Navigation

← Previous Changeset
Next Changeset →

Changeset 156

Timestamp:

12/29/02 18:17:30 (6 years ago)

Author:

jajcus

Message:

- use custom character conversion function instead of iconv() which is broken
in some implementations (at least in glibc 2.2)

Files:

trunk/configure.ac (modified) (3 diffs)
trunk/src/Makefile.am (modified) (1 diff)
trunk/src/enc_test.out (added)
trunk/src/enc_test.txt (added)
trunk/src/enc_uni2win.h (added)
trunk/src/enc_win2uni.h (added)
trunk/src/encoding.c (modified) (3 diffs)

Legend:

: Unmodified
: Added
: Removed
: Modified
: Copied
: Moved

trunk/configure.ac

r154	r156
16	16	# Checks for libraries.
17	17	AC_CHECK_LIB([gadu], [gg_change_info],[],AC_MSG_ERROR("You need libgadu (from ekg>=20020807 package) to compile this."))
18		~~AC_CHECK_LIB([iconv], [iconv_open])~~
19	18
20	19	AC_ARG_WITH(pthread,
…	…
78	77	AC_HEADER_STDC
79	78	AC_HEADER_SYS_WAIT
80		AC_CHECK_HEADERS([stdlib.h string.h ~~iconv.h~~])
	79	AC_CHECK_HEADERS([stdlib.h string.h])
81	80	AC_CHECK_HEADERS([sys/socket.h netdb.h netinet/in.h])
82	81	AC_CHECK_HEADERS([sys/time.h unistd.h errno.h fcntl.h])
…	…
96	95	AC_CHECK_FUNCS([select socket],[],AC_MSG_ERROR("Some functions needed are missing"))
97	96	AC_CHECK_FUNCS([memset strchr],[],AC_MSG_ERROR("Some functions needed are missing"))
98		~~AC_CHECK_FUNCS([iconv_open],[],AC_MSG_ERROR("Some functions needed are missing"))~~
99	97	AC_CHECK_LIB([gadu], [gg_event_free],[],AC_MSG_ERROR("You need libgadu>=20020807."))
100	98

trunk/src/Makefile.am

r139	r156
37	37	users.h
38	38
	39	check_PROGRAMS = encodingtest
	40	encodingtest_SOURCES = \
	41	encoding.c \
	42	encoding.h
	43	encodingtest_CFLAGS=-DENCODINGTEST
	44	encodingtest_LDADD=$(GLIB_LIBS)
	45
39	46	INCLUDES=$(GLIB_CFLAGS) -I$(top_srcdir)/libxode/include
40	47

trunk/src/encoding.c

r152	r156
1		/* $Id: encoding.c,v 1.1~~0 2002/12/25 11:03:48~~ jajcus Exp $ */
	1	/* $Id: encoding.c,v 1.11 2002/12/29 17:17:30 jajcus Exp $ */
2	2
3	3	/*
…	…
19	19
20	20	#include "ggtrans.h"
21		~~#include <iconv.h>~~
22	21	#include <errno.h>
23	22	#include <assert.h>
24	23	#include "encoding.h"
	24	#include "enc_win2uni.h"
	25	#include "enc_uni2win.h"
25	26
26		#define ENCODING "windows-1250"
27
28		static iconv_t to_utf8_c;
29		static iconv_t from_utf8_c;
30		static char *buf;
	27	static unsigned char *buf;
31	28	static int buf_len;
32
33	29
34	30	int encoding_init(){
35	31
36		~~to_utf8_c=iconv_open("utf-8",ENCODING);~~
37		~~if (to_utf8_c==(iconv_t)-1)~~
38		~~g_error("Couldn't open 'to Unicode' converter (%s)",g_strerror(errno));~~
39		~~from_utf8_c=iconv_open(ENCODING,"utf-8");~~
40		~~if (from_utf8_c==(iconv_t)-1)~~
41		~~g_error("Couldn't open 'from Unicode' converter (%s)",g_strerror(errno));~~
42	32	buf_len=16;
43	33	buf=g_new(char,buf_len);
…	…
47	37	void encoding_done(){
48	38
49		~~iconv_close(to_utf8_c);~~
50		~~iconv_close(from_utf8_c);~~
51	39	g_free(buf);
52	40	}
53	41
54		static char convert(iconv_t conv,const char str){
55		char *inbuf;
56		size_t inbytesleft;
57		char *outbuf;
58		char *oldbuf;
59		size_t outbytesleft;
60		int r;
	42	char to_utf8(const char str){
	43	int o=0;
	44	int i;
	45	unsigned char c;
	46	unsigned u;
61	47
62	48	if (str==NULL) return NULL;
63		*buf=0;
64		inbuf=(char *)str;
65		inbytesleft=strlen(str);
66		outbuf=buf;
67		outbytesleft=buf_len-1;
68		iconv(conv,NULL,&inbytesleft,&outbuf,&outbytesleft);
69		while(inbytesleft>0){
70		r=iconv(conv,&inbuf,&inbytesleft,&outbuf,&outbytesleft);
71		if (r>=0){
72		*outbuf=0;
73		break;
	49	if (buf_len<(3*strlen(str)+1)){
	50	buf_len=3strlen(str)+1; / this should always be enough */
	51	buf=(char *)g_realloc(buf,buf_len);
	52	assert(buf!=NULL);
	53	}
	54	for(i=0;str[i];i++){
	55	c=(unsigned char)str[i];
	56	if (c<128){
	57	buf[o++]=c;
	58	continue;
74	59	}
75		switch(errno){
76		case EILSEQ:
77		if (!*inbuf) break;
78		inbuf++;
79		*(outbuf++)='?';
80		outbytesleft--;
81		inbytesleft--;
82		if (outbytesleft>0) break;
83		case E2BIG:
84		buf_len+=1024;
85		oldbuf=buf;
86		buf=(char *)g_realloc(oldbuf,buf_len);
87		assert(buf!=NULL);
88		outbytesleft+=1024;
89		outbuf=buf+(outbuf-oldbuf);
90		break;
91		case EINVAL:
92		inbytesleft=0;
93		break;
94		default:
95		*buf=0;
96		inbytesleft=0;
97		break;
	60	u=win1250_to_unicode[c-128];
	61	if (u==0\|\|u>0x10000){ /* we don't need character > U+0x10000 */
	62	buf[o++]='\xef';
	63	buf[o++]='\xbf';
	64	buf[o++]='\xbd';
	65	}
	66	else if (u<0x800){
	67	buf[o++]=0xc0\|(u>>6);
	68	buf[o++]=0x80\|(u&0x3f);
	69	}
	70	else {
	71	buf[o++]=0xe0\|(u>>12);
	72	buf[o++]=0x80\|((u>>6)&0x3f);
	73	buf[o++]=0x80\|(u&0x3f);
98	74	}
99	75	}
	76	buf[o]=0;
	77	return (char *)buf;
	78	}
	79
	80	char from_utf8(const char str){
	81	unsigned char b,c;
	82	unsigned u;
	83	int o=0;
	84	int i;
	85
	86	if (str==NULL) return NULL;
	87	if (buf_len<(strlen(str)+1)){
	88	buf_len=strlen(str)+1; /* this should always be enough */
	89	buf=(char *)g_realloc(buf,buf_len);
	90	assert(buf!=NULL);
	91	}
	92	for(i=0;str[i];i++){
	93	b=(unsigned char)str[i];
	94	if ((b&0x80)==0) { /* ASCII */
	95	buf[o++]=b;
	96	continue;
	97	}
	98	if ((b&0xc0)==0x80) { /* middle of UTF-8 char */
	99	continue;
	100	}
	101	if ((b&0xe0)==0xc0) {
	102	u=b&0x1f;
	103	i++;
	104	b=(unsigned char)str[i];
	105	if (b==0){
	106	buf[o++]='?';
	107	break;
	108	}
	109	if ((b&0xc0)!=0x80){
	110	buf[o++]='?';
	111	continue;
	112	}
	113	u=(u<<6)\|(b&0x3f);
	114	}
	115	else if ((b&0xf0)==0xe0) {
	116	u=b&0x0f;
	117	b=(unsigned char)str[++i];
	118	if (b==0){
	119	buf[o++]='?';
	120	break;
	121	}
	122	if ((b&0xc0)!=0x80){
	123	buf[o++]='?';
	124	continue;
	125	}
	126	u=(u<<6)\|(b&0x3f);
	127	b=(unsigned char)str[++i];
	128	if (b==0){
	129	buf[o++]='?';
	130	break;
	131	}
	132	if ((b&0xc0)!=0x80){
	133	buf[o++]='?';
	134	continue;
	135	}
	136	u=(u<<6)\|(b&0x3f);
	137	}
	138	else{
	139	buf[o++]='?';
	140	continue;
	141	}
	142	if (u<0x00a0)
	143	buf[o++]='?';
	144	else if (u<0x0180)
	145	buf[o++]=unicode_to_win1250_a0_17f[u-0x00a0];
	146	else if (u==0x02c7)
	147	buf[o++]=0xa1;
	148	else if (u<0x02d8)
	149	buf[o++]='?';
	150	else if (u<0x02de)
	151	buf[o++]=unicode_to_win1250_2d8_2dd[u-0x02d8];
	152	else if (u<0x2013)
	153	buf[o++]='?';
	154	else if (u<0x203b)
	155	buf[o++]=unicode_to_win1250_2013_203a[u-0x2013];
	156	else if (u==0x20ac)
	157	buf[o++]=0x80;
	158	else if (u==0x2122)
	159	buf[o++]=0x99;
	160	else
	161	buf[o++]='?';
	162	}
	163	buf[o]=0;
100	164	return buf;
101	165	}
102	166
103		char to_utf8(const char str){
	167	#ifdef ENCODINGTEST
	168	#include <stdio.h>
104	169
105		return convert(to_utf8_c,str);
	170	int main(int argc,char *argv[]){
	171	char buf[1024],*p;
	172
	173	encoding_init();
	174	while(1){
	175	p=fgets(buf,1024,stdin);
	176	if (p==NULL \|\| buf[0]=='\n') break;
	177	printf("To UTF8: %s",to_utf8(buf));
	178	printf("From UTF8: %s",from_utf8(buf));
	179	}
	180	encoding_done();
	181	return 0;
106	182	}
107	183
108		char from_utf8(const char str){
109
110		return convert(from_utf8_c,str);
111		}
112
	184	#endif