=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.20 retrieving revision 1.59 diff -u -p -r1.20 -r1.59 --- mandoc/roff.c 2008/11/30 12:41:45 1.20 +++ mandoc/roff.c 2008/12/10 12:05:33 1.59 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.20 2008/11/30 12:41:45 kristaps Exp $ */ +/* $Id: roff.c,v 1.59 2008/12/10 12:05:33 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -16,6 +16,9 @@ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR * PERFORMANCE OF THIS SOFTWARE. */ +#include +#include + #include #include #include @@ -25,336 +28,105 @@ #include #include -#include "libmdocml.h" #include "private.h" +#include "roff.h" -/* FIXME: warn if Pp occurs before/after Sh etc. (see mdoc.samples). */ +/* 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 empty lists. */ +/* TODO: (warn) some sections need specific elements. */ +/* TODO: (warn) NAME section has particular order. */ +/* TODO: macros with a set number of arguments? */ +/* TODO: validate Dt macro arguments. */ +/* FIXME: Bl -diag supposed to ignore callable children. */ -/* FIXME: warn about "X section only" macros. */ - -/* FIXME: warn about empty lists. */ - -/* FIXME: ; : } ) (etc.) after text macros? */ - -/* FIXME: NAME section needs specific elements. */ - -/* FIXME: don't print Os, just do roffbegin. */ - -#define ROFF_MAXARG 32 - -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; - 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_QUOTES (1 << 2) /* Quoted args. */ -#define ROFF_SHALLOW (1 << 3) /* Nesting block. */ -}; - -struct roffarg { - int flags; -#define ROFF_VALUE (1 << 0) /* Has a value. */ -}; - -struct roffnode { - int tok; /* Token id. */ - struct roffnode *parent; /* Parent (or NULL). */ -}; - struct rofftree { struct roffnode *last; /* Last parsed node. */ - char *cur; - - time_t date; /* `Dd' results. */ + 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. */ - + enum roffmsec section; + enum roffvol volume; int state; -#define ROFF_PRELUDE (1 << 1) /* In roff prelude. */ +#define ROFF_PRELUDE (1 << 1) /* In roff prelude. */ /* FIXME: put into asec. */ #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. */ + 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_layout(ROFFCALL_ARGS); -static int roff_text(ROFFCALL_ARGS); -static int roff_comment(ROFFCALL_ARGS); -static int roff_close(ROFFCALL_ARGS); -static int roff_special(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); static int roffargs(const struct rofftree *, int, char *, char **); static int roffargok(int, int); 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 roffexit(struct rofftree *, int); 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_Oo[] = { ROFF_Op, ROFF_Oc, 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_Oc, 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_comment, 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, ROFF_QUOTES }, /* Os */ - { roff_layout, NULL, NULL, NULL, ROFF_Sh, ROFF_LAYOUT, ROFF_PARSED }, /* Sh */ - { roff_layout, NULL, NULL, NULL, ROFF_Ss, ROFF_LAYOUT, ROFF_PARSED }, /* Ss */ - { roff_text, NULL, NULL, NULL, ROFF_Pp, ROFF_TEXT, 0 }, /* Pp */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* D1 */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Dl */ - { roff_layout, roffarg_Bd, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Bd */ - { roff_close, NULL, NULL, NULL, ROFF_Bd, ROFF_LAYOUT, 0 }, /* Ed */ - { roff_layout, roffarg_Bl, NULL, roffchild_Bl, 0, ROFF_LAYOUT, 0 }, /* Bl */ - { roff_close, NULL, roffparent_El, NULL, ROFF_Bl, ROFF_LAYOUT, 0 }, /* El */ - { roff_layout, NULL, roffparent_It, NULL, ROFF_It, ROFF_LAYOUT, 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, ROFF_QUOTES }, /* 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, ROFF_QUOTES }, /* 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 }, /* Op */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ot */ /* XXX deprecated */ - { 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 }, /* 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 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 }, /* Bq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bsx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bx */ - {roff_special, 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 }, /* Dq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ec */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ef */ /* FIXME */ - { 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 */ - { NULL, 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 */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Pf */ - { roff_text, NULL, NULL, NULL, 0, ROFF_LAYOUT, ROFF_PARSED | ROFF_CALLABLE }, /* Po */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* 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_layout, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qo */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qq */ - { roff_close, 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 }, /* Sq */ - {roff_special, 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_close, NULL, roffparent_Fc, NULL, ROFF_Fo, ROFF_LAYOUT, 0 }, /* Fc */ - { roff_layout, NULL, NULL, roffchild_Oo, 0, ROFF_LAYOUT, 0 }, /* Oo */ - { roff_close, NULL, roffparent_Oc, NULL, ROFF_Oo, ROFF_LAYOUT, 0 }, /* Oc */ - { roff_layout, roffarg_Bk, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Bk */ - { roff_close, NULL, NULL, NULL, ROFF_Bk, ROFF_LAYOUT, 0 }, /* Ek */ - }; - -/* 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", - "Xr", "\%A", "\%B", "\%D", - "\%I", "\%J", "\%N", "\%O", - "\%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", - }; - -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) { @@ -369,24 +141,34 @@ roff_free(struct rofftree *tree, int flush) error = 1; if (ROFF_PRELUDE & tree->state) { - roff_warn(tree, NULL, "prelude never finished"); + (void)roff_err(tree, NULL, "prelude never finished"); goto end; - } + } else 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->parent; n = n->parent) { + for (n = tree->last; n; n = n->parent) { if (0 != tokens[n->tok].ctx) continue; - roff_warn(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, &tree->tm, + tree->os, tree->title, + tree->section, tree->volume)) + goto end; + error = 0; end: @@ -413,6 +195,7 @@ roff_alloc(const struct roffcb *cb, void *args) tree->state = ROFF_PRELUDE; tree->arg = args; + tree->section = ROFF_MSEC_MAX; (void)memcpy(&tree->cb, cb, sizeof(struct roffcb)); @@ -427,10 +210,9 @@ roff_engine(struct rofftree *tree, char *buf) tree->cur = buf; assert(buf); - if (0 == *buf) { - roff_warn(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)); @@ -438,10 +220,41 @@ roff_engine(struct rofftree *tree, char *buf) static int -textparse(const struct rofftree *tree, char *buf) +textparse(struct rofftree *tree, char *buf) { + char *bufp; - return((*tree->cb.roffdata)(tree->arg, buf)); + /* TODO: literal parsing. */ + + if ( ! (ROFF_BODY & tree->state)) + return(roff_err(tree, buf, "data not in body")); + + /* 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); } @@ -457,23 +270,35 @@ 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) 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 && ! isspace(*buf)) - buf++; + while (*buf) { + if ( ! isspace(*buf)) { + buf++; + continue; + } + if (*(buf - 1) == '\\') { + buf++; + continue; + } + break; + } if (0 == *buf) continue; } @@ -481,20 +306,16 @@ roffargs(const struct rofftree *tree, while (*buf && isspace(*buf)) buf++; } - + assert(i > 0); - if (ROFF_MAXARG == 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); } -/* XXX */ static int roffscan(int tok, const int *tokv) { @@ -515,21 +336,17 @@ roffparse(struct rofftree *tree, char *buf) { int tok, t; struct roffnode *n; - char *argv[ROFF_MAXARG]; + char *argv[ROFF_MAXLINEARG]; char **argvp; - if (ROFF_MAX == (tok = rofffindtok(buf + 1))) { - roff_err(tree, buf + 1, "bogus line macro"); + if (0 != *buf && 0 != *(buf + 1) && 0 != *(buf + 2)) + if (0 == strncmp(buf, ".\\\"", 3)) + return(1); + + 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); - } else if (ROFF_COMMENT == tokens[tok].type) - return(1); - - if ( ! roffargs(tree, tok, buf, argv)) - return(0); argvp = (char **)argv; @@ -539,9 +356,8 @@ roffparse(struct rofftree *tree, char *buf) if (ROFF_PRELUDE & tree->state) { assert(NULL == tree->last); - return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); - } else - assert(tree->last); + return(roffcall(tree, tok, argvp)); + } assert(ROFF_BODY & tree->state); @@ -550,33 +366,26 @@ roffparse(struct rofftree *tree, char *buf) * children are satisfied. */ - if ( ! roffscan(tree->last->tok, tokens[tok].parents)) { - roff_err(tree, *argvp, "`%s' has invalid parent `%s'", - toknames[tok], - toknames[tree->last->tok]); - return(0); - } + if (tree->last && ! roffscan + (tree->last->tok, tokens[tok].parents)) + return(roff_err(tree, *argvp, "`%s' has invalid " + "parent `%s'", toknames[tok], + toknames[tree->last->tok])); - if ( ! 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); - } + if (tree->last && ! roffscan + (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)); - - /* - * Check our scope rules. - */ - + 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: @@ -605,6 +414,12 @@ roffparse(struct rofftree *tree, char *buf) n = NULL; break; } + if (tokens[n->tok].ctx == n->tok) + continue; + return(roff_err(tree, *argv, "`%s' breaks " + "scope of prior`%s'", + toknames[tok], + toknames[n->tok])); } /* @@ -613,7 +428,7 @@ 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 @@ -622,11 +437,11 @@ 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)); } /* @@ -638,17 +453,32 @@ roffparse(struct rofftree *tree, char *buf) * In this, the `El' tag closes out the scope of `Bl'. */ - assert(tree->last); 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) + continue; + 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])); + + /* 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); - assert(tree->last); return(1); } @@ -674,7 +504,7 @@ static int rofffindtok(const char *buf) { char token[4]; - size_t i; + int i; for (i = 0; *buf && ! isspace(*buf) && i < 3; i++, buf++) token[i] = *buf; @@ -696,8 +526,78 @@ rofffindtok(const char *buf) } -#if notyet static int +roffchecksec(struct rofftree *tree, const char *start, int sec) +{ + int prior; + + switch (sec) { + case(ROFFSec_SYNOP): + if ((prior = ROFFSec_NAME) & tree->asec) + return(1); + break; + case(ROFFSec_DESC): + if ((prior = ROFFSec_SYNOP) & tree->asec) + return(1); + break; + case(ROFFSec_RETVAL): + if ((prior = ROFFSec_DESC) & tree->asec) + return(1); + break; + case(ROFFSec_ENV): + if ((prior = ROFFSec_RETVAL) & tree->asec) + return(1); + break; + case(ROFFSec_FILES): + if ((prior = ROFFSec_ENV) & tree->asec) + return(1); + break; + case(ROFFSec_EX): + if ((prior = ROFFSec_FILES) & tree->asec) + return(1); + break; + case(ROFFSec_DIAG): + if ((prior = ROFFSec_EX) & tree->asec) + return(1); + break; + case(ROFFSec_ERRS): + if ((prior = ROFFSec_DIAG) & tree->asec) + return(1); + break; + case(ROFFSec_SEEALSO): + if ((prior = ROFFSec_ERRS) & tree->asec) + return(1); + break; + case(ROFFSec_STAND): + if ((prior = ROFFSec_SEEALSO) & tree->asec) + return(1); + break; + case(ROFFSec_HIST): + if ((prior = ROFFSec_STAND) & tree->asec) + return(1); + break; + case(ROFFSec_AUTH): + if ((prior = ROFFSec_HIST) & tree->asec) + return(1); + break; + case(ROFFSec_CAVEATS): + if ((prior = ROFFSec_AUTH) & tree->asec) + return(1); + break; + case(ROFFSec_BUGS): + if ((prior = ROFFSec_CAVEATS) & tree->asec) + return(1); + break; + default: + return(1); + } + + return(roff_warnp(tree, start, ROFF_Sh, WRN_SECORD)); +} + + +/* FIXME: move this into literals.c (or similar). */ +static int roffispunct(const char *p) { @@ -737,7 +637,6 @@ roffispunct(const char *p) return(0); } -#endif static int @@ -798,6 +697,118 @@ 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 (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) + 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); + } + } + + return((*tokens[tok].cb)(tok, tree, argv, ROFF_ENTER)); +} + + +static int roffnextopt(const struct rofftree *tree, int tok, char ***in, char **val) { @@ -814,14 +825,16 @@ 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[1], 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); } @@ -831,7 +844,7 @@ 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); } @@ -840,10 +853,72 @@ roffnextopt(const struct rofftree *tree, int tok, } +static int +roffpurgepunct(struct rofftree *tree, char **argv) +{ + int i; + + i = 0; + while (argv[i]) + i++; + assert(i > 0); + if ( ! roffispunct(argv[--i])) + return(1); + while (i >= 0 && roffispunct(argv[i])) + i--; + i++; + + /* LINTED */ + while (argv[i]) + if ( ! roffdata(tree, 0, argv[i++])) + return(0); + return(1); +} + + +static int +roffparseopts(struct rofftree *tree, int tok, + char ***args, int *argc, char **argv) +{ + int i, c; + char *v; + + i = 0; + + while (-1 != (c = roffnextopt(tree, tok, args, &v))) { + if (ROFF_ARGMAX == c) + return(0); + + argc[i] = c; + argv[i] = v; + i++; + *args = *args + 1; + } + + argc[i] = ROFF_ARGMAX; + argv[i] = NULL; + return(1); +} + + +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) { + time_t t; + char *p, buf[32]; + size_t sz; if (ROFF_BODY & tree->state) { assert( ! (ROFF_PRELUDE & tree->state)); @@ -854,19 +929,61 @@ roff_Dd(ROFFCALL_ARGS) 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); + if (ROFF_PRELUDE_Dd & tree->state) + return(roff_errp(tree, *argv, tok, ERR_PR_REP)); + if (ROFF_PRELUDE_Dt & tree->state) + return(roff_errp(tree, *argv, tok, ERR_PR_OOO)); + + assert(NULL == tree->last); + + argv++; + + /* + * 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 $")) { + t = time(NULL); + if (NULL == localtime_r(&t, &tree->tm)) + err(1, "localtime_r"); + tree->state |= ROFF_PRELUDE_Dd; + return(1); + } + + buf[0] = 0; + p = *argv; + sz = sizeof(buf); + + if (0 != strcmp(*argv, "$Mdocdate:")) { + while (*argv) { + if (strlcat(buf, *argv++, sz) < sz) + continue; + return(roff_errp(tree, p, tok, ERR_BADARG)); + } + if (strptime(buf, "%b%d,%Y", &tree->tm)) { + tree->state |= ROFF_PRELUDE_Dd; + return(1); + } + return(roff_errp(tree, p, tok, ERR_BADARG)); } - /* TODO: parse date. */ + argv++; - assert(NULL == tree->last); - tree->state |= ROFF_PRELUDE_Dd; + while (*argv && **argv != '$') { + 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) + 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); } @@ -875,6 +992,7 @@ roff_Dd(ROFFCALL_ARGS) static int roff_Dt(ROFFCALL_ARGS) { + size_t sz; if (ROFF_BODY & tree->state) { assert( ! (ROFF_PRELUDE & tree->state)); @@ -885,16 +1003,68 @@ roff_Dt(ROFFCALL_ARGS) assert(ROFF_PRELUDE & tree->state); assert( ! (ROFF_BODY & tree->state)); - 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 ( ! (ROFF_PRELUDE_Dd & tree->state)) + return(roff_errp(tree, *argv, tok, ERR_PR_OOO)); + if (ROFF_PRELUDE_Dt & tree->state) + return(roff_errp(tree, *argv, tok, ERR_PR_REP)); - /* TODO: parse date. */ + argv++; + sz = sizeof(tree->title); + 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) + return(roff_errp(tree, *argv, tok, ERR_ARGGE2)); + + if (ROFF_MSEC_MAX == (tree->section = roff_msec(*argv))) + return(roff_errp(tree, *argv, tok, ERR_BADARG)); + + argv++; + + if (NULL == *argv) { + 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; @@ -902,15 +1072,80 @@ roff_Dt(ROFFCALL_ARGS) } +static int +roffsetname(struct rofftree *tree, char **ordp) +{ + size_t sz; + + assert(*ordp); + + /* FIXME: not all sections can set this. */ + + 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)); + + return(1); +} + + /* ARGSUSED */ static int +roff_Ns(ROFFCALL_ARGS) +{ + int j, c, first; + char *morep[1]; + + first = (*argv++ == tree->cur); + morep[0] = NULL; + + 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 ( ! roffispunct(*argv)) { + if ( ! roffdata(tree, 1, *argv++)) + return(0); + continue; + } + + 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)); +} + + +/* ARGSUSED */ +static int roff_Os(ROFFCALL_ARGS) { + char *p; + size_t sz; - if (ROFF_EXIT == type) { - roffnode_free(tree); - return((*tree->cb.rofftail)(tree->arg)); - } else if (ROFF_BODY & tree->state) { + if (ROFF_BODY & tree->state) { assert( ! (ROFF_PRELUDE & tree->state)); assert(ROFF_PRELUDE_Os & tree->state); return(roff_text(tok, tree, argv, type)); @@ -918,23 +1153,35 @@ roff_Os(ROFFCALL_ARGS) 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); - } + ! (ROFF_PRELUDE_Dd & tree->state)) + return(roff_errp(tree, *argv, tok, ERR_PR_OOO)); - /* TODO: extract OS. */ + tree->os[0] = 0; + p = *++argv; + sz = sizeof(tree->os); + + while (*argv) + if (strlcat(tree->os, *argv++, sz) >= sz) + return(roff_errp(tree, p, tok, ERR_ARGLEN)); + + if (0 == tree->os[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); - if (NULL == roffnode_new(tok, tree)) - return(0); - - return((*tree->cb.roffhead)(tree->arg)); + return((*tree->cb.roffhead)(tree->arg, &tree->tm, + tree->os, tree->title, + tree->section, tree->volume)); } @@ -942,149 +1189,352 @@ roff_Os(ROFFCALL_ARGS) static int roff_layout(ROFFCALL_ARGS) { - int i, c, argcp[ROFF_MAXARG]; - char *v, *argvp[ROFF_MAXARG]; + int i, 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); - } + /* + * 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 (ROFF_PRELUDE & tree->state) + 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)); } - i = 0; argv++; + assert( ! (ROFF_CALLABLE & tokens[tok].flags)); - while (-1 != (c = roffnextopt(tree, tok, &argv, &v))) { - if (ROFF_ARGMAX == c) - return(0); - - argcp[i] = c; - argvp[i] = v; - i++; - argv++; - } - - argcp[i] = ROFF_ARGMAX; - argvp[i] = NULL; - + if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) + return(0); if (NULL == roffnode_new(tok, tree)) return(0); - if ( ! (*tree->cb.roffblkin)(tree->arg, tok, argcp, argvp)) + /* + * 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, + (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)) + return(0); + + tree->asec |= tree->csec; + break; + default: + break; + } + + /* --- End run macro-specific hooks over argv. */ + if (NULL == *argv) - return(1); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, + (const char **)argvp)); - if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) + if ( ! (*tree->cb.roffblkheadin)(tree->arg, tok, argcp, + (const char **)argvp)) return(0); + /* + * If there are no parsable parts, then write remaining tokens + * into the layout header and exit. + */ + if ( ! (ROFF_PARSED & tokens[tok].flags)) { - while (*argv) { - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + i = 0; + while (*argv) + if ( ! roffdata(tree, i++, *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, + (const char **)argvp)); } + /* + * Parsable elements may be in the header (or be the header, for + * that matter). Follow the regular parsing rules for these. + */ + + i = 0; while (*argv) { - if (ROFF_MAX != (c = rofffindcallable(*argv))) { - if (NULL == tokens[c].cb) { - roff_err(tree, *argv, "unsupported " - "macro `%s'", - toknames[c]); + if (ROFF_MAX == (c = rofffindcallable(*argv))) { + assert(tree->arg); + if ( ! roffdata(tree, i++, *argv++)) return(0); - } - if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) - return(0); - break; + continue; } + if ( ! roffcall(tree, c, argv)) + return(0); + break; + } - assert(tree->arg); - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + /* + * If there's trailing punctuation in the header, then write it + * out now. Here we mimic the behaviour of a line-dominant text + * macro. + */ + + if (NULL == *argv) { + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) return(0); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, + (const char **)argvp)); } - return((*tree->cb.roffout)(tree->arg, tok)); + /* + * Expensive. Scan to the end of line then work backwards until + * a token isn't punctuation. + */ + + if ( ! roffpurgepunct(tree, argv)) + return(0); + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkbodyin)(tree->arg, + tok, argcp, (const char **)argvp)); } /* ARGSUSED */ static int -roff_text(ROFFCALL_ARGS) +roff_ordered(ROFFCALL_ARGS) { - int i, c, argcp[ROFF_MAXARG]; - char *v, *argvp[ROFF_MAXARG]; + 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]); + /* + * 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 (ROFF_PRELUDE & tree->state) + return(roff_errp(tree, *argv, tok, ERR_NOT_PR)); + + first = (*argv == tree->cur); + p = *argv++; + ordp[0] = NULL; + + 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; - argv++; + while (*argv && i < ROFF_MAXLINEARG) { + c = ROFF_PARSED & tokens[tok].flags ? + rofffindcallable(*argv) : ROFF_MAX; - while (-1 != (c = roffnextopt(tree, tok, &argv, &v))) { - if (ROFF_ARGMAX == c) + 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); - argcp[i] = c; - argvp[i] = v; - i++; - argv++; + return(roffcall(tree, c, argv)); } - argcp[i] = ROFF_ARGMAX; - argvp[i] = NULL; + assert(i != ROFF_MAXLINEARG); + ordp[i] = NULL; - if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) + 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_MAXLINEARG]; + char *argvp[ROFF_MAXLINEARG]; + + /* + * 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 (ROFF_PRELUDE & tree->state) + 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, + (const char **)argvp)) + return(0); + if (NULL == *argv) + return((*tree->cb.roffout)(tree->arg, tok)); + if ( ! (ROFF_PARSED & tokens[tok].flags)) { - while (*argv) { - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + i = 0; + while (*argv) + if ( ! roffdata(tree, i++, *argv++)) return(0); - } return((*tree->cb.roffout)(tree->arg, tok)); } + /* + * Deal with punctuation. Ugly. Work ahead until we encounter + * 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; while (*argv) { - if (ROFF_MAX == (c = rofffindcallable(*argv))) { - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + if (ROFF_MAX != (c = rofffindcallable(*argv))) { + if ( ! (ROFF_LSCOPE & tokens[tok].flags)) + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); + + if ( ! roffcall(tree, c, argv)) return(0); + if (ROFF_LSCOPE & tokens[tok].flags) + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); + break; + } + + if ( ! roffispunct(*argv)) { + if ( ! roffdata(tree, i++, *argv++)) + return(0); continue; } - /* - * A sub-command has been found. Execute it and - * discontinue parsing for arguments. - */ + 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 ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) - return(0); + 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; + } + + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); break; } - return((*tree->cb.roffout)(tree->arg, tok)); + if (NULL == *argv) + return((*tree->cb.roffout)(tree->arg, tok)); + if ( ! first) + return(1); + + return(roffpurgepunct(tree, argv)); } /* ARGSUSED */ static int -roff_comment(ROFFCALL_ARGS) +roff_noop(ROFFCALL_ARGS) { return(1); @@ -1093,23 +1543,33 @@ roff_comment(ROFFCALL_ARGS) /* ARGSUSED */ static int -roff_close(ROFFCALL_ARGS) +roff_depr(ROFFCALL_ARGS) { - return(1); + return(roff_errp(tree, *argv, tok, ERR_DEPREC)); } -/* ARGSUSED */ static int -roff_special(ROFFCALL_ARGS) +roff_warnp(const struct rofftree *tree, const char *pos, + int tok, enum rofferr type) { + char *p; - return((*tree->cb.roffspecial)(tree->arg, tok)); + 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 void +static int roff_warn(const struct rofftree *tree, const char *pos, char *fmt, ...) { va_list ap; @@ -1119,21 +1579,75 @@ 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)); } +