=================================================================== RCS file: /cvs/mandoc/Attic/argv.c,v retrieving revision 1.30 retrieving revision 1.38 diff -u -p -r1.30 -r1.38 --- mandoc/Attic/argv.c 2009/02/24 11:43:13 1.30 +++ mandoc/Attic/argv.c 2009/03/06 14:13:47 1.38 @@ -1,4 +1,4 @@ -/* $Id: argv.c,v 1.30 2009/02/24 11:43:13 kristaps Exp $ */ +/* $Id: argv.c,v 1.38 2009/03/06 14:13:47 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -34,14 +34,22 @@ #define ARGS_QUOTED (1 << 0) #define ARGS_DELIM (1 << 1) #define ARGS_TABSEP (1 << 2) +#define ARGS_ARGVLIKE (1 << 3) +#define ARGV_NONE (1 << 0) +#define ARGV_SINGLE (1 << 1) +#define ARGV_MULTI (1 << 2) +#define ARGV_OPT_SINGLE (1 << 3) + static int argv_a2arg(int, const char *); static int args(struct mdoc *, int, int *, char *, int, char **); -static int argv(struct mdoc *, int, +static int argv(struct mdoc *, int, int, struct mdoc_arg *, int *, char *); static int argv_single(struct mdoc *, int, struct mdoc_arg *, int *, char *); +static int argv_opt_single(struct mdoc *, int, + struct mdoc_arg *, int *, char *); static int argv_multi(struct mdoc *, int, struct mdoc_arg *, int *, char *); static int pwarn(struct mdoc *, int, int, int); @@ -60,6 +68,36 @@ static int perr(struct mdoc *, int, int, int); #define EARGVAL (1) #define EARGMANY (2) +/* Per-argument flags. */ + +static int mdoc_argvflags[MDOC_ARG_MAX] = { + ARGV_NONE, /* MDOC_Split */ + ARGV_NONE, /* MDOC_Nosplit */ + ARGV_NONE, /* MDOC_Ragged */ + ARGV_NONE, /* MDOC_Unfilled */ + ARGV_NONE, /* MDOC_Literal */ + ARGV_NONE, /* MDOC_File */ + ARGV_SINGLE, /* MDOC_Offset */ + ARGV_NONE, /* MDOC_Bullet */ + ARGV_NONE, /* MDOC_Dash */ + ARGV_NONE, /* MDOC_Hyphen */ + ARGV_NONE, /* MDOC_Item */ + ARGV_NONE, /* MDOC_Enum */ + ARGV_NONE, /* MDOC_Tag */ + ARGV_NONE, /* MDOC_Diag */ + ARGV_NONE, /* MDOC_Hang */ + ARGV_NONE, /* MDOC_Ohang */ + ARGV_NONE, /* MDOC_Inset */ + ARGV_MULTI, /* MDOC_Column */ + ARGV_SINGLE, /* MDOC_Width */ + ARGV_NONE, /* MDOC_Compact */ + ARGV_SINGLE, /* MDOC_Std */ + ARGV_NONE, /* MDOC_Filled */ + ARGV_NONE, /* MDOC_Words */ + ARGV_NONE, /* MDOC_Emphasis */ + ARGV_NONE /* MDOC_Symbolic */ +}; + static int mdoc_argflags[MDOC_MAX] = { 0, /* \" */ 0, /* Dd */ @@ -74,7 +112,7 @@ static int mdoc_argflags[MDOC_MAX] = { 0, /* Ed */ 0, /* Bl */ 0, /* El */ - ARGS_DELIM, /* It */ + 0, /* It */ ARGS_DELIM, /* Ad */ ARGS_DELIM, /* An */ ARGS_DELIM, /* Ar */ @@ -98,7 +136,7 @@ static int mdoc_argflags[MDOC_MAX] = { 0, /* Ot */ ARGS_DELIM, /* Pa */ 0, /* Rv */ - ARGS_DELIM, /* St */ + ARGS_DELIM | ARGS_ARGVLIKE, /* St */ ARGS_DELIM, /* Va */ ARGS_DELIM, /* Vt */ ARGS_DELIM, /* Xr */ @@ -167,6 +205,7 @@ static int mdoc_argflags[MDOC_MAX] = { 0, /* Hf */ 0, /* Fr */ 0, /* Ud */ + 0, /* Lb */ }; @@ -236,29 +275,49 @@ mdoc_args(struct mdoc *mdoc, int line, fl = (0 == tok) ? 0 : mdoc_argflags[tok]; /* - * First see if we should use TABSEP (Bl -column). This - * invalidates the use of ARGS_DELIM. + * Override per-macro argument flags with context-specific ones. + * As of now, this is only valid for `It' depending on its list + * context. */ - if (MDOC_It == tok) { + switch (tok) { + case (MDOC_It): for (n = mdoc->last; n; n = n->parent) - if (MDOC_BLOCK == n->type) - if (MDOC_Bl == n->tok) - break; + if (MDOC_BLOCK == n->type && MDOC_Bl == n->tok) + break; + assert(n); c = (int)n->data.block.argc; assert(c > 0); + /* + * Using `Bl -column' adds ARGS_TABSEP to the arguments + * and invalidates ARGS_DELIM. Using `Bl -diag' allows + * for quoted arguments. + */ + /* LINTED */ for (i = 0; i < c; i++) { - if (MDOC_Column != n->data.block.argv[i].arg) - continue; - fl |= ARGS_TABSEP; - fl &= ~ARGS_DELIM; - break; + switch (n->data.block.argv[i].arg) { + case (MDOC_Column): + fl |= ARGS_TABSEP; + fl &= ~ARGS_DELIM; + i = c; + break; + case (MDOC_Diag): + fl |= ARGS_QUOTED; + i = c; + break; + default: + break; + } } + default: + break; } + /* Continue parsing the arguments themselves... */ + return(args(mdoc, line, pos, buf, fl, v)); } @@ -267,7 +326,7 @@ static int args(struct mdoc *mdoc, int line, int *pos, char *buf, int fl, char **v) { - int i, c; + int i; char *p, *pp; assert(*pos > 0); @@ -279,7 +338,7 @@ args(struct mdoc *mdoc, int line, if ( ! pwarn(mdoc, line, *pos, WQUOTPARM)) return(ARGS_ERROR); - if ('-' == buf[*pos]) + if ( ! (fl & ARGS_ARGVLIKE) && '-' == buf[*pos]) if ( ! pwarn(mdoc, line, *pos, WARGVPARM)) return(ARGS_ERROR); @@ -290,14 +349,15 @@ args(struct mdoc *mdoc, int line, */ if ((fl & ARGS_DELIM) && mdoc_iscdelim(buf[*pos])) { - for (i = *pos; (c = buf[i]); ) { - if ( ! mdoc_iscdelim(c)) + for (i = *pos; buf[i]; ) { + if ( ! mdoc_iscdelim(buf[i])) break; i++; - if (0 == buf[i] || ! isspace(c)) + /* There must be at least one space... */ + if (0 == buf[i] || ! isspace((u_char)buf[i])) break; i++; - while (buf[i] && isspace(c)) + while (buf[i] && isspace((u_char)buf[i])) i++; } if (0 == buf[i]) { @@ -375,7 +435,7 @@ args(struct mdoc *mdoc, int line, p++; if (0 != *p) *(p - 1) = 0; - *pos += p - *v; + *pos += (int)(p - *v); } if (p && 0 == *p) @@ -396,7 +456,7 @@ args(struct mdoc *mdoc, int line, if (p > *v && ' ' == *(p - 1)) if ( ! pwarn(mdoc, line, *pos, WTAILWS)) return(0); - *pos += p - *v; + *pos += (int)(p - *v); return(ARGS_WORD); } @@ -404,8 +464,8 @@ args(struct mdoc *mdoc, int line, /* Do non-tabsep look-ahead here. */ if ( ! (ARGS_TABSEP & fl)) - while ((c = buf[*pos])) { - if (isspace(c)) + while (buf[*pos]) { + if (isspace((u_char)buf[*pos])) if ('\\' != buf[*pos - 1]) break; (*pos)++; @@ -420,7 +480,7 @@ args(struct mdoc *mdoc, int line, return(ARGS_WORD); if ( ! (ARGS_TABSEP & fl)) - while (buf[*pos] && isspace((int)buf[*pos])) + while (buf[*pos] && isspace((u_char)buf[*pos])) (*pos)++; if (buf[*pos]) @@ -452,7 +512,7 @@ args(struct mdoc *mdoc, int line, if (0 == buf[*pos]) return(ARGS_QWORD); - while (buf[*pos] && isspace((int)buf[*pos])) + while (buf[*pos] && isspace((u_char)buf[*pos])) (*pos)++; if (buf[*pos]) @@ -469,6 +529,14 @@ static int argv_a2arg(int tok, const char *argv) { + /* + * Parse an argument identifier from its text. XXX - this + * should really be table-driven to clarify the code. + * + * If you add an argument to the list, make sure that you + * register it here with its one or more macros! + */ + switch (tok) { case (MDOC_An): if (xstrcmp(argv, "split")) @@ -543,84 +611,6 @@ argv_a2arg(int tok, const char *argv) if (xstrcmp(argv, "std")) return(MDOC_Std); break; - - case (MDOC_St): - if (xstrcmp(argv, "p1003.1-88")) - return(MDOC_p1003_1_88); - else if (xstrcmp(argv, "p1003.1-90")) - return(MDOC_p1003_1_90); - else if (xstrcmp(argv, "p1003.1-96")) - return(MDOC_p1003_1_96); - else if (xstrcmp(argv, "p1003.1-2001")) - return(MDOC_p1003_1_2001); - else if (xstrcmp(argv, "p1003.1-2004")) - return(MDOC_p1003_1_2004); - else if (xstrcmp(argv, "p1003.1")) - return(MDOC_p1003_1); - else if (xstrcmp(argv, "p1003.1b")) - return(MDOC_p1003_1b); - else if (xstrcmp(argv, "p1003.1b-93")) - return(MDOC_p1003_1b_93); - else if (xstrcmp(argv, "p1003.1c-95")) - return(MDOC_p1003_1c_95); - else if (xstrcmp(argv, "p1003.1g-2000")) - return(MDOC_p1003_1g_2000); - else if (xstrcmp(argv, "p1003.2-92")) - return(MDOC_p1003_2_92); - else if (xstrcmp(argv, "p1003.2-95")) - return(MDOC_p1387_2_95); - else if (xstrcmp(argv, "p1003.2")) - return(MDOC_p1003_2); - else if (xstrcmp(argv, "p1387.2-95")) - return(MDOC_p1387_2); - else if (xstrcmp(argv, "isoC-90")) - return(MDOC_isoC_90); - else if (xstrcmp(argv, "isoC-amd1")) - return(MDOC_isoC_amd1); - else if (xstrcmp(argv, "isoC-tcor1")) - return(MDOC_isoC_tcor1); - else if (xstrcmp(argv, "isoC-tcor2")) - return(MDOC_isoC_tcor2); - else if (xstrcmp(argv, "isoC-99")) - return(MDOC_isoC_99); - else if (xstrcmp(argv, "ansiC")) - return(MDOC_ansiC); - else if (xstrcmp(argv, "ansiC-89")) - return(MDOC_ansiC_89); - else if (xstrcmp(argv, "ansiC-99")) - return(MDOC_ansiC_99); - else if (xstrcmp(argv, "ieee754")) - return(MDOC_ieee754); - else if (xstrcmp(argv, "iso8802-3")) - return(MDOC_iso8802_3); - else if (xstrcmp(argv, "xpg3")) - return(MDOC_xpg3); - else if (xstrcmp(argv, "xpg4")) - return(MDOC_xpg4); - else if (xstrcmp(argv, "xpg4.2")) - return(MDOC_xpg4_2); - else if (xstrcmp(argv, "xpg4.3")) - return(MDOC_xpg4_3); - else if (xstrcmp(argv, "xbd5")) - return(MDOC_xbd5); - else if (xstrcmp(argv, "xcu5")) - return(MDOC_xcu5); - else if (xstrcmp(argv, "xsh5")) - return(MDOC_xsh5); - else if (xstrcmp(argv, "xns5")) - return(MDOC_xns5); - else if (xstrcmp(argv, "xns5.2d2.0")) - return(MDOC_xns5_2d2_0); - else if (xstrcmp(argv, "xcurses4.2")) - return(MDOC_xcurses4_2); - else if (xstrcmp(argv, "susv2")) - return(MDOC_susv2); - else if (xstrcmp(argv, "susv3")) - return(MDOC_susv3); - else if (xstrcmp(argv, "svid4")) - return(MDOC_svid4); - break; - default: break; } @@ -650,7 +640,7 @@ argv_multi(struct mdoc *mdoc, int line, return(0); } else if (ARGS_EOLN == c) break; - v->value[v->sz] = p; + v->value[(int)v->sz] = p; } if (0 < v->sz && v->sz < MDOC_LINEARG_MAX) @@ -665,6 +655,34 @@ argv_multi(struct mdoc *mdoc, int line, static int +argv_opt_single(struct mdoc *mdoc, int line, + struct mdoc_arg *v, int *pos, char *buf) +{ + int c, ppos; + char *p; + + ppos = *pos; + + if ('-' == buf[*pos]) + return(1); + + c = args(mdoc, line, pos, buf, ARGS_QUOTED, &p); + if (ARGS_ERROR == c) + return(0); + if (ARGS_EOLN == c) + return(1); + + v->sz = 1; + v->value = xcalloc(1, sizeof(char *)); + v->value[0] = p; + return(1); +} + + +/* + * Parse a single, mandatory value from the stream. + */ +static int argv_single(struct mdoc *mdoc, int line, struct mdoc_arg *v, int *pos, char *buf) { @@ -677,7 +695,7 @@ argv_single(struct mdoc *mdoc, int line, if (ARGS_ERROR == c) return(0); if (ARGS_EOLN == c) - return(perr(mdoc, line, ppos, EARGVAL)); + return(perr(mdoc, line, ppos, EARGVAL)); v->sz = 1; v->value = xcalloc(1, sizeof(char *)); @@ -686,24 +704,42 @@ argv_single(struct mdoc *mdoc, int line, } +/* + * Determine rules for parsing arguments. Arguments can either accept + * no parameters, an optional single parameter, one parameter, or + * multiple parameters. + */ static int -argv(struct mdoc *mdoc, int line, +argv(struct mdoc *mdoc, int tok, int line, struct mdoc_arg *v, int *pos, char *buf) { + int fl; v->sz = 0; v->value = NULL; + fl = mdoc_argvflags[v->arg]; - switch (v->arg) { - case(MDOC_Std): - /* FALLTHROUGH */ - case(MDOC_Width): - /* FALLTHROUGH */ - case(MDOC_Offset): + /* + * Override the default per-argument value. + */ + + switch (tok) { + case (MDOC_Ex): + fl = ARGV_OPT_SINGLE; + break; + default: + break; + } + + switch (fl) { + case (ARGV_SINGLE): return(argv_single(mdoc, line, v, pos, buf)); - case(MDOC_Column): + case (ARGV_MULTI): return(argv_multi(mdoc, line, v, pos, buf)); + case (ARGV_OPT_SINGLE): + return(argv_opt_single(mdoc, line, v, pos, buf)); default: + /* ARGV_NONE */ break; } @@ -711,6 +747,11 @@ argv(struct mdoc *mdoc, int line, } +/* + * Parse an argument from line text. This comes in the form of -key + * [value0...], which may either have a single mandatory value, at least + * one mandatory value, an optional single value, or no value. + */ int mdoc_argv(struct mdoc *mdoc, int line, int tok, struct mdoc_arg *v, int *pos, char *buf) @@ -723,7 +764,7 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok, if (0 == buf[*pos]) return(ARGV_EOLN); - assert( ! isspace((int)buf[*pos])); + assert( ! isspace((u_char)buf[*pos])); if ('-' != buf[*pos]) return(ARGV_WORD); @@ -738,7 +779,7 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok, /* LINTED */ while (buf[*pos]) { - if (isspace((int)buf[*pos])) + if (isspace((u_char)buf[*pos])) if ('\\' != buf[*pos - 1]) break; (*pos)++; @@ -747,18 +788,24 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok, if (buf[*pos]) buf[(*pos)++] = 0; + /* + * We now parse out the per-macro arguments. XXX - this can be + * made much cleaner using per-argument tables. See argv_a2arg + * for details. + */ + if (MDOC_ARG_MAX == (v->arg = argv_a2arg(tok, p))) { if ( ! pwarn(mdoc, line, i, WARGVPARM)) return(ARGV_ERROR); return(ARGV_WORD); } - while (buf[*pos] && isspace((int)buf[*pos])) + while (buf[*pos] && isspace((u_char)buf[*pos])) (*pos)++; /* FIXME: whitespace if no value. */ - if ( ! argv(mdoc, line, v, pos, buf)) + if ( ! argv(mdoc, tok, line, v, pos, buf)) return(ARGV_ERROR); return(ARGV_ARG);