=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.20 retrieving revision 1.32 diff -u -p -r1.20 -r1.32 --- mandoc/roff.c 2008/11/30 12:41:45 1.20 +++ mandoc/roff.c 2008/12/02 00:15:41 1.32 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.20 2008/11/30 12:41:45 kristaps Exp $ */ +/* $Id: roff.c,v 1.32 2008/12/02 00:15:41 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -16,6 +16,8 @@ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR * PERFORMANCE OF THIS SOFTWARE. */ +#include + #include #include #include @@ -28,18 +30,14 @@ #include "libmdocml.h" #include "private.h" -/* FIXME: warn if Pp occurs before/after Sh etc. (see mdoc.samples). */ +/* 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? */ -/* 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 { @@ -63,15 +61,15 @@ struct rofftree; struct rofftok { int (*cb)(ROFFCALL_ARGS); /* Callback. */ const int *args; /* Args (or NULL). */ - const int *parents; - const int *children; - int ctx; + 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_QUOTES (1 << 2) /* Quoted args. */ -#define ROFF_SHALLOW (1 << 3) /* Nesting block. */ +#define ROFF_SHALLOW (1 << 2) /* Nesting block. */ +#define ROFF_LSCOPE (1 << 3) /* Line scope. */ }; struct roffarg { @@ -86,43 +84,37 @@ struct roffnode { 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 os[64]; /* `Os' results. */ char title[64]; /* `Dt' results. */ char section[64]; /* `Dt' results. */ char volume[64]; /* `Dt' results. */ - int state; #define ROFF_PRELUDE (1 << 1) /* In roff prelude. */ #define ROFF_PRELUDE_Os (1 << 2) /* `Os' is parsed. */ #define ROFF_PRELUDE_Dt (1 << 3) /* `Dt' is parsed. */ #define ROFF_PRELUDE_Dd (1 << 4) /* `Dd' is parsed. */ #define ROFF_BODY (1 << 5) /* In roff body. */ - - struct roffcb cb; - void *arg; + struct roffcb cb; /* Callbacks. */ + void *arg; /* Callbacks' arg. */ }; static int roff_Dd(ROFFCALL_ARGS); static int roff_Dt(ROFFCALL_ARGS); static int roff_Os(ROFFCALL_ARGS); - +static int roff_Ns(ROFFCALL_ARGS); static int roff_layout(ROFFCALL_ARGS); static int roff_text(ROFFCALL_ARGS); -static int roff_comment(ROFFCALL_ARGS); -static int roff_close(ROFFCALL_ARGS); -static int roff_special(ROFFCALL_ARGS); - +static int roff_noop(ROFFCALL_ARGS); +static int roff_depr(ROFFCALL_ARGS); static struct roffnode *roffnode_new(int, struct rofftree *); static void roffnode_free(struct rofftree *); - static void roff_warn(const struct rofftree *, const char *, char *, ...); static void roff_err(const struct rofftree *, const char *, char *, ...); - +static int roffpurgepunct(struct rofftree *, char **); static int roffscan(int, const int *); static int rofffindtok(const char *); static int rofffindarg(const char *); @@ -132,68 +124,63 @@ static int roffargs(const struct rofftree *, 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 roffparse(struct rofftree *, char *); static int textparse(const struct rofftree *, char *); -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_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, +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 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_Oc[] = { ROFF_Oo, ROFF_MAX }; static const int roffparent_It[] = { ROFF_Bl, ROFF_It, ROFF_MAX }; static const int roffparent_Re[] = { ROFF_Rs, ROFF_MAX }; -/* Table of all known tokens. */ static const struct rofftok tokens[ROFF_MAX] = { - {roff_comment, NULL, NULL, NULL, 0, ROFF_COMMENT, 0 }, /* \" */ + { 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, 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_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_close, NULL, NULL, NULL, ROFF_Bd, ROFF_LAYOUT, 0 }, /* Ed */ + { roff_noop, 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_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, ROFF_QUOTES }, /* Cd */ /* XXX man.4 only */ + { 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 */ @@ -207,10 +194,10 @@ static const struct rofftok tokens[ROFF_MAX] = { { 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, 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 }, /* Op */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ot */ /* XXX deprecated */ + { 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 */ @@ -230,42 +217,42 @@ static const struct rofftok tokens[ROFF_MAX] = { { 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, 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Bf */ /* FIXME */ + { 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 }, /* Bq */ + { 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 */ - {roff_special, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Db */ + { 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 }, /* Dq */ + { 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ef */ /* FIXME */ + { 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ns */ + { 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 */ - { 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 */ + { 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_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_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 }, /* Sq */ - {roff_special, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Sm */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE | ROFF_LSCOPE }, /* Sq */ + { NULL, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Sm */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sx */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sy */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Tn */ @@ -273,14 +260,17 @@ static const struct rofftok tokens[ROFF_MAX] = { { 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_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_close, NULL, NULL, NULL, ROFF_Bk, ROFF_LAYOUT, 0 }, /* Ek */ + { roff_noop, NULL, NULL, NULL, ROFF_Bk, ROFF_LAYOUT, 0 }, /* Ek */ + { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Bt */ + { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Hf */ + { roff_depr, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Fr */ + { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ud */ }; -/* Table of all known token arguments. */ static const int tokenargs[ROFF_ARGMAX] = { 0, 0, 0, 0, 0, ROFF_VALUE, ROFF_VALUE, 0, @@ -310,8 +300,11 @@ const char *const toknamesp[ROFF_MAX] = { "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", @@ -325,7 +318,8 @@ const char *const toknamesp[ROFF_MAX] = { "Sq", "Sm", "Sx", "Sy", "Tn", "Ux", "Xc", "Xo", "Fo", "Fc", "Oo", "Oc", - "Bk", "Ek", + "Bk", "Ek", "Bt", "Hf", + "Fr", "Ud", }; const char *const tokargnamesp[ROFF_ARGMAX] = { @@ -369,14 +363,14 @@ roff_free(struct rofftree *tree, int flush) error = 1; if (ROFF_PRELUDE & tree->state) { - roff_warn(tree, NULL, "prelude never finished"); + roff_err(tree, NULL, "prelude never finished"); goto end; } - 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'", + roff_err(tree, NULL, "closing explicit scope `%s'", toknames[n->tok]); goto end; } @@ -387,6 +381,9 @@ roff_free(struct rofftree *tree, int flush) goto end; } + if ( ! (*tree->cb.rofftail)(tree->arg)) + goto end; + error = 0; end: @@ -428,7 +425,7 @@ roff_engine(struct rofftree *tree, char *buf) assert(buf); if (0 == *buf) { - roff_warn(tree, buf, "blank line"); + roff_err(tree, buf, "blank line"); return(0); } else if ('.' != *buf) return(textparse(tree, buf)); @@ -441,7 +438,11 @@ static int textparse(const struct rofftree *tree, char *buf) { - return((*tree->cb.roffdata)(tree->arg, buf)); + if ( ! (ROFF_BODY & tree->state)) { + roff_err(tree, buf, "data not in body"); + return(0); + } + return((*tree->cb.roffdata)(tree->arg, 1, buf)); } @@ -494,7 +495,6 @@ roffargs(const struct rofftree *tree, } -/* XXX */ static int roffscan(int tok, const int *tokv) { @@ -518,6 +518,10 @@ roffparse(struct rofftree *tree, char *buf) char *argv[ROFF_MAXARG]; char **argvp; + if (0 != *buf && 0 != *(buf + 1) && 0 != *(buf + 2)) + if (0 == strncmp(buf, ".\\\"", 3)) + return(1); + if (ROFF_MAX == (tok = rofffindtok(buf + 1))) { roff_err(tree, buf + 1, "bogus line macro"); return(0); @@ -525,9 +529,9 @@ roffparse(struct rofftree *tree, char *buf) roff_err(tree, buf + 1, "unsupported macro `%s'", toknames[tok]); return(0); - } else if (ROFF_COMMENT == tokens[tok].type) - return(1); - + } + + assert(ROFF___ != tok); if ( ! roffargs(tree, tok, buf, argv)) return(0); @@ -540,8 +544,7 @@ 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); + } assert(ROFF_BODY & tree->state); @@ -550,14 +553,16 @@ roffparse(struct rofftree *tree, char *buf) * children are satisfied. */ - if ( ! roffscan(tree->last->tok, tokens[tok].parents)) { + 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); } - if ( ! roffscan(tok, tokens[tree->last->tok].children)) { + 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]); @@ -570,11 +575,6 @@ roffparse(struct rofftree *tree, char *buf) if (ROFF_LAYOUT != tokens[tok].type) return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); - - /* - * Check our scope rules. - */ - if (0 == tokens[tok].ctx) return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); @@ -605,6 +605,11 @@ roffparse(struct rofftree *tree, char *buf) n = NULL; break; } + if (tokens[n->tok].ctx == n->tok) + continue; + roff_err(tree, *argv, "`%s' breaks `%s' scope", + toknames[tok], toknames[n->tok]); + return(0); } /* @@ -638,17 +643,34 @@ 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; + roff_err(tree, *argv, "`%s' breaks `%s' scope", + toknames[tok], toknames[n->tok]); + return(0); + } else + break; + + + 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)) return(0); } while (t != tokens[tok].ctx); - assert(tree->last); return(1); } @@ -674,7 +696,7 @@ 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; @@ -696,7 +718,6 @@ rofffindtok(const char *buf) } -#if notyet static int roffispunct(const char *p) { @@ -737,7 +758,6 @@ roffispunct(const char *p) return(0); } -#endif static int @@ -840,10 +860,60 @@ 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 ( ! (*tree->cb.roffdata)(tree->arg, 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); +} + + /* ARGSUSED */ static int roff_Dd(ROFFCALL_ARGS) { + time_t t; + char *p, buf[32]; if (ROFF_BODY & tree->state) { assert( ! (ROFF_PRELUDE & tree->state)); @@ -862,11 +932,63 @@ roff_Dd(ROFFCALL_ARGS) return(0); } - /* TODO: parse date. */ - assert(NULL == tree->last); - tree->state |= ROFF_PRELUDE_Dd; + argv++; + + if (0 == strcmp(*argv, "$Mdocdate: December 2 2008 $")) { + t = time(NULL); + if (NULL == localtime_r(&t, &tree->tm)) + err(1, "localtime_r"); + tree->state |= ROFF_PRELUDE_Dd; + return(1); + } + + /* Build this from Mdocdate or raw date. */ + + buf[0] = 0; + p = *argv; + + if (0 != strcmp(*argv, "$Mdocdate:")) { + while (*argv) { + if (strlcat(buf, *argv++, sizeof(buf)) + < sizeof(buf)) + continue; + roff_err(tree, p, "bad `Dd' date"); + return(0); + } + if (strptime(buf, "%b%d,%Y", &tree->tm)) { + tree->state |= ROFF_PRELUDE_Dd; + return(1); + } + roff_err(tree, *argv, "bad `Dd' date"); + return(0); + } + + 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 (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); + } + + tree->state |= ROFF_PRELUDE_Dd; return(1); } @@ -893,8 +1015,35 @@ roff_Dt(ROFFCALL_ARGS) return(0); } - /* TODO: parse date. */ + 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); + } + 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); + } + + 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); + } + assert(NULL == tree->last); tree->state |= ROFF_PRELUDE_Dt; @@ -904,13 +1053,84 @@ roff_Dt(ROFFCALL_ARGS) /* ARGSUSED */ static int +roff_Ns(ROFFCALL_ARGS) +{ + int j, c, first; + + first = (*argv == tree->cur); + + argv++; + + if (ROFF_MAX != (c = rofffindcallable(*argv))) { + if (NULL == tokens[c].cb) { + roff_err(tree, *argv, "unsupported macro `%s'", + toknames[c]); + return(0); + } + if ( ! (*tree->cb.roffspecial)(tree->arg, tok)) + return(0); + if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) + return(0); + if ( ! first) + return(1); + return(roffpurgepunct(tree, argv)); + } + + if ( ! (*tree->cb.roffdata)(tree->arg, 0, *argv++)) + return(0); + + while (*argv) { + if (ROFF_MAX == (c = rofffindcallable(*argv))) { + if ( ! roffispunct(*argv)) { + if ( ! (*tree->cb.roffdata) + (tree->arg, 1, *argv++)) + return(0); + continue; + } + + /* FIXME: this is identical to that of + * roff_text. */ + + /* See if only punctuation remains. */ + + for (j = 0; argv[j]; j++) + if ( ! roffispunct(argv[j])) + break; + + if (argv[j]) { + if ( ! (*tree->cb.roffdata) + (tree->arg, 0, *argv++)) + return(0); + continue; + } + + /* Only punctuation remains. */ + + 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); + break; + } + + if ( ! first) + return(1); + return(roffpurgepunct(tree, argv)); +} + + +/* ARGSUSED */ +static int roff_Os(ROFFCALL_ARGS) { + char *p; - 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)); @@ -923,17 +1143,31 @@ roff_Os(ROFFCALL_ARGS) return(0); } - /* TODO: extract OS. */ + tree->os[0] = 0; + p = *++argv; + + 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); + } + tree->state |= ROFF_PRELUDE_Os; tree->state &= ~ROFF_PRELUDE; tree->state |= ROFF_BODY; assert(NULL == tree->last); - if (NULL == roffnode_new(tok, tree)) - return(0); - return((*tree->cb.roffhead)(tree->arg)); } @@ -943,73 +1177,99 @@ static int roff_layout(ROFFCALL_ARGS) { int i, c, argcp[ROFF_MAXARG]; - char *v, *argvp[ROFF_MAXARG]; + char *argvp[ROFF_MAXARG]; if (ROFF_PRELUDE & tree->state) { - roff_err(tree, *argv, "`%s' disallowed in prelude", + roff_err(tree, *argv, "bad `%s' in prelude", toknames[tok]); return(0); - } - - if (ROFF_EXIT == type) { + } else if (ROFF_EXIT == type) { roffnode_free(tree); 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); + ++argv; - 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); + /* + * 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. + */ + if ( ! (*tree->cb.roffblkin)(tree->arg, tok, argcp, argvp)) return(0); - if (NULL == *argv) return(1); - if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) return(0); + /* + * If there are no parsable parts, then write remaining tokens + * into the layout header and exit. + */ + if ( ! (ROFF_PARSED & tokens[tok].flags)) { + i = 0; while (*argv) { - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++)) return(0); + i = 1; } return((*tree->cb.roffout)(tree->arg, tok)); } + /* + * 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 ( ! (*tree->cb.roffdata) + (tree->arg, i, *argv++)) return(0); - } - if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) - return(0); - break; + i = 1; + continue; } - assert(tree->arg); - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + 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); + + break; } + /* + * 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) + 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); + return((*tree->cb.roffout)(tree->arg, tok)); } @@ -1018,8 +1278,8 @@ roff_layout(ROFFCALL_ARGS) static int roff_text(ROFFCALL_ARGS) { - int i, c, argcp[ROFF_MAXARG]; - char *v, *argvp[ROFF_MAXARG]; + int i, j, first, c, argcp[ROFF_MAXARG]; + char *argvp[ROFF_MAXARG]; if (ROFF_PRELUDE & tree->state) { roff_err(tree, *argv, "`%s' disallowed in prelude", @@ -1027,43 +1287,70 @@ roff_text(ROFFCALL_ARGS) return(0); } - i = 0; + first = (*argv == tree->cur); argv++; - 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 ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) return(0); + if (NULL == *argv) + return((*tree->cb.roffout)(tree->arg, tok)); if ( ! (ROFF_PARSED & tokens[tok].flags)) { + i = 0; while (*argv) { - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++)) return(0); + i = 1; } 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). + */ + + i = 0; while (*argv) { if (ROFF_MAX == (c = rofffindcallable(*argv))) { - if ( ! (*tree->cb.roffdata)(tree->arg, *argv++)) + if ( ! roffispunct(*argv)) { + if ( ! (*tree->cb.roffdata) + (tree->arg, i, *argv++)) + return(0); + i = 1; + continue; + } + + /* See if only punctuation remains. */ + + i = 1; + for (j = 0; argv[j]; j++) + if ( ! roffispunct(argv[j])) + break; + + if (argv[j]) { + if ( ! (*tree->cb.roffdata) + (tree->arg, 0, *argv++)) + return(0); + continue; + } + + /* Only punctuation remains. */ + + if ( ! (*tree->cb.roffout)(tree->arg, tok)) return(0); - continue; + break; } /* * A sub-command has been found. Execute it and - * discontinue parsing for arguments. + * discontinue parsing for arguments. If we're + * line-scoped, then close out after it returns; if we + * aren't, then close out before. */ if (NULL == tokens[c].cb) { @@ -1072,28 +1359,32 @@ roff_text(ROFFCALL_ARGS) return(0); } + if ( ! (ROFF_LSCOPE & tokens[tok].flags)) + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); + if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER)) return(0); + if (ROFF_LSCOPE & tokens[tok].flags) + 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); - -/* ARGSUSED */ -static int -roff_comment(ROFFCALL_ARGS) -{ - - return(1); + return(roffpurgepunct(tree, argv)); } /* ARGSUSED */ static int -roff_close(ROFFCALL_ARGS) +roff_noop(ROFFCALL_ARGS) { return(1); @@ -1102,10 +1393,11 @@ roff_close(ROFFCALL_ARGS) /* ARGSUSED */ static int -roff_special(ROFFCALL_ARGS) +roff_depr(ROFFCALL_ARGS) { - return((*tree->cb.roffspecial)(tree->arg, tok)); + roff_err(tree, *argv, "`%s' is deprecated", toknames[tok]); + return(0); }