=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.5 retrieving revision 1.45 diff -u -p -r1.5 -r1.45 --- mandoc/roff.c 2008/11/25 16:49:57 1.5 +++ mandoc/roff.c 2008/12/06 16:50:18 1.45 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.5 2008/11/25 16:49:57 kristaps Exp $ */ +/* $Id: roff.c,v 1.45 2008/12/06 16:50:18 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -16,9 +16,13 @@ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR * PERFORMANCE OF THIS SOFTWARE. */ +#include +#include + #include #include #include +#include #include #include #include @@ -26,606 +30,448 @@ #include "libmdocml.h" #include "private.h" +#include "roff.h" -#define ROFF_MAXARG 10 +/* FIXME: First letters of quoted-text interpreted in rofffindtok. */ +/* FIXME: `No' not implemented. */ +/* TODO: warn if Pp occurs before/after Sh etc. (see mdoc.samples). */ +/* TODO: warn about "X section only" macros. */ +/* TODO: warn about empty lists. */ +/* TODO: (warn) some sections need specific elements. */ +/* TODO: (warn) NAME section has particular order. */ +/* TODO: unify empty-content tags a la
. */ +/* TODO: macros with a set number of arguments? */ +/* TODO: validate Dt macro arguments. */ +/* FIXME: Bl -diag supposed to ignore callable children. */ +/* FIXME: Nm has newline when used in NAME section. */ -/* Whether we're entering or leaving a roff scope. */ -enum roffd { - ROFF_ENTER = 0, - ROFF_EXIT -}; - -/* The type of a macro (see mdoc(7) for more). */ -enum rofftype { - ROFF_COMMENT, - ROFF_TEXT, - ROFF_LAYOUT -}; - -/* Arguments passed to a macro callback. */ -#define ROFFCALL_ARGS \ - int tok, struct rofftree *tree, \ - const char *argv[], enum roffd type - -struct rofftree; - -/* Describes a roff token (like D1 or Sh). */ -struct rofftok { - int (*cb)(ROFFCALL_ARGS); /* Callback. */ - const int *args; /* Args (or NULL). */ - enum rofftype type; /* Type of macro. */ - int symm; /* FIXME */ - int flags; -#define ROFF_NESTED (1 << 0) /* Nested-layout. */ -#define ROFF_PARSED (1 << 1) /* "Parsed". */ -#define ROFF_CALLABLE (1 << 2) /* "Callable". */ -#define ROFF_QUOTES (1 << 3) /* Quoted args. */ -}; - -/* An argument to a roff token (like -split or -enum). */ -struct roffarg { - int flags; -#define ROFF_VALUE (1 << 0) /* Has a value. */ -}; - -/* mdocml remembers only the current parse node and the chain leading to - * the document root (scopes). - */ struct roffnode { int tok; /* Token id. */ struct roffnode *parent; /* Parent (or NULL). */ - size_t line; /* Parsed at line. */ }; -/* State of file parse. */ struct rofftree { struct roffnode *last; /* Last parsed node. */ - 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. */ int state; #define ROFF_PRELUDE (1 << 1) /* In roff prelude. */ - /* FIXME: if we had prev ptrs, this wouldn't be necessary. */ #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. */ - roffin roffin; /* Text-macro cb. */ - roffblkin roffblkin; /* Block-macro cb. */ - roffout roffout; /* Text-macro cb. */ - roffblkout roffblkout; /* Block-macro cb. */ - struct md_mbuf *mbuf; /* Output (or NULL). */ - const struct md_args *args; /* Global args. */ - const struct md_rbuf *rbuf; /* Input. */ + 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_layout(ROFFCALL_ARGS); -static int roff_text(ROFFCALL_ARGS); - static struct roffnode *roffnode_new(int, struct rofftree *); -static void roffnode_free(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 *); 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 roffparse(struct rofftree *, char *, size_t); -static int textparse(const struct rofftree *, - const char *, size_t); +static int roffnextopt(const struct rofftree *, + int, char ***, char **); +static int roffparseopts(struct rofftree *, int, + char ***, int *, char **); +static int roffcall(struct rofftree *, int, char **); +static int roffparse(struct rofftree *, char *); +static int textparse(struct rofftree *, char *); +static int roffdata(struct rofftree *, int, char *); +static int roffspecial(struct rofftree *, int, + const char *, size_t, char **); +static int roffsetname(struct rofftree *, char **); -/* Arguments for `An' macro. */ -static const int roffarg_An[] = { - ROFF_Split, ROFF_Nosplit, ROFF_ARGMAX }; -/* Arguments for `Bd' macro. */ -static const int roffarg_Bd[] = { - ROFF_Ragged, ROFF_Unfilled, ROFF_Literal, ROFF_File, - ROFF_Offset, ROFF_ARGMAX }; -/* Arguments for `Bl' macro. */ -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 }; +#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 -/* FIXME: a big list of fixes that must occur. - * - * (1) Distinction not between ROFF_TEXT and ROFF_LAYOUT, but instead - * ROFF_ATOM and ROFF_NODE, which designate line spacing. If - * ROFF_ATOM, we need not remember any state. - * - * (2) Have a maybe-NULL list of possible subsequent children for each - * node. Bl, e.g., can only have It children (roffparse). - * - * (3) Have a maybe-NULL list of possible parents for each node. It, - * e.g., can only have Bl as a parent (roffparse). - * - * (N.B. If (2) were complete, (3) wouldn't be necessary.) - * - * (4) Scope rules. If Pp exists, it closes the scope out from the - * previous Pp (if it exists). Same with Sh and Ss. If El exists, - * it closes out Bl and interim It. - * - * (5) Nesting. Sh cannot be any descendant of Sh. Bl, however, can be - * nested within an It. - * - * Once that's done, we're golden. - */ +int +roff_free(struct rofftree *tree, int flush) +{ + int error, t; + struct roffnode *n; -/* Table of all known tokens. */ -static const struct rofftok tokens[ROFF_MAX] = { - { NULL, NULL, 0, ROFF_COMMENT, 0 }, /* \" */ - { roff_Dd, NULL, 0, ROFF_TEXT, 0 }, /* Dd */ - { roff_Dt, NULL, 0, ROFF_TEXT, 0 }, /* Dt */ - { roff_Os, NULL, 0, ROFF_TEXT, 0 }, /* Os */ - { roff_layout, NULL, ROFF_Sh, ROFF_LAYOUT, ROFF_PARSED }, /* Sh */ - { roff_layout, NULL, ROFF_Ss, ROFF_LAYOUT, ROFF_PARSED }, /* Ss */ - { roff_layout, NULL, ROFF_Pp, ROFF_LAYOUT, 0 }, /* Pp */ - { roff_layout, NULL, 0, ROFF_TEXT, 0 }, /* D1 */ - { roff_layout, NULL, 0, ROFF_TEXT, 0 }, /* Dl */ - { roff_layout, roffarg_Bd, 0, ROFF_LAYOUT, 0 }, /* Bd */ - { roff_layout, NULL, ROFF_Bd, ROFF_LAYOUT, 0 }, /* Ed */ - { roff_layout, roffarg_Bl, 0, ROFF_LAYOUT, 0 }, /* Bl */ - { roff_layout, NULL, ROFF_Bl, ROFF_LAYOUT, 0 }, /* El */ - { roff_layout, NULL, ROFF_It, ROFF_LAYOUT, 0 }, /* It */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ad */ - { roff_text, roffarg_An, 0, ROFF_TEXT, ROFF_PARSED }, /* An */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ar */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Cd */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Cm */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dv */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Er */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ev */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ex */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fa */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Fd */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fl */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fn */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ft */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ic */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* In */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Li */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Nd */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Nm */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Op */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ot */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pa */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Rv */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* St */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Va */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Vt */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xr */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* %A */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %B */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* %D */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %I */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %J */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* %N */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* %O */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* %P */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* %R */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* %T */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* %V */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ac */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ao */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Aq */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* At */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bc */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Bf */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bo */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bq */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bsx */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bx */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Db */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dc */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Do */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dq */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ec */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Ef */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Em */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Eo */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Fx */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ms */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* No */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ns */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Nx */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ox */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pc */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Pf */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Po */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pq */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qc */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ql */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qo */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qq */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Re */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Rs */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sc */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* So */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sq */ - { roff_text, NULL, 0, ROFF_TEXT, 0 }, /* Sm */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sx */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sy */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Tn */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ux */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xc */ - { roff_text, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xo */ -}; + error = 0; -/* 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 */ -}; + if ( ! flush) + goto end; -const char *const toknamesp[ROFF_MAX] = - { - "\\\"", - "Dd", /* Title macros. */ - "Dt", - "Os", - "Sh", /* Layout macros */ - "Ss", - "Pp", - "D1", - "Dl", - "Bd", - "Ed", - "Bl", - "El", - "It", - "Ad", /* Text macros. */ - "An", - "Ar", - "Cd", - "Cm", - "Dr", - "Er", - "Ev", - "Ex", - "Fa", - "Fd", - "Fl", - "Fn", - "Ft", - "Ex", - "Ic", - "In", - "Li", - "Nd", - "Nm", - "Op", - "Ot", - "Pa", - "Rv", - "St", - "Va", - "Vt", - "Xr", - "\%A", /* General text macros. */ - "\%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", /* FIXME: do not support! */ - "Xo", /* FIXME: do not support! */ - }; + error = 1; -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", - }; + if (ROFF_PRELUDE & tree->state) { + roff_err(tree, NULL, "prelude never finished"); + goto end; + } -const char *const *toknames = toknamesp; -const char *const *tokargnames = tokargnamesp; + for (n = tree->last; n; n = n->parent) { + if (0 != tokens[n->tok].ctx) + continue; + roff_err(tree, NULL, "closing explicit scope `%s'", + toknames[n->tok]); + goto end; + } + while (tree->last) { + t = tree->last->tok; + if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) + goto end; + } -int -roff_free(struct rofftree *tree, int flush) -{ - int error; + if ( ! (*tree->cb.rofftail)(tree->arg)) + goto end; - assert(tree->mbuf); - if ( ! flush) - tree->mbuf = NULL; + error = 0; - /* LINTED */ - while (tree->last) - if ( ! (*tokens[tree->last->tok].cb) - (tree->last->tok, tree, NULL, ROFF_EXIT)) - /* Disallow flushing. */ - tree->mbuf = NULL; +end: - error = tree->mbuf ? 0 : 1; + while (tree->last) + roffnode_free(tree); - if (tree->mbuf && (ROFF_PRELUDE & tree->state)) { - warnx("%s: prelude never finished", - tree->rbuf->name); - error = 1; - } - free(tree); + return(error ? 0 : 1); } struct rofftree * -roff_alloc(const struct md_args *args, struct md_mbuf *out, - const struct md_rbuf *in, roffin textin, - roffout textout, roffblkin blkin, roffblkout blkout) +roff_alloc(const struct roffcb *cb, void *args) { struct rofftree *tree; - if (NULL == (tree = calloc(1, sizeof(struct rofftree)))) { - warn("malloc"); - return(NULL); - } + assert(args); + assert(cb); + if (NULL == (tree = calloc(1, sizeof(struct rofftree)))) + err(1, "calloc"); + tree->state = ROFF_PRELUDE; - tree->args = args; - tree->mbuf = out; - tree->rbuf = in; - tree->roffin = textin; - tree->roffout = textout; - tree->roffblkin = blkin; - tree->roffblkout = blkout; + tree->arg = args; + (void)memcpy(&tree->cb, cb, sizeof(struct roffcb)); + return(tree); } int -roff_engine(struct rofftree *tree, char *buf, size_t sz) +roff_engine(struct rofftree *tree, char *buf) { - if (0 == sz) { - warnx("%s: blank line (line %zu)", - tree->rbuf->name, - tree->rbuf->line); + tree->cur = buf; + assert(buf); + + if (0 == *buf) { + roff_err(tree, buf, "blank line"); return(0); } else if ('.' != *buf) - return(textparse(tree, buf, sz)); + return(textparse(tree, buf)); - return(roffparse(tree, buf, sz)); + return(roffparse(tree, buf)); } static int -textparse(const struct rofftree *tree, const char *buf, size_t sz) +textparse(struct rofftree *tree, char *buf) { - - if (NULL == tree->last) { - warnx("%s: unexpected text (line %zu)", - tree->rbuf->name, - tree->rbuf->line); + char *bufp; + + /* TODO: literal parsing. */ + + if ( ! (ROFF_BODY & tree->state)) { + roff_err(tree, buf, "data not in body"); return(0); - } else if (NULL == tree->last->parent) { - warnx("%s: disallowed text (line %zu)", - tree->rbuf->name, - tree->rbuf->line); - return(0); } - /* Print text. */ + /* 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); } 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; + + /* + * 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++) { - argv[i] = buf++; - while (*buf && ! isspace(*buf)) - buf++; - if (0 == *buf) { - continue; + 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); + } + } else { + argv[i] = buf++; + while (*buf) { + if ( ! isspace(*buf)) { + buf++; + continue; + } + if (*(buf - 1) == '\\') { + buf++; + continue; + } + break; + } + if (0 == *buf) + continue; } *buf++ = 0; while (*buf && isspace(*buf)) buf++; } - + assert(i > 0); - if (i < ROFF_MAXARG) - argv[i] = NULL; + if (ROFF_MAXLINEARG == i && *buf) { + roff_err(tree, p, "too many arguments for `%s'", toknames + [tok]); + return(0); + } - return(ROFF_MAXARG > i); + argv[i] = NULL; + return(1); } static int -roffparse(struct rofftree *tree, char *buf, size_t sz) +roffscan(int tok, const int *tokv) { + + if (NULL == tokv) + return(1); + + for ( ; ROFF_MAX != *tokv; tokv++) + if (tok == *tokv) + return(1); + + return(0); +} + + +static int +roffparse(struct rofftree *tree, char *buf) +{ int tok, t; - struct roffnode *node; - char *argv[ROFF_MAXARG]; - const char **argvp; + struct roffnode *n; + char *argv[ROFF_MAXLINEARG]; + char **argvp; - assert(sz > 0); + if (0 != *buf && 0 != *(buf + 1) && 0 != *(buf + 2)) + if (0 == strncmp(buf, ".\\\"", 3)) + return(1); - /* - * Extract the token identifier from the buffer. If there's no - * callback for the token (comment, etc.) then exit immediately. - * We don't do any error handling (yet), so if the token doesn't - * exist, die. - */ + if (ROFF_MAX == (tok = rofffindtok(buf + 1))) { + roff_err(tree, buf + 1, "bogus line macro"); + return(0); + } else if (NULL == tokens[tok].cb) { + roff_err(tree, buf + 1, "unsupported macro `%s'", + toknames[tok]); + return(0); + } - if (3 > sz) { - warnx("%s: malformed line (line %zu)", - tree->rbuf->name, - tree->rbuf->line); + assert(ROFF___ != tok); + if ( ! roffargs(tree, tok, buf, argv)) return(0); + + argvp = (char **)argv; + + /* + * Prelude macros break some assumptions, so branch now. + */ - /* FIXME: .Bsx is three letters! */ - } else if (ROFF_MAX == (tok = rofffindtok(buf + 1))) { - warnx("%s: unknown line token `%c%c' (line %zu)", - tree->rbuf->name, - *(buf + 1), *(buf + 2), - tree->rbuf->line); + if (ROFF_PRELUDE & tree->state) { + assert(NULL == tree->last); + return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + } + + assert(ROFF_BODY & tree->state); + + /* + * First check that our possible parents and parent's possible + * children are satisfied. + */ + + if (tree->last && ! roffscan + (tree->last->tok, tokens[tok].parents)) { + roff_err(tree, *argvp, "`%s' has invalid parent `%s'", + toknames[tok], + toknames[tree->last->tok]); return(0); - } else if (ROFF_COMMENT == tokens[tok].type) - /* Ignore comment tokens. */ - return(1); - - if ( ! roffargs(tok, buf, argv)) { - warnx("%s: too many arguments to `%s' (line %zu)", - tree->rbuf->name, toknames[tok], - tree->rbuf->line); + } + + if (tree->last && ! roffscan + (tok, tokens[tree->last->tok].children)) { + roff_err(tree, *argvp, "`%s' is invalid child of `%s'", + toknames[tok], + toknames[tree->last->tok]); return(0); } /* - * If this is a non-nestable layout token and we're below a - * token of the same type, then recurse upward to the token, - * closing out the interim scopes. - * - * If there's a nested token on the chain, then raise an error - * as nested tokens have corresponding "ending" tokens and we're - * breaking their scope. + * Branch if we're not a layout token. */ - node = NULL; + if (ROFF_LAYOUT != tokens[tok].type) + return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + if (0 == tokens[tok].ctx) + return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); - if (ROFF_LAYOUT == tokens[tok].type && - ! (ROFF_NESTED & tokens[tok].flags)) { - for (node = tree->last; node; node = node->parent) { - if (node->tok == tok) - break; + /* + * 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. + */ - /* Don't break nested scope. */ + 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. + */ - if ( ! (ROFF_NESTED & tokens[node->tok].flags)) + 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 (tokens[n->tok].ctx == n->tok) continue; - warnx("%s: scope of %s (line %zu) broken by " - "%s (line %zu)", - tree->rbuf->name, - toknames[tok], node->line, - toknames[node->tok], - tree->rbuf->line); + roff_err(tree, *argv, "`%s' breaks `%s' scope", + toknames[tok], toknames[n->tok]); return(0); } - } - if (node) { - assert(ROFF_LAYOUT == tokens[tok].type); - assert( ! (ROFF_NESTED & tokens[tok].flags)); - assert(node->tok == tok); + /* + * Create a new scope, as no previous one exists to + * close out. + */ - /* Clear up to last scoped token. */ + if (NULL == n) + return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); - /* LINTED */ + /* + * Close out all intermediary scoped blocks, then hang + * the current scope from our predecessor's parent. + */ + do { t = tree->last->tok; - if ( ! (*tokens[tree->last->tok].cb) - (tree->last->tok, tree, NULL, ROFF_EXIT)) + if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) return(0); } while (t != tok); + + return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); } - /* Proceed with actual token processing. */ + /* + * 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'. + */ - argvp = (const char **)&argv[1]; - return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); + 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); + + return(1); } @@ -647,29 +493,81 @@ rofffindarg(const char *name) static int -rofffindtok(const char *name) +rofffindtok(const char *buf) { - size_t i; + char token[4]; + int i; + for (i = 0; *buf && ! isspace(*buf) && i < 3; i++, buf++) + token[i] = *buf; + + if (i == 3) + return(ROFF_MAX); + + token[i] = 0; + /* FIXME: use a table, this is slow but ok for now. */ /* LINTED */ for (i = 0; i < ROFF_MAX; i++) /* LINTED */ - if (0 == strncmp(name, toknames[i], 2)) + if (0 == strcmp(toknames[i], token)) return((int)i); - + return(ROFF_MAX); } static int +roffispunct(const char *p) +{ + + if (0 == *p) + return(0); + if (0 != *(p + 1)) + return(0); + + switch (*p) { + case('{'): + /* FALLTHROUGH */ + case('.'): + /* FALLTHROUGH */ + case(','): + /* FALLTHROUGH */ + case(';'): + /* FALLTHROUGH */ + case(':'): + /* FALLTHROUGH */ + case('?'): + /* FALLTHROUGH */ + case('!'): + /* FALLTHROUGH */ + case('('): + /* FALLTHROUGH */ + case(')'): + /* FALLTHROUGH */ + case('['): + /* FALLTHROUGH */ + case(']'): + /* FALLTHROUGH */ + case('}'): + return(1); + default: + break; + } + + return(0); +} + + +static int rofffindcallable(const char *name) { int c; if (ROFF_MAX == (c = rofffindtok(name))) return(ROFF_MAX); + assert(c >= 0 && c < ROFF_MAX); return(ROFF_CALLABLE & tokens[c].flags ? c : ROFF_MAX); } @@ -679,15 +577,13 @@ 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); } @@ -709,12 +605,11 @@ roffargok(int tokid, int argid) static void -roffnode_free(int tokid, struct rofftree *tree) +roffnode_free(struct rofftree *tree) { struct roffnode *p; assert(tree->last); - assert(tree->last->tok == tokid); p = tree->last; tree->last = tree->last->parent; @@ -722,10 +617,191 @@ roffnode_free(int tokid, struct rofftree *tree) } +static int +roffspecial(struct rofftree *tree, int tok, + const char *start, size_t sz, char **ordp) +{ + + switch (tok) { + case (ROFF_Nm): + if (0 == sz) { + if (0 == tree->name[0]) { + roff_err(tree, start, "`Nm' not set"); + return(0); + } + ordp[0] = tree->name; + ordp[1] = NULL; + } else if ( ! roffsetname(tree, ordp)) + return(0); + break; + + case (ROFF_Ex): + if (0 == sz) { + roff_err(tree, start, "`Ex' expects an arg"); + return(0); + } else if (1 != sz) { + roff_err(tree, start, "`Ex' expects one arg"); + return(0); + } + break; + + case (ROFF_Sm): + if (0 == sz) { + roff_err(tree, start, "`Sm' expects an arg"); + return(0); + } else if (1 != sz) { + roff_err(tree, start, "`Sm' expects one arg"); + return(0); + } + + if (0 != strcmp(ordp[0], "on") && + 0 != strcmp(ordp[0], "off")) { + roff_err(tree, start, "`Sm' has invalid argument"); + return(0); + } + break; + + case (ROFF_Ud): + /* FALLTHROUGH */ + case (ROFF_Bt): + if (0 != sz) { + roff_err(tree, start, "`%s' expects no args", + toknames[tok]); + return(0); + } + break; + default: + break; + } + + return((*tree->cb.roffspecial) + (tree->arg, tok, tree->cur, ordp)); +} + + +static int +roffcall(struct rofftree *tree, int tok, char **argv) +{ + + if (NULL == tokens[tok].cb) { + roff_err(tree, *argv, "unsupported macro `%s'", + toknames[tok]); + return(0); + } + if ( ! (*tokens[tok].cb)(tok, tree, argv, ROFF_ENTER)) + return(0); + return(1); +} + + +static int +roffnextopt(const struct rofftree *tree, int tok, + char ***in, char **val) +{ + char *arg, **argv; + int v; + + *val = NULL; + argv = *in; + assert(argv); + + if (NULL == (arg = *argv)) + return(-1); + if ('-' != *arg) + return(-1); + + if (ROFF_ARGMAX == (v = rofffindarg(arg + 1))) { + roff_warn(tree, arg, "argument-like parameter `%s' to " + "`%s'", arg, toknames[tok]); + return(-1); + } + + 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])) + return(v); + + *in = ++argv; + + if (NULL == *argv) { + roff_err(tree, arg, "empty value of `%s' for `%s'", + tokargnames[v], toknames[tok]); + return(ROFF_ARGMAX); + } + + return(v); +} + + +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]; if (ROFF_BODY & tree->state) { assert( ! (ROFF_PRELUDE & tree->state)); @@ -736,18 +812,71 @@ roff_Dd(ROFFCALL_ARGS) assert(ROFF_PRELUDE & tree->state); assert( ! (ROFF_BODY & tree->state)); - if (ROFF_PRELUDE_Dd & tree->state || - ROFF_PRELUDE_Dt & tree->state) { - warnx("%s: prelude `Dd' out-of-order (line %zu)", - tree->rbuf->name, tree->rbuf->line); + 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); } - /* TODO: parse date. */ - assert(NULL == tree->last); - tree->state |= ROFF_PRELUDE_Dd; + argv++; + + if (0 == strcmp(*argv, "$Mdocdate: December 6 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); } @@ -766,15 +895,43 @@ roff_Dt(ROFFCALL_ARGS) assert(ROFF_PRELUDE & tree->state); assert( ! (ROFF_BODY & tree->state)); - if ( ! (ROFF_PRELUDE_Dd & tree->state) || - (ROFF_PRELUDE_Dt & tree->state)) { - warnx("%s: prelude `Dt' out-of-order (line %zu)", - tree->rbuf->name, tree->rbuf->line); + 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); } - /* 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; @@ -782,15 +939,95 @@ roff_Dt(ROFFCALL_ARGS) } +static int +roffsetname(struct rofftree *tree, char **ordp) +{ + + assert(*ordp); + + /* FIXME: not all sections can set this. */ + + if (NULL != *(ordp + 1)) { + roff_err(tree, *ordp, "too many `Nm' args"); + return(0); + } + + if (strlcpy(tree->name, *ordp, sizeof(tree->name)) + >= sizeof(tree->name)) { + roff_err(tree, *ordp, "`Nm' arg too long"); + return(0); + } + + return(1); +} + + /* ARGSUSED */ +static int +roff_Sm(ROFFCALL_ARGS) +{ + char *ordp[1], *p; + + p = *argv++; + *ordp = *argv; + + return(roffspecial(tree, tok, p, *ordp ? 1 : 0, ordp)); +} + + +/* 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, 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; - if (ROFF_EXIT == type) { - assert(ROFF_PRELUDE_Os & tree->state); - return(roff_layout(tok, tree, argv, type)); - } 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)); @@ -799,189 +1036,335 @@ 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); } - /* 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); - return(roff_layout(tok, tree, argv, type)); + return((*tree->cb.roffhead)(tree->arg, &tree->tm, + tree->os, tree->title, tree->section, + tree->volume)); } -/* ARGUSED */ +/* ARGSUSED */ static int -roffnextopt(int tok, const char ***in, char **val) +roff_layout(ROFFCALL_ARGS) { - const char *arg, **argv; - int v; + int i, c, argcp[ROFF_MAXLINEARG]; + char *argvp[ROFF_MAXLINEARG]; - *val = NULL; - argv = *in; - assert(argv); + if (ROFF_PRELUDE & tree->state) { + roff_err(tree, *argv, "bad `%s' in prelude", + toknames[tok]); + return(0); + } else if (ROFF_EXIT == type) { + roffnode_free(tree); + if ( ! (*tree->cb.roffblkbodyout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkout)(tree->arg, tok)); + } - if (NULL == (arg = *argv)) - return(-1); - if ('-' != *arg) - return(-1); + assert( ! (ROFF_CALLABLE & tokens[tok].flags)); - /* FIXME: should we let this slide... ? */ + ++argv; - if (ROFF_ARGMAX == (v = rofffindarg(&arg[1]))) - return(-1); + if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) + return(0); + if (NULL == roffnode_new(tok, tree)) + return(0); - /* FIXME: should we let this slide... ? */ + /* + * 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 ( ! roffargok(tok, v)) - return(-1); - if ( ! (ROFF_VALUE & tokenargs[v].flags)) - return(v); + if ( ! (*tree->cb.roffblkin)(tree->arg, tok, argcp, argvp)) + return(0); + if (NULL == *argv) + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); - *in = ++argv; + if ( ! (*tree->cb.roffblkheadin)(tree->arg, tok, argcp, argvp)) + return(0); - /* FIXME: what if this looks like a roff token or argument? */ + /* + * If there are no parsable parts, then write remaining tokens + * into the layout header and exit. + */ - return(*argv ? v : ROFF_ARGMAX); + if ( ! (ROFF_PARSED & tokens[tok].flags)) { + i = 0; + while (*argv) + if ( ! roffdata(tree, i++, *argv++)) + return(0); + + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, 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))) { + assert(tree->arg); + if ( ! roffdata(tree, i++, *argv++)) + return(0); + continue; + } + if ( ! roffcall(tree, c, argv)) + 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) { + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); + } + + /* + * Expensive. Scan to the end of line then work backwards until + * 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, argvp)); } /* ARGSUSED */ static int -roff_layout(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) { - 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->roffblkout)(tok)); - } + first = (*argv == tree->cur); + p = *argv++; + if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) + return(0); + + if (NULL == *argv) + return(roffspecial(tree, tok, p, 0, ordp)); + i = 0; + while (*argv && i < ROFF_MAXLINEARG) { + c = rofffindcallable(*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); - 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); + if (ROFF_MAX == c && ! roffispunct(*argv)) { + ordp[i++] = *argv++; + continue; } - argcp[i] = c; - argvp[i] = v; - i++; - argv++; + ordp[i] = NULL; + + if (ROFF_MAX == c) + break; + + if ( ! roffspecial(tree, tok, p, (size_t)i, ordp)) + return(0); + + return(roffcall(tree, c, ordp)); } - argcp[i] = ROFF_ARGMAX; - argvp[i] = NULL; + assert(i != ROFF_MAXLINEARG); + ordp[i] = NULL; - if (NULL == roffnode_new(tok, tree)) + if ( ! roffspecial(tree, tok, p, (size_t)i, ordp)) return(0); - if ( ! (*tree->roffin)(tok, argcp, argvp)) + /* 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]; + + if (ROFF_PRELUDE & tree->state) { + roff_err(tree, *argv, "`%s' disallowed in prelude", + toknames[tok]); return(0); + } + first = (*argv == tree->cur); + argv++; + + 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)) { - /* TODO: print all tokens. */ + i = 0; + while (*argv) + if ( ! roffdata(tree, i++, *argv++)) + return(0); - if ( ! ((*tree->roffout)(tok))) - return(0); - return((*tree->roffblkin)(tok)); + 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 (2 >= strlen(*argv) && ROFF_MAX != - (c = rofffindcallable(*argv))) - if ( ! (*tokens[c].cb)(c, tree, - argv + 1, ROFF_ENTER)) + 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; + } - /* TODO: print token. */ - argv++; + if ( ! roffispunct(*argv)) { + if ( ! roffdata(tree, i++, *argv++)) + return(0); + continue; + } + + i = 1; + for (j = 0; argv[j]; j++) + if ( ! roffispunct(argv[j])) + break; + + if (argv[j]) { + if ( ! roffdata(tree, 0, *argv++)) + return(0); + continue; + } + + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); + break; } - if ( ! ((*tree->roffout)(tok))) - return(0); + if (NULL == *argv) + return((*tree->cb.roffout)(tree->arg, tok)); + if ( ! first) + return(1); - return((*tree->roffblkin)(tok)); + return(roffpurgepunct(tree, argv)); } /* ARGSUSED */ static int -roff_text(ROFFCALL_ARGS) +roff_noop(ROFFCALL_ARGS) { - int i, c, argcp[ROFF_MAXARG]; - 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); - return(0); - } + return(1); +} - i = 0; - 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); - return(0); - } - argcp[i] = c; - argvp[i] = v; - i++; - argv++; - } +/* ARGSUSED */ +static int +roff_depr(ROFFCALL_ARGS) +{ - argcp[i] = ROFF_ARGMAX; - argvp[i] = NULL; + roff_err(tree, *argv, "`%s' is deprecated", toknames[tok]); + return(0); +} - if ( ! (*tree->roffin)(tok, argcp, argvp)) - return(0); - if ( ! (ROFF_PARSED & tokens[tok].flags)) { - /* TODO: print all tokens. */ - return((*tree->roffout)(tok)); - } +static void +roff_warn(const struct rofftree *tree, const char *pos, char *fmt, ...) +{ + va_list ap; + char buf[128]; - while (*argv) { - if (2 >= strlen(*argv) && ROFF_MAX != - (c = rofffindcallable(*argv))) - if ( ! (*tokens[c].cb)(c, tree, - argv + 1, ROFF_ENTER)) - return(0); + va_start(ap, fmt); + (void)vsnprintf(buf, sizeof(buf), fmt, ap); + va_end(ap); - /* TODO: print token. */ - argv++; - } + (*tree->cb.roffmsg)(tree->arg, + ROFF_WARN, tree->cur, pos, buf); +} - return((*tree->roffout)(tok)); + +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->arg, + ROFF_ERROR, tree->cur, pos, buf); } +