=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.31 retrieving revision 1.48 diff -u -p -r1.31 -r1.48 --- mandoc/roff.c 2008/12/02 00:10:37 1.31 +++ mandoc/roff.c 2008/12/07 14:38:57 1.48 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.31 2008/12/02 00:10:37 kristaps Exp $ */ +/* $Id: roff.c,v 1.48 2008/12/07 14:38:57 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -17,6 +17,7 @@ * PERFORMANCE OF THIS SOFTWARE. */ #include +#include #include #include @@ -29,60 +30,21 @@ #include "libmdocml.h" #include "private.h" +#include "roff.h" +/* FIXME: First letters of quoted-text interpreted in rofffindtok. */ +/* FIXME: `No' not implemented. */ /* TODO: warn if Pp occurs before/after Sh etc. (see mdoc.samples). */ - /* TODO: warn about "X section only" macros. */ - /* TODO: warn about empty lists. */ - /* TODO: (warn) some sections need specific elements. */ - /* TODO: (warn) NAME section has particular order. */ - /* TODO: unify empty-content tags a la
. */ - /* TODO: macros with a set number of arguments? */ +/* TODO: validate Dt macro arguments. */ +/* FIXME: Bl -diag supposed to ignore callable children. */ +/* FIXME: Nm has newline when used in NAME section. */ -#define ROFF_MAXARG 32 - -enum roffd { - ROFF_ENTER = 0, - ROFF_EXIT -}; - -enum rofftype { - ROFF_COMMENT, - ROFF_TEXT, - ROFF_LAYOUT, - ROFF_SPECIAL -}; - -#define ROFFCALL_ARGS \ - int tok, struct rofftree *tree, \ - char *argv[], enum roffd type - -struct rofftree; - -struct rofftok { - int (*cb)(ROFFCALL_ARGS); /* Callback. */ - const int *args; /* Args (or NULL). */ - const int *parents; - const int *children; - int ctx; - enum rofftype type; /* Type of macro. */ - int flags; -#define ROFF_PARSED (1 << 0) /* "Parsed". */ -#define ROFF_CALLABLE (1 << 1) /* "Callable". */ -#define ROFF_SHALLOW (1 << 2) /* Nesting block. */ -#define ROFF_LSCOPE (1 << 3) -}; - -struct roffarg { - int flags; -#define ROFF_VALUE (1 << 0) /* Has a value. */ -}; - struct roffnode { int tok; /* Token id. */ struct roffnode *parent; /* Parent (or NULL). */ @@ -90,43 +52,29 @@ struct roffnode { struct rofftree { struct roffnode *last; /* Last parsed node. */ - char *cur; - + char *cur; /* Line start. */ struct tm tm; /* `Dd' results. */ + char name[64]; /* `Nm' results. */ char os[64]; /* `Os' results. */ char title[64]; /* `Dt' results. */ char section[64]; /* `Dt' results. */ char volume[64]; /* `Dt' results. */ - int state; #define ROFF_PRELUDE (1 << 1) /* In roff prelude. */ #define ROFF_PRELUDE_Os (1 << 2) /* `Os' is parsed. */ #define ROFF_PRELUDE_Dt (1 << 3) /* `Dt' is parsed. */ #define ROFF_PRELUDE_Dd (1 << 4) /* `Dd' is parsed. */ #define ROFF_BODY (1 << 5) /* In roff body. */ - - struct roffcb cb; - void *arg; + struct roffcb cb; /* Callbacks. */ + void *arg; /* Callbacks' arg. */ }; -static int roff_Dd(ROFFCALL_ARGS); -static int roff_Dt(ROFFCALL_ARGS); -static int roff_Os(ROFFCALL_ARGS); -static int roff_Ns(ROFFCALL_ARGS); - -static int roff_layout(ROFFCALL_ARGS); -static int roff_text(ROFFCALL_ARGS); -static int roff_noop(ROFFCALL_ARGS); -static int roff_depr(ROFFCALL_ARGS); - static struct roffnode *roffnode_new(int, struct rofftree *); static void roffnode_free(struct rofftree *); - static void roff_warn(const struct rofftree *, const char *, char *, ...); static void roff_err(const struct rofftree *, const char *, char *, ...); - static int roffpurgepunct(struct rofftree *, char **); static int roffscan(int, const int *); static int rofffindtok(const char *); @@ -139,236 +87,24 @@ static int roffnextopt(const struct rofftree *, int, char ***, char **); static int roffparseopts(struct rofftree *, int, char ***, int *, char **); +static int roffcall(struct rofftree *, int, char **); static int roffparse(struct rofftree *, char *); -static int textparse(const struct rofftree *, char *); +static int textparse(struct rofftree *, char *); +static int roffdata(struct rofftree *, int, char *); +static int roffspecial(struct rofftree *, int, + const char *, const int *, + const char **, size_t, char **); +static int roffsetname(struct rofftree *, char **); +#ifdef __linux__ +extern size_t strlcat(char *, const char *, size_t); +extern size_t strlcpy(char *, const char *, size_t); +extern int vsnprintf(char *, size_t, + const char *, va_list); +extern char *strptime(const char *, const char *, + struct tm *); +#endif -static const int roffarg_An[] = { ROFF_Split, ROFF_Nosplit, - ROFF_ARGMAX }; -static const int roffarg_Bd[] = { ROFF_Ragged, ROFF_Unfilled, - ROFF_Literal, ROFF_File, ROFF_Offset, ROFF_Filled, - ROFF_Compact, ROFF_ARGMAX }; -static const int roffarg_Bk[] = { ROFF_Words, ROFF_ARGMAX }; -static const int roffarg_Ex[] = { ROFF_Std, ROFF_ARGMAX }; -static const int roffarg_Rv[] = { ROFF_Std, ROFF_ARGMAX }; -static const int roffarg_Bl[] = { ROFF_Bullet, ROFF_Dash, - ROFF_Hyphen, ROFF_Item, ROFF_Enum, ROFF_Tag, ROFF_Diag, - ROFF_Hang, ROFF_Ohang, ROFF_Inset, ROFF_Column, ROFF_Offset, - ROFF_Width, ROFF_Compact, ROFF_ARGMAX }; -static const int roffarg_St[] = { - ROFF_p1003_1_88, ROFF_p1003_1_90, ROFF_p1003_1_96, - ROFF_p1003_1_2001, ROFF_p1003_1_2004, ROFF_p1003_1, - ROFF_p1003_1b, ROFF_p1003_1b_93, ROFF_p1003_1c_95, - ROFF_p1003_1g_2000, ROFF_p1003_2_92, ROFF_p1387_2_95, - ROFF_p1003_2, ROFF_p1387_2, ROFF_isoC_90, ROFF_isoC_amd1, - ROFF_isoC_tcor1, ROFF_isoC_tcor2, ROFF_isoC_99, ROFF_ansiC, - ROFF_ansiC_89, ROFF_ansiC_99, ROFF_ieee754, ROFF_iso8802_3, - ROFF_xpg3, ROFF_xpg4, ROFF_xpg4_2, ROFF_xpg4_3, ROFF_xbd5, - ROFF_xcu5, ROFF_xsh5, ROFF_xns5, ROFF_xns5_2d2_0, - ROFF_xcurses4_2, ROFF_susv2, ROFF_susv3, ROFF_svid4, - ROFF_ARGMAX }; - -static const int roffchild_Bl[] = { ROFF_It, ROFF_El, ROFF_MAX }; -static const int roffchild_Fo[] = { ROFF_Fa, ROFF_Fc, ROFF_MAX }; -static const int roffchild_Rs[] = { ROFF_Re, ROFF__A, ROFF__B, - ROFF__D, ROFF__I, ROFF__J, ROFF__N, ROFF__O, ROFF__P, - ROFF__R, ROFF__T, ROFF__V, ROFF_MAX }; - -static const int roffparent_El[] = { ROFF_Bl, ROFF_It, ROFF_MAX }; -static const int roffparent_Fc[] = { ROFF_Fo, ROFF_Fa, ROFF_MAX }; -static const int roffparent_Oc[] = { ROFF_Oo, ROFF_MAX }; -static const int roffparent_It[] = { ROFF_Bl, ROFF_It, ROFF_MAX }; -static const int roffparent_Re[] = { ROFF_Rs, ROFF_MAX }; - -/* Table of all known tokens. */ -static const struct rofftok tokens[ROFF_MAX] = { - { roff_noop, NULL, NULL, NULL, 0, ROFF_COMMENT, 0 }, /* \" */ - { roff_Dd, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Dd */ - { roff_Dt, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Dt */ - { roff_Os, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Os */ - { roff_layout, NULL, NULL, NULL, ROFF_Sh, ROFF_LAYOUT, 0 }, /* Sh */ - { roff_layout, NULL, NULL, NULL, ROFF_Ss, ROFF_LAYOUT, 0 }, /* Ss */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Pp */ /* XXX 0 args */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_LSCOPE }, /* D1 */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_LSCOPE }, /* Dl */ - { roff_layout, roffarg_Bd, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Bd */ - { roff_noop, NULL, NULL, NULL, ROFF_Bd, ROFF_LAYOUT, 0 }, /* Ed */ - { roff_layout, roffarg_Bl, NULL, roffchild_Bl, 0, ROFF_LAYOUT, 0 }, /* Bl */ - { roff_noop, NULL, roffparent_El, NULL, ROFF_Bl, ROFF_LAYOUT, 0 }, /* El */ - { roff_layout, NULL, roffparent_It, NULL, ROFF_It, ROFF_LAYOUT, ROFF_PARSED | ROFF_SHALLOW }, /* It */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ad */ /* FIXME */ - { roff_text, roffarg_An, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* An */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ar */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Cd */ /* XXX man.4 only */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Cm */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dv */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Er */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ev */ /* XXX needs arg */ - { roff_text, roffarg_Ex, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ex */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fa */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Fd */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fl */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fn */ /* XXX needs arg */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ft */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ic */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* In */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Li */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Nd */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Nm */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_LSCOPE }, /* Op */ - { roff_depr, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ot */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pa */ - { roff_text, roffarg_Rv, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Rv */ - { roff_text, roffarg_St, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* St */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Va */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Vt */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xr */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* %A */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %B */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %D */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %I */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %J */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %N */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %O */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %P */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %R */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* %T */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %V */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ac */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ao */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_LSCOPE }, /* Aq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* At */ /* XXX at most 2 args */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bc */ - { roff_layout, NULL, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Bf */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bo */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_LSCOPE }, /* Bq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bsx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bx */ - { NULL, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Db */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dc */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Do */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_LSCOPE }, /* Dq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ec */ - { roff_noop, NULL, NULL, NULL, ROFF_Bf, ROFF_LAYOUT, 0 }, /* Ef */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Em */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Eo */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Fx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ms */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* No */ - { roff_Ns, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ns */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Nx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ox */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pc */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Pf */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Po */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_LSCOPE }, /* Pq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qc */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ql */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qo */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_LSCOPE }, /* Qq */ - { roff_noop, NULL, roffparent_Re, NULL, ROFF_Rs, ROFF_LAYOUT, 0 }, /* Re */ - { roff_layout, NULL, NULL, roffchild_Rs, 0, ROFF_LAYOUT, 0 }, /* Rs */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sc */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* So */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_LSCOPE }, /* Sq */ - { NULL, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Sm */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sy */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Tn */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ux */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xc */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xo */ - { roff_layout, NULL, NULL, roffchild_Fo, 0, ROFF_LAYOUT, 0 }, /* Fo */ - { roff_noop, NULL, roffparent_Fc, NULL, ROFF_Fo, ROFF_LAYOUT, 0 }, /* Fc */ - { roff_layout, NULL, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Oo */ - { roff_noop, NULL, roffparent_Oc, NULL, ROFF_Oo, ROFF_LAYOUT, 0 }, /* Oc */ - { roff_layout, roffarg_Bk, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Bk */ - { roff_noop, NULL, NULL, NULL, ROFF_Bk, ROFF_LAYOUT, 0 }, /* Ek */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Bt */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Hf */ - { roff_depr, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Fr */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ud */ - }; - -/* Table of all known token arguments. */ -static const int tokenargs[ROFF_ARGMAX] = { - 0, 0, 0, 0, - 0, ROFF_VALUE, ROFF_VALUE, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, ROFF_VALUE, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - }; - -const char *const toknamesp[ROFF_MAX] = { - "\\\"", "Dd", "Dt", "Os", - "Sh", "Ss", "Pp", "D1", - "Dl", "Bd", "Ed", "Bl", - "El", "It", "Ad", "An", - "Ar", "Cd", "Cm", "Dv", - "Er", "Ev", "Ex", "Fa", - "Fd", "Fl", "Fn", "Ft", - "Ic", "In", "Li", "Nd", - "Nm", "Op", "Ot", "Pa", - "Rv", "St", "Va", "Vt", - /* LINTED */ - "Xr", "\%A", "\%B", "\%D", - /* LINTED */ - "\%I", "\%J", "\%N", "\%O", - /* LINTED */ - "\%P", "\%R", "\%T", "\%V", - "Ac", "Ao", "Aq", "At", - "Bc", "Bf", "Bo", "Bq", - "Bsx", "Bx", "Db", "Dc", - "Do", "Dq", "Ec", "Ef", - "Em", "Eo", "Fx", "Ms", - "No", "Ns", "Nx", "Ox", - "Pc", "Pf", "Po", "Pq", - "Qc", "Ql", "Qo", "Qq", - "Re", "Rs", "Sc", "So", - "Sq", "Sm", "Sx", "Sy", - "Tn", "Ux", "Xc", "Xo", - "Fo", "Fc", "Oo", "Oc", - "Bk", "Ek", "Bt", "Hf", - "Fr", "Ud", - }; - -const char *const tokargnamesp[ROFF_ARGMAX] = { - "split", "nosplit", "ragged", - "unfilled", "literal", "file", - "offset", "bullet", "dash", - "hyphen", "item", "enum", - "tag", "diag", "hang", - "ohang", "inset", "column", - "width", "compact", "std", - "p1003.1-88", "p1003.1-90", "p1003.1-96", - "p1003.1-2001", "p1003.1-2004", "p1003.1", - "p1003.1b", "p1003.1b-93", "p1003.1c-95", - "p1003.1g-2000", "p1003.2-92", "p1387.2-95", - "p1003.2", "p1387.2", "isoC-90", - "isoC-amd1", "isoC-tcor1", "isoC-tcor2", - "isoC-99", "ansiC", "ansiC-89", - "ansiC-99", "ieee754", "iso8802-3", - "xpg3", "xpg4", "xpg4.2", - "xpg4.3", "xbd5", "xcu5", - "xsh5", "xns5", "xns5.2d2.0", - "xcurses4.2", "susv2", "susv3", - "svid4", "filled", "words", - }; - -const char *const *toknames = toknamesp; -const char *const *tokargnames = tokargnamesp; - - int roff_free(struct rofftree *tree, int flush) { @@ -455,14 +191,43 @@ roff_engine(struct rofftree *tree, char *buf) static int -textparse(const struct rofftree *tree, char *buf) +textparse(struct rofftree *tree, char *buf) { + char *bufp; + /* TODO: literal parsing. */ + if ( ! (ROFF_BODY & tree->state)) { roff_err(tree, buf, "data not in body"); return(0); } - return((*tree->cb.roffdata)(tree->arg, 1, buf)); + + /* LINTED */ + while (*buf) { + while (*buf && isspace(*buf)) + buf++; + + if (0 == *buf) + break; + + bufp = buf++; + + while (*buf && ! isspace(*buf)) + buf++; + + if (0 != *buf) { + *buf++ = 0; + if ( ! roffdata(tree, 1, bufp)) + return(0); + continue; + } + + if ( ! roffdata(tree, 1, bufp)) + return(0); + break; + } + + return(1); } @@ -478,8 +243,15 @@ roffargs(const struct rofftree *tree, p = buf; + /* + * This is an ugly little loop. It parses a line into + * space-delimited tokens. If a quote mark is encountered, a + * token is alloted the entire quoted text. If whitespace is + * escaped, it's included in the prior alloted token. + */ + /* LINTED */ - for (i = 0; *buf && i < ROFF_MAXARG; i++) { + for (i = 0; *buf && i < ROFF_MAXLINEARG; i++) { if ('\"' == *buf) { argv[i] = ++buf; while (*buf && '\"' != *buf) @@ -493,8 +265,17 @@ roffargs(const struct rofftree *tree, } } else { argv[i] = buf++; - while (*buf && ! isspace(*buf)) - buf++; + while (*buf) { + if ( ! isspace(*buf)) { + buf++; + continue; + } + if (*(buf - 1) == '\\') { + buf++; + continue; + } + break; + } if (0 == *buf) continue; } @@ -502,9 +283,9 @@ roffargs(const struct rofftree *tree, while (*buf && isspace(*buf)) buf++; } - + assert(i > 0); - if (ROFF_MAXARG == i && *buf) { + if (ROFF_MAXLINEARG == i && *buf) { roff_err(tree, p, "too many arguments for `%s'", toknames [tok]); return(0); @@ -535,7 +316,7 @@ roffparse(struct rofftree *tree, char *buf) { int tok, t; struct roffnode *n; - char *argv[ROFF_MAXARG]; + char *argv[ROFF_MAXLINEARG]; char **argvp; if (0 != *buf && 0 != *(buf + 1) && 0 != *(buf + 2)) @@ -665,6 +446,7 @@ roffparse(struct rofftree *tree, char *buf) assert(tok != tokens[tok].ctx && 0 != tokens[tok].ctx); + /* LINTED */ for (n = tree->last; n; n = n->parent) if (n->tok != tokens[tok].ctx) { if (n->tok == tokens[n->tok].ctx) @@ -837,6 +619,100 @@ roffnode_free(struct rofftree *tree) static int +roffspecial(struct rofftree *tree, int tok, const char *start, + const int *argc, const char **argv, + size_t sz, char **ordp) +{ + + switch (tok) { + case (ROFF_At): + if (0 == sz) + break; + if (0 == strcmp(*ordp, "v6")) + break; + else if (0 == strcmp(*ordp, "v7")) + break; + else if (0 == strcmp(*ordp, "32v")) + break; + else if (0 == strcmp(*ordp, "V.1")) + break; + else if (0 == strcmp(*ordp, "V.4")) + break; + roff_err(tree, start, "invalid `At' arg"); + return(0); + + case (ROFF_Nm): + if (0 == sz) { + if (0 == tree->name[0]) { + roff_err(tree, start, "`Nm' not set"); + return(0); + } + ordp[0] = tree->name; + ordp[1] = NULL; + } else if ( ! roffsetname(tree, ordp)) + return(0); + break; + + case (ROFF_Rv): + /* FALLTHROUGH*/ + case (ROFF_Sx): + /* FALLTHROUGH*/ + case (ROFF_Ex): + if (1 == sz) + break; + roff_err(tree, start, "`%s' expects one arg", + toknames[tok]); + return(0); + + case (ROFF_Sm): + if (1 != sz) { + roff_err(tree, start, "`Sm' expects one arg"); + return(0); + } + + if (0 != strcmp(ordp[0], "on") && + 0 != strcmp(ordp[0], "off")) { + roff_err(tree, start, "`Sm' has invalid argument"); + return(0); + } + break; + + case (ROFF_Ud): + /* FALLTHROUGH */ + case (ROFF_Ux): + /* FALLTHROUGH */ + case (ROFF_Bt): + if (0 != sz) { + roff_err(tree, start, "`%s' expects no args", + toknames[tok]); + return(0); + } + break; + default: + break; + } + + return((*tree->cb.roffspecial)(tree->arg, tok, + tree->cur, argc, argv, ordp)); +} + + +static int +roffcall(struct rofftree *tree, int tok, char **argv) +{ + + if (NULL == tokens[tok].cb) { + roff_err(tree, *argv, "unsupported macro `%s'", + toknames[tok]); + return(0); + } + if ( ! (*tokens[tok].cb)(tok, tree, argv, ROFF_ENTER)) + return(0); + return(1); +} + + +static int roffnextopt(const struct rofftree *tree, int tok, char ***in, char **val) { @@ -854,7 +730,7 @@ roffnextopt(const struct rofftree *tree, int tok, if (ROFF_ARGMAX == (v = rofffindarg(arg + 1))) { roff_warn(tree, arg, "argument-like parameter `%s' to " - "`%s'", &arg[1], toknames[tok]); + "`%s'", arg, toknames[tok]); return(-1); } @@ -896,7 +772,7 @@ roffpurgepunct(struct rofftree *tree, char **argv) /* LINTED */ while (argv[i]) - if ( ! (*tree->cb.roffdata)(tree->arg, 0, argv[i++])) + if ( ! roffdata(tree, 0, argv[i++])) return(0); return(1); } @@ -927,6 +803,17 @@ roffparseopts(struct rofftree *tree, int tok, } +static int +roffdata(struct rofftree *tree, int space, char *buf) +{ + + if (0 == *buf) + return(1); + return((*tree->cb.roffdata)(tree->arg, + space != 0, tree->cur, buf)); +} + + /* ARGSUSED */ static int roff_Dd(ROFFCALL_ARGS) @@ -955,7 +842,7 @@ roff_Dd(ROFFCALL_ARGS) argv++; - if (0 == strcmp(*argv, "$Mdocdate: December 2 2008 $")) { + if (0 == strcmp(*argv, "$Mdocdate: December 7 2008 $")) { t = time(NULL); if (NULL == localtime_r(&t, &tree->tm)) err(1, "localtime_r"); @@ -1070,75 +957,71 @@ roff_Dt(ROFFCALL_ARGS) } -/* ARGSUSED */ -static int -roff_Ns(ROFFCALL_ARGS) +static int +roffsetname(struct rofftree *tree, char **ordp) { - int j, c, first; + + assert(*ordp); - first = (*argv == tree->cur); + /* FIXME: not all sections can set this. */ - argv++; - - if (ROFF_MAX != (c = rofffindcallable(*argv))) { - if (NULL == tokens[c].cb) { - roff_err(tree, *argv, "unsupported macro `%s'", - toknames[c]); - return(0); - } - if ( ! (*tree->cb.roffspecial)(tree->arg, tok)) - return(0); - if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) - return(0); - if ( ! first) - return(1); - return(roffpurgepunct(tree, argv)); + if (NULL != *(ordp + 1)) { + roff_err(tree, *ordp, "too many `Nm' args"); + return(0); } - - if ( ! (*tree->cb.roffdata)(tree->arg, 0, *argv++)) + + if (strlcpy(tree->name, *ordp, sizeof(tree->name)) + >= sizeof(tree->name)) { + roff_err(tree, *ordp, "`Nm' arg too long"); return(0); + } - while (*argv) { - if (ROFF_MAX == (c = rofffindcallable(*argv))) { - if ( ! roffispunct(*argv)) { - if ( ! (*tree->cb.roffdata) - (tree->arg, 1, *argv++)) - return(0); - continue; - } - - /* FIXME: this is identical to that of - * roff_text. */ + return(1); +} - /* See if only punctuation remains. */ - for (j = 0; argv[j]; j++) - if ( ! roffispunct(argv[j])) - break; +/* ARGSUSED */ +static int +roff_Ns(ROFFCALL_ARGS) +{ + int j, c, first; + char *morep[1]; - if (argv[j]) { - if ( ! (*tree->cb.roffdata) - (tree->arg, 0, *argv++)) - return(0); - continue; - } + first = (*argv++ == tree->cur); + morep[0] = NULL; - /* Only punctuation remains. */ + if ( ! roffspecial(tree, tok, *argv, NULL, NULL, 0, morep)) + return(0); + while (*argv) { + if (ROFF_MAX != (c = rofffindcallable(*argv))) { + if ( ! roffcall(tree, c, argv)) + return(0); break; } - if (NULL == tokens[c].cb) { - roff_err(tree, *argv, "unsupported macro `%s'", - toknames[c]); - return(0); + + if ( ! roffispunct(*argv)) { + if ( ! roffdata(tree, 1, *argv++)) + return(0); + continue; } - if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) - return(0); + + for (j = 0; argv[j]; j++) + if ( ! roffispunct(argv[j])) + break; + + if (argv[j]) { + if ( ! roffdata(tree, 0, *argv++)) + return(0); + continue; + } + break; } if ( ! first) return(1); + return(roffpurgepunct(tree, argv)); } @@ -1187,7 +1070,9 @@ roff_Os(ROFFCALL_ARGS) assert(NULL == tree->last); - return((*tree->cb.roffhead)(tree->arg)); + return((*tree->cb.roffhead)(tree->arg, &tree->tm, + tree->os, tree->title, tree->section, + tree->volume)); } @@ -1195,8 +1080,8 @@ roff_Os(ROFFCALL_ARGS) static int roff_layout(ROFFCALL_ARGS) { - int i, c, argcp[ROFF_MAXARG]; - char *argvp[ROFF_MAXARG]; + int i, c, argcp[ROFF_MAXLINEARG]; + char *argvp[ROFF_MAXLINEARG]; if (ROFF_PRELUDE & tree->state) { roff_err(tree, *argv, "bad `%s' in prelude", @@ -1204,6 +1089,8 @@ roff_layout(ROFFCALL_ARGS) return(0); } else if (ROFF_EXIT == type) { roffnode_free(tree); + if ( ! (*tree->cb.roffblkbodyout)(tree->arg, tok)) + return(0); return((*tree->cb.roffblkout)(tree->arg, tok)); } @@ -1225,8 +1112,10 @@ roff_layout(ROFFCALL_ARGS) if ( ! (*tree->cb.roffblkin)(tree->arg, tok, argcp, argvp)) return(0); if (NULL == *argv) - return(1); - if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); + + if ( ! (*tree->cb.roffblkheadin)(tree->arg, tok, argcp, argvp)) return(0); /* @@ -1236,12 +1125,14 @@ roff_layout(ROFFCALL_ARGS) if ( ! (ROFF_PARSED & tokens[tok].flags)) { i = 0; - while (*argv) { - if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++)) + while (*argv) + if ( ! roffdata(tree, i++, *argv++)) return(0); - i = 1; - } - return((*tree->cb.roffout)(tree->arg, tok)); + + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); } /* @@ -1253,22 +1144,12 @@ roff_layout(ROFFCALL_ARGS) while (*argv) { if (ROFF_MAX == (c = rofffindcallable(*argv))) { assert(tree->arg); - if ( ! (*tree->cb.roffdata) - (tree->arg, i, *argv++)) + if ( ! roffdata(tree, i++, *argv++)) return(0); - i = 1; continue; } - - if (NULL == tokens[c].cb) { - roff_err(tree, *argv, "unsupported macro `%s'", - toknames[c]); + if ( ! roffcall(tree, c, argv)) return(0); - } - - if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) - return(0); - break; } @@ -1278,8 +1159,12 @@ roff_layout(ROFFCALL_ARGS) * macro. */ - if (NULL == *argv) - return((*tree->cb.roffout)(tree->arg, tok)); + if (NULL == *argv) { + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); + } /* * Expensive. Scan to the end of line then work backwards until @@ -1289,16 +1174,82 @@ roff_layout(ROFFCALL_ARGS) if ( ! roffpurgepunct(tree, argv)) return(0); - return((*tree->cb.roffout)(tree->arg, tok)); + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); } /* ARGSUSED */ static int +roff_ordered(ROFFCALL_ARGS) +{ + int i, first, c, argcp[ROFF_MAXLINEARG]; + char *ordp[ROFF_MAXLINEARG], *p, + *argvp[ROFF_MAXLINEARG]; + + if (ROFF_PRELUDE & tree->state) { + roff_err(tree, *argv, "`%s' disallowed in prelude", + toknames[tok]); + return(0); + } + + first = (*argv == tree->cur); + p = *argv++; + + if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) + return(0); + + if (NULL == *argv) + return(roffspecial(tree, tok, p, argcp, + (const char **)argvp, 0, ordp)); + + i = 0; + while (*argv && i < ROFF_MAXLINEARG) { + c = ROFF_PARSED & tokens[tok].flags ? + rofffindcallable(*argv) : ROFF_MAX; + + if (ROFF_MAX == c && ! roffispunct(*argv)) { + ordp[i++] = *argv++; + continue; + } + ordp[i] = NULL; + + if (ROFF_MAX == c) + break; + + if ( ! roffspecial(tree, tok, p, argcp, + (const char **)argvp, + (size_t)i, ordp)) + return(0); + + return(roffcall(tree, c, argv)); + } + + assert(i != ROFF_MAXLINEARG); + ordp[i] = NULL; + + if ( ! roffspecial(tree, tok, p, argcp, + (const char**)argvp, + (size_t)i, ordp)) + return(0); + + /* FIXME: error if there's stuff after the punctuation. */ + + if ( ! first || NULL == *argv) + return(1); + + return(roffpurgepunct(tree, argv)); +} + + +/* ARGSUSED */ +static int roff_text(ROFFCALL_ARGS) { - int i, j, first, c, argcp[ROFF_MAXARG]; - char *argvp[ROFF_MAXARG]; + int i, j, first, c, argcp[ROFF_MAXLINEARG]; + char *argvp[ROFF_MAXLINEARG]; if (ROFF_PRELUDE & tree->state) { roff_err(tree, *argv, "`%s' disallowed in prelude", @@ -1318,11 +1269,10 @@ roff_text(ROFFCALL_ARGS) if ( ! (ROFF_PARSED & tokens[tok].flags)) { i = 0; - while (*argv) { - if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++)) + while (*argv) + if ( ! roffdata(tree, i++, *argv++)) return(0); - i = 1; - } + return((*tree->cb.roffout)(tree->arg, tok)); } @@ -1335,60 +1285,40 @@ roff_text(ROFFCALL_ARGS) i = 0; while (*argv) { - if (ROFF_MAX == (c = rofffindcallable(*argv))) { - if ( ! roffispunct(*argv)) { - if ( ! (*tree->cb.roffdata) - (tree->arg, i, *argv++)) + if (ROFF_MAX != (c = rofffindcallable(*argv))) { + if ( ! (ROFF_LSCOPE & tokens[tok].flags)) + if ( ! (*tree->cb.roffout)(tree->arg, tok)) return(0); - i = 1; - continue; - } - - /* See if only punctuation remains. */ - - i = 1; - for (j = 0; argv[j]; j++) - if ( ! roffispunct(argv[j])) - break; - - if (argv[j]) { - if ( ! (*tree->cb.roffdata) - (tree->arg, 0, *argv++)) + + if ( ! roffcall(tree, c, argv)) + return(0); + + if (ROFF_LSCOPE & tokens[tok].flags) + if ( ! (*tree->cb.roffout)(tree->arg, tok)) return(0); - continue; - } + + break; + } - /* Only punctuation remains. */ - - if ( ! (*tree->cb.roffout)(tree->arg, tok)) + if ( ! roffispunct(*argv)) { + if ( ! roffdata(tree, i++, *argv++)) return(0); - break; + continue; } - /* - * A sub-command has been found. Execute it and - * discontinue parsing for arguments. If we're - * line-scoped, then close out after it returns; if we - * aren't, then close out before. - */ + i = 1; + for (j = 0; argv[j]; j++) + if ( ! roffispunct(argv[j])) + break; - if (NULL == tokens[c].cb) { - roff_err(tree, *argv, "unsupported macro `%s'", - toknames[c]); - return(0); - } - - if ( ! (ROFF_LSCOPE & tokens[tok].flags)) - if ( ! (*tree->cb.roffout)(tree->arg, tok)) + if (argv[j]) { + if ( ! roffdata(tree, 0, *argv++)) return(0); - - if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) - return(0); + continue; + } - if (ROFF_LSCOPE & tokens[tok].flags) - if ( ! (*tree->cb.roffout)(tree->arg, tok)) - return(0); - + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); break; } @@ -1448,3 +1378,4 @@ roff_err(const struct rofftree *tree, const char *pos, (*tree->cb.roffmsg)(tree->arg, ROFF_ERROR, tree->cur, pos, buf); } +