diff -ur zephyr-2.1.20070719.SNAPSHOT/clients/zaway/zaway.c zephyr-patched/clients/zaway/zaway.c --- zephyr-2.1.20070719.SNAPSHOT/clients/zaway/zaway.c 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/clients/zaway/zaway.c 2009-03-31 17:40:10.000000000 -0400 @@ -15,6 +15,7 @@ #include #include #include +#include #ifndef lint static const char rcsid_zaway_c[] = "$Id: zaway.c 404 2009-02-28 23:30:45Z kcr $"; @@ -56,6 +57,9 @@ #ifdef _POSIX_VERSION struct sigaction sa; #endif + + /* Initialize locale from environment variables */ + setlocale( LC_ALL, "" ); if ((retval = ZInitialize()) != ZERR_NONE) { com_err(argv[0],retval,"while initializing"); @@ -135,7 +139,7 @@ } for (;;) { - if ((retval = ZReceiveNotice(¬ice, (struct sockaddr_in *)0)) != ZERR_NONE) { + if ((retval = ZReceiveNoticeText(¬ice, (struct sockaddr_in *)0)) != ZERR_NONE) { if (retval != ETIMEDOUT) com_err(argv[0],retval,"while receiving notice"); continue; @@ -195,7 +199,7 @@ msg[1] = ptr; notice.z_message_len = strlen(notice.z_message)+1; - if ((retval = ZSendList(¬ice,msg,2,ZNOAUTH)) != ZERR_NONE) { + if ((retval = ZSendListText(¬ice,msg,2,ZNOAUTH)) != ZERR_NONE) { com_err(argv[0],retval,"while sending notice"); } free(ptr); diff -ur zephyr-2.1.20070719.SNAPSHOT/clients/zwrite/zwrite.c zephyr-patched/clients/zwrite/zwrite.c --- zephyr-2.1.20070719.SNAPSHOT/clients/zwrite/zwrite.c 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/clients/zwrite/zwrite.c 2009-03-31 17:40:10.000000000 -0400 @@ -15,6 +15,7 @@ #include #include #include +#include #ifndef lint static const char rcsid_zwrite_c[] = "$Id: zwrite.c 404 2009-02-28 23:30:45Z kcr $"; @@ -48,6 +49,9 @@ whoami = argv[0]; + /* Initialize locale from environment variables */ + setlocale( LC_ALL, "" ); + if ((retval = ZInitialize()) != ZERR_NONE) { com_err(whoami, retval, "while initializing"); exit(1); @@ -368,7 +372,7 @@ auth?"authenticated ":"", class, inst, nrecips?notice->z_recipient:"everyone"); - if ((retval = ZSendNotice(notice, auth)) != ZERR_NONE) { + if ((retval = ZSendNoticeText(notice, auth)) != ZERR_NONE) { (void) sprintf(bfr, "while sending notice to %s", dest); com_err(whoami, retval, bfr); break; diff -ur zephyr-2.1.20070719.SNAPSHOT/h/zephyr/zephyr.h zephyr-patched/h/zephyr/zephyr.h --- zephyr-2.1.20070719.SNAPSHOT/h/zephyr/zephyr.h 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/h/zephyr/zephyr.h 2009-03-31 17:40:10.000000000 -0400 @@ -42,7 +42,7 @@ #define ZVERSIONHDR "ZEPH" #define ZVERSIONMAJOR 0 -#define ZVERSIONMINOR 2 +#define ZVERSIONMINOR 3 /* Zephyr messages are now UTF-8. */ #define Z_MAXPKTLEN 1024 #define Z_MAXHEADERLEN 800 @@ -54,6 +54,10 @@ #define ZAUTH_YES 1 #define ZAUTH_NO 0 +/* Directions for charset conversion */ +#define Z_CONVERT_INCOMING 0 +#define Z_CONVERT_OUTGOING 1 + typedef char ZPacket_t[Z_MAXPKTLEN]; /* Packet type */ @@ -154,9 +158,12 @@ Code_t ZReadAscii16 ZP((char *, int, unsigned short *)); Code_t ZSendPacket ZP((char*, int, int)); Code_t ZSendList ZP((ZNotice_t*, char *[], int, Z_AuthProc)); -Code_t ZSrvSendList ZP((ZNotice_t*, char*[], int, Z_AuthProc, Code_t (*)())); +Code_t ZSendListText ZP((ZNotice_t*, char *[], int, Z_AuthProc)); +Code_t ZSrvSendList ZP((ZNotice_t*, char*[], int, Z_AuthProc, Code_t (*)(), + int)); Code_t ZSendNotice ZP((ZNotice_t *, Z_AuthProc)); -Code_t ZSrvSendNotice ZP((ZNotice_t*, Z_AuthProc, Code_t (*)())); +Code_t ZSendNoticeText ZP((ZNotice_t *, Z_AuthProc)); +Code_t ZSrvSendNotice ZP((ZNotice_t*, Z_AuthProc, Code_t (*)(), int)); Code_t ZFormatNotice ZP((ZNotice_t*, char**, int*, Z_AuthProc)); Code_t ZFormatSmallNotice ZP((ZNotice_t*, ZPacket_t, int*, Z_AuthProc)); Code_t ZFormatRawNoticeList ZP((ZNotice_t *notice, char *list[], int nitems, @@ -211,6 +218,8 @@ Code_t ZCancelSubscriptions ZP((unsigned int port)); int ZPending ZP((void)); Code_t ZReceiveNotice ZP((ZNotice_t *notice, struct sockaddr_in *from)); +Code_t ZReceiveNoticeText ZP((ZNotice_t *notice, struct sockaddr_in *from)); +Code_t ZCharacterConvert ZP((char **, int *, ZNotice_t *, int)); #ifdef Z_DEBUG void Z_debug ZP((ZCONST char *, ...)); #endif diff -ur zephyr-2.1.20070719.SNAPSHOT/lib/Makefile.in zephyr-patched/lib/Makefile.in --- zephyr-2.1.20070719.SNAPSHOT/lib/Makefile.in 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/lib/Makefile.in 2009-03-31 17:40:10.000000000 -0400 @@ -39,7 +39,7 @@ ZPeekIfNot.lo ZPeekNot.lo ZPeekPkt.lo ZPending.lo ZReadAscii.lo \ ZRecvNot.lo ZRecvPkt.lo ZRetSubs.lo ZSendList.lo ZSendNot.lo \ ZSendPkt.lo ZSendRaw.lo ZSendRLst.lo ZSetDest.lo ZSetFD.lo ZSetSrv.lo \ - ZSubs.lo ZVariables.lo ZWait4Not.lo Zinternal.lo + ZSubs.lo ZVariables.lo ZWait4Not.lo Zinternal.lo ZCharConv.lo .SUFFIXES: .lo diff -ur zephyr-2.1.20070719.SNAPSHOT/lib/ZCharConv.c zephyr-patched/lib/ZCharConv.c --- zephyr-2.1.20070719.SNAPSHOT/lib/ZCharConv.c 2009-03-31 18:12:03.000000000 -0400 +++ zephyr-patched/lib/ZCharConv.c 2009-03-31 17:40:40.000000000 -0400 @@ -0,0 +1,332 @@ +/* This file is part of the Project Athena Zephyr Notification System. + * It contains source for ZCharacterConvert function. + * + * Created by: Keith Winstein + * + * Copyright (c) 2009 by the Massachusetts Institute of Technology. + * For copying and distribution information, see the file + * "mit-copyright.h". + */ + +#include + +#include +#include +#include +#include +#include +#include +#include + +Code_t zconvert( char **bufptr, int *len, ZNotice_t *notice, + char *src, char *dst, + size_t prepend_len, char *prepend_str ); +int get_output_length( char *destencoding, char *srcencoding, + char *message, size_t len ); +int do_convert( char *destencoding, char *srcencoding, + char *dst, size_t dst_len, + char *src, size_t src_len ); + +Code_t ZCharacterConvert( bufptr, len, notice, direction ) + char **bufptr; + int *len; + ZNotice_t *notice; + int direction; +{ + Code_t retval; + char wire[16] = "UTF-8//TRANSLIT"; + char *local, *src, *dst; + char *saved_ctype; + int old_zephyr_version = 0; + + /* Step 1: What are the source and destination charsets? */ + local = alloca( sizeof( nl_langinfo( CODESET ) + 11 ) ); + strcpy( local, nl_langinfo( CODESET ) ); + strcat( local, "//TRANSLIT" ); + + switch ( direction ) { + case Z_CONVERT_INCOMING: + src = wire; dst = local; + break; + case Z_CONVERT_OUTGOING: + src = local; dst = wire; + break; + default: + return EINVAL; + } + + /* For 0.2 packets, convert silently without reporting invalid char. */ + if ( !strcmp( notice->z_version, "ZEPH0.2" ) ) + old_zephyr_version = 1; + + /* Step 2: If our LC_CTYPE is C or POSIX, tell iconv() to assume English. + Rationale: iconv() produces better results if it knows what + language we're in. E.g., with a destination of ASCII//TRANSLIT, + in the C locale, a-with-circumflex becomes a question mark. + In en_US en_US.UTF-8, it becomes a normal "a". */ + saved_ctype = strdup( setlocale( LC_CTYPE, NULL ) ); + if ( ctype_is_posix() ) { + if ( NULL == setlocale( LC_CTYPE, "en_US" ) ) { + if ( NULL == setlocale( LC_CTYPE, "en_US.UTF-8" ) ) { + /* Give up on finding an English locale */ + setlocale( LC_CTYPE, saved_ctype ); + } + } + } + + /* Step 3: Do the conversion. */ + retval = zconvert( bufptr, len, notice, src, dst, 0, "" ); + + /* Step 4: If conversion fails because of invalid sequence, try + again with src encoding = Latin-1. This is a heuristic. */ + + /* If this happens in a locale-aware zephyrgram, tell the user. */ + if ( (retval == EILSEQ) || (retval == EINVAL) ) { + char invalid_tag[32]; + + if ( old_zephyr_version ) { + strcpy( invalid_tag, "" ); + } else if ( direction == Z_CONVERT_INCOMING ) { + strcpy( invalid_tag, "[Invalid incoming character] " ); + } else { + strcpy( invalid_tag, "[Invalid outgoing character] " ); + } + + retval = zconvert( bufptr, len, notice, "ISO-8859-1", dst, + strlen( invalid_tag ), invalid_tag ); + } + + setlocale( LC_CTYPE, saved_ctype ); + free( saved_ctype ); + + return retval; +} + +int ctype_is_posix( void ) +{ + char *current_ctype, *c_ctype; + int retval; + + current_ctype = strdup( setlocale( LC_CTYPE, NULL ) ); + + /* Temporarily change ctype */ + c_ctype = strdup( setlocale( LC_CTYPE, "C" ) ); + + /* Restore real ctype */ + setlocale( LC_CTYPE, current_ctype ); + + retval = !strcmp( current_ctype, c_ctype ); + + free( current_ctype ); + free( c_ctype ); + + return retval; +} + +Code_t zconvert( bufptr, len, notice, src, dst, prepend_len, prepend_str ) + char **bufptr; + int *len; + ZNotice_t *notice; + char *src; + char *dst; + size_t prepend_len; + char *prepend_str; +{ + Code_t retval; + + char *orig_packet; + int new_message_length, old_message_length, header_length; + + /* Find out how much space we'll need */ + header_length = notice->z_message - notice->z_packet; + old_message_length = notice->z_message_len; + new_message_length = get_output_length( dst, src, + notice->z_message, + notice->z_message_len ); + + if ( new_message_length < 0 ) + return errno; + + new_message_length += prepend_len; + + /* Store original packet elsewhere */ + if ( notice->z_packet != *bufptr ) + return EINVAL; + + orig_packet = (char *) malloc( *len ); + + if ( orig_packet == NULL ) + return ENOMEM; + + memcpy( orig_packet, *bufptr, *len ); + + /* Make a new buffer */ + *bufptr = (char *) realloc( *bufptr, header_length + new_message_length ); + if ( NULL == *bufptr ) { + free( orig_packet ); + return ENOMEM; + } + + /* Patch up part of the notice */ + notice->z_packet = *bufptr; + notice->z_message = *bufptr + header_length; + notice->z_message_len = new_message_length; + *len = header_length + new_message_length; + + strncpy( notice->z_message, prepend_str, prepend_len ); + + /* Convert to user locale */ + if ( do_convert( dst, src, + notice->z_message + prepend_len, + new_message_length - prepend_len, + orig_packet + header_length, old_message_length ) < 0 ) { + free( orig_packet ); + return EINVAL; + } + + free( orig_packet ); + + /* Fix rest of notice fields */ + return ZParseNotice( *bufptr, *len, notice ); +} + +int get_output_length( char *destencoding, char *srcencoding, + char *message, size_t len ) +{ + int the_length = 0; + int saveerrno; + iconv_t converter; + size_t inbytesleft, outlen = 1, outbytesleft; + /* Better length would be 32768, but use 1 here to stress */ + + char *outbuf; + char *outptr, *inptr; + + converter = iconv_open( destencoding, srcencoding ); + if ( converter == (iconv_t) -1 ) { + perror( "iconv_open" ); + return -1; + } + + while( 1 ) { + /* Attempt conversion with increasingly-larger buffers */ + /* This shouldn't be necessary, but GNU iconv() did not handle + E2BIG according to POSIX until 2009 CVS bugfix. */ + + size_t ret; + + outbuf = (char *) malloc( outlen ); + if ( outbuf == NULL ) { + perror( "malloc" ); + if ( iconv_close( converter ) < 0 ) { + perror( "iconv_close" ); + } + return -1; + } + + inptr = message; + outptr = outbuf; + + inbytesleft = len; + outbytesleft = outlen; + + ret = iconv( converter, &inptr, &inbytesleft, &outptr, &outbytesleft ); + + if ( ret == -1 ) { + if ( errno == EILSEQ ) { + // fprintf( stderr, "Invalid multibyte sequence in input.\n" ); + goto badstate; + } else if ( errno == EINVAL ) { + // fprintf( stderr, "Incomplete input.\n" ); + goto badstate; + } else if ( errno == E2BIG ) { + /* Need a bigger buffer */ + outlen *= 2; + free( outbuf ); + } else { + perror( "iconv" ); + goto badstate; + } + } else { + /* Completed conversion */ + break; + } + } + + /* Flush output buffer */ + size_t additional = outptr - outbuf; + the_length += additional; + + free( outbuf ); + + if ( iconv_close( converter ) < 0 ) { + perror( "iconv_close" ); + } + + return the_length; + + badstate: + saveerrno = errno; + + free( outbuf ); + + if ( iconv_close( converter ) < 0 ) { + perror( "iconv_close" ); + } + + errno = saveerrno; + return -1; +} + +int do_convert( char *destencoding, char *srcencoding, + char *dst, size_t dst_len, + char *src, size_t src_len ) +{ + iconv_t converter; + Code_t retval; + + converter = iconv_open( destencoding, srcencoding ); + if ( converter == (iconv_t) -1 ) { + perror( "iconv_open" ); + return -1; + } + + if ( iconv( converter, &src, &src_len, &dst, &dst_len ) == -1 ) { + if ( errno == EILSEQ ) { + fprintf( stderr, "BUG: Invalid multibyte sequence in input!\n" ); + goto badstate; + } else if ( errno == EINVAL ) { + fprintf( stderr, "BUG: Incomplete input!\n" ); + goto badstate; + } else if ( errno == E2BIG ) { + fprintf( stderr, "BUG: Ran out of room!\n" ); + goto badstate; + } else { + perror( "iconv" ); + goto badstate; + } + } + + if ( dst_len != 0 ) { + fprintf( stderr, "BUG: Ended up with dst_len = %d\n", dst_len ); + goto badstate; + } + + if ( src_len != 0 ) { + fprintf( stderr, "BUG: Ended up with src_len = %d\n", dst_len ); + goto badstate; + } + + if ( iconv_close( converter ) < 0 ) { + perror( "iconv_close" ); + } + + return 0; + + badstate: + if ( iconv_close( converter ) < 0 ) { + perror( "iconv_close" ); + } + + return -1; +} diff -ur zephyr-2.1.20070719.SNAPSHOT/lib/ZRecvNot.c zephyr-patched/lib/ZRecvNot.c --- zephyr-2.1.20070719.SNAPSHOT/lib/ZRecvNot.c 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/lib/ZRecvNot.c 2009-03-31 17:40:10.000000000 -0400 @@ -16,9 +16,26 @@ #include -Code_t ZReceiveNotice(notice, from) - ZNotice_t *notice; - struct sockaddr_in *from; +Code_t Int_ReceiveNotice(notice,from,text); + +Code_t ZReceiveNotice(notice,from) + ZNotice_t *notice; + struct sockaddr_in *from; +{ + return Int_ReceiveNotice( notice, from, 0 ); +} + +Code_t ZReceiveNoticeText(notice,from) + ZNotice_t *notice; + struct sockaddr_in *from; +{ + return Int_ReceiveNotice( notice, from, 1 ); +} + +Code_t Int_ReceiveNotice(notice,from,text) + ZNotice_t *notice; + struct sockaddr_in *from; + int text; { char *buffer; struct _Z_InputQ *nextq; @@ -46,5 +63,13 @@ if ((retval = ZParseNotice(buffer, len, notice)) != ZERR_NONE) return (retval); notice->z_checked_auth = auth; + + if ( text ) { + if ((retval = ZCharacterConvert( &buffer, &len, notice, + Z_CONVERT_INCOMING )) != ZERR_NONE) + return (retval); + notice->z_checked_auth = auth; + } + return ZERR_NONE; } diff -ur zephyr-2.1.20070719.SNAPSHOT/lib/ZSendList.c zephyr-patched/lib/ZSendList.c --- zephyr-2.1.20070719.SNAPSHOT/lib/ZSendList.c 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/lib/ZSendList.c 2009-03-31 17:40:10.000000000 -0400 @@ -22,15 +22,25 @@ int nitems; Z_AuthProc cert_routine; { - return(ZSrvSendList(notice, list, nitems, cert_routine, Z_XmitFragment)); + return(ZSrvSendList(notice, list, nitems, cert_routine, Z_XmitFragment, 0)); } -Code_t ZSrvSendList(notice, list, nitems, cert_routine, send_routine) +Code_t ZSendListText(notice, list, nitems, cert_routine) + ZNotice_t *notice; + char *list[]; + int nitems; + Z_AuthProc cert_routine; +{ + return(ZSrvSendList(notice, list, nitems, cert_routine, Z_XmitFragment, 1)); +} + +Code_t ZSrvSendList(notice, list, nitems, cert_routine, send_routine, text) ZNotice_t *notice; char *list[]; int nitems; Z_AuthProc cert_routine; Code_t (*send_routine)(); + int text; { Code_t retval; ZNotice_t newnotice; @@ -44,6 +54,12 @@ if ((retval = ZParseNotice(buffer, len, &newnotice)) != ZERR_NONE) return (retval); + if ( text ) { + if ((retval = ZCharacterConvert( &buffer, &len, &newnotice, + Z_CONVERT_OUTGOING )) != ZERR_NONE) + return (retval); + } + retval = Z_SendFragmentedNotice(&newnotice, len, cert_routine, send_routine); diff -ur zephyr-2.1.20070719.SNAPSHOT/lib/ZSendNot.c zephyr-patched/lib/ZSendNot.c --- zephyr-2.1.20070719.SNAPSHOT/lib/ZSendNot.c 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/lib/ZSendNot.c 2009-03-31 17:40:10.000000000 -0400 @@ -20,13 +20,21 @@ ZNotice_t *notice; Z_AuthProc cert_routine; { - return(ZSrvSendNotice(notice, cert_routine, Z_XmitFragment)); + return(ZSrvSendNotice(notice, cert_routine, Z_XmitFragment, 0)); } -Code_t ZSrvSendNotice(notice, cert_routine, send_routine) +Code_t ZSendNoticeText(notice, cert_routine) + ZNotice_t *notice; + Z_AuthProc cert_routine; +{ + return(ZSrvSendNotice(notice, cert_routine, Z_XmitFragment, 1)); +} + +Code_t ZSrvSendNotice(notice, cert_routine, send_routine, text) ZNotice_t *notice; Z_AuthProc cert_routine; Code_t (*send_routine)(); + int text; { Code_t retval; ZNotice_t newnotice; @@ -39,7 +47,13 @@ if ((retval = ZParseNotice(buffer, len, &newnotice)) != ZERR_NONE) return (retval); - + + if ( text ) { + if ((retval = ZCharacterConvert( &buffer, &len, &newnotice, + Z_CONVERT_OUTGOING )) != ZERR_NONE) + return (retval); + } + retval = Z_SendFragmentedNotice(&newnotice, len, cert_routine, send_routine); diff -ur zephyr-2.1.20070719.SNAPSHOT/zwgc/main.c zephyr-patched/zwgc/main.c --- zephyr-2.1.20070719.SNAPSHOT/zwgc/main.c 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/zwgc/main.c 2009-03-31 17:40:10.000000000 -0400 @@ -25,6 +25,7 @@ #include #include #include +#include #include "new_memory.h" #include "zwgc.h" @@ -277,6 +278,9 @@ } #endif + /* Initialize locale from environment variables */ + setlocale( LC_ALL, "" ); + /* * Initialize various subsystems in proper order: */ diff -ur zephyr-2.1.20070719.SNAPSHOT/zwgc/X_driver.c zephyr-patched/zwgc/X_driver.c --- zephyr-2.1.20070719.SNAPSHOT/zwgc/X_driver.c 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/zwgc/X_driver.c 2009-03-31 17:40:10.000000000 -0400 @@ -38,6 +38,8 @@ #include "xselect.h" #include "unsigned_long_dictionary.h" +#include + char *app_instance; /* @@ -375,6 +377,11 @@ mux_add_input_source(ConnectionNumber(dpy), x_get_input, dpy); + /* Use Latin-1 or C charset, since X driver can't handle UTF-8 yet */ + if ( NULL == setlocale( LC_CTYPE, "en_US" ) ) { + setlocale( LC_CTYPE, "C" ); + } + return(0); } diff -ur zephyr-2.1.20070719.SNAPSHOT/zwgc/zephyr.c zephyr-patched/zwgc/zephyr.c --- zephyr-2.1.20070719.SNAPSHOT/zwgc/zephyr.c 2009-03-31 18:11:29.000000000 -0400 +++ zephyr-patched/zwgc/zephyr.c 2009-03-31 17:40:10.000000000 -0400 @@ -112,7 +112,7 @@ notice = malloc(sizeof(ZNotice_t)); - TRAP( ZReceiveNotice(notice, &from), "while getting zephyr notice" ); + TRAP( ZReceiveNoticeText(notice, &from), "while getting zephyr notice" ); if (!error_code) { notice->z_auth = ZCheckAuthentication(notice, &from); notice_handler(notice);