X-Git-Url: http://git.marmaro.de/?p=mmh;a=blobdiff_plain;f=sbr%2Fdtimep.lex;h=db2b19397e26b589a37cfe98c45101c522cfd5ee;hp=693925e909e4487571ee5e7ad03886448bc82bf3;hb=fc9279e818dfc96c63a5d75a89080cc68cfe1170;hpb=5f80fdd1ec6888f34af829208fd90fb9716ac289 diff --git a/sbr/dtimep.lex b/sbr/dtimep.lex index 693925e..db2b193 100644 --- a/sbr/dtimep.lex +++ b/sbr/dtimep.lex @@ -1,36 +1,68 @@ -%option noyywrap +/* +** dtimep.lex exceeds the default table capacities for some old versions +** of lex (and the minimum defaults as specified by POSIX). The following +** choices meet or exceed the lex defaults for older SunOS4.x, Solaris, +** HPUX, and AIX. +*/ +%e4000 +%p7000 +%n2500 +%a5000 %{ -#include +#include +#include #include -#define YY_NO_UNPUT -#define YY_DECL struct tws * dparsetime(char *lexstr) +/* +** Since we're looking at a string at a time, don't worry about +** wrapping to the next buffer. +*/ +#define yywrap() 1 +#define YY_SKIP_YYWRAP + +#define YY_NO_INPUT + +/* +** This is the tricky thing that makes this function cool. We +** replace the traditional int yylex(void) declaration with our +** dparsetime() declaration, essentially piggy-backing off the +** utility of the yylex() function and adding what we need to make +** the parsing function useful to us. +*/ +#define YY_DECL struct tws *dparsetime(char *lexstr) + +/* +** yyerminate() is called after the input string is matched to +** completion (actually, when the lexer reaches an EOF). The only +** thing that really needs to be in this macro function is the +** return call, which must be substituted inline into dparsetime. +*/ #define yyterminate() (void)yy_delete_buffer(lexhandle); \ - if(!(tw.tw_flags & TW_SUCC)) { \ - return (struct tws *)NULL; \ - } \ - if(tw.tw_year < 1960) \ - tw.tw_year += 1900; \ - if(tw.tw_year < 1960) \ - tw.tw_year += 100; \ - return(&tw) + if(!(tw.tw_flags & TW_SUCC)) { \ + return (struct tws *)NULL; \ + } \ + if(tw.tw_year < 1970) \ + tw.tw_year += 1900; \ + if(tw.tw_year < 1970) \ + tw.tw_year += 100; \ + return(&tw) /* - * Patchable flag that says how to interpret NN/NN/NN dates. When - * true, we do it European style: DD/MM/YY. When false, we do it - * American style: MM/DD/YY. Of course, these are all non-RFC822 - * compliant. - */ +** Patchable flag that says how to interpret NN/NN/NN dates. When +** true, we do it European style: DD/MM/YY. When false, we do it +** American style: MM/DD/YY. Of course, these are all non-RFC822 +** compliant. +*/ int europeandate = 0; /* - * Table to convert month names to numeric month. We use the - * fact that the low order 5 bits of the sum of the 2nd & 3rd - * characters of the name is a hash with no collisions for the 12 - * valid month names. (The mask to 5 bits maps any combination of - * upper and lower case into the same hash value). - */ +** Table to convert month names to numeric month. We use the +** fact that the low order 5 bits of the sum of the 2nd & 3rd +** characters of the name is a hash with no collisions for the 12 +** valid month names. (The mask to 5 bits maps any combination of +** upper and lower case into the same hash value). +*/ static int month_map[] = { 0, 6, /* 1 - Jul */ @@ -64,9 +96,12 @@ static int month_map[] = { }; /* - * Same trick for day-of-week using the hash function - * (c1 & 7) + (c2 & 4) - */ +** Lookup table for day-of-week using the same hash trick as for above +** name-of-month table, but using the first and second character, not +** second and third. +** +** Compute index into table using: (day_name[0] & 7) + (day_name[1] & 4) +*/ static int day_map[] = { 0, 0, @@ -82,251 +117,277 @@ static int day_map[] = { 3 /*11 - Wed */ }; -#define INIT() { cp = yytext;} -#define SETWDAY() { cp++; \ - tw.tw_wday= day_map[(cp[0] & 7) + (cp[1] & 4)]; \ - tw.tw_flags &= ~TW_SDAY; tw.tw_flags |= TW_SEXP; \ - SKIPA(); } -#define SETMON() { cp++; \ - tw.tw_mon = month_map[(cp[0] + cp[1]) & 0x1f]; \ - SKIPA(); } -#define SETMON_NUM() { tw.tw_mon = atoi(cp)-1; \ - SKIPD(); } -#define SETYEAR() { tw.tw_year = atoi(cp); \ - SKIPD(); } -#define SETDAY() { tw.tw_mday = atoi(cp); \ - tw.tw_flags |= TW_YES; \ - SKIPD(); } -#define SETTIME() { tw.tw_hour = atoi(cp); \ - cp += 2; \ - SKIPTOD(); \ - tw.tw_min = atoi(cp); \ - cp += 2; \ - if(*cp == ':') { tw.tw_sec = atoi(++cp); SKIPD(); } \ - } -#define SETZONE(x) { tw.tw_zone = ((x)/100)*60+(x)%100; \ - tw.tw_flags |= TW_SZEXP; \ - SKIPD(); } -#define SETDST() { tw.tw_flags |= TW_DST; } -#define SKIPD() { while ( isdigit(*cp++) ) ; \ - --cp; } -#define SKIPTOD() { while ( !isdigit(*cp++) ) ; \ - --cp; } -#define SKIPA() { while ( isalpha(*cp++) ) ; \ - --cp; } -#define SKIPTOA() { while ( !isalpha(*cp++) ) ; \ - --cp; } -#define SKIPSP() { while ( isspace(*cp++) ) ; \ - --cp; } -#define SKIPTOSP() { while ( !isspace(*cp++) ) ; \ - --cp; } +/* +** The SET* macros will parse for the appropriate field, and leave the +** cp pointer at the first character after the desired field. Be +** careful with variable-length fields or alpha-num mixes. +** +** The SKIP* macros skip over characters of a particular class and +** leave cp at the position of the first character that doesn't match +** that class. Correspondingly, SKIPTO* skips until it reaches a +** character of a particular class. +*/ + +#define INIT() { cp = yytext;} +#define SETWDAY() { tw.tw_wday= day_map[(cp[0] & 7) + (cp[1] & 4)]; \ + tw.tw_flags &= ~TW_SDAY; tw.tw_flags |= TW_SEXP; SKIPA(); } +#define SETMON() { cp++; tw.tw_mon = month_map[(cp[0] + cp[1]) & 0x1f]; \ + SKIPA(); } +#define SETMON_NUM() { tw.tw_mon = atoi(cp)-1; SKIPD(); } +#define SETYEAR() { tw.tw_year = atoi(cp); SKIPD(); } +#define SETDAY() { tw.tw_mday = atoi(cp); tw.tw_flags |= TW_YES; SKIPD(); } +#define SETTIME() { tw.tw_hour = atoi(cp); cp += 2; SKIPTOD(); \ + tw.tw_min = atoi(cp); cp += 2; if(*cp == ':') { \ + tw.tw_sec = atoi(++cp); SKIPD(); } } +#define SETZONE(x) { tw.tw_zone = ((x)/100)*60+(x)%100; \ + tw.tw_flags |= TW_SZEXP; SKIPD(); } +#define SETDST() { tw.tw_flags |= TW_DST; } +#define SKIPD() { while ( isdigit(*cp++) ) ; --cp; } +#define SKIPTOD() { while ( !isdigit(*cp++) ) ; --cp; } +#define SKIPA() { while ( isalpha(*cp++) ) ; --cp; } +#define SKIPTOA() { while ( !isalpha(*cp++) ) ; --cp; } +#define SKIPSP() { while ( isspace(*cp++) ) ; --cp; } +#define SKIPTOSP() { while ( !isspace(*cp++) ) ; --cp; } + +#ifdef ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST +# ifdef HAVE_SYS_TIME_H +# include +# endif +#include + +static void +zonehack (struct tws *tw) +{ + struct tm *tm; + + if (dmktime (tw) == (time_t) -1) + return; + + tm = localtime (&tw->tw_clock); + if (tm->tm_isdst) { + tw->tw_flags |= TW_DST; + tw->tw_zone -= 60; + } +} +#endif /* ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST */ %} -sun (sun(day)?) -mon (mon(day)?) -tue (tue(sday)?) -wed (wed(nesday)?) -thu (thu(rsday)?) -fri (fri(day)?) -sat (sat(urday)?) +sun ([Ss]un(day)?) +mon ([Mm]on(day)?) +tue ([Tt]ue(sday)?) +wed ([Ww]ed(nesday)?) +thu ([Tt]hu(rsday)?) +fri ([Ff]ri(day)?) +sat ([Ss]at(urday)?) DAY ({sun}|{mon}|{tue}|{wed}|{thu}|{fri}|{sat}) -jan (jan(uary)?) -feb (feb(ruary)?) -mar (mar(ch)?) -apr (apr(il)?) -may (may) -jun (jun(e)?) -jul (jul(y)?) -aug (aug(ust)?) -sep (sep(tember)?) -oct (oct(ober)?) -nov (nov(ember)?) -dec (dec(ember)?) +jan ([Jj]an(uary)?) +feb ([Ff]eb(ruary)?) +mar ([Mm]ar(ch)?) +apr ([Aa]pr(il)?) +may ([Mm]ay) +jun ([Jj]un(e)?) +jul ([Jj]ul(y)?) +aug ([Aa]ug(ust)?) +sep ([Ss]ep(tember)?) +oct ([Oo]ct(ober)?) +nov ([Nn]ov(ember)?) +dec ([Dd]ec(ember)?) MONTH ({jan}|{feb}|{mar}|{apr}|{may}|{jun}|{jul}|{aug}|{sep}|{oct}|{nov}|{dec}) -TIME ({D}:{d}{d}(:{d}{d})?) +TIME ({D}:{d}{d}(:{d}{d})?) + +/* +** The year can either be 2 digits, or 4. However, after +** Y2K, we found that some MUA were reporting the year 100, hence +** the middle term here. yyterminate() resolves the actual +** issues with 2-digit years. +*/ -YEAR (({d}{d})|(1{d}{d})|({d}{4})) +YEAR (({d}{d})|(1{d}{d})|({d}{4})) w ([ \t]*) W ([ \t]+) D ([0-9]?[0-9]) d [0-9] +nl [ \t\n()] %% %{ - - YY_BUFFER_STATE lexhandle; + /* + ** This section begins the definition of dparsetime(). + ** Put here any local variable definitions and initializations + */ + YY_BUFFER_STATE lexhandle; - register char *cp; /* *cp is internal to the lexing function yylex() */ - static struct tws tw; + unsigned char *cp; + static struct tws tw; - memset(&tw,0,sizeof(struct tws)); - lexhandle = yy_scan_string(lexstr); + memset(&tw,0,sizeof(struct tws)); + + lexhandle = yy_scan_string(lexstr); %} -{DAY}","?{W}{MONTH}{W}{D}{W}{TIME}{W}{YEAR} { - INIT(); - SETWDAY(); - SKIPTOA(); - SETMON(); - SKIPTOD(); - SETDAY(); - SKIPTOD(); - SETTIME(); - SKIPTOD(); - SETYEAR(); - } +{DAY}","?{W}{MONTH}{W}{D}{W}{TIME}{W}{YEAR} { + INIT(); + SETWDAY(); + SKIPTOA(); + SETMON(); + SKIPTOD(); + SETDAY(); + SKIPTOD(); + SETTIME(); + SKIPTOD(); + SETYEAR(); +} {DAY}","?{W}{D}{W}{MONTH}{W}{YEAR}{W}{TIME} { - INIT(); - SETWDAY(); - SKIPTOD(); - SETDAY(); - SKIPTOA(); - SETMON(); - SKIPTOD(); - SETYEAR(); - SKIPTOD(); - SETTIME(); - } -{D}{W}{MONTH}{W}{YEAR}{W}{TIME} { - INIT(); - SETDAY(); - SKIPTOA(); - SETMON(); - SKIPTOD(); - SETYEAR(); - SKIPTOD(); - SETTIME(); - } + INIT(); + SETWDAY(); + SKIPTOD(); + SETDAY(); + SKIPTOA(); + SETMON(); + SKIPTOD(); + SETYEAR(); + SKIPTOD(); + SETTIME(); +} +{D}{W}{MONTH}{W}{YEAR}{W}{TIME} { + INIT(); + SETDAY(); + SKIPTOA(); + SETMON(); + SKIPTOD(); + SETYEAR(); + SKIPTOD(); + SETTIME(); +} {DAY}","?{W}{MONTH}{W}{D}","?{W}{YEAR}","?{W}{TIME} { - INIT(); - SETWDAY(); - SKIPTOA(); - SETMON(); - SKIPTOD(); - SETDAY(); - SKIPTOD(); - SETYEAR(); - SKIPTOD(); - SETTIME(); - } + INIT(); + SETWDAY(); + SKIPTOA(); + SETMON(); + SKIPTOD(); + SETDAY(); + SKIPTOD(); + SETYEAR(); + SKIPTOD(); + SETTIME(); +} {DAY}","?{W}{MONTH}{W}{D}","?{W}{YEAR} { - INIT(); - SETWDAY(); - SKIPTOA(); - SETMON(); - SKIPTOD(); - SETDAY(); - SKIPTOD(); - SETYEAR(); - } -{MONTH}{W}{D}","?{W}{YEAR}","?{W}{DAY} { - INIT(); - SETMON(); - SKIPTOD(); - SETDAY(); - SKIPTOD(); - SETYEAR(); - SKIPTOA(); - SETWDAY(); - } -{MONTH}{W}{D}","?{W}{YEAR} { - INIT(); - SETMON(); - SKIPTOD(); - SETDAY(); - SKIPTOD(); - SETYEAR(); - } -{D}("-"|"/"){D}("-"|"/"){YEAR}{W}{TIME} { - INIT(); - if(europeandate) { - /* DD/MM/YY */ - SETDAY(); - SKIPTOD(); - SETMON_NUM(); - } else { - /* MM/DD/YY */ - SETMON_NUM(); - SKIPTOD(); - SETDAY(); - } - SKIPTOD(); - SETYEAR(); - SKIPTOD(); - SETTIME(); - } -{D}("-"|"/"){D}("-"|"/"){YEAR} { - INIT(); - if(europeandate) { - /* DD/MM/YY */ - SETDAY(); - SKIPTOD(); - SETMON_NUM(); - } else { - /* MM/DD/YY */ - SETMON_NUM(); - SKIPTOD(); - SETDAY(); - } - SKIPTOD(); - SETYEAR(); - } - -"[Aa][Mm]" -"[Pp][Mm]" tw.tw_hour += 12; - -"+"{D}{d}{d} { - INIT(); - SKIPTOD(); - SETZONE(atoi(cp)); - } -"-"{D}{d}{d} { - INIT(); - SKIPTOD(); - SETZONE(-atoi(cp)); - } -"-"?("ut"|"UT") INIT(); SETZONE(0); -"-"?("gmt"|"GMT") INIT(); SETZONE(0); -"-"?("jst"|"JST") INIT(); SETZONE(200); -"-"?("jdt"|"JDT") INIT(); SETDST(); SETZONE(2); -"-"?("est"|"EST") INIT(); SETZONE(-500); -"-"?("edt"|"EDT") INIT(); SETDST(); SETZONE(-500); -"-"?("cst"|"CST") INIT(); SETZONE(-600); -"-"?("cdt"|"CDT") INIT(); SETDST(); SETZONE(-600); -"-"?("mst"|"MST") INIT(); SETZONE(-700); -"-"?("mdt"|"MDT") INIT(); SETDST(); SETZONE(-700); -"-"?("pst"|"PST") INIT(); SETZONE(-800); -"-"?("pdt"|"PDT") INIT(); SETDST(); SETZONE(-800); -"-"?("nst"|"NST") INIT(); SETZONE(-330); -"-"?("ast"|"AST") INIT(); SETZONE(-400); -"-"?("adt"|"ADT") INIT(); SETDST(); SETZONE(-400); -"-"?("yst"|"YST") INIT(); SETZONE(-900); -"-"?("ydt"|"YDT") INIT(); SETDST(); SETZONE(-900); -"-"?("hst"|"HST") INIT(); SETZONE(-1000); -"-"?("hdt"|"HDT") INIT(); SETDST(); SETZONE(-1000); -"-"?("bst"|"BST") INIT(); SETDST(); SETZONE(-100); -[a-i] { - INIT(); - SETZONE(100*(('a'-1) - tolower(*cp))); - } -[k-m] { - INIT(); - SETZONE(100*('a' - tolower(*cp))); - } -[n-y] { - INIT(); - SETZONE(100*(tolower(*cp) - 'm')); - } - - -\n -. + INIT(); + SETWDAY(); + SKIPTOA(); + SETMON(); + SKIPTOD(); + SETDAY(); + SKIPTOD(); + SETYEAR(); +} +{MONTH}{W}{D}","?{W}{YEAR}","?{W}{DAY} { + INIT(); + SETMON(); + SKIPTOD(); + SETDAY(); + SKIPTOD(); + SETYEAR(); + SKIPTOA(); + SETWDAY(); +} +{MONTH}{W}{D}","?{W}{YEAR} { + INIT(); + SETMON(); + SKIPTOD(); + SETDAY(); + SKIPTOD(); + SETYEAR(); +} +{D}("-"|"/"){D}("-"|"/"){YEAR}{W}{TIME} { + INIT(); + if(europeandate) { + /* DD/MM/YY */ + SETDAY(); + SKIPTOD(); + SETMON_NUM(); + } else { + /* MM/DD/YY */ + SETMON_NUM(); + SKIPTOD(); + SETDAY(); + } + SKIPTOD(); + SETYEAR(); + SKIPTOD(); + SETTIME(); +} +{D}("-"|"/"){D}("-"|"/"){YEAR} { + INIT(); + if(europeandate) { + /* DD/MM/YY */ + SETDAY(); + SKIPTOD(); + SETMON_NUM(); + } else { + /* MM/DD/YY */ + SETMON_NUM(); + SKIPTOD(); + SETDAY(); + } + SKIPTOD(); + SETYEAR(); +} + +"[Aa][Mm]" +"[Pp][Mm]" tw.tw_hour += 12; +"+"{D}{d}{d} { + INIT(); + SKIPTOD(); + SETZONE(atoi(cp)); +#ifdef ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST + zonehack (&tw); +#endif /* ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST */ + yyterminate(); +} +"-"{D}{d}{d} { + INIT(); + SKIPTOD(); + SETZONE(-atoi(cp)); +#ifdef ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST + zonehack (&tw); +#endif /* ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST */ + yyterminate(); +} +{nl}("ut"|"UT") INIT(); SETZONE(0); yyterminate(); +{nl}("gmt"|"GMT") INIT(); SETZONE(0); yyterminate(); +{nl}("est"|"EST") INIT(); SETZONE(-500); yyterminate(); +{nl}("edt"|"EDT") { INIT(); SETDST(); SETZONE(-500); yyterminate(); } +{nl}("cst"|"CST") INIT(); SETZONE(-600); yyterminate(); +{nl}("cdt"|"CDT") { INIT(); SETDST(); SETZONE(-600); yyterminate(); } +{nl}("mst"|"MST") INIT(); SETZONE(-700); yyterminate(); +{nl}("mdt"|"MDT") { INIT(); SETDST(); SETZONE(-700); yyterminate(); } +{nl}("pst"|"PST") INIT(); SETZONE(-800); yyterminate(); +{nl}("pdt"|"PDT") { INIT(); SETDST(); SETZONE(-800); yyterminate(); } +{nl}("nst"|"NST") INIT(); SETZONE(-330); yyterminate(); +{nl}("ast"|"AST") INIT(); SETZONE(-400); yyterminate(); +{nl}("adt"|"ADT") { INIT(); SETDST(); SETZONE(-400); yyterminate(); } +{nl}("hst"|"HST") INIT(); SETZONE(-1000); yyterminate(); +{nl}("hdt"|"HDT") { INIT(); SETDST(); SETZONE(-1000); yyterminate(); } +.|\n +%% +/* +** This is a portable way to squash a warning about the yyunput() +** function being static but never used. It costs us a tiny amount +** of extra code in the binary but the other options are: +** "%option nounput" which is flex-specific +** makefile hackery just to compile dtimep.c with different flags +*/ +void dtimep_yyunput(int c) +{ + unput(c); +}