diff options
-rw-r--r-- | include/wordsplit.h | 18 | ||||
-rw-r--r-- | src/wordsplit.c | 185 | ||||
-rw-r--r-- | tests/wordsplit.at | 122 | ||||
-rw-r--r-- | tests/wsp.c | 818 |
4 files changed, 688 insertions, 455 deletions
diff --git a/include/wordsplit.h b/include/wordsplit.h index a175275..d7eb26f 100644 --- a/include/wordsplit.h +++ b/include/wordsplit.h @@ -76,6 +76,16 @@ struct wordsplit size_t ws_envidx; /* Index of first free slot */ size_t ws_envsiz; /* Size of the ws_envbuf array */ + char const **ws_paramv; /* [WRDSO_PARAMV] User-supplied positional + parameters */ + size_t ws_paramc; /* Number of positional parameters */ + + /* Temporary storage for parameters. Works similarly to ws_enbuf. + */ + char **ws_parambuf; + size_t ws_paramidx; + size_t ws_paramsiz; + int (*ws_getvar) (char **ret, const char *var, size_t len, void *clos); /* [Input] (WRDSF_GETVAR, !WRDSF_NOVAR) Looks up the name VAR (LEN bytes long) in the table of @@ -226,6 +236,12 @@ struct wordsplit $(echo foo bar) */ #define WRDSO_NOCMDSPLIT 0x00002000 +/* Enable positional parameters */ +#define WRDSO_PARAMV 0x00004000 +/* Enable negative positional indices (${-1} is the last positional + parameter) */ +#define WRDSO_PARAM_NEGIDX 0x00008000 + #define WRDSO_BSKEEP WRDSO_BSKEEP_WORD #define WRDSO_OESC WRDSO_OESC_WORD #define WRDSO_XESC WRDSO_XESC_WORD @@ -250,12 +266,14 @@ struct wordsplit #define WRDSE_PAREN 7 #define WRDSE_GLOBERR 8 #define WRDSE_USERERR 9 +#define WRDSE_BADPARAM 10 int wordsplit (const char *s, wordsplit_t *ws, int flags); int wordsplit_len (const char *s, size_t len, wordsplit_t *ws, int flags); void wordsplit_free (wordsplit_t *ws); void wordsplit_free_words (wordsplit_t *ws); void wordsplit_free_envbuf (wordsplit_t *ws); +void wordsplit_free_parambuf (struct wordsplit *ws); int wordsplit_get_words (wordsplit_t *ws, size_t *wordc, char ***wordv); static inline void wordsplit_getwords (wordsplit_t *ws, size_t *wordc, char ***wordv) diff --git a/src/wordsplit.c b/src/wordsplit.c index 521a1eb..6a33636 100644 --- a/src/wordsplit.c +++ b/src/wordsplit.c @@ -1,5 +1,5 @@ /* wordsplit - a word splitter - Copyright (C) 2009-2018 Sergey Poznyakoff + Copyright (C) 2009-2019 Sergey Poznyakoff This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the @@ -27,6 +27,7 @@ #include <stdarg.h> #include <pwd.h> #include <glob.h> +#include <limits.h> #if ENABLE_NLS # include <gettext.h> @@ -56,6 +57,9 @@ #define WSP_RETURN_DELIMS(wsp) \ ((wsp)->ws_flags & WRDSF_RETURN_DELIMS || ((wsp)->ws_options & WRDSO_MAXWORDS)) +#define to_num(c) \ + (ISDIGIT(c) ? c - '0' : (ISXDIGIT(c) ? toupper(c) - 'A' + 10 : 255 )) + #define ALLOC_INIT 128 #define ALLOC_INCR 128 @@ -285,6 +289,14 @@ wordsplit_init (struct wordsplit *wsp, const char *input, size_t len, } } + if (!(wsp->ws_options & WRDSO_PARAMV)) + { + wsp->ws_paramv = NULL; + wsp->ws_paramc = 0; + } + wsp->ws_paramidx = wsp->ws_paramsiz = 0; + wsp->ws_parambuf = NULL; + wsp->ws_endp = 0; wsp->ws_wordi = 0; @@ -1036,7 +1048,7 @@ wordsplit_find_env (struct wordsplit *wsp, const char *name, size_t len, static int wsplt_assign_var (struct wordsplit *wsp, const char *name, size_t namelen, - char *value) + char const *value) { int n = (wsp->ws_flags & WRDSF_ENV_KV) ? 2 : 1; char *v; @@ -1071,7 +1083,7 @@ wsplt_assign_var (struct wordsplit *wsp, const char *name, size_t namelen, { for (; j > 1; j--) free (newenv[j-1]); - free (newenv[j-1]); + free (newenv); return _wsplt_nomem (wsp); } } @@ -1142,6 +1154,70 @@ wsplt_assign_var (struct wordsplit *wsp, const char *name, size_t namelen, return WRDSE_OK; } +int +wsplt_assign_param (struct wordsplit *wsp, int param_idx, char *value) +{ + char *v; + + if (param_idx < 0) + return WRDSE_BADPARAM; + if (param_idx == wsp->ws_paramc) + { + char **parambuf; + if (!wsp->ws_parambuf) + { + size_t i; + + parambuf = calloc ((size_t)param_idx + 1, sizeof (parambuf[0])); + if (!parambuf) + return _wsplt_nomem (wsp); + + for (i = 0; i < wsp->ws_paramc; i++) + { + parambuf[i] = strdup (wsp->ws_paramv[i]); + if (!parambuf[i]) + { + for (; i > 1; i--) + free (parambuf[i-1]); + free (parambuf); + return _wsplt_nomem (wsp); + } + } + + wsp->ws_parambuf = parambuf; + wsp->ws_paramidx = param_idx; + wsp->ws_paramsiz = param_idx + 1; + } + else + { + size_t n = wsp->ws_paramsiz; + + if ((size_t) -1 / 3 * 2 / sizeof (wsp->ws_parambuf[0]) <= n) + return _wsplt_nomem (wsp); + n += (n + 1) / 2; + parambuf = realloc (wsp->ws_parambuf, n * sizeof (wsp->ws_parambuf[0])); + if (!parambuf) + return _wsplt_nomem (wsp); + wsp->ws_parambuf = parambuf; + wsp->ws_paramsiz = n; + wsp->ws_parambuf[param_idx] = NULL; + } + + wsp->ws_paramv = (const char**) wsp->ws_parambuf; + wsp->ws_paramc = param_idx + 1; + } + else if (param_idx > wsp->ws_paramc) + return WRDSE_BADPARAM; + + v = strdup (value); + if (!v) + return _wsplt_nomem (wsp); + + free (wsp->ws_parambuf[param_idx]); + wsp->ws_parambuf[param_idx] = v; + return WRDSE_OK; +} + /* Recover from what looked like a variable reference, but turned out not to be one. STR points to first character after '$'. */ static int @@ -1177,6 +1253,8 @@ expvar (struct wordsplit *wsp, const char *str, size_t len, const char *start = str - 1; int rc; struct wordsplit ws; + int is_param = 0; + long param_idx = 0; if (ISVARBEG (str[0])) { @@ -1185,16 +1263,39 @@ expvar (struct wordsplit *wsp, const char *str, size_t len, break; *pend = str + i - 1; } - else if (ISDIGIT (str[0])) + else if ((wsp->ws_options & WRDSO_PARAMV) && ISDIGIT (str[0])) { i = 1; *pend = str; + is_param = 1; + param_idx = to_num (str[0]); } - else if (str[0] == '{' && (ISVARBEG (str[1]) || ISDIGIT (str[1]))) + else if ((wsp->ws_options & WRDSO_PARAMV) && str[0] == '#') + { + char b[16]; + snprintf (b, sizeof(b), "%d", (int) wsp->ws_paramc); + value = strdup (b); + if (!value) + return _wsplt_nomem (wsp); + if (wsnode_new (wsp, &newnode)) + return 1; + wsnode_insert (wsp, newnode, *ptail, 0); + *ptail = newnode; + newnode->flags = _WSNF_WORD | _WSNF_NOEXPAND | flg; + newnode->v.word = value; + return 0; + } + else if (str[0] == '{' + && (ISVARBEG (str[1]) + || (is_param = (((wsp->ws_options & WRDSO_PARAMV) + && ISDIGIT (str[1])) + || ((wsp->ws_options & WRDSO_PARAM_NEGIDX) + && (str[1] == '-' + && ISDIGIT (str[2]))))) != 0)) { str++; len--; - for (i = 1; i < len; i++) + for (i = str[0] == '-' ? 1 : 0; i < len; i++) { if (str[i] == ':') { @@ -1222,9 +1323,16 @@ expvar (struct wordsplit *wsp, const char *str, size_t len, *pend = str + j; break; } - else if (ISDIGIT (str[1])) + else if (is_param) + { + if (ISDIGIT (str[i])) { - if (!ISDIGIT (str[i])) + param_idx = param_idx * 10 + to_num (str[i]); + if ((str[0] == '-' && -param_idx < INT_MIN) + || param_idx > INT_MAX) + return expvar_recover (wsp, str - 1, ptail, pend, flg); + } + else { return expvar_recover (wsp, str - 1, ptail, pend, flg); } @@ -1234,6 +1342,10 @@ expvar (struct wordsplit *wsp, const char *str, size_t len, return expvar_recover (wsp, str - 1, ptail, pend, flg); } } + + if (is_param && str[0] == '-') + param_idx = wsp->ws_paramc - param_idx; + if (i == len) return _wsplt_seterr (wsp, WRDSE_CBRACE); } @@ -1254,6 +1366,21 @@ expvar (struct wordsplit *wsp, const char *str, size_t len, } else { + if (is_param) + { + if (param_idx >= 0 && param_idx < wsp->ws_paramc) + { + value = strdup (wsp->ws_paramv[param_idx]); + if (!value) + rc = WRDSE_NOSPACE; + else + rc = WRDSE_OK; + } + else + rc = WRDSE_UNDEF; + } + else + { rc = wordsplit_find_env (wsp, str, i, &vptr); if (rc == WRDSE_OK) { @@ -1270,6 +1397,7 @@ expvar (struct wordsplit *wsp, const char *str, size_t len, rc = wsp->ws_getvar (&value, str, i, wsp->ws_closure); else rc = WRDSE_UNDEF; + } if (rc == WRDSE_OK && (!value || value[0] == 0) @@ -1321,7 +1449,17 @@ expvar (struct wordsplit *wsp, const char *str, size_t len, wordsplit_free (&ws); if (defstr[-1] == '=') - wsplt_assign_var (wsp, str, i, value); + { + if (is_param) + rc = wsplt_assign_param (wsp, param_idx, value); + else + rc = wsplt_assign_var (wsp, str, i, value); + } + if (rc) + { + free (value); + return rc; + } } else { @@ -1458,7 +1596,7 @@ expvar (struct wordsplit *wsp, const char *str, size_t len, static int begin_var_p (int c) { - return c == '{' || ISVARBEG (c) || ISDIGIT (c); + return c == '{' || c == '#' || ISVARBEG (c) || ISDIGIT (c); } static int @@ -2106,9 +2244,6 @@ scan_word (struct wordsplit *wsp, size_t start, int consume_all) return _WRDS_OK; } -#define to_num(c) \ - (ISDIGIT(c) ? c - '0' : (ISXDIGIT(c) ? toupper(c) - 'A' + 10 : 255 )) - static int xtonum (int *pval, const char *src, int base, int cnt) { @@ -2498,7 +2633,7 @@ wordsplit_free_words (struct wordsplit *ws) void wordsplit_free_envbuf (struct wordsplit *ws) { - if (ws->ws_flags & WRDSF_NOCMD) + if (!(ws->ws_flags & WRDSF_ENV)) return; if (ws->ws_envbuf) { @@ -2513,6 +2648,23 @@ wordsplit_free_envbuf (struct wordsplit *ws) } void +wordsplit_free_parambuf (struct wordsplit *ws) +{ + if (!(ws->ws_options & WRDSO_PARAMV)) + return; + if (ws->ws_parambuf) + { + size_t i; + + for (i = 0; ws->ws_parambuf[i]; i++) + free (ws->ws_parambuf[i]); + free (ws->ws_parambuf); + ws->ws_paramidx = ws->ws_paramsiz = 0; + ws->ws_parambuf = NULL; + } +} + +void wordsplit_clearerr (struct wordsplit *ws) { if (ws->ws_errno == WRDSE_USERERR) @@ -2529,6 +2681,7 @@ wordsplit_free (struct wordsplit *ws) free (ws->ws_wordv); ws->ws_wordv = NULL; wordsplit_free_envbuf (ws); + wordsplit_free_parambuf (ws); } int @@ -2557,7 +2710,9 @@ const char *_wordsplit_errstr[] = { N_("undefined variable"), N_("input exhausted"), N_("unbalanced parenthesis"), - N_("globbing error") + N_("globbing error"), + N_("user-defined error"), + N_("invalid parameter number in assignment") }; int _wordsplit_nerrs = sizeof (_wordsplit_errstr) / sizeof (_wordsplit_errstr[0]); diff --git a/tests/wordsplit.at b/tests/wordsplit.at index 1f2e80d..ebd168b 100644 --- a/tests/wordsplit.at +++ b/tests/wordsplit.at @@ -108,7 +108,7 @@ WSPGROUP() dnl ------------------------------------------------------------ dnl Test worsplit-specific behavior dnl ------------------------------------------------------------ -TESTWSP([append],[wsp-append],[append], +TESTWSP([append],[wsp-append],[-append], [jeden dwa trzy cztery piec szesc], @@ -133,7 +133,7 @@ NF: 6 TOTAL: 2 ]) -TESTWSP([dooffs],[wsp-doofs ],[dooffs 3 jeden dwa trzy], +TESTWSP([dooffs],[wsp-doofs ],[-dooffs jeden dwa trzy], [cztery piec], [NF: 2 (3) (0): jeden @@ -214,7 +214,7 @@ TOTAL: 3 [], [unset FOO;]) -TESTWSP([undefined variables 2],[],[keepundef], +TESTWSP([undefined variables 2],[],[-keepundef], [a $FOO test a${FOO}b], [NF: 4 0: a @@ -226,7 +226,7 @@ TOTAL: 4 [], [unset FOO;]) -TESTWSP([warn about undefined variables],[],[warnundef], +TESTWSP([warn about undefined variables],[],[-warnundef], [$FOO], [NF: 0 TOTAL: 0 @@ -235,14 +235,14 @@ TOTAL: 0 ], [unset FOO;]) -TESTWSP([bail out on undefined variables],[],[undef], +TESTWSP([bail out on undefined variables],[],[-undef], [$FOO], [], [undefined variable ], [unset FOO;]) -TESTWSP([disable variable expansion],[],[novar], +TESTWSP([disable variable expansion],[],[-novar], [$FOO], [NF: 1 0: $FOO @@ -252,7 +252,7 @@ TOTAL: 1 [FOO=bar]) TESTWSP([K/V environment],[wsp-env-kv wsp-env_kv], -[env_kv], +[-env_kv], [$FOO a$BAZ], [NF: 2 0: bar @@ -262,7 +262,7 @@ TOTAL: 2 [], [FOO=bar BAZ=qux]) -TESTWSP([nosplit with variable expansion],[wsp-var-nosplit],[nosplit], +TESTWSP([nosplit with variable expansion],[wsp-var-nosplit],[-nosplit], [a $FOO test], [NF: 1 0: "a variable expansion test\n" @@ -271,7 +271,7 @@ TOTAL: 1 [], [FOO="variable expansion"]) -TESTWSP([nosplit without variable expansion],[],[nosplit novar], +TESTWSP([nosplit without variable expansion],[],[-nosplit -novar], [a $FOO test], [NF: 1 0: "a $FOO test\n" @@ -280,7 +280,7 @@ TOTAL: 1 [], [FOO="variable expansion"]) -TESTWSP([nosplit: empty expansion],[],[nosplit trimnl], +TESTWSP([nosplit: empty expansion],[],[-nosplit -trimnl], [$FOO], [NF: 1 0: "" @@ -289,15 +289,6 @@ TOTAL: 1 [], [FOO=""]) -TESTWSP([default value (defined)],[],[], -[${FOO:-bar}], -[NF: 1 -0: qux -TOTAL: 1 -], -[], -[FOO=qux]) - TESTWSP([default value],[],[], [${FOO:-bar}], [NF: 1 @@ -451,7 +442,7 @@ TOTAL: 2 WSPGROUP() -TESTWSP([ignore quotes],[wsp-ignore-quotes ],[-quote], +TESTWSP([ignore quotes],[wsp-ignore-quotes ],[-noquote], ["a text"], [NF: 2 0: "\"a" @@ -462,7 +453,7 @@ TOTAL: 2 WSPGROUP(wsp-delim) TESTWSP([custom delimiters (squeeze)],[], -[delim : -ws trimnl], +[-delim : -nows -trimnl], [semicolon: separated::list: of :words], [NF: 5 0: semicolon @@ -474,7 +465,7 @@ TOTAL: 5 ]) TESTWSP([custom delimiters (no squeeze)],[], -[delim : -ws -squeeze_delims trimnl], +[-delim : -nows -nosqueeze_delims -trimnl], [semicolon: separated::list: of :words], [NF: 6 0: semicolon @@ -487,7 +478,7 @@ TOTAL: 6 ]) TESTWSP([custom, with returned delimiters],[], -[delim : -ws trimnl return_delims], +[-delim : -nows -trimnl -return_delims], [semicolon: separated::list: of :words], [NF: 9 0: semicolon @@ -503,7 +494,7 @@ TOTAL: 9 ]) TESTWSP([custom, with returned & squeezed delimiters],[], -[delim : -ws trimnl return_delims -squeeze_delims], +[-delim : -nows -trimnl -return_delims -nosqueeze_delims], [semicolon: separated::list: of :words], [NF: 10 0: semicolon @@ -521,7 +512,7 @@ TOTAL: 10 WSPGROUP(wsp-sed) -TESTWSP([sed expressions],[],[sed], +TESTWSP([sed expressions],[],[-sed], [arg1 s/foo/bar/g;s/bar baz/quz quux/ arg2], [NF: 3 0: arg1 @@ -532,7 +523,7 @@ TOTAL: 3 WSPGROUP() -TESTWSP([C escapes on],[wcp-c-escape],[cescapes], +TESTWSP([C escapes on],[wcp-c-escape],[-cescapes], [a\ttab form\ffeed and new\nline], [NF: 4 0: a\ttab @@ -542,7 +533,7 @@ TESTWSP([C escapes on],[wcp-c-escape],[cescapes], TOTAL: 4 ]) -TESTWSP([C escapes off],[wcp-c-escape-off],[-cescapes], +TESTWSP([C escapes off],[wcp-c-escape-off],[-nocescapes], [a\ttab form\ffeed and new\nline], [NF: 4 0: attab @@ -552,7 +543,7 @@ TESTWSP([C escapes off],[wcp-c-escape-off],[-cescapes], TOTAL: 4 ]) -TESTWSP([ws elimination],[wsp-ws-elim],[delim ' ()' ws return_delims], +TESTWSP([ws elimination],[wsp-ws-elim],[-delim ' ()' -ws -return_delims], [( list items )], [NF: 4 0: ( @@ -563,7 +554,7 @@ TOTAL: 4 ]) TESTWSP([ws elimination + return delim],[wsp-ws-elim-ret], -[-default novar nocmd delim ":," return_delims ws dquote], +[-nodefault -novar -nocmd -delim ":," -return_delims -ws -dquote], ["foo" : "bar", "quux" : "baaz" ], [NF: 7 0: foo @@ -576,7 +567,7 @@ TESTWSP([ws elimination + return delim],[wsp-ws-elim-ret], TOTAL: 7 ]) -TESTWSP([empty quotes],[wsp-empty-quotes],[delim : ws return_delims], +TESTWSP([empty quotes],[wsp-empty-quotes],[-delim : -ws -return_delims], [t=""], [NF: 1 0: t= @@ -584,7 +575,7 @@ TOTAL: 1 ]) TESTWSP([delimiter following empty quotes], -[],[delim : ws return_delims], +[],[-delim : -ws -return_delims], [t="":r], [NF: 3 0: t= @@ -595,7 +586,7 @@ TOTAL: 3 TESTWSP([suppress ws trimming within quotes], [], -[default delim , ws return_delims], +[-default -delim , -ws -return_delims], [nocomponent,nonewline, formatfield="In message %{text}, "], [NF: 5 0: nocomponent @@ -608,7 +599,7 @@ TOTAL: 5 TESTWSP([unescape], [wsp-unescape wsp-unescape-simple], -[-default novar nocmd quote escape ':+:\\""'], +[-nodefault -novar -nocmd -quote -escape ':+:\\""'], [\Seen "quote \"" "bs \\"], [NF: 3 0: \\Seen @@ -619,7 +610,7 @@ TOTAL: 3 TESTWSP([unescape: word/quote], [wsp-unescape wsp-unescape-word], -[-default novar nocmd quote escape-word '\\""' escape-quote ':+0x:\\""'], +[-nodefault -novar -nocmd -quote -escape-word '\\""' -escape-quote ':+0x:\\""'], [\Seen "quote \"" "bs \\" "3\x31 \101" 3\x31 \101], [NF: 6 0: Seen @@ -631,7 +622,7 @@ TESTWSP([unescape: word/quote], TOTAL: 6 ]) -TESTWSP([dquote],[],[-default novar nocmd dquote], +TESTWSP([dquote],[],[-nodefault -novar -nocmd -dquote], [a "quoted example" isn't it], [NF: 4 0: a @@ -641,7 +632,7 @@ TESTWSP([dquote],[],[-default novar nocmd dquote], TOTAL: 4 ]) -TESTWSP([squote],[],[-default novar nocmd squote], +TESTWSP([squote],[],[-nodefault -novar -nocmd -squote], [a 'quoted example' isn"t it], [NF: 4 0: a @@ -653,7 +644,7 @@ TOTAL: 4 WSPGROUP(wsp-incr) -TESTWSP([incremental],[],[incremental], +TESTWSP([incremental],[],[-incremental], [incremental "input test" line @@ -671,7 +662,7 @@ TOTAL: 3 [input exhausted ]) -TESTWSP([incremental append],[],[incremental append], +TESTWSP([incremental append],[],[-incremental -append], [incremental "input test" line @@ -693,7 +684,7 @@ TOTAL: 3 ]) TESTWSP([incremental ws], -[],[return_delims -squeeze_delims incremental ws], +[],[-return_delims -nosqueeze_delims -incremental -ws], [a list test @@ -711,7 +702,7 @@ TOTAL: 3 [input exhausted ]) -TESTWSP([incremental nosplit],[],[incremental nosplit], +TESTWSP([incremental nosplit],[],[-incremental -nosplit], [incremental "input test" line ], [NF: 1 @@ -721,7 +712,7 @@ TOTAL: 1 [input exhausted ]) -TESTWSP([simple command substitution],[],[-nocmd], +TESTWSP([simple command substitution],[],[-cmd], [begin $(words a b) end], [NF: 4 0: begin @@ -731,7 +722,7 @@ TESTWSP([simple command substitution],[],[-nocmd], TOTAL: 4 ]) -TESTWSP([quoted command substitution],[],[-nocmd], +TESTWSP([quoted command substitution],[],[-cmd], [begin "$(words a b)" end], [NF: 3 0: begin @@ -740,7 +731,7 @@ TESTWSP([quoted command substitution],[],[-nocmd], TOTAL: 3 ]) -TESTWSP([coalesced command substitution],[],[-nocmd], +TESTWSP([coalesced command substitution],[],[-cmd], [begin($(words a b))end], [NF: 2 0: begin(a @@ -748,14 +739,14 @@ TESTWSP([coalesced command substitution],[],[-nocmd], TOTAL: 2 ]) -TESTWSP([quoted coalesced command substitution],[],[-nocmd], +TESTWSP([quoted coalesced command substitution],[],[-cmd], ["begin($(words a b))end"], [NF: 1 0: "begin(a b)end" TOTAL: 1 ]) -TESTWSP([variable and command substitution],[],[-nocmd -novar], +TESTWSP([variable and command substitution],[],[-cmd -var], [begin $X $(words $X $Y) end], [NF: 5 0: begin @@ -766,14 +757,14 @@ TESTWSP([variable and command substitution],[],[-nocmd -novar], TOTAL: 5 ],[],[X=a Y=b]) -TESTWSP([variable expansion and command substitution in quotes],[],[-nocmd -novar], +TESTWSP([variable expansion and command substitution in quotes],[],[-cmd -var], ["${BEGIN}($(words $X $Y))end"], [NF: 1 0: "begin(a b)end" TOTAL: 1 ],[],[X=a Y=b BEGIN=begin]) -TESTWSP([nested commands],[],[-nocmd -novar], +TESTWSP([nested commands],[],[-cmd -var], [$(words output $(words in$SUFFIX text) end)], [NF: 4 0: output @@ -792,7 +783,7 @@ mkdir dir > dir/2.c > dir/3.b -wsp pathexpand<<'EOT' +wsp -pathexpand<<'EOT' begin dir/*.c end EOT ], @@ -813,7 +804,7 @@ mkdir dir > dir/1.c > dir/2.b -wsp pathexpand<<'EOT' +wsp -pathexpand<<'EOT' begin dir/*.d end EOT ], @@ -833,7 +824,7 @@ mkdir dir > dir/1.c > dir/2.b -wsp pathexpand nullglob<<'EOT' +wsp -pathexpand -nullglob<<'EOT' begin dir/*.d end EOT ], @@ -852,7 +843,7 @@ mkdir dir > dir/1.c > dir/2.b -wsp pathexpand failglob<<'EOT' +wsp -pathexpand -failglob<<'EOT' begin dir/*.d end EOT ], @@ -862,7 +853,7 @@ EOT ]) AT_CLEANUP -TESTWSP([append],[],[-- extra arguments follow], +TESTWSP([append],[],[-append-args extra arguments follow], [some words and], [NF: 6 0: some @@ -875,7 +866,7 @@ TOTAL: 3 ]) TESTWSP([append + dooffs + env],[], -[-env dooffs 2 preface words V=2 -- extra arguments follow], +[-env none -dooffs preface words -- V=2 -append-args extra arguments follow], [some words and var=$V], [NF: 7 (2) (0): preface @@ -892,7 +883,7 @@ TOTAL: 4 # Maxwords TESTWSP([maxwords],[], -[trimnl maxwords 3], +[-trimnl -maxwords 3], [ws_maxwords limits the number of returned words], [NF: 3 0: ws_maxwords @@ -902,7 +893,7 @@ TOTAL: 3 ]) TESTWSP([maxwords return_delims],[], -[trimnl maxwords 8 return_delims delim :-], +[-trimnl -maxwords 8 -return_delims -delim :-], [foo:::bar-:baz-quux:ux:zu], [NF: 8 0: foo @@ -917,7 +908,7 @@ TOTAL: 8 ]) TESTWSP([maxwords return_delims -squeeze_delims],[], -[trimnl maxwords 8 return_delims -squeeze_delims delim :-], +[-trimnl -maxwords 8 -return_delims -nosqueeze_delims -delim :-], [foo:::bar-:baz:qux-], [NF: 8 0: foo @@ -932,7 +923,7 @@ TOTAL: 8 ]) TESTWSP([maxwords incremental],[], -[trimnl maxwords 3 incremental], +[-trimnl -maxwords 3 -incremental], [foo bar baz qux uz @@ -950,7 +941,7 @@ TOTAL: 3 [input exhausted ])) -TESTWSP([variable nosplit],[],[novar novarsplit], +TESTWSP([variable nosplit],[],[-novar -novarsplit], [begin ${VAR:- a b} end], [NF: 3 0: begin @@ -959,7 +950,7 @@ TESTWSP([variable nosplit],[],[novar novarsplit], TOTAL: 3 ]) -TESTWSP([command nosplit],[],[nocmd nocmdsplit], +TESTWSP([command nosplit],[],[-nocmd -nocmdsplit], [begin $(words a b) end], [NF: 3 0: begin @@ -968,6 +959,19 @@ TESTWSP([command nosplit],[],[nocmd nocmdsplit], TOTAL: 3 ]) +TESTWSP([positional parameters],[],[one two three four five six seven eight nine ten eleven twelve], +[$0 $5 ${10} +$#], +[NF: 3 +0: one +1: six +2: eleven +TOTAL: 3 +NF: 1 +0: 12 +TOTAL: 1 +]) + m4_popdef([TESTWSP]) m4_popdef([wspnum]) m4_popdef([wspid]) diff --git a/tests/wsp.c b/tests/wsp.c index bd13e63..958d01f 100644 --- a/tests/wsp.c +++ b/tests/wsp.c @@ -1,5 +1,5 @@ /* grecs - Gray's Extensible Configuration System - Copyright (C) 2014-2016 Sergey Poznyakoff + Copyright (C) 2014-2019 Sergey Poznyakoff Grecs is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the @@ -29,99 +29,408 @@ extern char **environ; char *progname; -struct kwd +/* Global options */ +enum { - const char *name; - int tok; + TRIMNL_OPTION = 0x01, /* Remove trailing newline */ + PLAINTEXT_OPTION = 0x02 /* Print intput verbatim (no escapes) */ }; -struct kwd bool_keytab[] = { - { "append", WRDSF_APPEND }, - /*{ "reuse", WRDSF_REUSE },*/ - { "undef", WRDSF_UNDEF }, - { "novar", WRDSF_NOVAR }, - { "nocmd", WRDSF_NOCMD }, - { "ws", WRDSF_WS }, - { "quote", WRDSF_QUOTE }, - { "squote", WRDSF_SQUOTE }, - { "dquote", WRDSF_DQUOTE }, - { "squeeze_delims", WRDSF_SQUEEZE_DELIMS }, - { "return_delims", WRDSF_RETURN_DELIMS }, - { "sed", WRDSF_SED_EXPR }, - { "debug", WRDSF_SHOWDBG }, - { "nosplit", WRDSF_NOSPLIT }, - { "keepundef", WRDSF_KEEPUNDEF }, - { "warnundef", WRDSF_WARNUNDEF }, - { "cescapes", WRDSF_CESCAPES }, - { "default", WRDSF_DEFFLAGS }, - { "env_kv", WRDSF_ENV_KV }, - { "incremental", WRDSF_INCREMENTAL }, - { "pathexpand", WRDSF_PATHEXPAND }, - { NULL, 0 } +/* Environment types */ +enum env_type + { + env_none, /* No environment */ + env_null, /* Null environment */ + env_sys /* Use system environment */ }; -struct kwd opt_keytab[] = { - { "nullglob", WRDSO_NULLGLOB }, - { "failglob", WRDSO_FAILGLOB }, - { "dotglob", WRDSO_DOTGLOB }, - { "bskeep_words", WRDSO_BSKEEP_WORD }, - { "bskeep_quote", WRDSO_BSKEEP_QUOTE }, - { "bskeep", WRDSO_BSKEEP_WORD|WRDSO_BSKEEP_QUOTE }, - { "novarsplit", WRDSO_NOVARSPLIT }, - { "nocmdsplit", WRDSO_NOCMDSPLIT }, - { NULL, 0 } +struct wsclosure +{ + int options; /* Global options */ + struct wordsplit ws; /* The wordsplit structure */ + int wsflags; /* Wordsplit flags */ + enum env_type env_type; /* Environment type */ + int offarg; /* Index of the first of the initial words in + the argv array. The ws.ws_dooffs field gives + the number of such variables. Forces the + WRDSF_DOOFFS flag. */ + char **fenvbase; /* Environment for testing the ws_getenv function */ + int fenvidx; /* Number of variables in fenvbase */ + int fenvmax; /* Size of fenbase (entries) */ + int append_start; /* First argument to append (index in argv) */ + int append_count; /* Number of arguments to append */ }; -struct kwd string_keytab[] = { - { "delim", WRDSF_DELIM }, - { "comment", WRDSF_COMMENT }, - { "escape", WRDSF_ESCAPE }, - { NULL, 0 } +/* Command line option types */ +enum + { + ws_no_argument, /* Option requires no arguments */ + ws_boolean, /* Option is boolean (can be prefixed with -no) */ + ws_required_argument, /* Option requires one argument */ + ws_multiple_arguments /* Option takes multiple arguments, terminated with + "--" or end of argument list */ }; +/* Structure describing a single command-line option */ +struct wsopt +{ + const char *name; /* Option name */ + int tok; /* Corresponding flag */ + int arg; /* Option type (see the enum above) */ + void (*setfn) (int tok, int neg, char *arg, struct wsclosure *wsc); + /* Setter function */ +}; + +/* Index of the next argument in the argv */ +static int wsoptind = -1; + +/* Parse next argument from the command line. Return EOF on end of arguments + or when the "--" argument is seen. */ static int -kwxlat (struct kwd *kwp, const char *str, int *res) +getwsopt (int argc, char **argv, struct wsopt *wso, struct wsclosure *wsc) +{ + int negate = 0; + char *opt; + + if (wsoptind == -1) + wsoptind = 1; + if (wsoptind == argc) + return EOF; + + opt = argv[wsoptind++]; + if (strcmp (opt, "--") == 0) + return EOF; + if (*opt != '-') { - for (; kwp->name; kwp++) - if (strcmp (kwp->name, str) == 0) + if (strchr (opt, '=')) { - *res = kwp->tok; + assert (wsc->fenvidx < wsc->fenvmax - 1); + wsc->fenvbase[wsc->fenvidx++] = opt; return 0; } - return -1; + wsoptind--; + return EOF; + } + opt++; /* skip past initial dash */ + if (strncmp (opt, "no-", 3) == 0) + { + negate = 1; + opt += 3; + } + else if (strncmp (opt, "no", 2) == 0) + { + negate = 1; + opt += 2; + } + + for (; wso->name; wso++) + { + if (wso->arg == ws_boolean && wso->name[0] == 'n' && wso->name[1] == 'o' + && strcmp (wso->name + 2, opt) == 0) + { + negate ^= 1; + break; + } + if (strcmp (wso->name, opt) == 0) + break; + } + + if (wso->name) + { + char *arg; + if (wso->arg == ws_multiple_arguments) + { + while (1) + { + if (wsoptind == argc) + break; + arg = argv[wsoptind++]; + if (strcmp (arg, "--") == 0) + break; + wso->setfn (wso->tok, negate, arg, wsc); + } + } + else + { + if (wso->arg == ws_required_argument) + { + if (wsoptind == argc) + { + fprintf (stderr, "%s: missing arguments for -%s\n", + progname, opt); + exit (1); + } + arg = argv[wsoptind++]; + } + wso->setfn (wso->tok, negate, arg, wsc); + } + return 0; + } + + fprintf (stderr, "%s: unrecognized option: -%s\n", + progname, opt); + fprintf (stderr, "%s: try %s -help for more detail\n", + progname, progname); + exit (1); +} + +/* Setter functions for various options */ + +static void +setfn_flag (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + if (neg) + wsc->wsflags &= ~flag; + else + wsc->wsflags |= flag; +} + +static void +setfn_option (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + wsc->wsflags |= WRDSF_OPTIONS; + if (neg) + wsc->ws.ws_options &= ~flag; + else + wsc->ws.ws_options |= flag; } static void -help () +setfn_delim (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + wsc->wsflags |= flag; + wsc->ws.ws_delim = arg; +} + +static void +setfn_comment (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + wsc->wsflags |= flag; + wsc->ws.ws_comment = arg; +} + +static void +set_escape_string (wordsplit_t *ws, int *wsflags, int q, const char *str) +{ + if (*str == ':') + { + while (*++str != ':') + { + int f; + switch (*str) + { + case '+': + f = WRDSO_BSKEEP; + break; + + case '0': + f = WRDSO_OESC; + break; + + case 'x': + f = WRDSO_XESC; + break; + + default: + fprintf (stderr, "%s: invalid escape flag near %s\n", + progname, str); + abort (); + } + WRDSO_ESC_SET (ws, q, f); + } + *wsflags |= WRDSF_OPTIONS; + ++str; + } + ws->ws_escape[q] = str; +} + +static void +setfn_escape (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + wsc->wsflags |= flag; + set_escape_string (&wsc->ws, &wsc->wsflags, 0, arg); + set_escape_string (&wsc->ws, &wsc->wsflags, 1, arg); +} + +static void +setfn_escape_qw (char *arg, int quote, struct wsclosure *wsc) +{ + if (!(wsc->wsflags & WRDSF_ESCAPE)) + { + wsc->wsflags |= WRDSF_ESCAPE; + wsc->ws.ws_escape[!quote] = NULL; + } + set_escape_string (&wsc->ws, &wsc->wsflags, quote, arg); +} + +static void +setfn_escape_word (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + setfn_escape_qw (arg, 0, wsc); +} + +static void +setfn_escape_quote (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + setfn_escape_qw (arg, 1, wsc); +} + +static void +setfn_maxwords (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + char *p; + + wsc->wsflags |= WRDSF_OPTIONS; + wsc->ws.ws_options |= WRDSO_MAXWORDS; + + wsc->ws.ws_maxwords = strtoul (arg, &p, 10); + if (*p) + { + fprintf (stderr, "%s: invalid number: %s\n", progname, arg); + exit (1); + } +} + +static void +setfn_global (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + if (neg) + wsc->options &= ~flag; + else + wsc->options |= flag; +} + +static void +setfn_env (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + if (strcmp (arg, "none") == 0) + wsc->env_type = env_none; + else if (strcmp (arg, "null") == 0) + wsc->env_type = env_null; + else if (strcmp (arg, "sys") == 0) + wsc->env_type = env_sys; + else + { + fprintf (stderr, "%s: environment flag: %s\n", progname, arg); + exit (1); + } +} + +static void +setfn_dooffs (int flag, int neg, char *arg, struct wsclosure *wsc) +{ + if (!(wsc->wsflags & flag)) + { + wsc->wsflags |= flag; + wsc->offarg = wsoptind - 1; |