF record found\n",filename);
exit(1);
}
while(itemsread){
unsigned char buffer[2];
itemsread = catdoc_read(buffer, 2, 1, input);
if (catdoc_eof(input)) {
process_item(MSEOF,0,NULL);
return;
}
if(itemsread == 0)
break;
rectype=getshort(buffer,0);
itemsread = catdoc_read(buffer, 2, 1, input);
if(itemsread == 0)
break;
reclen=getshort(buffer,0);
if (reclen && reclen <MAX_MS_RECSIZE &&reclen >0){
itemsread = catdoc_read(rec, 1, reclen, input);
rec[reclen] = '\0';
}
if(eof_flag) {
if (rectype != BOF) {
break;
}
}
/* fprintf(stderr,"Rectype 0x%04X reclen=%d\n",rectype, reclen); */
process_item(rectype,reclen,rec);
if (rectype == MSEOF) {
eof_flag=1;
} else {
eof_flag=0;
}
}
return;
}
/**
*
*
* @param rectype
* @param reclen
* @param input
*/
static void process_item (int rectype, long reclen, FILE* input) {
int i=0, u;
static char buf[2];
switch(rectype) {
case DOCUMENT_END:
/* fprintf(stderr,"End of document, ended at %ld\n",catdoc_tell(input)); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case DOCUMENT:
/* fprintf(stderr,"Start of document, reclen=%ld, started at %ld\n", reclen, */
/* catdoc_tell(input)); */
break;
case DOCUMENT_ATOM:
/* fprintf(stderr,"DocumentAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case SLIDE:
/* fprintf(stderr,"Slide, reclen=%ld\n", reclen); */
/* fputs("---------------------------------------\n",stderr); */
break;
case SLIDE_ATOM:
/* fprintf(stderr,"SlideAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case SLIDE_BASE:
/* fprintf(stderr,"SlideBase, reclen=%ld\n", reclen); */
break;
case SLIDE_BASE_ATOM:
/* fprintf(stderr,"SlideBaseAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case NOTES:
/* fprintf(stderr,"Notes, reclen=%ld\n", reclen); */
break;
case NOTES_ATOM:
/* fprintf(stderr,"NotesAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case HEADERS_FOOTERS:
/* fprintf(stderr,"HeadersFooters, reclen=%ld\n", reclen); */
break;
case HEADERS_FOOTERS_ATOM:
/* fprintf(stderr,"HeadersFootersAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case MAIN_MASTER:
/* fprintf(stderr,"MainMaster, reclen=%ld\n", reclen); */
break;
case TEXT_BYTES_ATOM: {
/* fprintf(stderr,"TextBytes, reclen=%ld\n", reclen); */
for(i=0; i < reclen; i++) {
catdoc_read(buf,1,1,input);
if((unsigned char)*buf!=0x0d)
fputs(convert_char((unsigned char)*buf),stdout);
else
fputc('\n',stdout);
}
fputc('\n',stdout);
}
break;
case TEXT_CHARS_ATOM:
case CSTRING: {
long text_len;
/* fprintf(stderr,"CString, reclen=%ld\n", reclen); */
text_len=reclen/2;
for(i=0; i < text_len; i++) {
catdoc_read(buf,2,1,input);
u=(unsigned short)getshort(buf,0);
if(u!=0x0d)
fputs(convert_char(u),stdout);
else
fputc('\n',stdout);
}
fputc('\n',stdout);
}
break;
case USER_EDIT_ATOM:
/* fprintf(stderr,"UserEditAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case COLOR_SCHEME_ATOM:
/* fprintf(stderr,"ColorSchemeAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case PPDRAWING:
/* fprintf(stderr,"PPDrawing, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case ENVIRONMENT:
/* fprintf(stderr,"Environment, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case SSDOC_INFO_ATOM:
/* fprintf(stderr,"SSDocInfoAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case SSSLIDE_INFO_ATOM:
/* fprintf(stderr,"SSSlideInfoAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case PROG_TAGS:
/* fprintf(stderr,"ProgTags, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case PROG_STRING_TAG:
/* fprintf(stderr,"ProgStringTag, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case PROG_BINARY_TAG:
/* fprintf(stderr,"ProgBinaryTag, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case LIST:
/* fprintf(stderr,"List, reclen=%ld\n", reclen); */
break;
case SLIDE_LIST_WITH_TEXT:
/* fprintf(stderr,"SlideListWithText, reclen=%ld\n", reclen); */
/* fputs("---------------------------------------\n",stderr); */
break;
case PERSIST_PTR_INCREMENTAL_BLOCK:
/* fprintf(stderr,"PersistPtrIncrementalBlock, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case EX_OLE_OBJ_STG:
/* fprintf(stderr,"ExOleObjStg, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case PPDRAWING_GROUP:
/* fprintf(stderr,"PpdrawingGroup, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case EX_OBJ_LIST:
/* fprintf(stderr,"ExObjList, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case TX_MASTER_STYLE_ATOM:
/* fprintf(stderr,"TxMasterStyleAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case HANDOUT:
/* fprintf(stderr,"Handout, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case SLIDE_PERSIST_ATOM:
/* fprintf(stderr,"SlidePersistAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case TEXT_HEADER_ATOM:
/* fprintf(stderr,"TextHeaderAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case TEXT_SPEC_INFO:
/* fprintf(stderr,"TextSpecInfo, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
case STYLE_TEXT_PROP_ATOM:
/* fprintf(stderr,"StyleTextPropAtom, reclen=%ld\n", reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
break;
/* case :
fprintf(stderr,", reclen=%ld\n", reclen);
catdoc_seek(input, reclen, SEEK_CUR);
break;*/
/* case :
fprintf(stderr,", reclen=%ld\n", reclen);
catdoc_seek(input, reclen, SEEK_CUR);
break;*/
default:
/* fprintf(stderr,"Default action for rectype=%d reclen=%ld\n", */
/* rectype, reclen); */
catdoc_seek(input, reclen, SEEK_CUR);
}
}
void copy_out (FILE *f,char *header) {
char *buf=(char *)buffer;
int count,i;
long offset;
if (get_unicode_char == get_word8_char) {
/* non-word file and -u specified. Trying to guess which kind of
* unicode is used
*/
if ((unsigned char)header[0]==0xFE && (unsigned char)header[1]==0xFF) {
get_unicode_char = get_utf16msb;
fputs(convert_char(header[2]<<8|header[3]),output_file);
fputs(convert_char(header[4]<<8|header[5]),output_file);
fputs(convert_char(header[6]<<8|header[7]),output_file);
} else if ((unsigned char)header[0]!=0xFF ||
(unsigned char)header[1]!=0xFE) {
int c,j,d;
/* if it is not utf16, assume it is UTF8. We are told -u,
* aren't we */
get_unicode_char = get_utf8;
i=0;
while (i<8) {
c=(unsigned char)header[i++];
if (c >=0x80) {
if ( c<0xE0) {
c=(c & 0x1F);
count =1;
} else {
c=(c & 0xF);
count = 2;
}
for (j=0;j<count;j++) {
if (i<7) {
d=(unsigned char) header[i++];
} else {
d=fgetc(f);
}
c=c<<6 | (d & 0x3F);
}
}
fputs (convert_char(c),output_file);
}
} else {
get_unicode_char = get_utf16lsb;
fputs(convert_char(header[3]<<8|header[2]),output_file);
fputs(convert_char(header[5]<<8|header[4]),output_file);
fputs(convert_char(header[7]<<8|header[6]),output_file);
}
while (!catdoc_eof(f)) {
i=get_unicode_char(f,&offset,0x7FFFFFFF);
if (i!=EOF) fputs(convert_char(i),output_file);
}
} else {
for (i=0;i<8;i++) {
fputs(convert_char(to_unicode(source_charset,(unsigned char)header[i])),output_file);
}
/* Assuming 8-bit input text */
while ((count = catdoc_read(buf,1,PARAGRAPH_BUFFER,f))) {
for (i=0;i<count;i++) {
fputs(convert_char(to_unicode(source_charset,
(unsigned char)buf[i])),output_file);
}
}
}
}
