=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.40 retrieving revision 1.61 diff -u -p -r1.40 -r1.61 --- mandoc/roff.c 2008/12/04 23:10:51 1.40 +++ mandoc/roff.c 2008/12/10 13:15:55 1.61 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.40 2008/12/04 23:10:51 kristaps Exp $ */ +/* $Id: roff.c,v 1.61 2008/12/10 13:15:55 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -28,96 +28,73 @@ #include #include -#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. */ -enum roffd { - ROFF_ENTER = 0, - ROFF_EXIT +struct roffnode { + int tok; /* Token id. */ + struct roffnode *parent; /* Parent (or NULL). */ }; -enum rofftype { - ROFF_COMMENT, - ROFF_TEXT, - ROFF_LAYOUT, - ROFF_SPECIAL +enum rofferr { + ERR_ARGEQ1, /* Macro requires arg == 1. */ + ERR_ARGEQ0, /* Macro requires arg == 0. */ + ERR_ARGGE1, /* Macro requires arg >= 1. */ + ERR_ARGGE2, /* Macro requires arg >= 2. */ + ERR_ARGLEN, /* Macro argument too long. */ + ERR_BADARG, /* Macro has bad arg. */ + ERR_ARGMNY, /* Too many macro arguments. */ + ERR_NOTSUP, /* Macro not supported. */ + ERR_DEPREC, /* Macro deprecated. */ + ERR_PR_OOO, /* Prelude macro bad order. */ + ERR_PR_REP, /* Prelude macro repeated. */ + ERR_NOT_PR, /* Not allowed in prelude. */ + WRN_SECORD /* Sections out-of-order. */ }; -#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; /* Limit to parents. */ - const int *children; /* Limit to kids. */ - int ctx; /* Blk-close node. */ - 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) /* Line scope. */ -}; - -struct roffnode { - int tok; /* Token id. */ - struct roffnode *parent; /* Parent (or NULL). */ -}; - struct rofftree { struct roffnode *last; /* Last parsed node. */ 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. */ + enum roffmsec section; + enum roffvol volume; struct roffcb cb; /* Callbacks. */ void *arg; /* Callbacks' arg. */ + int csec; /* Current section. */ + int asec; /* Thus-far sections. */ }; -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_Sm(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 int roff_ordered(ROFFCALL_ARGS); static struct roffnode *roffnode_new(int, struct rofftree *); static void roffnode_free(struct rofftree *); -static void roff_warn(const struct rofftree *, +static int roff_warn(const struct rofftree *, const char *, char *, ...); -static void roff_err(const struct rofftree *, +static int roff_warnp(const struct rofftree *, + const char *, int, enum rofferr); +static int roff_err(const struct rofftree *, const char *, char *, ...); +static int roff_errp(const struct rofftree *, + const char *, int, enum rofferr); static int roffpurgepunct(struct rofftree *, char **); static int roffscan(int, const int *); static int rofffindtok(const char *); static int rofffindarg(const char *); static int rofffindcallable(const char *); +static int roffispunct(const char *); +static int roffchecksec(struct rofftree *, + const char *, int, int); static int roffargs(const struct rofftree *, int, char *, char **); static int roffargok(int, int); @@ -126,10 +103,14 @@ static int roffnextopt(const struct rofftree *, static int roffparseopts(struct rofftree *, int, char ***, int *, char **); static int roffcall(struct rofftree *, int, char **); +static int roffexit(struct rofftree *, int); static int roffparse(struct rofftree *, char *); static int textparse(struct rofftree *, char *); static int roffdata(struct rofftree *, int, char *); -static int roffspecial(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); @@ -140,228 +121,6 @@ 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 }; - -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_ordered, NULL, NULL, NULL, 0, ROFF_TEXT, /*XXX*/ -1 }, /* Fn */ - { 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_ordered, NULL, NULL, NULL, 0, ROFF_TEXT, /*XXX*/ -1 }, /* Xr */ - { 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 */ - { roff_Sm, NULL, NULL, NULL, 0, ROFF_TEXT, 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 */ - }; - -#define ROFF_VALUE (1 << 0) - -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) { @@ -375,26 +134,30 @@ roff_free(struct rofftree *tree, int flush) error = 1; - if (ROFF_PRELUDE & tree->state) { - roff_err(tree, NULL, "prelude never finished"); + if ( ! (ROFFSec_NAME & tree->asec)) { + (void)roff_err(tree, NULL, "missing `NAME' section"); goto end; - } + } else if ( ! (ROFFSec_NMASK & tree->asec)) + (void)roff_warn(tree, NULL, "missing suggested `NAME', " + "`SYNOPSIS', `DESCRIPTION' sections"); for (n = tree->last; n; n = n->parent) { if (0 != tokens[n->tok].ctx) continue; - roff_err(tree, NULL, "closing explicit scope `%s'", - toknames[n->tok]); + (void)roff_err(tree, NULL, "closing explicit scope " + "`%s'", toknames[n->tok]); goto end; } while (tree->last) { t = tree->last->tok; - if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) + if ( ! roffexit(tree, t)) goto end; } - if ( ! (*tree->cb.rofftail)(tree->arg)) + if ( ! (*tree->cb.rofftail)(tree->arg, &tree->tm, + tree->os, tree->title, + tree->section, tree->volume)) goto end; error = 0; @@ -421,8 +184,8 @@ roff_alloc(const struct roffcb *cb, void *args) if (NULL == (tree = calloc(1, sizeof(struct rofftree)))) err(1, "calloc"); - tree->state = ROFF_PRELUDE; tree->arg = args; + tree->section = ROFF_MSEC_MAX; (void)memcpy(&tree->cb, cb, sizeof(struct roffcb)); @@ -437,10 +200,9 @@ roff_engine(struct rofftree *tree, char *buf) tree->cur = buf; assert(buf); - if (0 == *buf) { - roff_err(tree, buf, "blank line"); - return(0); - } else if ('.' != *buf) + if (0 == *buf) + return(roff_err(tree, buf, "blank line")); + else if ('.' != *buf) return(textparse(tree, buf)); return(roffparse(tree, buf)); @@ -454,10 +216,8 @@ textparse(struct rofftree *tree, char *buf) /* TODO: literal parsing. */ - if ( ! (ROFF_BODY & tree->state)) { - roff_err(tree, buf, "data not in body"); - return(0); - } + if ( ! (ROFFSec_NAME & tree->asec)) + return(roff_err(tree, buf, "data before `NAME' section")); /* LINTED */ while (*buf) { @@ -513,13 +273,9 @@ roffargs(const struct rofftree *tree, argv[i] = ++buf; while (*buf && '\"' != *buf) buf++; - if (0 == *buf) { - roff_err(tree, argv[i], "unclosed " - "quote in argument " - "list for `%s'", - toknames[tok]); - return(0); - } + if (0 == *buf) + return(roff_err(tree, argv[i], + "unclosed quote in arg list")); } else { argv[i] = buf++; while (*buf) { @@ -542,11 +298,8 @@ roffargs(const struct rofftree *tree, } assert(i > 0); - if (ROFF_MAXLINEARG == i && *buf) { - roff_err(tree, p, "too many arguments for `%s'", toknames - [tok]); - return(0); - } + if (ROFF_MAXLINEARG == i && *buf) + return(roff_err(tree, p, "too many args")); argv[i] = NULL; return(1); @@ -580,61 +333,47 @@ roffparse(struct rofftree *tree, char *buf) if (0 == strncmp(buf, ".\\\"", 3)) return(1); - if (ROFF_MAX == (tok = rofffindtok(buf + 1))) { - roff_err(tree, buf + 1, "bogus line macro"); + if (ROFF_MAX == (tok = rofffindtok(buf + 1))) + return(roff_err(tree, buf, "bogus line macro")); + else if ( ! roffargs(tree, tok, buf, argv)) return(0); - } else if (NULL == tokens[tok].cb) { - roff_err(tree, buf + 1, "unsupported macro `%s'", - toknames[tok]); - return(0); - } - assert(ROFF___ != tok); - if ( ! roffargs(tree, tok, buf, argv)) - return(0); - argvp = (char **)argv; /* * Prelude macros break some assumptions, so branch now. */ - if (ROFF_PRELUDE & tree->state) { + if ( ! (ROFFSec_PR_Dd & tree->asec)) { assert(NULL == tree->last); - return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + return(roffcall(tree, tok, argvp)); } - assert(ROFF_BODY & tree->state); - /* * First check that our possible parents and parent's possible * children are satisfied. */ if (tree->last && ! roffscan - (tree->last->tok, tokens[tok].parents)) { - roff_err(tree, *argvp, "`%s' has invalid parent `%s'", - toknames[tok], - toknames[tree->last->tok]); - return(0); - } + (tree->last->tok, tokens[tok].parents)) + return(roff_err(tree, *argvp, "`%s' has invalid " + "parent `%s'", toknames[tok], + toknames[tree->last->tok])); if (tree->last && ! roffscan - (tok, tokens[tree->last->tok].children)) { - roff_err(tree, *argvp, "`%s' is invalid child of `%s'", - toknames[tok], - toknames[tree->last->tok]); - return(0); - } + (tok, tokens[tree->last->tok].children)) + return(roff_err(tree, *argvp, "`%s' has invalid " + "child `%s'", toknames[tok], + toknames[tree->last->tok])); /* * Branch if we're not a layout token. */ if (ROFF_LAYOUT != tokens[tok].type) - return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + return(roffcall(tree, tok, argvp)); if (0 == tokens[tok].ctx) - return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + return(roffcall(tree, tok, argvp)); /* * First consider implicit-end tags, like as follows: @@ -665,9 +404,10 @@ roffparse(struct rofftree *tree, char *buf) } if (tokens[n->tok].ctx == n->tok) continue; - roff_err(tree, *argv, "`%s' breaks `%s' scope", - toknames[tok], toknames[n->tok]); - return(0); + return(roff_err(tree, *argv, "`%s' breaks " + "scope of prior`%s'", + toknames[tok], + toknames[n->tok])); } /* @@ -676,7 +416,7 @@ roffparse(struct rofftree *tree, char *buf) */ if (NULL == n) - return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + return(roffcall(tree, tok, argvp)); /* * Close out all intermediary scoped blocks, then hang @@ -685,11 +425,11 @@ roffparse(struct rofftree *tree, char *buf) do { t = tree->last->tok; - if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) + if ( ! roffexit(tree, t)) return(0); } while (t != tok); - return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + return(roffcall(tree, tok, argvp)); } /* @@ -708,24 +448,22 @@ roffparse(struct rofftree *tree, char *buf) if (n->tok != tokens[tok].ctx) { if (n->tok == tokens[n->tok].ctx) continue; - roff_err(tree, *argv, "`%s' breaks `%s' scope", - toknames[tok], toknames[n->tok]); - return(0); + return(roff_err(tree, *argv, "`%s' breaks " + "scope of prior `%s'", + toknames[tok], + toknames[n->tok])); } else break; + if (NULL == n) + return(roff_err(tree, *argv, "`%s' has no starting " + "tag `%s'", toknames[tok], + toknames[tokens[tok].ctx])); - if (NULL == n) { - roff_err(tree, *argv, "`%s' has no starting tag `%s'", - toknames[tok], - toknames[tokens[tok].ctx]); - return(0); - } - /* LINTED */ do { t = tree->last->tok; - if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) + if ( ! roffexit(tree, t)) return(0); } while (t != tokens[tok].ctx); @@ -777,6 +515,92 @@ rofffindtok(const char *buf) static int +roffchecksec(struct rofftree *tree, + const char *start, int sec, int fail) +{ + + switch (sec) { + case(ROFFSec_PR_Dd): + return(1); + case(ROFFSec_PR_Dt): + if (ROFFSec_PR_Dd & tree->asec) + return(1); + break; + case(ROFFSec_PR_Os): + if (ROFFSec_PR_Dt & tree->asec) + return(1); + break; + case(ROFFSec_NAME): + if (ROFFSec_PR_Os & tree->asec) + return(1); + break; + case(ROFFSec_SYNOP): + if (ROFFSec_NAME & tree->asec) + return(1); + break; + case(ROFFSec_DESC): + if (ROFFSec_SYNOP & tree->asec) + return(1); + break; + case(ROFFSec_RETVAL): + if (ROFFSec_DESC & tree->asec) + return(1); + break; + case(ROFFSec_ENV): + if (ROFFSec_RETVAL & tree->asec) + return(1); + break; + case(ROFFSec_FILES): + if (ROFFSec_ENV & tree->asec) + return(1); + break; + case(ROFFSec_EX): + if (ROFFSec_FILES & tree->asec) + return(1); + break; + case(ROFFSec_DIAG): + if (ROFFSec_EX & tree->asec) + return(1); + break; + case(ROFFSec_ERRS): + if (ROFFSec_DIAG & tree->asec) + return(1); + break; + case(ROFFSec_SEEALSO): + if (ROFFSec_ERRS & tree->asec) + return(1); + break; + case(ROFFSec_STAND): + if (ROFFSec_SEEALSO & tree->asec) + return(1); + break; + case(ROFFSec_HIST): + if (ROFFSec_STAND & tree->asec) + return(1); + break; + case(ROFFSec_AUTH): + if (ROFFSec_HIST & tree->asec) + return(1); + break; + case(ROFFSec_CAVEATS): + if (ROFFSec_AUTH & tree->asec) + return(1); + break; + case(ROFFSec_BUGS): + if (ROFFSec_CAVEATS & tree->asec) + return(1); + break; + default: + return(1); + } + + if (fail) + return(0); + return(roff_warnp(tree, start, ROFF_Sh, WRN_SECORD)); +} + + +static int roffispunct(const char *p) { @@ -876,26 +700,114 @@ roffnode_free(struct rofftree *tree) static int -roffspecial(struct rofftree *tree, int tok, char **ordp) +roffspecial(struct rofftree *tree, int tok, const char *start, + const int *argc, const char **argv, + size_t sz, char **ordp) { - return((*tree->cb.roffspecial)(tree->arg, tok, - tree->cur, ordp)); + switch (tok) { + case (ROFF_At): + if (0 == sz) + break; + if (ROFF_ATT_MAX != roff_att(*ordp)) + break; + return(roff_errp(tree, *ordp, tok, ERR_BADARG)); + + case (ROFF_Xr): + if (2 == sz) { + assert(ordp[1]); + if (ROFF_MSEC_MAX != roff_msec(ordp[1])) + break; + if ( ! roff_warn(tree, start, "invalid `%s' manual " + "section", toknames[tok])) + return(0); + } + /* FALLTHROUGH */ + + case (ROFF_Sx): + /* FALLTHROUGH*/ + case (ROFF_Fn): + if (0 != sz) + break; + return(roff_errp(tree, start, tok, ERR_ARGGE1)); + + case (ROFF_Nm): + if (0 == sz) { + if (0 != tree->name[0]) { + ordp[0] = tree->name; + ordp[1] = NULL; + break; + } + return(roff_err(tree, start, "`Nm' not set")); + } else if ( ! roffsetname(tree, ordp)) + return(0); + break; + + case (ROFF_Rv): + /* FALLTHROUGH*/ + case (ROFF_Ex): + if (1 == sz) + break; + return(roff_errp(tree, start, tok, ERR_ARGEQ1)); + + case (ROFF_Sm): + if (1 != sz) + return(roff_errp(tree, start, tok, ERR_ARGEQ1)); + else if (0 == strcmp(ordp[0], "on") || + 0 == strcmp(ordp[0], "off")) + break; + return(roff_errp(tree, *ordp, tok, ERR_BADARG)); + + case (ROFF_Ud): + /* FALLTHROUGH */ + case (ROFF_Ux): + /* FALLTHROUGH */ + case (ROFF_Bt): + if (0 == sz) + break; + return(roff_errp(tree, start, tok, ERR_ARGEQ0)); + default: + break; + } + + return((*tree->cb.roffspecial)(tree->arg, tok, tree->cur, + argc, argv, (const char **)ordp)); } static int +roffexit(struct rofftree *tree, int tok) +{ + + assert(tokens[tok].cb); + return((*tokens[tok].cb)(tok, tree, NULL, ROFF_EXIT)); +} + + +static int roffcall(struct rofftree *tree, int tok, char **argv) { + int i; + enum roffmsec c; - if (NULL == tokens[tok].cb) { - roff_err(tree, *argv, "unsupported macro `%s'", - toknames[tok]); - return(0); + if (NULL == tokens[tok].cb) + return(roff_errp(tree, *argv, tok, ERR_NOTSUP)); + + if (tokens[tok].sections && ROFF_MSEC_MAX != tree->section) { + i = 0; + while (ROFF_MSEC_MAX != + (c = tokens[tok].sections[i++])) + if (c == tree->section) + break; + if (ROFF_MSEC_MAX == c) { + if ( ! roff_warn(tree, *argv, "`%s' is not a valid " + "macro in this manual section", + toknames[tok])) + return(0); + } } - if ( ! (*tokens[tok].cb)(tok, tree, argv, ROFF_ENTER)) - return(0); - return(1); + + return((*tokens[tok].cb)(tok, tree, argv, ROFF_ENTER)); } @@ -916,14 +828,16 @@ roffnextopt(const struct rofftree *tree, int tok, return(-1); if (ROFF_ARGMAX == (v = rofffindarg(arg + 1))) { - roff_warn(tree, arg, "argument-like parameter `%s' to " - "`%s'", arg, toknames[tok]); + if ( ! roff_warn(tree, arg, "argument-like parameter `%s' to " + "`%s'", arg, toknames[tok])) + return(ROFF_ARGMAX); return(-1); } if ( ! roffargok(tok, v)) { - roff_warn(tree, arg, "invalid argument parameter `%s' to " - "`%s'", tokargnames[v], toknames[tok]); + if ( ! roff_warn(tree, arg, "invalid argument parameter `%s' to " + "`%s'", tokargnames[v], toknames[tok])) + return(ROFF_ARGMAX); return(-1); } @@ -933,7 +847,7 @@ roffnextopt(const struct rofftree *tree, int tok, *in = ++argv; if (NULL == *argv) { - roff_err(tree, arg, "empty value of `%s' for `%s'", + (void)roff_err(tree, arg, "empty value of `%s' for `%s'", tokargnames[v], toknames[tok]); return(ROFF_ARGMAX); } @@ -1007,81 +921,61 @@ roff_Dd(ROFFCALL_ARGS) { time_t t; char *p, buf[32]; + size_t sz; - if (ROFF_BODY & tree->state) { - assert( ! (ROFF_PRELUDE & tree->state)); - assert(ROFF_PRELUDE_Dd & tree->state); + if (ROFFSec_PR_Os & tree->asec) return(roff_text(tok, tree, argv, type)); - } + if (ROFFSec_PR_Dd & tree->asec) + return(roff_errp(tree, *argv, tok, ERR_PR_REP)); + if ( ! roffchecksec(tree, *argv, ROFFSec_PR_Dd, 1)) + return(roff_errp(tree, *argv, tok, ERR_PR_OOO)); - assert(ROFF_PRELUDE & tree->state); - assert( ! (ROFF_BODY & tree->state)); - - if (ROFF_PRELUDE_Dd & tree->state) { - roff_err(tree, *argv, "repeated `Dd' in prelude"); - return(0); - } else if (ROFF_PRELUDE_Dt & tree->state) { - roff_err(tree, *argv, "out-of-order `Dd' in prelude"); - return(0); - } - assert(NULL == tree->last); - argv++; + tree->asec |= (tree->csec = ROFFSec_PR_Dd); - if (0 == strcmp(*argv, "$Mdocdate: December 4 2008 $")) { + /* + * This is a bit complex because there are many forms the date + * can be in: it can be simply $Mdocdate: December 10 2008 $, $Mdocdate $, + * or a raw date. Process accordingly. + */ + + if (0 == strcmp(*argv, "$Mdocdate: December 10 2008 $")) { t = time(NULL); if (NULL == localtime_r(&t, &tree->tm)) err(1, "localtime_r"); - tree->state |= ROFF_PRELUDE_Dd; return(1); } - /* Build this from Mdocdate or raw date. */ - buf[0] = 0; p = *argv; + sz = sizeof(buf); if (0 != strcmp(*argv, "$Mdocdate:")) { while (*argv) { - if (strlcat(buf, *argv++, sizeof(buf)) - < sizeof(buf)) + if (strlcat(buf, *argv++, sz) < sz) continue; - roff_err(tree, p, "bad `Dd' date"); - return(0); + return(roff_errp(tree, p, tok, ERR_BADARG)); } - if (strptime(buf, "%b%d,%Y", &tree->tm)) { - tree->state |= ROFF_PRELUDE_Dd; + if (strptime(buf, "%b%d,%Y", &tree->tm)) return(1); - } - roff_err(tree, *argv, "bad `Dd' date"); - return(0); + return(roff_errp(tree, p, tok, ERR_BADARG)); } argv++; + while (*argv && **argv != '$') { - if (strlcat(buf, *argv++, sizeof(buf)) - >= sizeof(buf)) { - roff_err(tree, p, "bad `Dd' Mdocdate"); - return(0); - } - if (strlcat(buf, " ", sizeof(buf)) - >= sizeof(buf)) { - roff_err(tree, p, "bad `Dd' Mdocdate"); - return(0); - } + if (strlcat(buf, *argv++, sz) >= sz) + return(roff_errp(tree, p, tok, ERR_BADARG)); + if (strlcat(buf, " ", sz) >= sz) + return(roff_errp(tree, p, tok, ERR_BADARG)); } - if (NULL == *argv) { - roff_err(tree, p, "bad `Dd' Mdocdate"); - return(0); - } - if (NULL == strptime(buf, "%b %d %Y", &tree->tm)) { - roff_err(tree, *argv, "bad `Dd' Mdocdate"); - return(0); - } + if (NULL == *argv) + return(roff_errp(tree, p, tok, ERR_BADARG)); + if (NULL == strptime(buf, "%b %d %Y", &tree->tm)) + return(roff_errp(tree, p, tok, ERR_BADARG)); - tree->state |= ROFF_PRELUDE_Dd; return(1); } @@ -1090,87 +984,95 @@ roff_Dd(ROFFCALL_ARGS) static int roff_Dt(ROFFCALL_ARGS) { + size_t sz; - if (ROFF_BODY & tree->state) { - assert( ! (ROFF_PRELUDE & tree->state)); - assert(ROFF_PRELUDE_Dt & tree->state); + if (ROFFSec_PR_Os & tree->asec) return(roff_text(tok, tree, argv, type)); - } + if (ROFFSec_PR_Dt & tree->asec) + return(roff_errp(tree, *argv, tok, ERR_PR_REP)); + if ( ! roffchecksec(tree, *argv, ROFFSec_PR_Dt, 1)) + return(roff_errp(tree, *argv, tok, ERR_PR_OOO)); - assert(ROFF_PRELUDE & tree->state); - assert( ! (ROFF_BODY & tree->state)); + argv++; + tree->asec |= (tree->csec = ROFFSec_PR_Dt); + sz = sizeof(tree->title); - if ( ! (ROFF_PRELUDE_Dd & tree->state)) { - roff_err(tree, *argv, "out-of-order `Dt' in prelude"); - return(0); - } else if (ROFF_PRELUDE_Dt & tree->state) { - roff_err(tree, *argv, "repeated `Dt' in prelude"); - return(0); - } + if (NULL == *argv) + return(roff_errp(tree, *argv, tok, ERR_ARGGE2)); + if (strlcpy(tree->title, *argv, sz) >= sz) + return(roff_errp(tree, *argv, tok, ERR_ARGLEN)); argv++; - if (NULL == *argv) { - roff_err(tree, *argv, "`Dt' needs document title"); - return(0); - } else if (strlcpy(tree->title, *argv, sizeof(tree->title)) - >= sizeof(tree->title)) { - roff_err(tree, *argv, "`Dt' document title too long"); - return(0); - } + if (NULL == *argv) + return(roff_errp(tree, *argv, tok, ERR_ARGGE2)); - argv++; - if (NULL == *argv) { - roff_err(tree, *argv, "`Dt' needs section"); - return(0); - } else if (strlcpy(tree->section, *argv, sizeof(tree->section)) - >= sizeof(tree->section)) { - roff_err(tree, *argv, "`Dt' section too long"); - return(0); - } + if (ROFF_MSEC_MAX == (tree->section = roff_msec(*argv))) + return(roff_errp(tree, *argv, tok, ERR_BADARG)); argv++; + if (NULL == *argv) { - tree->volume[0] = 0; - } else if (strlcpy(tree->volume, *argv, sizeof(tree->volume)) - >= sizeof(tree->volume)) { - roff_err(tree, *argv, "`Dt' volume too long"); - return(0); - } + switch (tree->section) { + case(ROFF_MSEC_1): + /* FALLTHROUGH */ + case(ROFF_MSEC_6): + /* FALLTHROUGH */ + case(ROFF_MSEC_7): + tree->volume = ROFF_VOL_URM; + break; + case(ROFF_MSEC_2): + /* FALLTHROUGH */ + case(ROFF_MSEC_3): + /* FALLTHROUGH */ + case(ROFF_MSEC_3p): + /* FALLTHROUGH */ + case(ROFF_MSEC_4): + /* FALLTHROUGH */ + case(ROFF_MSEC_5): + tree->volume = ROFF_VOL_PRM; + break; + case(ROFF_MSEC_8): + tree->volume = ROFF_VOL_PRM; + break; + case(ROFF_MSEC_9): + tree->volume = ROFF_VOL_KM; + break; + case(ROFF_MSEC_UNASS): + /* FALLTHROUGH */ + case(ROFF_MSEC_DRAFT): + /* FALLTHROUGH */ + case(ROFF_MSEC_PAPER): + tree->volume = ROFF_VOL_NONE; + break; + default: + abort(); + /* NOTREACHED */ + } + } else if (ROFF_VOL_MAX == (tree->volume = roff_vol(*argv))) + return(roff_errp(tree, *argv, tok, ERR_BADARG)); assert(NULL == tree->last); - tree->state |= ROFF_PRELUDE_Dt; return(1); } -/* ARGSUSED */ -static int -roff_Sm(ROFFCALL_ARGS) +static int +roffsetname(struct rofftree *tree, char **ordp) { - char *morep[1], *p; + size_t sz; + + assert(*ordp); - p = *argv++; + /* FIXME: not all sections can set this. */ - if (NULL == (morep[0] = *argv++)) { - roff_err(tree, p, "`Sm' expects an argument"); - return(0); - } else if (0 != strcmp(morep[0], "on") && - 0 != strcmp(morep[0], "off")) { - roff_err(tree, p, "`Sm' has invalid argument"); - return(0); - } + if (NULL != *(ordp + 1)) + return(roff_errp(tree, *ordp, ROFF_Nm, ERR_ARGMNY)); + + sz = sizeof(tree->name); + if (strlcpy(tree->name, *ordp, sz) >= sz) + return(roff_errp(tree, *ordp, ROFF_Nm, ERR_ARGLEN)); - if (*argv) - roff_warn(tree, *argv, "`Sm' shouldn't have arguments"); - - if ( ! roffspecial(tree, tok, morep)) - return(0); - - while (*argv) - if ( ! roffdata(tree, 1, *argv++)) - return(0); - return(1); } @@ -1185,7 +1087,7 @@ roff_Ns(ROFFCALL_ARGS) first = (*argv++ == tree->cur); morep[0] = NULL; - if ( ! roffspecial(tree, tok, morep)) + if ( ! roffspecial(tree, tok, *argv, NULL, NULL, 0, morep)) return(0); while (*argv) { @@ -1226,48 +1128,36 @@ static int roff_Os(ROFFCALL_ARGS) { char *p; + size_t sz; - if (ROFF_BODY & tree->state) { - assert( ! (ROFF_PRELUDE & tree->state)); - assert(ROFF_PRELUDE_Os & tree->state); + if (ROFFSec_PR_Os & tree->asec) return(roff_text(tok, tree, argv, type)); - } + if ( ! roffchecksec(tree, *argv, ROFFSec_PR_Os, 1)) + return(roff_errp(tree, *argv, tok, ERR_PR_OOO)); - assert(ROFF_PRELUDE & tree->state); - if ( ! (ROFF_PRELUDE_Dt & tree->state) || - ! (ROFF_PRELUDE_Dd & tree->state)) { - roff_err(tree, *argv, "out-of-order `Os' in prelude"); - return(0); - } + p = *++argv; + sz = sizeof(tree->os); + tree->asec |= (tree->csec = ROFFSec_PR_Os); tree->os[0] = 0; - p = *++argv; + while (*argv) + if (strlcat(tree->os, *argv++, sz) >= sz) + return(roff_errp(tree, p, tok, ERR_ARGLEN)); - while (*argv) { - if (strlcat(tree->os, *argv++, sizeof(tree->os)) - < sizeof(tree->os)) - continue; - roff_err(tree, p, "`Os' value too long"); - return(0); - } - if (0 == tree->os[0]) - if (strlcpy(tree->os, "LOCAL", sizeof(tree->os)) - >= sizeof(tree->os)) { - roff_err(tree, p, "`Os' value too long"); - return(0); - } + if (strlcpy(tree->os, "LOCAL", sz) >= sz) + return(roff_errp(tree, p, tok, ERR_ARGLEN)); - tree->state |= ROFF_PRELUDE_Os; - tree->state &= ~ROFF_PRELUDE; - tree->state |= ROFF_BODY; + assert(ROFF_MSEC_MAX != tree->section); + assert(0 != tree->title[0]); + assert(0 != tree->os[0]); assert(NULL == tree->last); return((*tree->cb.roffhead)(tree->arg, &tree->tm, - tree->os, tree->title, tree->section, - tree->volume)); + tree->os, tree->title, + tree->section, tree->volume)); } @@ -1278,21 +1168,33 @@ roff_layout(ROFFCALL_ARGS) int i, c, argcp[ROFF_MAXLINEARG]; char *argvp[ROFF_MAXLINEARG]; - if (ROFF_PRELUDE & tree->state) { - roff_err(tree, *argv, "bad `%s' in prelude", - toknames[tok]); - return(0); - } else if (ROFF_EXIT == type) { + /* + * The roff_layout function is for multi-line macros. A layout + * has a start and end point, which is either declared + * explicitly or implicitly. An explicit start and end is + * embodied by `.Bl' and `.El', with the former being the start + * and the latter being an end. The `.Sh' and `.Ss' tags, on + * the other hand, are implicit. The scope of a layout is the + * space between start and end. Explicit layouts may not close + * out implicit ones and vice versa; implicit layouts may close + * out other implicit layouts. + */ + + assert( ! (ROFF_CALLABLE & tokens[tok].flags)); + + if ( ! ROFFSec_NAME & tree->asec) + return(roff_errp(tree, *argv, tok, ERR_NOT_PR)); + + if (ROFF_EXIT == type) { roffnode_free(tree); if ( ! (*tree->cb.roffblkbodyout)(tree->arg, tok)) return(0); return((*tree->cb.roffblkout)(tree->arg, tok)); } + argv++; assert( ! (ROFF_CALLABLE & tokens[tok].flags)); - ++argv; - if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) return(0); if (NULL == roffnode_new(tok, tree)) @@ -1302,15 +1204,65 @@ roff_layout(ROFFCALL_ARGS) * Layouts have two parts: the layout body and header. The * layout header is the trailing text of the line macro, while * the layout body is everything following until termination. + * Example: + * + * .It Fl f ) ; + * Bar. + * + * ...Produces... + * + * + * + * ; + * + * + * + * Bar. + * + * */ - if ( ! (*tree->cb.roffblkin)(tree->arg, tok, argcp, argvp)) + if ( ! (*tree->cb.roffblkin)(tree->arg, tok, argcp, + (const char **)argvp)) return(0); + + /* +++ Begin run macro-specific hooks over argv. */ + + switch (tok) { + case (ROFF_Sh): + if (NULL == *argv) { + argv--; + return(roff_errp(tree, *argv, tok, ERR_ARGGE1)); + } + + tree->csec = roff_sec((const char **)argv); + + if ( ! (ROFFSec_OTHER & tree->csec) && + tree->asec & tree->csec) + if ( ! roff_warn(tree, *argv, "section repeated")) + return(0); + + if (0 == tree->asec && ! (ROFFSec_NAME & tree->csec)) + return(roff_err(tree, *argv, "`NAME' section " + "must be first")); + if ( ! roffchecksec(tree, *argv, tree->csec, 0)) + return(0); + + tree->asec |= tree->csec; + break; + default: + break; + } + + /* --- End run macro-specific hooks over argv. */ + if (NULL == *argv) return((*tree->cb.roffblkbodyin) - (tree->arg, tok, argcp, argvp)); + (tree->arg, tok, argcp, + (const char **)argvp)); - if ( ! (*tree->cb.roffblkheadin)(tree->arg, tok, argcp, argvp)) + if ( ! (*tree->cb.roffblkheadin)(tree->arg, tok, argcp, + (const char **)argvp)) return(0); /* @@ -1326,8 +1278,8 @@ roff_layout(ROFFCALL_ARGS) if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) return(0); - return((*tree->cb.roffblkbodyin) - (tree->arg, tok, argcp, argvp)); + return((*tree->cb.roffblkbodyin)(tree->arg, tok, argcp, + (const char **)argvp)); } /* @@ -1358,7 +1310,8 @@ roff_layout(ROFFCALL_ARGS) if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) return(0); return((*tree->cb.roffblkbodyin) - (tree->arg, tok, argcp, argvp)); + (tree->arg, tok, argcp, + (const char **)argvp)); } /* @@ -1368,11 +1321,10 @@ roff_layout(ROFFCALL_ARGS) if ( ! roffpurgepunct(tree, argv)) return(0); - if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) return(0); - return((*tree->cb.roffblkbodyin) - (tree->arg, tok, argcp, argvp)); + return((*tree->cb.roffblkbodyin)(tree->arg, + tok, argcp, (const char **)argvp)); } @@ -1380,50 +1332,66 @@ roff_layout(ROFFCALL_ARGS) static int roff_ordered(ROFFCALL_ARGS) { - int i, first, c; - char *ordp[ROFF_MAXLINEARG]; + 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); - } + /* + * Ordered macros pass their arguments directly to handlers, + * instead of considering it free-form text. Thus, the + * following macro looks as follows: + * + * .Xr foo 1 ) , + * + * .Xr arg1 arg2 punctuation + */ + if ( ! ROFFSec_NAME & tree->asec) + return(roff_errp(tree, *argv, tok, ERR_NOT_PR)); + first = (*argv == tree->cur); - argv++; + p = *argv++; + ordp[0] = NULL; - if (NULL == *argv) { + if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) + return(0); - /* FIXME: satisfies number of args? */ + if (NULL == *argv) + return(roffspecial(tree, tok, p, argcp, + (const char **)argvp, 0, ordp)); - ordp[0] = NULL; - return(roffspecial(tree, tok, ordp)); - } - i = 0; while (*argv && i < ROFF_MAXLINEARG) { - if (ROFF_MAX != (c = rofffindcallable(*argv))) - return(roffcall(tree, c, argv)); - if (roffispunct(*argv)) + 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; - ordp[i++] = *argv++; + 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; - /* FIXME: too many or too few args? */ - - if (i == ROFF_MAXLINEARG && *argv) { - roff_err(tree, *argv, "too many args", toknames[tok]); + 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 ( ! roffspecial(tree, tok, ordp)) - return(0); - if ( ! first || NULL == *argv) return(1); @@ -1438,18 +1406,28 @@ roff_text(ROFFCALL_ARGS) 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", - toknames[tok]); - return(0); - } + /* + * Text macros are similar to special tokens, except that + * arguments are instead flushed as pure data: we're only + * concerned with the macro and its arguments. Example: + * + * .Fl v W f ; + * + * ...Produces... + * + * v W f ; + */ + if ( ! ROFFSec_NAME & tree->asec) + return(roff_errp(tree, *argv, tok, ERR_NOT_PR)); + first = (*argv == tree->cur); argv++; if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) return(0); - if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) + if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, + (const char **)argvp)) return(0); if (NULL == *argv) return((*tree->cb.roffout)(tree->arg, tok)); @@ -1459,7 +1437,6 @@ roff_text(ROFFCALL_ARGS) while (*argv) if ( ! roffdata(tree, i++, *argv++)) return(0); - return((*tree->cb.roffout)(tree->arg, tok)); } @@ -1468,6 +1445,8 @@ roff_text(ROFFCALL_ARGS) * terminating punctuation. If we encounter it and all * subsequent tokens are punctuation, then stop processing (the * line-dominant macro will print these tokens after closure). + * If the punctuation is followed by non-punctuation, then close + * and re-open our scope, then continue. */ i = 0; @@ -1479,11 +1458,9 @@ roff_text(ROFFCALL_ARGS) if ( ! roffcall(tree, c, argv)) return(0); - if (ROFF_LSCOPE & tokens[tok].flags) if ( ! (*tree->cb.roffout)(tree->arg, tok)) return(0); - break; } @@ -1499,8 +1476,21 @@ roff_text(ROFFCALL_ARGS) break; if (argv[j]) { + if (ROFF_LSCOPE & tokens[tok].flags) { + if ( ! roffdata(tree, 0, *argv++)) + return(0); + continue; + } + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); if ( ! roffdata(tree, 0, *argv++)) return(0); + if ( ! (*tree->cb.roffin)(tree->arg, tok, + argcp, + (const char **)argvp)) + return(0); + + i = 0; continue; } @@ -1532,12 +1522,30 @@ static int roff_depr(ROFFCALL_ARGS) { - roff_err(tree, *argv, "`%s' is deprecated", toknames[tok]); - return(0); + return(roff_errp(tree, *argv, tok, ERR_DEPREC)); } -static void +static int +roff_warnp(const struct rofftree *tree, const char *pos, + int tok, enum rofferr type) +{ + char *p; + + switch (type) { + case (WRN_SECORD): + p = "section at `%s' out of order"; + break; + default: + abort(); + /* NOTREACHED */ + } + + return(roff_warn(tree, pos, p, toknames[tok])); +} + + +static int roff_warn(const struct rofftree *tree, const char *pos, char *fmt, ...) { va_list ap; @@ -1547,22 +1555,75 @@ roff_warn(const struct rofftree *tree, const char *pos (void)vsnprintf(buf, sizeof(buf), fmt, ap); va_end(ap); - (*tree->cb.roffmsg)(tree->arg, - ROFF_WARN, tree->cur, pos, buf); + return((*tree->cb.roffmsg)(tree->arg, + ROFF_WARN, tree->cur, pos, buf)); } -static void +static int +roff_errp(const struct rofftree *tree, const char *pos, + int tok, enum rofferr type) +{ + char *p; + + switch (type) { + case (ERR_ARGEQ1): + p = "`%s' expects exactly one argument"; + break; + case (ERR_ARGEQ0): + p = "`%s' expects exactly zero arguments"; + break; + case (ERR_ARGGE1): + p = "`%s' expects one or more arguments"; + break; + case (ERR_ARGGE2): + p = "`%s' expects two or more arguments"; + break; + case (ERR_BADARG): + p = "invalid argument for `%s'"; + break; + case (ERR_NOTSUP): + p = "macro `%s' is not supported"; + break; + case(ERR_PR_OOO): + p = "prelude macro `%s' is out of order"; + break; + case(ERR_PR_REP): + p = "prelude macro `%s' repeated"; + break; + case(ERR_ARGLEN): + p = "macro argument for `%s' is too long"; + break; + case(ERR_DEPREC): + p = "macro `%s' is deprecated"; + break; + case(ERR_NOT_PR): + p = "macro `%s' disallowed in prelude"; + break; + case(ERR_ARGMNY): + p = "too many arguments for macro `%s'"; + break; + default: + abort(); + /* NOTREACHED */ + } + + return(roff_err(tree, pos, p, toknames[tok])); +} + + +static int roff_err(const struct rofftree *tree, const char *pos, char *fmt, ...) { va_list ap; char buf[128]; va_start(ap, fmt); - (void)vsnprintf(buf, sizeof(buf), fmt, ap); + if (-1 == vsnprintf(buf, sizeof(buf), fmt, ap)) + err(1, "vsnprintf"); va_end(ap); - (*tree->cb.roffmsg)(tree->arg, - ROFF_ERROR, tree->cur, pos, buf); + return((*tree->cb.roffmsg) + (tree->arg, ROFF_ERROR, tree->cur, pos, buf)); }