tcs.c

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

#ifndef PLAN9
#include	<sys/types.h>
#include	<stdio.h>
#include	<unistd.h>
#include	<stdlib.h>
#include	<fcntl.h>
#include	<string.h>
#include	<errno.h>
#include	"plan9.h"
#else /* PLAN9 */
#include	<u.h>
#include	<libc.h>
#include	<bio.h>
#endif /* PLAN9 */
#include	"cyrillic.h"
#include	"misc.h"
#include	"ms.h"
#include	"8859.h"
#include	"big5.h"
#include	"gb.h"
#include	"hdr.h"
#include	"conv.h"

void usage(void);
void list(void);
int squawk = 1;
int clean = 0;
int verbose = 0;
long ninput, noutput, nrunes, nerrors;
char *file = "stdin";
char *argv0;
Rune runes[N];
char obuf[UTFmax*N];	/* maximum bloat from N runes */
long tab[NRUNE];
#ifndef	PLAN9
extern char version[];
#endif

void intable(int, long *, struct convert *);
void unicode_in(int, long *, struct convert *);
void unicode_out(Rune *, int, long *);

int
main(int argc, char **argv)
{
	char *from = "utf";
	char *to = "utf";
	int fd;
	int listem = 0;
	struct convert *t, *f;

	ARGBEGIN {
	case 'c':
		clean = 1;
		break;
	case 'f':
		from = EARGF(usage());
		break;	
	case 'l':
		listem = 1;
		break;
	case 's':
		squawk = 0;
		break;
	case 't':
		to = EARGF(usage());
		break;
	case 'v':
		verbose = 1;
		break;
	default:
		usage();
		break;
	} ARGEND

	USED(argc);
	if(verbose)
		squawk = 1;
	if(listem){
		list();
		EXIT(0, 0);
	}
	if(!from || !to)
		usage();
	f = conv(from, 1);
	t = conv(to, 0);
#define	PROC	{if(f->flags&Table)\
			intable(fd, (long *)f->data, t);\
		else\
			((Infn)(f->fn))(fd, (long *)0, t);}
	if(*argv){
		while(*argv){
			file = *argv;
#ifndef PLAN9
			if((fd = open(*argv, 0)) < 0){
				EPR "%s: %s: %s\n", argv0, *argv, strerror(errno));
#else /* PLAN9 */
			if((fd = open(*argv, OREAD)) < 0){
				EPR "%s: %s: %r\n", argv0, *argv);
#endif /* PLAN9 */
				EXIT(1, "open failure");
			}
			PROC
			close(fd);
			argv++;
		}
	} else {
		fd = 0;
		PROC
	}
	if(verbose)
		EPR "%s: %ld input bytes, %ld runes, %ld output bytes (%ld errors)\n", argv0,
			ninput, nrunes, noutput, nerrors);
	EXIT(((nerrors && squawk)? 1:0), ((nerrors && squawk)? "conversion error":0));
	return(0);	/* shut up compiler */
}

void
usage(void)
{
	EPR "Usage: %s [-slv] [-f cs] [-t cs] [file ...]\n", argv0);
	verbose = 1;
	list();
	EXIT(1, "usage");
}

void
list(void)
{
	struct convert *c;
	char ch = verbose?'\t':' ';

#ifndef	PLAN9
	EPR "%s version = '%s'\n", argv0, version);
#endif
	if(verbose)
		EPR "character sets:\n");
	else
		EPR "cs:");
	for(c = convert; c->name; c++){
		if((c->flags&From) && c[1].name && (strcmp(c[1].name, c->name) == 0)){
			EPR "%c%s", ch, c->name);
			c++;
		} else if(c->flags&Table)
			EPR "%c%s", ch, c->name);
		else if(c->flags&From)
			EPR "%c%s(from)", ch, c->name);
		else
			EPR "%c%s(to)", ch, c->name);
		if(verbose)
			EPR "\t%s\n", c->chatter);
	}
	if(!verbose)
		EPR "\n");
}


struct convert *
conv(char *name, int from)
{
	struct convert *c;

	for(c = convert; c->name; c++){
		if(cistrcmp(c->name, name) != 0)
			continue;
		if(c->flags&Table)
			return(c);
		if(((c->flags&From) == 0) == (from == 0))
			return(c);
	}
	EPR "%s: charset `%s' unknown\n", argv0, name);
	EXIT(1, "unknown character set");
	return(0);	/* just shut the compiler up */
}

void
swab2(char *b, int n)
{
	char *e, p;

	for(e = b+n; b < e; b++){
		p = *b;
		*b = b[1];
		*++b = p;
	}
}

void
unicode_in(int fd, long *notused, struct convert *out)
{
	Rune buf[N];
	int n;
	int swabme;

	USED(notused);
	if(read(fd, (char *)buf, 2) != 2)
		return;
	ninput += 2;
	switch(buf[0])
	{
	default:
		OUT(out, buf, 1);
	case 0xFEFF:
		swabme = 0;
		break;
	case 0xFFFE:
		swabme = 1;
		break;
	}
	while((n = read(fd, (char *)buf, 2*N)) > 0){
		ninput += n;
		if(swabme)
			swab2((char *)buf, n);
		if(n&1){
			if(squawk)
				EPR "%s: odd byte count in %s\n", argv0, file);
			nerrors++;
			if(clean)
				n--;
			else
				buf[n++/2] = Runeerror;
		}
		OUT(out, buf, n/2);
	}
	OUT(out, buf, 0);
}

void
unicode_in_be(int fd, long *notused, struct convert *out)
{
	int i, n;
	Rune buf[N], r;
	uchar *p;

	USED(notused);
	while((n = read(fd, (char *)buf, 2*N)) > 0){
		ninput += n;
		p = (uchar*)buf;
		for(i=0; i<n/2; i++){
			r = *p++<<8;
			r |= *p++;
			buf[i] = r;
		}
		if(n&1){
			if(squawk)
				EPR "%s: odd byte count in %s\n", argv0, file);
			nerrors++;
			if(clean)
				n--;
			else
				buf[n++/2] = Runeerror;
		}
		OUT(out, buf, n/2);
	}
	OUT(out, buf, 0);
}

void
unicode_in_le(int fd, long *notused, struct convert *out)
{
	int i, n;
	Rune buf[N], r;
	uchar *p;

	USED(notused);
	while((n = read(fd, (char *)buf, 2*N)) > 0){
		ninput += n;
		p = (uchar*)buf;
		for(i=0; i<n/2; i++){
			r = *p++;
			r |= *p++<<8;
			buf[i] = r;
		}
		if(n&1){
			if(squawk)
				EPR "%s: odd byte count in %s\n", argv0, file);
			nerrors++;
			if(clean)
				n--;
			else
				buf[n++/2] = Runeerror;
		}
		OUT(out, buf, n/2);
	}
	OUT(out, buf, 0);
}

void
unicode_out(Rune *base, int n, long *notused)
{
	static int first = 1;

	USED(notused);
	nrunes += n;
	if(first){
		unsigned short x = 0xFEFF;