=================================================================== RCS file: /cvs/mandoc/mdoc.h,v retrieving revision 1.2 retrieving revision 1.35 diff -u -p -r1.2 -r1.35 --- mandoc/mdoc.h 2008/12/15 03:13:01 1.2 +++ mandoc/mdoc.h 2009/02/24 14:52:55 1.35 @@ -1,4 +1,4 @@ -/* $Id: mdoc.h,v 1.2 2008/12/15 03:13:01 kristaps Exp $ */ +/* $Id: mdoc.h,v 1.35 2009/02/24 14:52:55 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -19,8 +19,16 @@ #ifndef MDOC_H #define MDOC_H -#define MDOC_LINEARG_MAX 9 +/* + * This library implements a validating scanner/parser for ``mdoc'' roff + * macro documents, a.k.a. BSD manual page documents. The mdoc.c file + * drives the parser, while macro.c describes the macro ontologies. + * validate.c pre- and post-validates parsed macros, and action.c + * performs actions on parsed and validated macros. + */ +/* What follows is a list of ALL possible macros. */ + #define MDOC___ 0 #define MDOC_Dd 1 #define MDOC_Dt 2 @@ -129,6 +137,8 @@ #define MDOC_Ud 105 #define MDOC_MAX 106 +/* What follows is a list of ALL possible macro arguments. */ + #define MDOC_Split 0 #define MDOC_Nosplit 1 #define MDOC_Ragged 2 @@ -189,38 +199,56 @@ #define MDOC_svid4 57 #define MDOC_Filled 58 #define MDOC_Words 59 -#define MDOC_ARG_MAX 60 +#define MDOC_Emphasis 60 +#define MDOC_Symbolic 61 +#define MDOC_ARG_MAX 62 -enum mdoc_err { - ERR_SYNTAX_QUOTE, - ERR_SYNTAX_WS, - ERR_MACRO_NOTSUP, - ERR_MACRO_NOTCALL, - ERR_SCOPE_BREAK, - ERR_ARGS_GE1 +/* Warnings are either syntax or groff-compatibility. */ +enum mdoc_warn { + WARN_SYNTAX, + WARN_COMPAT }; -enum mdoc_warn { - WARN_SYNTAX_WS_EOLN, - WARN_SYNTAX_MACLIKE, - WARN_ARGS_GE1 +/* Possible values for the `At' macro. */ +enum mdoc_att { + ATT_DEFAULT = 0, + ATT_v1, + ATT_v2, + ATT_v3, + ATT_v4, + ATT_v5, + ATT_v6, + ATT_v7, + ATT_32v, + ATT_V1, + ATT_V2, + ATT_V3, + ATT_V4 }; +/* An argument to a macro (multiple values = `It -column'). */ struct mdoc_arg { int arg; + int line; + int pos; size_t sz; char **value; }; +/* Type of a syntax node. */ enum mdoc_type { MDOC_TEXT, MDOC_ELEM, MDOC_HEAD, + MDOC_TAIL, MDOC_BODY, - MDOC_BLOCK + MDOC_BLOCK, + MDOC_ROOT }; +/* Manual section. */ enum mdoc_msec { + MSEC_DEFAULT = 0, MSEC_1, MSEC_2, MSEC_3, @@ -238,12 +266,13 @@ enum mdoc_msec { MSEC_n, MSEC_unass, MSEC_draft, - MSEC_paper, - MSEC_NONE + MSEC_paper }; +/* Section (named/unnamed) of `Ss'. */ enum mdoc_sec { - SEC_PROLOGUE, + SEC_PROLOGUE = 0, + SEC_BODY, SEC_NAME, SEC_SYNOPSIS, SEC_DESCRIPTION, @@ -262,7 +291,9 @@ enum mdoc_sec { SEC_CUSTOM }; +/* Volume of `Dt'. */ enum mdoc_vol { + VOL_DEFAULT = 0, VOL_AMD, VOL_IND, VOL_KM, @@ -271,15 +302,17 @@ enum mdoc_vol { VOL_PS1, VOL_SMM, VOL_URM, - VOL_USD, - VOL_DEFAULT + VOL_USD }; +/* Architecture of `Dt'. */ enum mdoc_arch { + ARCH_DEFAULT = 0, ARCH_alpha, ARCH_amd64, ARCH_amiga, ARCH_arc, + ARCH_arm, ARCH_armish, ARCH_aviion, ARCH_hp300, @@ -300,79 +333,108 @@ enum mdoc_arch { ARCH_sparc64, ARCH_sun3, ARCH_vax, - ARCH_zaurus, - ARCH_DEFAULT + ARCH_zaurus }; +/* Information from prologue. */ struct mdoc_meta { enum mdoc_msec msec; enum mdoc_vol vol; enum mdoc_arch arch; - struct tm tm; + time_t date; + char *title; + char *os; + char *name; }; +/* Text-only node. */ struct mdoc_text { char *string; }; +/* Block (scoped) node. */ struct mdoc_block { - int tok; size_t argc; struct mdoc_arg *argv; + struct mdoc_node *head; + struct mdoc_node *body; + struct mdoc_node *tail; }; -struct mdoc_head { - size_t sz; - char **args; - int tok; -}; - -struct mdoc_body { - int tok; -}; - +/* In-line element node. */ struct mdoc_elem { - size_t sz; - char **args; - int tok; size_t argc; struct mdoc_arg *argv; }; +/* Typed nodes of an AST node. */ union mdoc_data { struct mdoc_text text; struct mdoc_elem elem; - struct mdoc_body body; - struct mdoc_head head; struct mdoc_block block; }; +/* Node in AST. */ struct mdoc_node { struct mdoc_node *parent; struct mdoc_node *child; struct mdoc_node *next; + struct mdoc_node *prev; + int line; + int pos; + int tok; + int flags; +#define MDOC_VALID (1 << 0) +#define MDOC_ACTED (1 << 1) enum mdoc_type type; union mdoc_data data; + enum mdoc_sec sec; }; +/* Call-backs for parse messages. */ struct mdoc_cb { - int (*mdoc_err)(void *, int, int, enum mdoc_err); - int (*mdoc_warn)(void *, int, int, enum mdoc_warn); - void (*mdoc_msg)(void *, int, const char *); + void (*mdoc_msg)(void *, int, int, const char *); + int (*mdoc_err)(void *, int, int, const char *); + int (*mdoc_warn)(void *, int, int, + enum mdoc_warn, const char *); }; +/* Global table of macro names (`Bd', `Ed', etc.). */ extern const char *const *mdoc_macronames; + +/* Global table of argument names (`column', `tag', etc.). */ extern const char *const *mdoc_argnames; __BEGIN_DECLS struct mdoc; +/* Free memory allocated with mdoc_alloc. */ void mdoc_free(struct mdoc *); + +/* Allocate a new parser instance. */ struct mdoc *mdoc_alloc(void *data, const struct mdoc_cb *); -int mdoc_parseln(struct mdoc *, char *buf); -const struct mdoc_node - *mdoc_result(struct mdoc *); + +/* Parse a single line in a stream (boolean retval). */ +int mdoc_parseln(struct mdoc *, int, char *buf); + +/* Get result first node (after mdoc_endparse!). */ +const struct mdoc_node *mdoc_node(const struct mdoc *); + +/* Get result meta-information (after mdoc_endparse!). */ +const struct mdoc_meta *mdoc_meta(const struct mdoc *); + +/* Signal end of parse sequence (boolean retval). */ +int mdoc_endparse(struct mdoc *); + +/* The following are utility functions. */ +const char *mdoc_arch2a(enum mdoc_arch); +const char *mdoc_vol2a(enum mdoc_vol); +const char *mdoc_msec2a(enum mdoc_msec); +const char *mdoc_att2a(enum mdoc_att); +enum mdoc_att mdoc_atoatt(const char *); +const char *mdoc_st2a(int); +int mdoc_isdelim(const char *); __END_DECLS