* [^ax-y] doesn't work but [^x-ya] does
@ 2017-02-24 20:38 Sebastian Gniazdowski
2017-02-25 7:32 ` Bart Schaefer
0 siblings, 1 reply; 14+ messages in thread
From: Sebastian Gniazdowski @ 2017-02-24 20:38 UTC (permalink / raw)
To: zsh-workers
Hello,
I couldn't create minimal test case interactively. In my big script, I
have:
disp_list=(
"${disp_list[@]//(#b)([$'\03'-$'\010'$'\022'])(${ZUI[PLUS]}([$'\011'-$'\020'])|)([^$'\03'-$'\010'${ZUI[COLOR_END]}]#)${ZUI[COLOR_END]}/${colormap[${match[1]}]}${colormap[${match[3]}]}$match[4]$RESET}"
)
Sorry, in vim it looks better. The point is about this part (this one
works):
[^$'\03'-$'\010'${ZUI[COLOR_END]}]#
So, no chars between code 3 .. 8 and no $ZUI[COLOR_END] (also single
char). The replaced line is in raw text:
" [Regenerate] this module ON-THE-FLY, without list restart"
The point is: "ON-THE-FLY" isn't replaced because it contains "-", and
this happens if the above bit is written instead as:
[^${ZUI[COLOR_END]}$'\03'-$'\010']#
This however works on zsh-5.2-dev-0 and starts not working with
zsh-5.2-dev-1 reaching zsh-5.3.1-dev-0.
--
Sebastian Gniazdowski
psprint2@fastmail.com
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [^ax-y] doesn't work but [^x-ya] does 2017-02-24 20:38 [^ax-y] doesn't work but [^x-ya] does Sebastian Gniazdowski @ 2017-02-25 7:32 ` Bart Schaefer 2017-02-25 8:21 ` Sebastian Gniazdowski 0 siblings, 1 reply; 14+ messages in thread From: Bart Schaefer @ 2017-02-25 7:32 UTC (permalink / raw) To: zsh-workers On Feb 24, 12:38pm, Sebastian Gniazdowski wrote: } Subject: [^ax-y] doesn't work but [^x-ya] does } } I couldn't create minimal test case interactively. I suspect therefore that value of ${ZUI[COLOR_END]} is going to matter. } The point is: "ON-THE-FLY" isn't replaced because it contains "-", and } } This however works on zsh-5.2-dev-0 and starts not working with } zsh-5.2-dev-1 reaching zsh-5.3.1-dev-0. What changed in there is that "-" started needing to be a tokenized special character in [charset] notation in order to match a range. This has several effects, including that in [${charset}] any hyphens in the expansion of $charset are treated as literal matches unles you use [${~charset}]. There are all sorts of possible side-effects here if the value of $charset begins or ends with hyphen. I understand you say that's not the case here, just mentioning in the context. So you've found something that causes the hyphen in $'\03'-$'\010' to remain untokenized. ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-02-25 7:32 ` Bart Schaefer @ 2017-02-25 8:21 ` Sebastian Gniazdowski 0 siblings, 0 replies; 14+ messages in thread From: Sebastian Gniazdowski @ 2017-02-25 8:21 UTC (permalink / raw) To: zsh-workers On Fri, Feb 24, 2017, at 11:32 PM, Bart Schaefer wrote: > On Feb 24, 12:38pm, Sebastian Gniazdowski wrote: > } Subject: [^ax-y] doesn't work but [^x-ya] does > } > } I couldn't create minimal test case interactively. > > I suspect therefore that value of ${ZUI[COLOR_END]} is going to matter. It is $'\037'. I can do more as time passes, commited error-revealing state into a branch. > (...) > in the expansion of $charset are treated as literal matches unles > you use [${~charset}]. There are all sorts of possible side-effects > here if the value of $charset begins or ends with hyphen. > > I understand you say that's not the case here, just mentioning in > the context. > > So you've found something that causes the hyphen in $'\03'-$'\010' to > remain untokenized. I knew you gonna spot it, but this morning started to bisect so finished it, and just for completeness it is 1c41f98 "37705: don't turn - to Dash after start of brace parameter". -- Sebastian Gniazdowski psprint2@fastmail.com ^ permalink raw reply [flat|nested] 14+ messages in thread
[parent not found: <CGME20170302073436epcas2p1ee2d6d723bc7b0865e11243777a08417@epcas2p1.samsung.com>]
* Re: [^ax-y] doesn't work but [^x-ya] does @ 2017-03-02 7:33 ` Sebastian Gniazdowski 2017-03-02 8:36 ` Sebastian Gniazdowski 2017-03-02 10:47 ` Peter Stephenson 0 siblings, 2 replies; 14+ messages in thread From: Sebastian Gniazdowski @ 2017-03-02 7:33 UTC (permalink / raw) To: zsh-workers [-- Attachment #1: Type: text/plain, Size: 728 bytes --] Hello, a minimal test case to investigate 1c41f98 that bisect pointed to. Dash "-" in character [range] is treated literally. The patch is: - if (!in_brace_param) - in_brace_param = bct; + if (!in_brace_param) { + if ((in_brace_param = bct)) + seen_brct = 0; + } The result of script is (first line – not matched, no ^G etc. replacement): this module ^GON-THE-FLY^_, without list restart this module ONTHEFLY, without list restart Script contains alternative [^x-ya] pattern that works (and [^ax-y] that doesn't). -- Sebastian Gniazdowski psprint3@fastmail.com [-- Attachment #2: toktest.zsh.txt --] [-- Type: text/plain, Size: 881 bytes --] #!/usr/local/bin/zsh-5.3.1 typeset -gAH ZUI ZUI[PLUS]=$'\21' ZUI[COLOR_END]=$'\037' button1='this module \aON-THE-FLY\x1f, \x04without\x1f list restart' # not working button2='this module \aONTHEFLY\x1f, \x04without\x1f list restart' # working local -A colormap colormap=( "" "" ) translate_color_marks() { # Working # disp_list=( "${disp_list[@]//(#b)([$'\03'-$'\010'$'\022'])(${ZUI[PLUS]}([$'\011'-$'\020'])|)([^$'\03'-$'\010'${ZUI[COLOR_END]}]#)${ZUI[COLOR_END]}/${colormap[${match[1]}]}${colormap[${match[3]}]}$match[4]$RESET}" ) # Not working disp_list=( "${disp_list[@]//(#b)([$'\03'-$'\010'$'\022'])(${ZUI[PLUS]}([$'\011'-$'\020'])|)([^${ZUI[COLOR_END]}$'\03'-$'\010']#)${ZUI[COLOR_END]}/${colormap[${match[1]}]}${colormap[${match[3]}]}$match[4]$RESET}" ) } disp_list=( "$button1" "$button2" ) translate_color_marks print -rl -- "${disp_list[@]}" "---" >> ~/result.txt ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-02 7:33 ` Sebastian Gniazdowski @ 2017-03-02 8:36 ` Sebastian Gniazdowski 2017-03-02 10:47 ` Peter Stephenson 1 sibling, 0 replies; 14+ messages in thread From: Sebastian Gniazdowski @ 2017-03-02 8:36 UTC (permalink / raw) To: zsh-workers [-- Attachment #1: Type: text/plain, Size: 146 bytes --] The same script, but instead: $'\03'-$'\010' it does: A-G etc. Behaves the same. -- Sebastian Gniazdowski psprint3@fastmail.com [-- Attachment #2: toktest.zsh.txt --] [-- Type: text/plain, Size: 789 bytes --] #!/usr/local/bin/zsh-5.3.1 typeset -A ZUI ZUI[PLUS]="X" ZUI[COLOR_END]="Z" button1='this module Gon-the-flyZ, DwithoutZ list restart' # not working button2='this module GontheflyZ, DwithoutZ list restart' # working local -A colormap colormap=( "" "" ) translate_color_marks() { # Working #disp_list=( "${disp_list[@]//(#b)([A-G])(${ZUI[PLUS]}([H-I])|)([^H-I${ZUI[COLOR_END]}]#)${ZUI[COLOR_END]}/${colormap[${match[1]}]}${colormap[${match[3]}]}$match[4]$RESET}" ) # Not working disp_list=( "${disp_list[@]//(#b)([A-G])(${ZUI[PLUS]}([H-I])|)([^${ZUI[COLOR_END]}H-I]#)${ZUI[COLOR_END]}/${colormap[${match[1]}]}${colormap[${match[3]}]}$match[4]$RESET}" ) } disp_list=( "$button1" "$button2" ) translate_color_marks print -rl -- "${disp_list[@]}" "---" >> ~/result.txt ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-02 7:33 ` Sebastian Gniazdowski 2017-03-02 8:36 ` Sebastian Gniazdowski @ 2017-03-02 10:47 ` Peter Stephenson 2017-03-02 11:22 ` Sebastian Gniazdowski 1 sibling, 1 reply; 14+ messages in thread From: Peter Stephenson @ 2017-03-02 10:47 UTC (permalink / raw) To: zsh-workers Isn't the same problem showing up here? Watch closely, I may be missing the point... foo=a value=ac % print ${value//[${foo}]/x} xc That's OK, only the a is replaced by an x. % print ${value//[b-z]/x} ax That's OK, only the c is replace by an x. % print ${value//[${foo}b-z]/x} xc Oops, should be a combination of the above effects, giving xx, which is what you get if you expand the ${foo} by hand... % print ${value//[ab-z]/x} xx To show it really is the dash... % value=ab % print ${value//[${foo}b-z]/x} xx Unfortunately tracking down where this came in isn't much help as we need to be able to get the correct overall effect of either Dash or literal '-' in all cases, rather than just moving the problem around. I wonder, however, if we're close to the point where we simply tokenise it when it's not quoted and try to track down all the places where we need to untokenise it. There have been more fixes to untokenisation since the previous attempt to guess where we need to tokenise went in. pws ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-02 10:47 ` Peter Stephenson @ 2017-03-02 11:22 ` Sebastian Gniazdowski 2017-03-02 11:53 ` Peter Stephenson 0 siblings, 1 reply; 14+ messages in thread From: Sebastian Gniazdowski @ 2017-03-02 11:22 UTC (permalink / raw) To: zsh-workers On Thu, Mar 2, 2017, at 02:47 AM, Peter Stephenson wrote: > Isn't the same problem showing up here? Watch closely, I may be missing > the point... (...) > % print ${value//[${foo}b-z]/x} > xc > > Oops, should be a combination of the above effects, giving xx, which is > what you get if you expand the ${foo} by hand... Yes exactly. With the addition that following works: # print ${value//[b-z${foo}]/x} xx > Unfortunately tracking down where this came in isn't much help as we > need to be able to get the correct overall effect of either Dash or > literal '-' in all cases, rather than just moving the problem around. Looked at code few times when debugging, had an unexamined impression that sister variable of "seen_brct" is needed: - if (!in_brace_param) - in_brace_param = bct; + if (!in_brace_param) { + if ((in_brace_param = bct)) + seen_brct = 0; + } if resetting seen_brct helps in one case but harms other one. Raw thought, tokenization is a single-"thread" process, this might be just absurd. -- Sebastian Gniazdowski psprint3@fastmail.com ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-02 11:22 ` Sebastian Gniazdowski @ 2017-03-02 11:53 ` Peter Stephenson 2017-03-04 23:21 ` Bart Schaefer 0 siblings, 1 reply; 14+ messages in thread From: Peter Stephenson @ 2017-03-02 11:53 UTC (permalink / raw) To: zsh-workers On Thu, 02 Mar 2017 03:22:13 -0800 Sebastian Gniazdowski <psprint3@fastmail.com> wrote: > Looked at code few times when debugging, had an unexamined impression > that sister variable of "seen_brct" is needed: > > - if (!in_brace_param) > - in_brace_param = bct; > + if (!in_brace_param) { > + if ((in_brace_param = bct)) > + seen_brct = 0; > + } We could certainly improve things along those lines, possibly using a count of braces which we use elsewhere, but I think we would then just find problems in expressions a stage more complicated. Here's the brute force approach --- it passes all tests, but I bet there's more fall out to come, and it's all over the place, so I'm not sure if there might be a better way. I won't commit it yet. pws diff --git a/Src/cond.c b/Src/cond.c index 8ab0193..0811c5c 100644 --- a/Src/cond.c +++ b/Src/cond.c @@ -138,13 +138,13 @@ evalcond(Estate state, char *fromtest) strs = arrdup(sbuf); l = 2; } - if (name && name[0] == '-') + if (name && (name[0] == '-' || name[0] == Dash)) errname = name; - else if (strs[0] && *strs[0] == '-') + else if (strs[0] && (*strs[0] == '-' || *strs[0] == Dash)) errname = strs[0]; else errname = "<null>"; - if (name && name[0] == '-' && + if (name && (name[0] == '-' || name[0] == Dash) && (cd = getconddef((ctype == COND_MODI), name + 1, 1))) { if (ctype == COND_MOD && (l < cd->min || (cd->max >= 0 && l > cd->max))) { @@ -171,7 +171,7 @@ evalcond(Estate state, char *fromtest) strs[0] = dupstring(name); name = s; - if (name && name[0] == '-' && + if (name && (name[0] == '-' || name[0] == Dash) && (cd = getconddef(0, name + 1, 1))) { if (l < cd->min || (cd->max >= 0 && l > cd->max)) { zwarnnam(fromtest, "unknown condition: %s", diff --git a/Src/exec.c b/Src/exec.c index 83d1513..8e4d8a3 100644 --- a/Src/exec.c +++ b/Src/exec.c @@ -2779,9 +2779,11 @@ execcmd_exec(Estate state, Execcmd_params eparams, char *argdata = (char *) getdata(argnode); char *cmdopt; int has_p = 0, has_vV = 0, has_other = 0; - while (*argdata == '-') { + while (*argdata == '-' || *argdata == Dash) { /* Just to be definite, stop on single "-", too, */ - if (!argdata[1] || (argdata[1] == '-' && !argdata[2])) + if (!argdata[1] || + ((argdata[1] == '-' || argdata[1] == Dash) + && !argdata[2])) break; for (cmdopt = argdata+1; *cmdopt; cmdopt++) { switch (*cmdopt) { @@ -2835,7 +2837,8 @@ execcmd_exec(Estate state, Execcmd_params eparams, * as if this is command [non-option-stuff]. This * isn't a good place for standard option handling. */ - if (!strcmp(argdata, "--")) + if ((argdata[0] == '-' || argdata[0] == Dash) && + (argdata[1] == '-' || argdata[1] == Dash) && !argdata[2]) uremnode(args, firstnode(args)); } if ((cflags & BINF_EXEC) && nextnode(firstnode(args))) { @@ -2855,7 +2858,8 @@ execcmd_exec(Estate state, Execcmd_params eparams, * people aren't likely to mix the option style * with the zsh style. */ - while (next && *next == '-' && strlen(next) >= 2) { + while (next && (*next == '-' || *next == Dash) && + strlen(next) >= 2) { if (!firstnode(args)) { zerr("exec requires a command to execute"); lastval = 1; @@ -2863,7 +2867,8 @@ execcmd_exec(Estate state, Execcmd_params eparams, goto done; } uremnode(args, firstnode(args)); - if (!strcmp(next, "--")) + if ((next[0] == '-' || next[0] == Dash) && + (next[1] == '-' || next[1] == Dash) && !next[2]) break; for (cmdopt = &next[1]; *cmdopt; ++cmdopt) { switch (*cmdopt) { diff --git a/Src/glob.c b/Src/glob.c index ff6b258..2e4a3ba 100644 --- a/Src/glob.c +++ b/Src/glob.c @@ -1314,6 +1314,7 @@ zglob(LinkList list, LinkNode np, int nountok) sense ^= 1; break; case '-': + case Dash: /* Toggle matching of symbolic links */ sense ^= 2; break; @@ -1608,7 +1609,8 @@ zglob(LinkList list, LinkNode np, int nountok) ++s; } /* See if it's greater than, equal to, or less than */ - if ((g_range = *s == '+' ? 1 : *s == '-' ? -1 : 0)) + if ((g_range = *s == '+' ? 1 : + (*s == '-' || *s == Dash) ? -1 : 0)) ++s; data = qgetnum(&s); break; @@ -2025,13 +2027,13 @@ hasbraces(char *str) if (bracechardots(str-1, NULL, NULL)) return 1; lbr = str - 1; - if (*str == '-') + if (*str == '-' || *str == Dash) str++; while (idigit(*str)) str++; if (*str == '.' && str[1] == '.') { str++; str++; - if (*str == '-') + if (*str == '-' || *str == Dash) str++; while (idigit(*str)) str++; @@ -2040,7 +2042,7 @@ hasbraces(char *str) return 1; else if (*str == '.' && str[1] == '.') { str++; str++; - if (*str == '-') + if (*str == '-' || *str == Dash) str++; while (idigit(*str)) str++; @@ -2123,7 +2125,7 @@ xpandredir(struct redir *fn, LinkList redirtab) fn->name = s; untokenize(s); if (fn->type == REDIR_MERGEIN || fn->type == REDIR_MERGEOUT) { - if (s[0] == '-' && !s[1]) + if ((s[0] == '-' || s[0] == Dash) && !s[1]) fn->type = REDIR_CLOSE; else if (s[0] == 'p' && !s[1]) fn->fd2 = -2; @@ -2329,12 +2331,16 @@ xpandbraces(LinkList list, LinkNode *np) * str+1 is the first number in the range, dots+2 the last, * and dots2+2 is the increment if that's given. */ /* TODO: sorry about this */ - int minw = (str[1] == '0' || (str[1] == '-' && str[2] == '0')) + int minw = (str[1] == '0' || + ((str[1] == '-' || str[1] == Dash) && str[2] == '0')) ? wid1 - : (dots[2] == '0' || (dots[2] == '-' && dots[3] == '0')) + : (dots[2] == '0' || + ((dots[2] == '-' || dots[2] == Dash) && + dots[3] == '0')) ? wid2 : (dots2 && (dots2[2] == '0' || - (dots2[2] == '-' && dots2[3] == '0'))) + ((dots2[2] == '-' || dots2[2] == Dash) && + dots2[3] == '0'))) ? wid3 : 0; if (rincr < 0) { @@ -2392,7 +2398,8 @@ xpandbraces(LinkList list, LinkNode *np) c2 = ztokens[c2 - STOUC(Pound)]; if ((char) c2 == Meta) c2 = 32 ^ p[1]; - if (c1 == '-' && lastch >= 0 && p < str2 && lastch <= (int)c2) { + if ((c1 == '-' || c1 == Dash) && + lastch >= 0 && p < str2 && lastch <= (int)c2) { while (lastch < (int)c2) ccl[lastch++] = 1; lastch = -1; @@ -3528,7 +3535,7 @@ zshtokenize(char *s, int flags) } t = s; while (idigit(*++s)); - if (*s != '-') + if (*s != '-' && *s != Dash) goto cont; while (idigit(*++s)); if (*s != '>') diff --git a/Src/lex.c b/Src/lex.c index 8896128..59e9d14 100644 --- a/Src/lex.c +++ b/Src/lex.c @@ -1359,17 +1359,13 @@ gettokstr(int c, int sub) case LX2_DASH: /* * - shouldn't be treated as a special character unless - * we're in a pattern. Howeve,simply counting "[" doesn't - * work as []a-z] is a valid expression and we don't know - * down here what this "[" is for as $foo[stuff] is valid - * in zsh. So just detect an opening [, which is enough - * to turn this into a pattern; the Dash will be harmlessly - * untokenised if not wanted. + * we're in a pattern. Unfortunately, working out for + * sure in complicated expressions whether we're in a + * pattern is tricky. So we'll make it special and + * turn it back any time we don't need it special. + * This is not ideal as it's a lot of work. */ - if (seen_brct) - c = Dash; - else - c = '-'; + c = Dash; break; case LX2_BANG: /* diff --git a/Src/math.c b/Src/math.c index f19c0ed..1f6f4d4 100644 --- a/Src/math.c +++ b/Src/math.c @@ -463,7 +463,7 @@ lexconstant(void) char *nptr; nptr = ptr; - if (*nptr == '-') + if (*nptr == '-' || *nptr == Dash) nptr++; if (*nptr == '0') { @@ -527,7 +527,7 @@ lexconstant(void) } if (*nptr == 'e' || *nptr == 'E') { nptr++; - if (*nptr == '+' || *nptr == '-') + if (*nptr == '+' || *nptr == '-' || *nptr == Dash) nptr++; while (idigit(*nptr) || *nptr == '_') nptr++; @@ -599,7 +599,8 @@ zzlex(void) } return (unary) ? UPLUS : PLUS; case '-': - if (*ptr == '-') { + case Dash: + if (*ptr == '-' || *ptr == Dash) { ptr++; return (unary) ? PREMINUS : POSTMINUS; } diff --git a/Src/parse.c b/Src/parse.c index 699ea49..0443237 100644 --- a/Src/parse.c +++ b/Src/parse.c @@ -2317,6 +2317,19 @@ par_cond_1(void) } /* + * Return 1 if condition matches. This also works for non-elided options. + * + * input is test string, may begin - or Dash. + * cond is condition following the -. + */ +static int check_cond(const char *input, const char *cond) +{ + if (input[0] != '-' && input[0] != Dash) + return 0; + return !strcmp(input + 1, cond); +} + +/* * cond_2 : BANG cond_2 | INPAR { SEPER } cond_2 { SEPER } OUTPAR | STRING STRING STRING @@ -2342,7 +2355,7 @@ par_cond_2(void) s1 = tokstr; condlex(); /* ksh behavior: [ -t ] means [ -t 1 ]; bash disagrees */ - if (unset(POSIXBUILTINS) && !strcmp(s1, "-t")) + if (unset(POSIXBUILTINS) && check_cond(s1, "t")) return par_cond_double(s1, dupstring("1")); return par_cond_double(dupstring("-n"), s1); } @@ -2352,7 +2365,8 @@ par_cond_2(void) if (!strcmp(*testargs, "=") || !strcmp(*testargs, "==") || !strcmp(*testargs, "!=") || - (**testargs == '-' && get_cond_num(*testargs + 1) >= 0)) { + ((**testargs == '-' || **testargs == Dash) + && get_cond_num(*testargs + 1) >= 0)) { s1 = tokstr; condlex(); s2 = tokstr; @@ -2374,8 +2388,8 @@ par_cond_2(void) * In "test" compatibility mode, "! -a ..." and "! -o ..." * are treated as "[string] [and] ..." and "[string] [or] ...". */ - if (!(n_testargs > 1 && - (!strcmp(*testargs, "-a") || !strcmp(*testargs, "-o")))) + if (!(n_testargs > 1 && (check_cond(*testargs, "a") || + check_cond(*testargs, "o")))) { condlex(); ecadd(WCB_COND(COND_NOT, 0)); @@ -2397,7 +2411,7 @@ par_cond_2(void) return r; } s1 = tokstr; - dble = (s1 && *s1 == '-' + dble = (s1 && (*s1 == '-' || *s1 == Dash) && (!n_testargs || strspn(s1+1, "abcdefghknoprstuvwxzLONGS") == 1) && !s1[2]); @@ -2411,7 +2425,8 @@ par_cond_2(void) YYERROR(ecused); } condlex(); - if (n_testargs == 2 && tok != STRING && tokstr && s1[0] == '-') { + if (n_testargs == 2 && tok != STRING && tokstr && + (s1[0] == '-' || s1[0] == Dash)) { /* * Something like "test -z" followed by a token. * We'll turn the token into a string (we've also @@ -2448,7 +2463,7 @@ par_cond_2(void) } s2 = tokstr; if (!n_testargs) - dble = (s2 && *s2 == '-' && !s2[2]); + dble = (s2 && (*s2 == '-' || *s2 == Dash) && !s2[2]); incond++; /* parentheses do globbing */ do condlex(); while (COND_SEP()); incond--; /* parentheses do grouping */ @@ -2476,7 +2491,7 @@ par_cond_2(void) static int par_cond_double(char *a, char *b) { - if (a[0] != '-' || !a[1]) + if ((a[0] != '-' && a[0] != Dash) || !a[1]) COND_ERROR("parse error: condition expected: %s", a); else if (!a[2] && strspn(a+1, "abcdefgknoprstuvwxzhLONGS") == 1) { ecadd(WCB_COND(a[1], 0)); @@ -2534,7 +2549,7 @@ par_cond_triple(char *a, char *b, char *c) ecadd(WCB_COND(COND_REGEX, 0)); ecstr(a); ecstr(c); - } else if (b[0] == '-') { + } else if (b[0] == '-' || b[0] == Dash) { if ((t0 = get_cond_num(b + 1)) > -1) { ecadd(WCB_COND(t0 + COND_NT, 0)); ecstr(a); @@ -2545,7 +2560,7 @@ par_cond_triple(char *a, char *b, char *c) ecstr(a); ecstr(c); } - } else if (a[0] == '-' && a[1]) { + } else if ((a[0] == '-' || a[0] == Dash) && a[1]) { ecadd(WCB_COND(COND_MOD, 2)); ecstr(a); ecstr(b); @@ -2560,7 +2575,7 @@ par_cond_triple(char *a, char *b, char *c) static int par_cond_multi(char *a, LinkList l) { - if (a[0] != '-' || !a[1]) + if ((a[0] != '-' && a[0] != Dash) || !a[1]) COND_ERROR("condition expected: %s", a); else { LinkNode n; @@ -3256,10 +3271,10 @@ build_dump(char *nam, char *dump, char **files, int ali, int map, int flags) for (hlen = FD_PRELEN, tlen = 0; *files; files++) { struct stat st; - if (!strcmp(*files, "-k")) { + if (check_cond(*files, "k")) { flags = (flags & ~(FDHF_KSHLOAD | FDHF_ZSHLOAD)) | FDHF_KSHLOAD; continue; - } else if (!strcmp(*files, "-z")) { + } else if (check_cond(*files, "z")) { flags = (flags & ~(FDHF_KSHLOAD | FDHF_ZSHLOAD)) | FDHF_ZSHLOAD; continue; } diff --git a/Src/subst.c b/Src/subst.c index 02dbe28..12a5283 100644 --- a/Src/subst.c +++ b/Src/subst.c @@ -481,6 +481,8 @@ multsub(char **s, int pf_flags, char ***a, int *isarr, char *sep, for ( ; *x; x += l) { int rawc = -1; convchar_t c; + if (*x == Dash) + *x = '-'; if (itok(STOUC(*x))) { /* token, can't be separator, must be single byte */ rawc = *x; @@ -1766,7 +1768,8 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, */ c = *s; if (itype_end(s, IIDENT, 1) == s && *s != '#' && c != Pound && - c != '-' && c != '!' && c != '$' && c != String && c != Qstring && + c != '-' && c != Dash && + c != '!' && c != '$' && c != String && c != Qstring && c != '?' && c != Quest && c != '*' && c != Star && c != '@' && c != '{' && c != Inbrace && c != '=' && c != Equals && c != Hat && @@ -1895,13 +1898,13 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, if (quotetype == QT_DOLLARS || quotetype == QT_BACKSLASH_PATTERN) goto flagerr; - if (s[1] == '-' || s[1] == '+') { + if (s[1] == '-' || s[1] == Dash || s[1] == '+') { if (quotemod) goto flagerr; s++; quotemod = 1; - quotetype = (*s == '-') ? QT_SINGLE_OPTIONAL : - QT_QUOTEDZPUTS; + quotetype = (*s == '+') ? QT_QUOTEDZPUTS : + QT_SINGLE_OPTIONAL; } else { if (quotetype == QT_SINGLE_OPTIONAL) { /* extra q's after '-' not allowed */ @@ -2210,7 +2213,8 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, */ || ((cc == '#' || cc == Pound) && s[2] == Outbrace) - || cc == '-' || (cc == ':' && s[2] == '-') + || cc == '-' || cc == Dash || + (cc == ':' && (s[2] == '-' || s[2] == Dash)) || (isstring(cc) && (s[2] == Inbrace || s[2] == Inpar)))) { getlen = 1 + whichlen, s++; /* @@ -2606,7 +2610,8 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, * examine properly later on. */ if (inbrace && - (c = *s) != '-' && c != '+' && c != ':' && c != '%' && c != '/' && + (c = *s) != '-' && c != Dash && + c != '+' && c != ':' && c != '%' && c != '/' && c != '=' && c != Equals && c != '#' && c != Pound && c != '?' && c != Quest && @@ -2690,7 +2695,7 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, /* Check for ${..?..} or ${..=..} or one of those. * * Only works if the name is in braces. */ - if (inbrace && ((c = *s) == '-' || + if (inbrace && ((c = *s) == '-' || c == Dash || c == '+' || c == ':' || /* i.e. a doubled colon */ c == '=' || c == Equals || @@ -2802,6 +2807,7 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, vunset = 1; /* Fall Through! */ case '-': + case Dash: if (vunset) { int split_flags; val = dupstring(s); diff --git a/Src/utils.c b/Src/utils.c index 7f3ddad..c016680 100644 --- a/Src/utils.c +++ b/Src/utils.c @@ -2376,7 +2376,7 @@ zstrtol_underscore(const char *s, char **t, int base, int underscore) while (inblank(*s)) s++; - if ((neg = (*s == '-'))) + if ((neg = (*s == '-' || *s == Dash))) s++; else if (*s == '+') s++; @@ -6118,7 +6118,9 @@ quotedzputs(char const *s, FILE *stream) } else *ptr++ = '\''; while(*s) { - if (*s == Meta) + if (*s == Dash) + c = '-'; + else if (*s == Meta) c = *++s ^ 32; else c = *s; @@ -6155,7 +6157,9 @@ quotedzputs(char const *s, FILE *stream) } else { /* use Bourne-style quoting, avoiding empty quoted strings */ while (*s) { - if (*s == Meta) + if (*s == Dash) + c = '-'; + else if (*s == Meta) c = *++s ^ 32; else c = *s; diff --git a/Test/D02glob.ztst b/Test/D02glob.ztst index 1385d57..413381f 100644 --- a/Test/D02glob.ztst +++ b/Test/D02glob.ztst @@ -686,3 +686,9 @@ rm glob.tmp/link 0:modifier ':P' resolves symlinks before '..' components *>*glob.tmp/hello/world + + foo=a + value="ac" + print ${value//[${foo}b-z]/x} +0:handling of - range in complicated pattern context +>xx ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-02 11:53 ` Peter Stephenson @ 2017-03-04 23:21 ` Bart Schaefer 2017-03-06 11:28 ` Peter Stephenson 0 siblings, 1 reply; 14+ messages in thread From: Bart Schaefer @ 2017-03-04 23:21 UTC (permalink / raw) To: zsh-workers On Mar 2, 11:53am, Peter Stephenson wrote: } } Here's the brute force approach --- it passes all tests, but I bet } there's more fall out to come, and it's all over the place, so I'm not } sure if there might be a better way. Maybe the better way is a macro along the lines of those in ztype.h? Perhaps idash(X) ? Doesn't change the number of places in the code that have to be touched, but perhaps cleaner. ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-04 23:21 ` Bart Schaefer @ 2017-03-06 11:28 ` Peter Stephenson 2017-03-07 10:51 ` Peter Stephenson 0 siblings, 1 reply; 14+ messages in thread From: Peter Stephenson @ 2017-03-06 11:28 UTC (permalink / raw) To: zsh-workers On Sat, 4 Mar 2017 15:21:43 -0800 Bart Schaefer <schaefer@brasslantern.com> wrote: > On Mar 2, 11:53am, Peter Stephenson wrote: > } > } Here's the brute force approach --- it passes all tests, but I bet > } there's more fall out to come, and it's all over the place, so I'm not > } sure if there might be a better way. > > Maybe the better way is a macro along the lines of those in ztype.h? > Perhaps idash(X) ? Doesn't change the number of places in the code > that have to be touched, but perhaps cleaner. A bit neater; I've avoided it looking too much the itype macros as it's a bit different, explicitly looking for either the raw or tokenized form rather than a type. But I suppose it would be straightforward to mark both with a bit; I don't think there's a huge difference either way. I may just commit this and see how it works. pws diff --git a/Src/cond.c b/Src/cond.c index 8ab0193..9b739f6 100644 --- a/Src/cond.c +++ b/Src/cond.c @@ -138,13 +138,13 @@ evalcond(Estate state, char *fromtest) strs = arrdup(sbuf); l = 2; } - if (name && name[0] == '-') + if (name && IS_DASH(name[0])) errname = name; - else if (strs[0] && *strs[0] == '-') + else if (strs[0] && IS_DASH(*strs[0])) errname = strs[0]; else errname = "<null>"; - if (name && name[0] == '-' && + if (name && IS_DASH(name[0]) && (cd = getconddef((ctype == COND_MODI), name + 1, 1))) { if (ctype == COND_MOD && (l < cd->min || (cd->max >= 0 && l > cd->max))) { @@ -171,7 +171,7 @@ evalcond(Estate state, char *fromtest) strs[0] = dupstring(name); name = s; - if (name && name[0] == '-' && + if (name && IS_DASH(name[0]) && (cd = getconddef(0, name + 1, 1))) { if (l < cd->min || (cd->max >= 0 && l > cd->max)) { zwarnnam(fromtest, "unknown condition: %s", diff --git a/Src/exec.c b/Src/exec.c index 83d1513..3a8c268 100644 --- a/Src/exec.c +++ b/Src/exec.c @@ -2779,9 +2779,10 @@ execcmd_exec(Estate state, Execcmd_params eparams, char *argdata = (char *) getdata(argnode); char *cmdopt; int has_p = 0, has_vV = 0, has_other = 0; - while (*argdata == '-') { + while (IS_DASH(*argdata)) { /* Just to be definite, stop on single "-", too, */ - if (!argdata[1] || (argdata[1] == '-' && !argdata[2])) + if (!argdata[1] || + (IS_DASH(argdata[1]) && !argdata[2])) break; for (cmdopt = argdata+1; *cmdopt; cmdopt++) { switch (*cmdopt) { @@ -2835,7 +2836,7 @@ execcmd_exec(Estate state, Execcmd_params eparams, * as if this is command [non-option-stuff]. This * isn't a good place for standard option handling. */ - if (!strcmp(argdata, "--")) + if (IS_DASH(argdata[0]) && IS_DASH(argdata[1]) && !argdata[2]) uremnode(args, firstnode(args)); } if ((cflags & BINF_EXEC) && nextnode(firstnode(args))) { @@ -2855,7 +2856,7 @@ execcmd_exec(Estate state, Execcmd_params eparams, * people aren't likely to mix the option style * with the zsh style. */ - while (next && *next == '-' && strlen(next) >= 2) { + while (next && IS_DASH(*next) && strlen(next) >= 2) { if (!firstnode(args)) { zerr("exec requires a command to execute"); lastval = 1; @@ -2863,7 +2864,7 @@ execcmd_exec(Estate state, Execcmd_params eparams, goto done; } uremnode(args, firstnode(args)); - if (!strcmp(next, "--")) + if (IS_DASH(next[0]) && IS_DASH(next[1]) && !next[2]) break; for (cmdopt = &next[1]; *cmdopt; ++cmdopt) { switch (*cmdopt) { diff --git a/Src/glob.c b/Src/glob.c index ff6b258..87127e1 100644 --- a/Src/glob.c +++ b/Src/glob.c @@ -1314,6 +1314,7 @@ zglob(LinkList list, LinkNode np, int nountok) sense ^= 1; break; case '-': + case Dash: /* Toggle matching of symbolic links */ sense ^= 2; break; @@ -1608,7 +1609,7 @@ zglob(LinkList list, LinkNode np, int nountok) ++s; } /* See if it's greater than, equal to, or less than */ - if ((g_range = *s == '+' ? 1 : *s == '-' ? -1 : 0)) + if ((g_range = *s == '+' ? 1 : IS_DASH(*s) ? -1 : 0)) ++s; data = qgetnum(&s); break; @@ -2025,13 +2026,13 @@ hasbraces(char *str) if (bracechardots(str-1, NULL, NULL)) return 1; lbr = str - 1; - if (*str == '-') + if (IS_DASH(*str)) str++; while (idigit(*str)) str++; if (*str == '.' && str[1] == '.') { str++; str++; - if (*str == '-') + if (IS_DASH(*str)) str++; while (idigit(*str)) str++; @@ -2040,7 +2041,7 @@ hasbraces(char *str) return 1; else if (*str == '.' && str[1] == '.') { str++; str++; - if (*str == '-') + if (IS_DASH(*str)) str++; while (idigit(*str)) str++; @@ -2123,7 +2124,7 @@ xpandredir(struct redir *fn, LinkList redirtab) fn->name = s; untokenize(s); if (fn->type == REDIR_MERGEIN || fn->type == REDIR_MERGEOUT) { - if (s[0] == '-' && !s[1]) + if (IS_DASH(s[0]) && !s[1]) fn->type = REDIR_CLOSE; else if (s[0] == 'p' && !s[1]) fn->fd2 = -2; @@ -2329,12 +2330,14 @@ xpandbraces(LinkList list, LinkNode *np) * str+1 is the first number in the range, dots+2 the last, * and dots2+2 is the increment if that's given. */ /* TODO: sorry about this */ - int minw = (str[1] == '0' || (str[1] == '-' && str[2] == '0')) + int minw = (str[1] == '0' || + (IS_DASH(str[1]) && str[2] == '0')) ? wid1 - : (dots[2] == '0' || (dots[2] == '-' && dots[3] == '0')) + : (dots[2] == '0' || + (IS_DASH(dots[2]) && dots[3] == '0')) ? wid2 : (dots2 && (dots2[2] == '0' || - (dots2[2] == '-' && dots2[3] == '0'))) + (IS_DASH(dots2[2]) && dots2[3] == '0'))) ? wid3 : 0; if (rincr < 0) { @@ -2392,7 +2395,7 @@ xpandbraces(LinkList list, LinkNode *np) c2 = ztokens[c2 - STOUC(Pound)]; if ((char) c2 == Meta) c2 = 32 ^ p[1]; - if (c1 == '-' && lastch >= 0 && p < str2 && lastch <= (int)c2) { + if (IS_DASH(c1) && lastch >= 0 && p < str2 && lastch <= (int)c2) { while (lastch < (int)c2) ccl[lastch++] = 1; lastch = -1; @@ -3528,7 +3531,7 @@ zshtokenize(char *s, int flags) } t = s; while (idigit(*++s)); - if (*s != '-') + if (!IS_DASH(*s)) goto cont; while (idigit(*++s)); if (*s != '>') diff --git a/Src/lex.c b/Src/lex.c index 8896128..59e9d14 100644 --- a/Src/lex.c +++ b/Src/lex.c @@ -1359,17 +1359,13 @@ gettokstr(int c, int sub) case LX2_DASH: /* * - shouldn't be treated as a special character unless - * we're in a pattern. Howeve,simply counting "[" doesn't - * work as []a-z] is a valid expression and we don't know - * down here what this "[" is for as $foo[stuff] is valid - * in zsh. So just detect an opening [, which is enough - * to turn this into a pattern; the Dash will be harmlessly - * untokenised if not wanted. + * we're in a pattern. Unfortunately, working out for + * sure in complicated expressions whether we're in a + * pattern is tricky. So we'll make it special and + * turn it back any time we don't need it special. + * This is not ideal as it's a lot of work. */ - if (seen_brct) - c = Dash; - else - c = '-'; + c = Dash; break; case LX2_BANG: /* diff --git a/Src/math.c b/Src/math.c index f19c0ed..f961300 100644 --- a/Src/math.c +++ b/Src/math.c @@ -463,7 +463,7 @@ lexconstant(void) char *nptr; nptr = ptr; - if (*nptr == '-') + if (IS_DASH(*nptr)) nptr++; if (*nptr == '0') { @@ -527,7 +527,7 @@ lexconstant(void) } if (*nptr == 'e' || *nptr == 'E') { nptr++; - if (*nptr == '+' || *nptr == '-') + if (*nptr == '+' || IS_DASH(*nptr)) nptr++; while (idigit(*nptr) || *nptr == '_') nptr++; @@ -599,7 +599,8 @@ zzlex(void) } return (unary) ? UPLUS : PLUS; case '-': - if (*ptr == '-') { + case Dash: + if (IS_DASH(*ptr)) { ptr++; return (unary) ? PREMINUS : POSTMINUS; } diff --git a/Src/parse.c b/Src/parse.c index 699ea49..6fe283d 100644 --- a/Src/parse.c +++ b/Src/parse.c @@ -2317,6 +2317,19 @@ par_cond_1(void) } /* + * Return 1 if condition matches. This also works for non-elided options. + * + * input is test string, may begin - or Dash. + * cond is condition following the -. + */ +static int check_cond(const char *input, const char *cond) +{ + if (!IS_DASH(input[0])) + return 0; + return !strcmp(input + 1, cond); +} + +/* * cond_2 : BANG cond_2 | INPAR { SEPER } cond_2 { SEPER } OUTPAR | STRING STRING STRING @@ -2342,7 +2355,7 @@ par_cond_2(void) s1 = tokstr; condlex(); /* ksh behavior: [ -t ] means [ -t 1 ]; bash disagrees */ - if (unset(POSIXBUILTINS) && !strcmp(s1, "-t")) + if (unset(POSIXBUILTINS) && check_cond(s1, "t")) return par_cond_double(s1, dupstring("1")); return par_cond_double(dupstring("-n"), s1); } @@ -2352,7 +2365,7 @@ par_cond_2(void) if (!strcmp(*testargs, "=") || !strcmp(*testargs, "==") || !strcmp(*testargs, "!=") || - (**testargs == '-' && get_cond_num(*testargs + 1) >= 0)) { + (IS_DASH(**testargs) && get_cond_num(*testargs + 1) >= 0)) { s1 = tokstr; condlex(); s2 = tokstr; @@ -2374,8 +2387,8 @@ par_cond_2(void) * In "test" compatibility mode, "! -a ..." and "! -o ..." * are treated as "[string] [and] ..." and "[string] [or] ...". */ - if (!(n_testargs > 1 && - (!strcmp(*testargs, "-a") || !strcmp(*testargs, "-o")))) + if (!(n_testargs > 1 && (check_cond(*testargs, "a") || + check_cond(*testargs, "o")))) { condlex(); ecadd(WCB_COND(COND_NOT, 0)); @@ -2397,7 +2410,7 @@ par_cond_2(void) return r; } s1 = tokstr; - dble = (s1 && *s1 == '-' + dble = (s1 && IS_DASH(*s1) && (!n_testargs || strspn(s1+1, "abcdefghknoprstuvwxzLONGS") == 1) && !s1[2]); @@ -2411,7 +2424,7 @@ par_cond_2(void) YYERROR(ecused); } condlex(); - if (n_testargs == 2 && tok != STRING && tokstr && s1[0] == '-') { + if (n_testargs == 2 && tok != STRING && tokstr && IS_DASH(s1[0])) { /* * Something like "test -z" followed by a token. * We'll turn the token into a string (we've also @@ -2446,9 +2459,9 @@ par_cond_2(void) } else YYERROR(ecused); } - s2 = tokstr; + s2 = tokstr; if (!n_testargs) - dble = (s2 && *s2 == '-' && !s2[2]); + dble = (s2 && IS_DASH(*s2) && !s2[2]); incond++; /* parentheses do globbing */ do condlex(); while (COND_SEP()); incond--; /* parentheses do grouping */ @@ -2476,7 +2489,7 @@ par_cond_2(void) static int par_cond_double(char *a, char *b) { - if (a[0] != '-' || !a[1]) + if (!IS_DASH(a[0]) || !a[1]) COND_ERROR("parse error: condition expected: %s", a); else if (!a[2] && strspn(a+1, "abcdefgknoprstuvwxzhLONGS") == 1) { ecadd(WCB_COND(a[1], 0)); @@ -2534,7 +2547,7 @@ par_cond_triple(char *a, char *b, char *c) ecadd(WCB_COND(COND_REGEX, 0)); ecstr(a); ecstr(c); - } else if (b[0] == '-') { + } else if (IS_DASH(b[0])) { if ((t0 = get_cond_num(b + 1)) > -1) { ecadd(WCB_COND(t0 + COND_NT, 0)); ecstr(a); @@ -2545,7 +2558,7 @@ par_cond_triple(char *a, char *b, char *c) ecstr(a); ecstr(c); } - } else if (a[0] == '-' && a[1]) { + } else if (IS_DASH(a[0]) && a[1]) { ecadd(WCB_COND(COND_MOD, 2)); ecstr(a); ecstr(b); @@ -2560,7 +2573,7 @@ par_cond_triple(char *a, char *b, char *c) static int par_cond_multi(char *a, LinkList l) { - if (a[0] != '-' || !a[1]) + if (!IS_DASH(a[0]) || !a[1]) COND_ERROR("condition expected: %s", a); else { LinkNode n; @@ -3256,10 +3269,10 @@ build_dump(char *nam, char *dump, char **files, int ali, int map, int flags) for (hlen = FD_PRELEN, tlen = 0; *files; files++) { struct stat st; - if (!strcmp(*files, "-k")) { + if (check_cond(*files, "k")) { flags = (flags & ~(FDHF_KSHLOAD | FDHF_ZSHLOAD)) | FDHF_KSHLOAD; continue; - } else if (!strcmp(*files, "-z")) { + } else if (check_cond(*files, "z")) { flags = (flags & ~(FDHF_KSHLOAD | FDHF_ZSHLOAD)) | FDHF_ZSHLOAD; continue; } diff --git a/Src/pattern.c b/Src/pattern.c index 928790f..75db016 100644 --- a/Src/pattern.c +++ b/Src/pattern.c @@ -1521,7 +1521,7 @@ patcomppiece(int *flagp, int paren) patparse = nptr; len |= 1; } - DPUTS(*patparse != '-', "BUG: - missing from numeric glob"); + DPUTS(!IS_DASH(*patparse), "BUG: - missing from numeric glob"); patparse++; if (idigit(*patparse)) { to = (zrange_t) zstrtol((char *)patparse, diff --git a/Src/subst.c b/Src/subst.c index 02dbe28..2214b3d 100644 --- a/Src/subst.c +++ b/Src/subst.c @@ -481,6 +481,8 @@ multsub(char **s, int pf_flags, char ***a, int *isarr, char *sep, for ( ; *x; x += l) { int rawc = -1; convchar_t c; + if (*x == Dash) + *x = '-'; if (itok(STOUC(*x))) { /* token, can't be separator, must be single byte */ rawc = *x; @@ -1766,7 +1768,8 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, */ c = *s; if (itype_end(s, IIDENT, 1) == s && *s != '#' && c != Pound && - c != '-' && c != '!' && c != '$' && c != String && c != Qstring && + !IS_DASH(c) && + c != '!' && c != '$' && c != String && c != Qstring && c != '?' && c != Quest && c != '*' && c != Star && c != '@' && c != '{' && c != Inbrace && c != '=' && c != Equals && c != Hat && @@ -1895,13 +1898,13 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, if (quotetype == QT_DOLLARS || quotetype == QT_BACKSLASH_PATTERN) goto flagerr; - if (s[1] == '-' || s[1] == '+') { + if (IS_DASH(s[1]) || s[1] == '+') { if (quotemod) goto flagerr; s++; quotemod = 1; - quotetype = (*s == '-') ? QT_SINGLE_OPTIONAL : - QT_QUOTEDZPUTS; + quotetype = (*s == '+') ? QT_QUOTEDZPUTS : + QT_SINGLE_OPTIONAL; } else { if (quotetype == QT_SINGLE_OPTIONAL) { /* extra q's after '-' not allowed */ @@ -2208,9 +2211,9 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, * properly in the first place we wouldn't * have this nonsense. */ - || ((cc == '#' || cc == Pound) && - s[2] == Outbrace) - || cc == '-' || (cc == ':' && s[2] == '-') + || ((cc == '#' || cc == Pound) && s[2] == Outbrace) + || IS_DASH(cc) + || (cc == ':' && IS_DASH(s[2])) || (isstring(cc) && (s[2] == Inbrace || s[2] == Inpar)))) { getlen = 1 + whichlen, s++; /* @@ -2605,14 +2608,17 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, * Again, this duplicates tests for characters we're about to * examine properly later on. */ - if (inbrace && - (c = *s) != '-' && c != '+' && c != ':' && c != '%' && c != '/' && - c != '=' && c != Equals && - c != '#' && c != Pound && - c != '?' && c != Quest && - c != '}' && c != Outbrace) { - zerr("bad substitution"); - return NULL; + if (inbrace) { + c = *s; + if (!IS_DASH(c) && + c != '+' && c != ':' && c != '%' && c != '/' && + c != '=' && c != Equals && + c != '#' && c != Pound && + c != '?' && c != Quest && + c != '}' && c != Outbrace) { + zerr("bad substitution"); + return NULL; + } } /* * Join arrays up if we're in quotes and there isn't some @@ -2690,8 +2696,8 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, /* Check for ${..?..} or ${..=..} or one of those. * * Only works if the name is in braces. */ - if (inbrace && ((c = *s) == '-' || - c == '+' || + if (inbrace && ((c = *s) == '+' || + IS_DASH(c) || c == ':' || /* i.e. a doubled colon */ c == '=' || c == Equals || c == '%' || @@ -2802,6 +2808,7 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags, vunset = 1; /* Fall Through! */ case '-': + case Dash: if (vunset) { int split_flags; val = dupstring(s); diff --git a/Src/utils.c b/Src/utils.c index 7f3ddad..9669944 100644 --- a/Src/utils.c +++ b/Src/utils.c @@ -2376,7 +2376,7 @@ zstrtol_underscore(const char *s, char **t, int base, int underscore) while (inblank(*s)) s++; - if ((neg = (*s == '-'))) + if ((neg = IS_DASH(*s))) s++; else if (*s == '+') s++; @@ -6118,7 +6118,9 @@ quotedzputs(char const *s, FILE *stream) } else *ptr++ = '\''; while(*s) { - if (*s == Meta) + if (*s == Dash) + c = '-'; + else if (*s == Meta) c = *++s ^ 32; else c = *s; @@ -6155,7 +6157,9 @@ quotedzputs(char const *s, FILE *stream) } else { /* use Bourne-style quoting, avoiding empty quoted strings */ while (*s) { - if (*s == Meta) + if (*s == Dash) + c = '-'; + else if (*s == Meta) c = *++s ^ 32; else c = *s; diff --git a/Src/zsh.h b/Src/zsh.h index f2c2790..ce41b17 100644 --- a/Src/zsh.h +++ b/Src/zsh.h @@ -238,6 +238,16 @@ struct mathfunc { #define PATCHARS "#^*()|[]<>?~\\" /* + * Check for a possibly tokenized dash. + * + * A dash only needs to be a token in a character range, [a-z], but + * it's difficult in general to ensure that. So it's turned into + * a token at the usual point in the lexer. However, we need + * to check for a literal dash at many opints. + */ +#define IS_DASH(x) ((x) == '-' || (x) == Dash) + +/* * Types of quote. This is used in various places, so care needs * to be taken when changing them. (Oooh, don't you look surprised.) * - Passed to quotestring() to indicate style. This is the ultimate diff --git a/Test/D02glob.ztst b/Test/D02glob.ztst index 1385d57..413381f 100644 --- a/Test/D02glob.ztst +++ b/Test/D02glob.ztst @@ -686,3 +686,9 @@ rm glob.tmp/link 0:modifier ':P' resolves symlinks before '..' components *>*glob.tmp/hello/world + + foo=a + value="ac" + print ${value//[${foo}b-z]/x} +0:handling of - range in complicated pattern context +>xx ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-06 11:28 ` Peter Stephenson @ 2017-03-07 10:51 ` Peter Stephenson 2017-03-08 19:12 ` Martijn Dekker 0 siblings, 1 reply; 14+ messages in thread From: Peter Stephenson @ 2017-03-07 10:51 UTC (permalink / raw) To: zsh-workers On Mon, 6 Mar 2017 11:28:30 +0000 Peter Stephenson <p.stephenson@samsung.com> wrote: > I may just commit this and see how it works. I've done so. It's entirely possible there will be more cases where a Dash token is encountered where the shell expects a '-' and consequently there's an error. These all ought to be fixable by simple local tests. pws ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-07 10:51 ` Peter Stephenson @ 2017-03-08 19:12 ` Martijn Dekker 2017-03-09 5:18 ` Bart Schaefer 0 siblings, 1 reply; 14+ messages in thread From: Martijn Dekker @ 2017-03-08 19:12 UTC (permalink / raw) To: Peter Stephenson, Zsh hackers list Op 07-03-17 om 10:51 schreef Peter Stephenson: > On Mon, 6 Mar 2017 11:28:30 +0000 > Peter Stephenson <p.stephenson@samsung.com> wrote: >> I may just commit this and see how it works. > > I've done so. > > It's entirely possible there will be more cases where a Dash token is > encountered where the shell expects a '-' and consequently there's an > error. These all ought to be fixable by simple local tests. Commit f3f8537c breaks unquoted $-: $ Src/zsh -c 'echo $-' (empty output) $ Src/zsh -c 'echo "$-"' 569X $ Src/zsh -u -c 'echo $-' zsh:1: : parameter not set $ Src/zsh -u -c 'echo "$-"' 569Xu $ Src/zsh -c 'echo ${-}' (empty output) $ Src/zsh -c 'echo "${-}"' 569X $ Src/zsh -u -c 'echo ${-}' (empty output) $ Src/zsh -u -c 'echo "${-}"' 569Xu ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-08 19:12 ` Martijn Dekker @ 2017-03-09 5:18 ` Bart Schaefer 2017-03-09 10:29 ` Peter Stephenson 0 siblings, 1 reply; 14+ messages in thread From: Bart Schaefer @ 2017-03-09 5:18 UTC (permalink / raw) To: Martijn Dekker, Peter Stephenson, Zsh hackers list On Mar 8, 7:12pm, Martijn Dekker wrote: } } Commit f3f8537c breaks unquoted $-: We get here: #0 fetchvalue (v=0xbff07f00, pptr=0xbff07f64, bracks=1, flags=0) at ../../zsh-5.0/Src/params.c:1969 #1 0x080de7db in paramsubst (l=0xb7d68778, n=0xb7d68790, str=0xbff07fd8, qt=0, pf_flags=0, ret_flags=0xbff0813c) at ../../zsh-5.0/Src/subst.c:2418 #2 0x080dacff in stringsubst (list=0xb7d68778, node=0xb7d68790, pf_flags=0, ret_flags=0xbff0813c, asssub=0) at ../../zsh-5.0/Src/subst.c:247 #3 0x080da47d in prefork (list=0xb7d68778, flags=0, ret_flags=0xbff0813c) at ../../zsh-5.0/Src/subst.c:85 With pptr still pointing at "\233" (Dash). Hopefully the following doesn't re-break anything else. diff --git a/Src/params.c b/Src/params.c index b89fb74..bee01c5 100644 --- a/Src/params.c +++ b/Src/params.c @@ -1987,7 +1987,9 @@ fetchvalue(Value v, char **pptr, int bracks, int flags) *s++ = '$'; else if (c == Star) *s++ = '*'; - else if (c == '#' || c == '-' || c == '?' || c == '$' || + else if (IS_DASH(c)) + *s++ = '-'; + else if (c == '#' || c == '?' || c == '$' || c == '!' || c == '@' || c == '*') s++; else ^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [^ax-y] doesn't work but [^x-ya] does 2017-03-09 5:18 ` Bart Schaefer @ 2017-03-09 10:29 ` Peter Stephenson 0 siblings, 0 replies; 14+ messages in thread From: Peter Stephenson @ 2017-03-09 10:29 UTC (permalink / raw) To: Zsh hackers list On Wed, 8 Mar 2017 21:18:22 -0800 Bart Schaefer <schaefer@brasslantern.com> wrote: > With pptr still pointing at "\233" (Dash). Hopefully the following > doesn't re-break anything else. > > diff --git a/Src/params.c b/Src/params.c > index b89fb74..bee01c5 100644 > --- a/Src/params.c > +++ b/Src/params.c > @@ -1987,7 +1987,9 @@ fetchvalue(Value v, char **pptr, int bracks, int flags) > *s++ = '$'; > else if (c == Star) > *s++ = '*'; > - else if (c == '#' || c == '-' || c == '?' || c == '$' || > + else if (IS_DASH(c)) > + *s++ = '-'; > + else if (c == '#' || c == '?' || c == '$' || > c == '!' || c == '@' || c == '*') > s++; > else > I'll tack this test onto D04 when I commit it. pws () { setopt localoptions extendedglob [[ $- = [[:alnum:]]## ]] || print Failed 1 [[ ${-} = [[:alnum:]]## ]] || print Failed 2 } 0:$- expansion correctly handles Dash token ^ permalink raw reply [flat|nested] 14+ messages in thread
end of thread, other threads:[~2017-03-09 10:30 UTC | newest]
Thread overview: 14+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-02-24 20:38 [^ax-y] doesn't work but [^x-ya] does Sebastian Gniazdowski
2017-02-25 7:32 ` Bart Schaefer
2017-02-25 8:21 ` Sebastian Gniazdowski
[not found] <CGME20170302073436epcas2p1ee2d6d723bc7b0865e11243777a08417@epcas2p1.samsung.com>
2017-03-02 7:33 ` Sebastian Gniazdowski
2017-03-02 8:36 ` Sebastian Gniazdowski
2017-03-02 10:47 ` Peter Stephenson
2017-03-02 11:22 ` Sebastian Gniazdowski
2017-03-02 11:53 ` Peter Stephenson
2017-03-04 23:21 ` Bart Schaefer
2017-03-06 11:28 ` Peter Stephenson
2017-03-07 10:51 ` Peter Stephenson
2017-03-08 19:12 ` Martijn Dekker
2017-03-09 5:18 ` Bart Schaefer
2017-03-09 10:29 ` Peter Stephenson
Code repositories for project(s) associated with this public inbox https://git.vuxu.org/mirror/zsh/ This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).