/* KSC 5601 < - > ISO-2022-KR code converter for Hangul Mail Exchange with non-localized MTA(e.g. sendmail ) version 1.0pl2 ( July 9, 1996) : 1.portability issue related to 'char' type range addressed. 2. another portability issue resolved : To compile with pre-ANSI C compiler, give '-DKNR' option. version 1.0pl1 ( June, 8, 1996) : supercedes one included in hmconv.tar.gz now is a bit more robust and works better in Emacs. version 1.0 ( May, 1996) By Jungshik Shin(jshin@minerva.cis.yale.edu) For details on Hangul mail exchange with this filter, see documents included in hmconv.tar.gz and references there. You may also subscribe to han.comp.mail and han.comp.hangul for further development in Hangul mail exchange including customization for Emacs+Rmail (.emacs setting) I plan to post. */ /* Uncomment following line if pre-ANSI C compiler is used */ /* #define KNR */ #include #define isksc(c) ( (unsigned char) (c) > (unsigned char) '\240' && \ (unsigned char) (c) < (unsigned char) '\377' ) #define is7ksc(c) ( (unsigned char) (c) > (unsigned char) '\040' && \ (unsigned char) (c) < (unsigned char) '\177' ) #define BUF 4096 #define SI '\017' #define SO '\016' FILE *in,*out; int isdecode; unsigned char line[BUF]; int ishangul; #ifndef KNR int openfile(char *,char *,FILE **); int convert(); #else int openfile(); int convert(); #endif #ifndef KNR main (int argc, char **argv) #else main (argc,argv) int argc; char **argv; #endif { isdecode = 0; ishangul = 0; in = stdin; out = stdout; if ( argc > 1 && ! strcmp(argv[1],"-h") ) { fprintf(stderr,"Usage: %s [-h] [-u] [input file] [output file]\n",\ argv[0]); fputs("\t -h : prints this help\n",stderr); fputs("\t -u : ISO-2022-KR -> KSC 5601\n",stderr); fputs("\t without '-u', KSC 5601 -> ISO-2022-KR\n",stderr); fputs("\t standard input(output) is assumed when\n",stderr); fputs("\t input(output) file isn't specified.\n",stderr); exit(0) ; } switch (argc) { case 4: if ( ! strcmp(argv[1], "-u")) { openfile(argv[2],"r",&in); openfile(argv[3],"w",&out); isdecode = 1; } else { openfile(argv[1],"r",&in); openfile(argv[2],"w",&out); } break; case 3: if ( ! strcmp(argv[1], "-u")) { openfile(argv[2],"r",&in); isdecode = 1; } else { openfile(argv[1],"r",&in); openfile(argv[2],"w",&out); } break; case 2: if ( ! strcmp(argv[1], "-u")) { isdecode = 1; } else { openfile(argv[1],"r",&in); } } while ( fgets((char *) line,BUF,in) != NULL ) convert(); fclose(in); fclose(out); } #define KSC 1 #define ASCII 0 int convert() { int mode=ASCII; int i=0; int c; if ( !isdecode ) { if ( !ishangul ) /* search for KSC 5601 character(s) in line */ while ( line[i] != '\n' && line[i] != (unsigned char) EOF && \ line[i] != '\0' ) { /* the last case for buffer fill or input from emacs which doesn't pad the region with EOF or '\n' when handing it over to a filter ( 'shell-command-on-region) */ if ( isksc(line[i]) ) { ishangul = 1; /* found KSC 5601 */ fprintf(out,"\033$)C\n"); /* put out the designator */ break; /* fprintf(out,"\033$)C"); */ /* RFC 1557 does not require '\n' after the designator and Hangul sendmail and cvt8.exe work fine without it, but 'hcode' expects '\n' and breaks a few characters after the designator if it's not followed by '\n' */ } i++; } if ( !ishangul) { /* KSC 5601 doesn't appear, yet */ fputs((char *) line,out); /* no conversion */ return; } i = 0 ; /* back to the beginning of the line */ while ( line[i] != '\n' && line[i] != (unsigned char) EOF && \ line[i] != '\0' ) { /* the last case for buffer fill or input from emacs which doesn't pad the region with EOF or '\n' when handing it over to a filter ( 'shell-command-on-region) */ if ( mode == ASCII && isksc(line[i])) { fputc(SO,out); fputc(0x7f & line[i],out); mode = KSC; } else if ( mode == ASCII && !isksc(line[i]) ) fputc(line[i],out); else if ( mode == KSC && isksc(line[i]) ) /* else if ( mode == KSC && ( isksc(line[i] || line[i] == ' ' ) ) */ fputc(0x7f & line[i],out); else { fputc(SI,out); fputc(line[i],out); mode = ASCII; } i++; } if ( mode == KSC) fputc(SI,out); if ( line[i] == '\n' || ( line[i] == '\0' && i == BUF ) ) fputc('\n',out); /* added after testing with emacs. EOF is added by fclose. no need to add it manually. '\0' is replaced by '\n' only in case of buffer fill */ } /* end of if for KSC -> ISO */ else { /* It is more economical and strictly conforms to ISO-2022-KR, but some programs (e.g. Mule) seem to use 'ESC $ ) C' as the character-set switcher instead of the designator. Hence, these lines were commented out, and routine to work with embeded designator was put in, instead */ /* if ( ! strncmp(line,"\033$)C",4) ) { ishangul = 1; i+=4; } if ( !ishangul) { fputs(line,out); return(0); } */ while ( line[i] != '\n' && line[i] != (unsigned char) EOF && \ line[i] != '\0' ) { /* the last case for buffer fill or input from emacs which doesn't pad the region with EOF or '\n' when handing it over to a filter (shell-command-on-region) */ if ( ! strncmp(&line[i],"\033$)C",4) ) { ishangul = 1; i+=4; if ( line[i] == '\n' && i == 4 ) /* remove '\n' from lines containing */ return(0); /* only the designator ESC+$)C */ continue; } if ( ! ishangul ) fputc(line[i],out); else { switch( line[i] ) { case SO: mode=KSC; break; case SI: mode=ASCII; break; default: if ( mode==ASCII) fputc(line[i],out); else /* space and tab can be embeded among KSC 5601 */ if ( line[i] != '\040' && line[i] != '\011' ) /* Or is it more reasonable to weed out characters outside [0x21,0x7e] ? */ /* if ( is7ksc(line[i]) ) */ fputc(line[i] | 0x80,out); else fputc(line[i],out); } /* end of switch */ } i++; } /* end of while */ if ( line[i] == '\n' ) /* added after testing with emacs */ fputc(line[i],out); /* EOF is added by fclose. no need to add it manually The same is true of '\0' */ } /* end of else for ISO -> KSC */ return(0); } #ifndef KNR int openfile(char *name,char *mode,FILE **fp) #else int openfile(name,mode,fp) char *name; char *mode; FILE **fp; #endif { if ( (*fp=fopen(name,mode)) == NULL ) { fprintf(stderr,"File %s open error !\n", name); exit(1); } return (0); }