=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.8 retrieving revision 1.13 diff -u -p -r1.8 -r1.13 --- mandoc/roff.c 2008/11/26 22:27:08 1.8 +++ mandoc/roff.c 2008/11/27 17:27:50 1.13 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.8 2008/11/26 22:27:08 kristaps Exp $ */ +/* $Id: roff.c,v 1.13 2008/11/27 17:27:50 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -19,6 +19,7 @@ #include #include #include +#include #include #include #include @@ -35,8 +36,10 @@ /* FIXME: ; : } ) (etc.) after text macros? */ -#define ROFF_MAXARG 10 +/* FIXME: NAME section needs specific elements. */ +#define ROFF_MAXARG 32 + enum roffd { ROFF_ENTER = 0, ROFF_EXIT @@ -45,7 +48,8 @@ enum roffd { enum rofftype { ROFF_COMMENT, ROFF_TEXT, - ROFF_LAYOUT + ROFF_LAYOUT, + ROFF_SPECIAL }; #define ROFFCALL_ARGS \ @@ -65,6 +69,7 @@ struct rofftok { #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 { @@ -81,6 +86,7 @@ struct roffnode { struct rofftree { struct roffnode *last; /* Last parsed node. */ time_t date; /* `Dd' results. */ + char *cur; char os[64]; /* `Os' results. */ char title[64]; /* `Dt' results. */ char section[64]; /* `Dt' results. */ @@ -105,17 +111,26 @@ 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(int, struct rofftree *); +static void roff_warn(const struct rofftree *, + const char *, char *, ...); +static void roff_err(const struct rofftree *, + const char *, char *, ...); + static int roffscan(int, const int *); static int rofffindtok(const char *); static int rofffindarg(const char *); static int rofffindcallable(const char *); -static int roffargs(int, char *, char **); +static int roffargs(const struct rofftree *, + int, char *, char **); static int roffargok(int, int); -static int roffnextopt(int, const char ***, char **); +static int roffnextopt(const struct rofftree *, + int, const char ***, char **); static int roffparse(struct rofftree *, char *, size_t); static int textparse(const struct rofftree *, const char *, size_t); @@ -124,13 +139,15 @@ static int textparse(const struct rofftree *, 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_ARGMAX }; + 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_ARGMAX }; + 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, @@ -144,33 +161,36 @@ static const int roffarg_St[] = { ROFF_xcurses4_2, ROFF_susv2, ROFF_susv3, ROFF_svid4, ROFF_ARGMAX }; -static const int roffchild_Bd[] = { ROFF_Ed, ROFF_MAX }; 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_Ed[] = { ROFF_Bd, 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, 0 }, /* Os */ +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, NULL, NULL, roffchild_Bd, 0, ROFF_LAYOUT, 0 }, /* Bd */ - { roff_layout, NULL, roffparent_Ed, NULL, ROFF_Bd, ROFF_LAYOUT, 0 }, /* Ed */ + { 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_layout, NULL, roffparent_El, NULL, ROFF_Bl, ROFF_LAYOUT, 0 }, /* El */ - { roff_layout, NULL, roffparent_It, NULL, ROFF_It, ROFF_LAYOUT, 0 }, /* It */ + { 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 */ { 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 */ @@ -181,107 +201,105 @@ static const struct rofftok tokens[ROFF_MAX] = { { 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, 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, 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 */ /* FIXME */ + { 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_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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* %A */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %B */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %D */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %I */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %J */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %N */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %O */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %P */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %R */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* %T */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %V */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ac */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ao */ - { NULL, 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bc */ + { 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bo */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bo */ { NULL, 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Db */ /* XXX */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dc */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Do */ + {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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ec */ + { 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Eo */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Eo */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Fx */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ms */ + { 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pc */ + { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pc */ { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Pf */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Po */ + { 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 | ROFF_CALLABLE }, /* Qc */ + { 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qo */ + { 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Re */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Rs */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sc */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* So */ + { 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 */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Sm */ + {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 | ROFF_QUOTES }, /* Sy */ + { 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, NULL, 0, ROFF_LAYOUT, 0 }, /* Fo */ - { roff_layout, NULL, roffparent_Fc, NULL, ROFF_Fo, ROFF_LAYOUT, 0 }, /* Fc */ + { roff_close, NULL, roffparent_Fc, NULL, ROFF_Fo, ROFF_LAYOUT, 0 }, /* Fc */ { roff_layout, NULL, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Oo */ - { roff_layout, NULL, roffparent_Oc, NULL, ROFF_Oo, ROFF_LAYOUT, 0 }, /* Oc */ -}; + { 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 struct roffarg tokenargs[ROFF_ARGMAX] = { - { 0 }, /* split */ - { 0 }, /* nosplit */ - { 0 }, /* ragged */ - { 0 }, /* unfilled */ - { 0 }, /* literal */ - { ROFF_VALUE }, /* file */ - { ROFF_VALUE }, /* offset */ - { 0 }, /* bullet */ - { 0 }, /* dash */ - { 0 }, /* hyphen */ - { 0 }, /* item */ - { 0 }, /* enum */ - { 0 }, /* tag */ - { 0 }, /* diag */ - { 0 }, /* hang */ - { 0 }, /* ohang */ - { 0 }, /* inset */ - { 0 }, /* column */ - { 0 }, /* width */ - { 0 }, /* compact */ -}; +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, 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, 0, + }; const char *const toknamesp[ROFF_MAX] = { @@ -309,33 +327,35 @@ const char *const toknamesp[ROFF_MAX] = "Re", "Rs", "Sc", "So", "Sq", "Sm", "Sx", "Sy", "Tn", "Ux", "Xc", "Xo", - "Fo", "Fc", "Oo", "Oc" + "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" + "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; @@ -359,8 +379,7 @@ roff_free(struct rofftree *tree, int flush) error = tree->mbuf ? 0 : 1; if (tree->mbuf && (ROFF_PRELUDE & tree->state)) { - warnx("%s: prelude never finished", - tree->rbuf->name); + /*roff_warn(tree, "prelude never finished");*/ error = 1; } @@ -375,10 +394,8 @@ roff_alloc(const struct md_args *args, struct md_mbuf { struct rofftree *tree; - if (NULL == (tree = calloc(1, sizeof(struct rofftree)))) { - warn("malloc"); - return(NULL); - } + if (NULL == (tree = calloc(1, sizeof(struct rofftree)))) + err(1, "calloc"); tree->state = ROFF_PRELUDE; tree->args = args; @@ -394,10 +411,10 @@ int roff_engine(struct rofftree *tree, char *buf, size_t sz) { + tree->cur = buf; + if (0 == sz) { - warnx("%s: blank line (line %zu)", - tree->rbuf->name, - tree->rbuf->line); + roff_warn(tree, buf, "blank line"); return(0); } else if ('.' != *buf) return(textparse(tree, buf, sz)); @@ -410,41 +427,42 @@ static int textparse(const struct rofftree *tree, const char *buf, size_t sz) { - if (NULL == tree->last) { - warnx("%s: unexpected text (line %zu)", - tree->rbuf->name, - tree->rbuf->line); - return(0); - } else if (NULL == tree->last->parent) { - warnx("%s: disallowed text (line %zu)", - tree->rbuf->name, - tree->rbuf->line); - return(0); - } - /* Print text. */ - return(1); } static int -roffargs(int tok, char *buf, char **argv) +roffargs(const struct rofftree *tree, + int tok, char *buf, char **argv) { int i; + char *p; - (void)tok;/* FIXME: quotable strings? */ - assert(tok >= 0 && tok < ROFF_MAX); assert('.' == *buf); + p = buf; + /* LINTED */ for (i = 0; *buf && i < ROFF_MAXARG; i++) { - argv[i] = buf++; - while (*buf && ! isspace(*buf)) - buf++; - if (0 == *buf) { - continue; + 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); + } + } else { + argv[i] = buf++; + while (*buf && ! isspace(*buf)) + buf++; + if (0 == *buf) + continue; } *buf++ = 0; while (*buf && isspace(*buf)) @@ -452,10 +470,14 @@ roffargs(int tok, char *buf, char **argv) } assert(i > 0); - if (i < ROFF_MAXARG) - argv[i] = NULL; + if (ROFF_MAXARG == i && *buf) { + roff_err(tree, p, "too many arguments for `%s'", toknames + [tok]); + return(0); + } - return(ROFF_MAXARG > i); + argv[i] = NULL; + return(1); } @@ -486,27 +508,22 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) assert(sz > 0); if (ROFF_MAX == (tok = rofffindtok(buf + 1))) { - warnx("%s: unknown line macro (line %zu)", - tree->rbuf->name, tree->rbuf->line); + roff_err(tree, buf + 1, "bogus line macro"); return(0); } else if (NULL == tokens[tok].cb) { - warnx("%s: macro `%s' not supported (line %zu)", - tree->rbuf->name, toknames[tok], - tree->rbuf->line); + roff_err(tree, buf + 1, "unsupported macro `%s'", + toknames[tok]); return(0); } else if (ROFF_COMMENT == tokens[tok].type) return(1); - if ( ! roffargs(tok, buf, argv)) { - warnx("%s: too many args to `%s' (line %zu)", - tree->rbuf->name, toknames[tok], - tree->rbuf->line); + if ( ! roffargs(tree, tok, buf, argv)) return(0); - } else - argvp = (const char **)argv + 1; + argvp = (const char **)argv; + /* - * Prelude macros break some assumptions: branch now. + * Prelude macros break some assumptions, so branch now. */ if (ROFF_PRELUDE & tree->state) { @@ -552,26 +569,50 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) if (0 == tokens[tok].ctx) return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + /* + * First consider implicit-end tags, like as follows: + * .Sh SECTION 1 + * .Sh SECTION 2 + * In this, we want to close the scope of the NAME section. If + * there's an intermediary implicit-end tag, such as + * .Sh SECTION 1 + * .Ss Subsection 1 + * .Sh SECTION 2 + * then it must be closed as well. + */ + if (tok == tokens[tok].ctx) { + /* + * First search up to the point where we must close. + * If one doesn't exist, then we can open a new scope. + */ + for (n = tree->last; n; n = n->parent) { assert(0 == tokens[n->tok].ctx || n->tok == tokens[n->tok].ctx); if (n->tok == tok) break; + if (ROFF_SHALLOW & tokens[tok].flags) { + n = NULL; + break; + } } - if (NULL == n) { -#ifdef DEBUG - (void)printf("scope: new `%s'\n", - toknames[tok]); -#endif + + /* + * Create a new scope, as no previous one exists to + * close out. + */ + + if (NULL == n) return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); - } + + /* + * Close out all intermediary scoped blocks, then hang + * the current scope from our predecessor's parent. + */ + do { t = tree->last->tok; -#ifdef DEBUG - (void)printf("scope: closing `%s'\n", - toknames[t]); -#endif if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) return(0); } while (t != tok); @@ -579,18 +620,26 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); } + /* + * Now consider explicit-end tags, where we want to close back + * to a specific tag. Example: + * .Bl + * .It Item. + * .El + * In this, the `El' tag closes out the scope of `Bl'. + */ + + assert(tree->last); assert(tok != tokens[tok].ctx && 0 != tokens[tok].ctx); do { t = tree->last->tok; -#ifdef DEBUG - (void)printf("scope: closing `%s'\n", toknames[t]); -#endif if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) return(0); } while (t != tokens[tok].ctx); - return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + assert(tree->last); + return(1); } @@ -620,12 +669,8 @@ rofffindtok(const char *buf) for (i = 0; *buf && ! isspace(*buf) && i < 3; i++, buf++) token[i] = *buf; - if (i == 3) { -#ifdef DEBUG - (void)printf("lookup: macro too long: `%s'\n", buf); -#endif + if (i == 3) return(ROFF_MAX); - } token[i] = 0; @@ -634,18 +679,9 @@ rofffindtok(const char *buf) /* LINTED */ for (i = 0; i < ROFF_MAX; i++) /* LINTED */ - if (0 == strcmp(toknames[i], token)) { -#ifdef DEBUG - (void)printf("lookup (good): `%s' (%d)\n", - token, (int)i); -#endif + if (0 == strcmp(toknames[i], token)) return((int)i); - } -#ifdef DEBUG - (void)printf("lookup (bad): `%s'\n", token); -#endif - return(ROFF_MAX); } @@ -667,15 +703,14 @@ roffnode_new(int tokid, struct rofftree *tree) { struct roffnode *p; - if (NULL == (p = malloc(sizeof(struct roffnode)))) { - warn("malloc"); - return(NULL); - } + if (NULL == (p = malloc(sizeof(struct roffnode)))) + err(1, "malloc"); p->line = tree->rbuf->line; p->tok = tokid; p->parent = tree->last; tree->last = p; + return(p); } @@ -711,7 +746,8 @@ roffnode_free(int tokid, struct rofftree *tree) static int -roffnextopt(int tok, const char ***in, char **val) +roffnextopt(const struct rofftree *tree, int tok, + const char ***in, char **val) { const char *arg, **argv; int v; @@ -725,23 +761,30 @@ roffnextopt(int tok, const char ***in, char **val) if ('-' != *arg) return(-1); - /* FIXME: should we let this slide... ? */ - - if (ROFF_ARGMAX == (v = rofffindarg(&arg[1]))) + if (ROFF_ARGMAX == (v = rofffindarg(arg + 1))) { + roff_warn(tree, arg, "argument-like parameter `%s' to " + "`%s'", &arg[1], toknames[tok]); return(-1); - - /* FIXME: should we let this slide... ? */ - - if ( ! roffargok(tok, v)) + } + + if ( ! roffargok(tok, v)) { + roff_warn(tree, arg, "invalid argument parameter `%s' to " + "`%s'", tokargnames[v], toknames[tok]); return(-1); - if ( ! (ROFF_VALUE & tokenargs[v].flags)) + } + + if ( ! (ROFF_VALUE & tokenargs[v])) return(v); *in = ++argv; - /* FIXME: what if this looks like a roff token or argument? */ + if (NULL == *argv) { + roff_err(tree, arg, "empty value of `%s' for `%s'", + tokargnames[v], toknames[tok]); + return(ROFF_ARGMAX); + } - return(*argv ? v : ROFF_ARGMAX); + return(v); } @@ -760,12 +803,10 @@ roff_Dd(ROFFCALL_ARGS) assert( ! (ROFF_BODY & tree->state)); if (ROFF_PRELUDE_Dd & tree->state) { - warnx("%s: prelude `Dd' repeated (line %zu)", - tree->rbuf->name, tree->rbuf->line); + roff_err(tree, *argv, "repeated `Dd' in prelude"); return(0); } else if (ROFF_PRELUDE_Dt & tree->state) { - warnx("%s: prelude `Dd' out-of-order (line %zu)", - tree->rbuf->name, tree->rbuf->line); + roff_err(tree, *argv, "out-of-order `Dd' in prelude"); return(0); } @@ -793,12 +834,10 @@ roff_Dt(ROFFCALL_ARGS) assert( ! (ROFF_BODY & tree->state)); if ( ! (ROFF_PRELUDE_Dd & tree->state)) { - warnx("%s: prelude `Dt' out-of-order (line %zu)", - tree->rbuf->name, tree->rbuf->line); + roff_err(tree, *argv, "out-of-order `Dt' in prelude"); return(0); } else if (ROFF_PRELUDE_Dt & tree->state) { - warnx("%s: prelude `Dt' repeated (line %zu)", - tree->rbuf->name, tree->rbuf->line); + roff_err(tree, *argv, "repeated `Dt' in prelude"); return(0); } @@ -828,8 +867,7 @@ roff_Os(ROFFCALL_ARGS) assert(ROFF_PRELUDE & tree->state); if ( ! (ROFF_PRELUDE_Dt & tree->state) || ! (ROFF_PRELUDE_Dd & tree->state)) { - warnx("%s: prelude `Os' out-of-order (line %zu)", - tree->rbuf->name, tree->rbuf->line); + roff_err(tree, *argv, "out-of-order `Os' in prelude"); return(0); } @@ -853,35 +891,23 @@ roff_layout(ROFFCALL_ARGS) char *v, *argvp[ROFF_MAXARG]; if (ROFF_PRELUDE & tree->state) { - warnx("%s: macro `%s' called in prelude (line %zu)", - tree->rbuf->name, - toknames[tok], - tree->rbuf->line); + roff_err(tree, *argv, "`%s' disallowed in prelude", + toknames[tok]); return(0); } if (ROFF_EXIT == type) { roffnode_free(tok, tree); - return((*tree->cb->roffblkout)(tok)); + return((*tree->cb->roffblkout)(tree->args, tok)); } i = 0; + argv++; - while (-1 != (c = roffnextopt(tok, &argv, &v))) { - if (ROFF_ARGMAX == c) { - warnx("%s: error parsing `%s' args (line %zu)", - tree->rbuf->name, - toknames[tok], - tree->rbuf->line); + while (-1 != (c = roffnextopt(tree, tok, &argv, &v))) { + if (ROFF_ARGMAX == c) return(0); - } else if ( ! roffargok(tok, c)) { - warnx("%s: arg `%s' not for `%s' (line %zu)", - tree->rbuf->name, - tokargnames[c], - toknames[tok], - tree->rbuf->line); - return(0); - } + argcp[i] = c; argvp[i] = v; i++; @@ -894,39 +920,40 @@ roff_layout(ROFFCALL_ARGS) if (NULL == roffnode_new(tok, tree)) return(0); - if ( ! (*tree->cb->roffin)(tok, argcp, argvp)) + if ( ! (*tree->cb->roffin)(tree->args, tok, argcp, argvp)) return(0); if ( ! (ROFF_PARSED & tokens[tok].flags)) { + /* TODO: print all tokens. */ - if ( ! ((*tree->cb->roffout)(tok))) + if ( ! ((*tree->cb->roffout)(tree->args, tok))) return(0); - return((*tree->cb->roffblkin)(tok)); + return((*tree->cb->roffblkin)(tree->args, tok)); } while (*argv) { if (ROFF_MAX != (c = rofffindcallable(*argv))) { if (NULL == tokens[c].cb) { - warnx("%s: macro `%s' not supported " - "(line %zu)", - tree->rbuf->name, - toknames[c], - tree->rbuf->line); + roff_err(tree, *argv, "unsupported " + "macro `%s'", + toknames[c]); return(0); } if ( ! (*tokens[c].cb)(c, tree, - argv + 1, ROFF_ENTER)) + argv, ROFF_ENTER)) return(0); } + /* TODO: print token. */ + argv++; } - if ( ! ((*tree->cb->roffout)(tok))) + if ( ! ((*tree->cb->roffout)(tree->args, tok))) return(0); - return((*tree->cb->roffblkin)(tok)); + return((*tree->cb->roffblkin)(tree->args, tok)); } @@ -938,23 +965,18 @@ roff_text(ROFFCALL_ARGS) char *v, *argvp[ROFF_MAXARG]; if (ROFF_PRELUDE & tree->state) { - warnx("%s: macro `%s' called in prelude (line %zu)", - tree->rbuf->name, - toknames[tok], - tree->rbuf->line); + roff_err(tree, *argv, "`%s' disallowed in prelude", + toknames[tok]); return(0); } i = 0; + argv++; - while (-1 != (c = roffnextopt(tok, &argv, &v))) { - if (ROFF_ARGMAX == c) { - warnx("%s: error parsing `%s' args (line %zu)", - tree->rbuf->name, - toknames[tok], - tree->rbuf->line); + while (-1 != (c = roffnextopt(tree, tok, &argv, &v))) { + if (ROFF_ARGMAX == c) return(0); - } + argcp[i] = c; argvp[i] = v; i++; @@ -964,40 +986,90 @@ roff_text(ROFFCALL_ARGS) argcp[i] = ROFF_ARGMAX; argvp[i] = NULL; - if ( ! (*tree->cb->roffin)(tok, argcp, argvp)) + if ( ! (*tree->cb->roffin)(tree->args, tok, argcp, argvp)) return(0); if ( ! (ROFF_PARSED & tokens[tok].flags)) { + /* TODO: print all tokens. */ - return((*tree->cb->roffout)(tok)); + + return((*tree->cb->roffout)(tree->args, tok)); } while (*argv) { if (ROFF_MAX != (c = rofffindcallable(*argv))) { if (NULL == tokens[c].cb) { - warnx("%s: macro `%s' not supported " - "(line %zu)", - tree->rbuf->name, - toknames[c], - tree->rbuf->line); + roff_err(tree, *argv, "unsupported " + "macro `%s'", + toknames[c]); return(0); } if ( ! (*tokens[c].cb)(c, tree, - argv + 1, ROFF_ENTER)) + argv, ROFF_ENTER)) return(0); } + /* TODO: print token. */ + argv++; } - return((*tree->cb->roffout)(tok)); + return((*tree->cb->roffout)(tree->args, tok)); } -/* ARGUSED */ +/* ARGSUSED */ static int roff_comment(ROFFCALL_ARGS) { return(1); +} + + +/* ARGSUSED */ +static int +roff_close(ROFFCALL_ARGS) +{ + + return(1); +} + + +/* ARGSUSED */ +static int +roff_special(ROFFCALL_ARGS) +{ + + return((*tree->cb->roffspecial)(tok)); +} + + +static void +roff_warn(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); + va_end(ap); + + (*tree->cb->roffmsg)(tree->args, ROFF_WARN, tree->cur, pos, + tree->rbuf->name, tree->rbuf->line, buf); +} + + +static void +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); + va_end(ap); + + (*tree->cb->roffmsg)(tree->args, ROFF_ERROR, tree->cur, pos, + tree->rbuf->name, tree->rbuf->line, buf); }