=================================================================== RCS file: /cvs/mandoc/term.c,v retrieving revision 1.70 retrieving revision 1.71 diff -u -p -r1.70 -r1.71 --- mandoc/term.c 2009/03/26 14:38:11 1.70 +++ mandoc/term.c 2009/03/31 13:50:19 1.71 @@ -1,4 +1,4 @@ -/* $Id: term.c,v 1.70 2009/03/26 14:38:11 kristaps dead $ */ +/* $Id: term.c,v 1.71 2009/03/31 13:50:19 kristaps Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -16,1911 +16,580 @@ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR * PERFORMANCE OF THIS SOFTWARE. */ -#include - #include -#include #include #include #include #include #include "term.h" +#include "man.h" +#include "mdoc.h" -/* - * Performs actions on nodes of the abstract syntax tree. Both pre- and - * post-fix operations are defined here. - */ +extern int man_run(struct termp *, + const struct man *); +extern int mdoc_run(struct termp *, + const struct mdoc *); -/* FIXME: macro arguments can be escaped. */ +static struct termp *term_alloc(enum termenc); +static void term_free(struct termp *); +static void term_pword(struct termp *, const char *, int); +static void term_pescape(struct termp *, + const char *, int *, int); +static void term_nescape(struct termp *, + const char *, size_t); +static void term_chara(struct termp *, char); +static void term_stringa(struct termp *, + const char *, size_t); +static int term_isopendelim(const char *, int); +static int term_isclosedelim(const char *, int); -#define TTYPE_PROG 0 -#define TTYPE_CMD_FLAG 1 -#define TTYPE_CMD_ARG 2 -#define TTYPE_SECTION 3 -#define TTYPE_FUNC_DECL 4 -#define TTYPE_VAR_DECL 5 -#define TTYPE_FUNC_TYPE 6 -#define TTYPE_FUNC_NAME 7 -#define TTYPE_FUNC_ARG 8 -#define TTYPE_LINK 9 -#define TTYPE_SSECTION 10 -#define TTYPE_FILE 11 -#define TTYPE_EMPH 12 -#define TTYPE_CONFIG 13 -#define TTYPE_CMD 14 -#define TTYPE_INCLUDE 15 -#define TTYPE_SYMB 16 -#define TTYPE_SYMBOL 17 -#define TTYPE_DIAG 18 -#define TTYPE_LINK_ANCHOR 19 -#define TTYPE_LINK_TEXT 20 -#define TTYPE_REF_JOURNAL 21 -#define TTYPE_LIST 22 -#define TTYPE_NMAX 23 -/* - * These define "styles" for element types, like command arguments or - * executable names. This is useful when multiple macros must decorate - * the same thing (like .Ex -std cmd and .Nm cmd). - */ - -/* TODO: abstract this into mdocterm.c. */ - -const int ttypes[TTYPE_NMAX] = { - TERMP_BOLD, /* TTYPE_PROG */ - TERMP_BOLD, /* TTYPE_CMD_FLAG */ - TERMP_UNDER, /* TTYPE_CMD_ARG */ - TERMP_BOLD, /* TTYPE_SECTION */ - TERMP_BOLD, /* TTYPE_FUNC_DECL */ - TERMP_UNDER, /* TTYPE_VAR_DECL */ - TERMP_UNDER, /* TTYPE_FUNC_TYPE */ - TERMP_BOLD, /* TTYPE_FUNC_NAME */ - TERMP_UNDER, /* TTYPE_FUNC_ARG */ - TERMP_UNDER, /* TTYPE_LINK */ - TERMP_BOLD, /* TTYPE_SSECTION */ - TERMP_UNDER, /* TTYPE_FILE */ - TERMP_UNDER, /* TTYPE_EMPH */ - TERMP_BOLD, /* TTYPE_CONFIG */ - TERMP_BOLD, /* TTYPE_CMD */ - TERMP_BOLD, /* TTYPE_INCLUDE */ - TERMP_BOLD, /* TTYPE_SYMB */ - TERMP_BOLD, /* TTYPE_SYMBOL */ - TERMP_BOLD, /* TTYPE_DIAG */ - TERMP_UNDER, /* TTYPE_LINK_ANCHOR */ - TERMP_BOLD, /* TTYPE_LINK_TEXT */ - TERMP_UNDER, /* TTYPE_REF_JOURNAL */ - TERMP_BOLD /* TTYPE_LIST */ -}; - -static int arg_hasattr(int, const struct mdoc_node *); -static int arg_getattrs(const int *, int *, size_t, - const struct mdoc_node *); -static int arg_getattr(int, const struct mdoc_node *); -static size_t arg_offset(const struct mdoc_argv *); -static size_t arg_width(const struct mdoc_argv *, int); -static int arg_listtype(const struct mdoc_node *); -static int fmt_block_vspace(struct termp *, - const struct mdoc_node *, - const struct mdoc_node *); - -/* - * What follows describes prefix and postfix operations for the abstract - * syntax tree descent. - */ - -#define DECL_ARGS \ - struct termp *p, \ - struct termpair *pair, \ - const struct mdoc_meta *meta, \ - const struct mdoc_node *node - -#define DECL_PRE(name) \ -static int name##_pre(DECL_ARGS) -#define DECL_POST(name) \ -static void name##_post(DECL_ARGS) -#define DECL_PREPOST(name) \ -DECL_PRE(name); \ -DECL_POST(name); - -DECL_PREPOST(termp__t); -DECL_PREPOST(termp_aq); -DECL_PREPOST(termp_bd); -DECL_PREPOST(termp_bq); -DECL_PREPOST(termp_brq); -DECL_PREPOST(termp_d1); -DECL_PREPOST(termp_dq); -DECL_PREPOST(termp_fd); -DECL_PREPOST(termp_fn); -DECL_PREPOST(termp_fo); -DECL_PREPOST(termp_ft); -DECL_PREPOST(termp_in); -DECL_PREPOST(termp_it); -DECL_PREPOST(termp_lb); -DECL_PREPOST(termp_op); -DECL_PREPOST(termp_pf); -DECL_PREPOST(termp_pq); -DECL_PREPOST(termp_qq); -DECL_PREPOST(termp_sh); -DECL_PREPOST(termp_ss); -DECL_PREPOST(termp_sq); -DECL_PREPOST(termp_vt); - -DECL_PRE(termp__j); -DECL_PRE(termp_ap); -DECL_PRE(termp_ar); -DECL_PRE(termp_at); -DECL_PRE(termp_bf); -DECL_PRE(termp_bsx); -DECL_PRE(termp_bt); -DECL_PRE(termp_cd); -DECL_PRE(termp_cm); -DECL_PRE(termp_dx); -DECL_PRE(termp_em); -DECL_PRE(termp_ex); -DECL_PRE(termp_fa); -DECL_PRE(termp_fl); -DECL_PRE(termp_fx); -DECL_PRE(termp_ic); -DECL_PRE(termp_lk); -DECL_PRE(termp_ms); -DECL_PRE(termp_mt); -DECL_PRE(termp_nd); -DECL_PRE(termp_nm); -DECL_PRE(termp_ns); -DECL_PRE(termp_nx); -DECL_PRE(termp_ox); -DECL_PRE(termp_pa); -DECL_PRE(termp_pp); -DECL_PRE(termp_rs); -DECL_PRE(termp_rv); -DECL_PRE(termp_sm); -DECL_PRE(termp_st); -DECL_PRE(termp_sx); -DECL_PRE(termp_sy); -DECL_PRE(termp_ud); -DECL_PRE(termp_ux); -DECL_PRE(termp_va); -DECL_PRE(termp_xr); - -DECL_POST(termp___); -DECL_POST(termp_bl); -DECL_POST(termp_bx); - -const struct termact __termacts[MDOC_MAX] = { - { NULL, NULL }, /* \" */ - { NULL, NULL }, /* Dd */ - { NULL, NULL }, /* Dt */ - { NULL, NULL }, /* Os */ - { termp_sh_pre, termp_sh_post }, /* Sh */ - { termp_ss_pre, termp_ss_post }, /* Ss */ - { termp_pp_pre, NULL }, /* Pp */ - { termp_d1_pre, termp_d1_post }, /* D1 */ - { termp_d1_pre, termp_d1_post }, /* Dl */ - { termp_bd_pre, termp_bd_post }, /* Bd */ - { NULL, NULL }, /* Ed */ - { NULL, termp_bl_post }, /* Bl */ - { NULL, NULL }, /* El */ - { termp_it_pre, termp_it_post }, /* It */ - { NULL, NULL }, /* Ad */ - { NULL, NULL }, /* An */ - { termp_ar_pre, NULL }, /* Ar */ - { termp_cd_pre, NULL }, /* Cd */ - { termp_cm_pre, NULL }, /* Cm */ - { NULL, NULL }, /* Dv */ - { NULL, NULL }, /* Er */ - { NULL, NULL }, /* Ev */ - { termp_ex_pre, NULL }, /* Ex */ - { termp_fa_pre, NULL }, /* Fa */ - { termp_fd_pre, termp_fd_post }, /* Fd */ - { termp_fl_pre, NULL }, /* Fl */ - { termp_fn_pre, termp_fn_post }, /* Fn */ - { termp_ft_pre, termp_ft_post }, /* Ft */ - { termp_ic_pre, NULL }, /* Ic */ - { termp_in_pre, termp_in_post }, /* In */ - { NULL, NULL }, /* Li */ - { termp_nd_pre, NULL }, /* Nd */ - { termp_nm_pre, NULL }, /* Nm */ - { termp_op_pre, termp_op_post }, /* Op */ - { NULL, NULL }, /* Ot */ - { termp_pa_pre, NULL }, /* Pa */ - { termp_rv_pre, NULL }, /* Rv */ - { termp_st_pre, NULL }, /* St */ - { termp_va_pre, NULL }, /* Va */ - { termp_vt_pre, termp_vt_post }, /* Vt */ - { termp_xr_pre, NULL }, /* Xr */ - { NULL, termp____post }, /* %A */ - { NULL, termp____post }, /* %B */ - { NULL, termp____post }, /* %D */ - { NULL, termp____post }, /* %I */ - { termp__j_pre, termp____post }, /* %J */ - { NULL, termp____post }, /* %N */ - { NULL, termp____post }, /* %O */ - { NULL, termp____post }, /* %P */ - { NULL, termp____post }, /* %R */ - { termp__t_pre, termp__t_post }, /* %T */ - { NULL, termp____post }, /* %V */ - { NULL, NULL }, /* Ac */ - { termp_aq_pre, termp_aq_post }, /* Ao */ - { termp_aq_pre, termp_aq_post }, /* Aq */ - { termp_at_pre, NULL }, /* At */ - { NULL, NULL }, /* Bc */ - { termp_bf_pre, NULL }, /* Bf */ - { termp_bq_pre, termp_bq_post }, /* Bo */ - { termp_bq_pre, termp_bq_post }, /* Bq */ - { termp_bsx_pre, NULL }, /* Bsx */ - { NULL, termp_bx_post }, /* Bx */ - { NULL, NULL }, /* Db */ - { NULL, NULL }, /* Dc */ - { termp_dq_pre, termp_dq_post }, /* Do */ - { termp_dq_pre, termp_dq_post }, /* Dq */ - { NULL, NULL }, /* Ec */ - { NULL, NULL }, /* Ef */ - { termp_em_pre, NULL }, /* Em */ - { NULL, NULL }, /* Eo */ - { termp_fx_pre, NULL }, /* Fx */ - { termp_ms_pre, NULL }, /* Ms */ - { NULL, NULL }, /* No */ - { termp_ns_pre, NULL }, /* Ns */ - { termp_nx_pre, NULL }, /* Nx */ - { termp_ox_pre, NULL }, /* Ox */ - { NULL, NULL }, /* Pc */ - { termp_pf_pre, termp_pf_post }, /* Pf */ - { termp_pq_pre, termp_pq_post }, /* Po */ - { termp_pq_pre, termp_pq_post }, /* Pq */ - { NULL, NULL }, /* Qc */ - { termp_sq_pre, termp_sq_post }, /* Ql */ - { termp_qq_pre, termp_qq_post }, /* Qo */ - { termp_qq_pre, termp_qq_post }, /* Qq */ - { NULL, NULL }, /* Re */ - { termp_rs_pre, NULL }, /* Rs */ - { NULL, NULL }, /* Sc */ - { termp_sq_pre, termp_sq_post }, /* So */ - { termp_sq_pre, termp_sq_post }, /* Sq */ - { termp_sm_pre, NULL }, /* Sm */ - { termp_sx_pre, NULL }, /* Sx */ - { termp_sy_pre, NULL }, /* Sy */ - { NULL, NULL }, /* Tn */ - { termp_ux_pre, NULL }, /* Ux */ - { NULL, NULL }, /* Xc */ - { NULL, NULL }, /* Xo */ - { termp_fo_pre, termp_fo_post }, /* Fo */ - { NULL, NULL }, /* Fc */ - { termp_op_pre, termp_op_post }, /* Oo */ - { NULL, NULL }, /* Oc */ - { NULL, NULL }, /* Bk */ - { NULL, NULL }, /* Ek */ - { termp_bt_pre, NULL }, /* Bt */ - { NULL, NULL }, /* Hf */ - { NULL, NULL }, /* Fr */ - { termp_ud_pre, NULL }, /* Ud */ - { termp_lb_pre, termp_lb_post }, /* Lb */ - { termp_ap_pre, NULL }, /* Lb */ - { termp_pp_pre, NULL }, /* Pp */ - { termp_lk_pre, NULL }, /* Lk */ - { termp_mt_pre, NULL }, /* Mt */ - { termp_brq_pre, termp_brq_post }, /* Brq */ - { termp_brq_pre, termp_brq_post }, /* Bro */ - { NULL, NULL }, /* Brc */ - { NULL, NULL }, /* %C */ - { NULL, NULL }, /* Es */ - { NULL, NULL }, /* En */ - { termp_dx_pre, NULL }, /* Dx */ - { NULL, NULL }, /* %Q */ -}; - -const struct termact *termacts = __termacts; - - -static size_t -arg_width(const struct mdoc_argv *arg, int pos) +void * +ascii_alloc(void) { - size_t v; - int i, len; - assert(pos < (int)arg->sz && pos >= 0); - assert(arg->value[pos]); - if (0 == strcmp(arg->value[pos], "indent")) - return(INDENT); - if (0 == strcmp(arg->value[pos], "indent-two")) - return(INDENT * 2); - - if (0 == (len = (int)strlen(arg->value[pos]))) - return(0); - - for (i = 0; i < len - 1; i++) - if ( ! isdigit((u_char)arg->value[pos][i])) - break; - - if (i == len - 1) { - if ('n' == arg->value[pos][len - 1]) { - v = (size_t)atoi(arg->value[pos]); - return(v); - } - - } - return(strlen(arg->value[pos]) + 1); + return(term_alloc(TERMENC_ASCII)); } -static int -arg_listtype(const struct mdoc_node *n) +int +terminal_run(void *arg, const struct man *man, + const struct mdoc *mdoc) { - int i, len; + struct termp *p; - assert(MDOC_BLOCK == n->type); + p = (struct termp *)arg; - len = (int)(n->args ? n->args->argc : 0); + if (NULL == p->symtab) + p->symtab = term_ascii2htab(); - for (i = 0; i < len; i++) - switch (n->args->argv[i].arg) { - case (MDOC_Bullet): - /* FALLTHROUGH */ - case (MDOC_Dash): - /* FALLTHROUGH */ - case (MDOC_Enum): - /* FALLTHROUGH */ - case (MDOC_Hyphen): - /* FALLTHROUGH */ - case (MDOC_Tag): - /* FALLTHROUGH */ - case (MDOC_Inset): - /* FALLTHROUGH */ - case (MDOC_Diag): - /* FALLTHROUGH */ - case (MDOC_Item): - /* FALLTHROUGH */ - case (MDOC_Column): - /* FALLTHROUGH */ - case (MDOC_Ohang): - return(n->args->argv[i].arg); - default: - break; - } + if (man) + return(man_run(p, man)); + if (mdoc) + return(mdoc_run(p, mdoc)); - errx(1, "list type not supported"); - /* NOTREACHED */ + return(1); } -static size_t -arg_offset(const struct mdoc_argv *arg) +void +terminal_free(void *arg) { - /* TODO */ - assert(*arg->value); - if (0 == strcmp(*arg->value, "indent")) - return(INDENT); - if (0 == strcmp(*arg->value, "indent-two")) - return(INDENT * 2); - return(strlen(*arg->value)); + term_free((struct termp *)arg); } -static int -arg_hasattr(int arg, const struct mdoc_node *n) +static void +term_free(struct termp *p) { - return(-1 != arg_getattr(arg, n)); + if (p->buf) + free(p->buf); + if (TERMENC_ASCII == p->enc && p->symtab) + term_asciifree(p->symtab); + + free(p); } -static int -arg_getattr(int v, const struct mdoc_node *n) +static struct termp * +term_alloc(enum termenc enc) { - int val; + struct termp *p; - return(arg_getattrs(&v, &val, 1, n) ? val : -1); + if (NULL == (p = malloc(sizeof(struct termp)))) + err(1, "malloc"); + bzero(p, sizeof(struct termp)); + p->maxrmargin = 78; + p->enc = enc; + return(p); } static int -arg_getattrs(const int *keys, int *vals, - size_t sz, const struct mdoc_node *n) +term_isclosedelim(const char *p, int len) { - int i, j, k; - if (NULL == n->args) + if (1 != len) return(0); - for (k = i = 0; i < (int)n->args->argc; i++) - for (j = 0; j < (int)sz; j++) - if (n->args->argv[i].arg == keys[j]) { - vals[j] = i; - k++; - } - return(k); -} - - -/* ARGSUSED */ -static int -fmt_block_vspace(struct termp *p, - const struct mdoc_node *bl, - const struct mdoc_node *node) -{ - const struct mdoc_node *n; - - term_newln(p); - - if (arg_hasattr(MDOC_Compact, bl)) - return(1); - - for (n = node; n; n = n->parent) { - if (MDOC_BLOCK != n->type) - continue; - if (MDOC_Ss == n->tok) - break; - if (MDOC_Sh == n->tok) - break; - if (NULL == n->prev) - continue; - term_vspace(p); - break; - } - - return(1); -} - - -/* ARGSUSED */ -static int -termp_dq_pre(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return(1); - - term_word(p, "\\(lq"); - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ -static void -termp_dq_post(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return; - - p->flags |= TERMP_NOSPACE; - term_word(p, "\\(rq"); -} - - -/* ARGSUSED */ -static int -termp_it_pre(DECL_ARGS) -{ - const struct mdoc_node *bl, *n; - char buf[7]; - int i, type, keys[3], vals[3]; - size_t width, offset; - - if (MDOC_BLOCK == node->type) - return(fmt_block_vspace(p, node->parent->parent, node)); - - bl = node->parent->parent->parent; - - /* Save parent attributes. */ - - pair->offset = p->offset; - pair->rmargin = p->rmargin; - pair->flag = p->flags; - - /* Get list width and offset. */ - - keys[0] = MDOC_Width; - keys[1] = MDOC_Offset; - keys[2] = MDOC_Column; - - vals[0] = vals[1] = vals[2] = -1; - - width = offset = 0; - - (void)arg_getattrs(keys, vals, 3, bl); - - type = arg_listtype(bl); - - /* Calculate real width and offset. */ - - switch (type) { - case (MDOC_Column): - if (MDOC_BODY == node->type) - break; - for (i = 0, n = node->prev; n; n = n->prev, i++) - offset += arg_width - (&bl->args->argv[vals[2]], i); - assert(i < (int)bl->args->argv[vals[2]].sz); - width = arg_width(&bl->args->argv[vals[2]], i); - if (vals[1] >= 0) - offset += arg_offset(&bl->args->argv[vals[1]]); - break; - default: - if (vals[0] >= 0) - width = arg_width(&bl->args->argv[vals[0]], 0); - if (vals[1] >= 0) - offset = arg_offset(&bl->args->argv[vals[1]]); - break; - } - - /* - * List-type can override the width in the case of fixed-head - * values (bullet, dash/hyphen, enum). Tags need a non-zero - * offset. - */ - - switch (type) { - case (MDOC_Bullet): + switch (*p) { + case('.'): /* FALLTHROUGH */ - case (MDOC_Dash): + case(','): /* FALLTHROUGH */ - case (MDOC_Enum): + case(';'): /* FALLTHROUGH */ - case (MDOC_Hyphen): - if (width < 4) - width = 4; - break; - case (MDOC_Tag): - if (0 == width) - width = 10; - break; - default: - break; - } - - /* - * Whitespace control. Inset bodies need an initial space. - */ - - switch (type) { - case (MDOC_Diag): + case(':'): /* FALLTHROUGH */ - case (MDOC_Inset): - if (MDOC_BODY == node->type) - p->flags &= ~TERMP_NOSPACE; - else - p->flags |= TERMP_NOSPACE; - break; - default: - p->flags |= TERMP_NOSPACE; - break; - } - - /* - * Style flags. Diagnostic heads need TTYPE_DIAG. - */ - - switch (type) { - case (MDOC_Diag): - if (MDOC_HEAD == node->type) - p->flags |= ttypes[TTYPE_DIAG]; - break; - default: - break; - } - - /* - * Pad and break control. This is the tricker part. Lists with - * set right-margins for the head get TERMP_NOBREAK because, if - * they overrun the margin, they wrap to the new margin. - * Correspondingly, the body for these types don't left-pad, as - * the head will pad out to to the right. - */ - - switch (type) { - case (MDOC_Bullet): + case('?'): /* FALLTHROUGH */ - case (MDOC_Dash): + case('!'): /* FALLTHROUGH */ - case (MDOC_Enum): + case(')'): /* FALLTHROUGH */ - case (MDOC_Hyphen): + case(']'): /* FALLTHROUGH */ - case (MDOC_Tag): - if (MDOC_HEAD == node->type) - p->flags |= TERMP_NOBREAK; - else - p->flags |= TERMP_NOLPAD; - if (MDOC_HEAD == node->type && MDOC_Tag == type) - if (NULL == node->next || - NULL == node->next->child) - p->flags |= TERMP_NONOBREAK; - break; - case (MDOC_Column): - if (MDOC_HEAD == node->type) { - assert(node->next); - if (MDOC_BODY == node->next->type) - p->flags &= ~TERMP_NOBREAK; - else - p->flags |= TERMP_NOBREAK; - if (node->prev) - p->flags |= TERMP_NOLPAD; - } - break; - case (MDOC_Diag): - if (MDOC_HEAD == node->type) - p->flags |= TERMP_NOBREAK; - break; + case('}'): + return(1); default: break; } - /* - * Margin control. Set-head-width lists have their right - * margins shortened. The body for these lists has the offset - * necessarily lengthened. Everybody gets the offset. - */ - - p->offset += offset; - - switch (type) { - case (MDOC_Bullet): - /* FALLTHROUGH */ - case (MDOC_Dash): - /* FALLTHROUGH */ - case (MDOC_Enum): - /* FALLTHROUGH */ - case (MDOC_Hyphen): - /* FALLTHROUGH */ - case (MDOC_Tag): - if (MDOC_HEAD == node->type) - p->rmargin = p->offset + width; - else - p->offset += width; - break; - case (MDOC_Column): - p->rmargin = p->offset + width; - break; - default: - break; - } - - /* - * The dash, hyphen, bullet and enum lists all have a special - * HEAD character. Print it now. - */ - - if (MDOC_HEAD == node->type) - switch (type) { - case (MDOC_Bullet): - term_word(p, "\\[bu]"); - break; - case (MDOC_Dash): - /* FALLTHROUGH */ - case (MDOC_Hyphen): - term_word(p, "\\-"); - break; - case (MDOC_Enum): - (pair->ppair->ppair->count)++; - (void)snprintf(buf, sizeof(buf), "%d.", - pair->ppair->ppair->count); - term_word(p, buf); - break; - default: - break; - } - - /* - * If we're not going to process our children, indicate so here. - */ - - switch (type) { - case (MDOC_Bullet): - /* FALLTHROUGH */ - case (MDOC_Item): - /* FALLTHROUGH */ - case (MDOC_Dash): - /* FALLTHROUGH */ - case (MDOC_Hyphen): - /* FALLTHROUGH */ - case (MDOC_Enum): - if (MDOC_HEAD == node->type) - return(0); - break; - case (MDOC_Column): - if (MDOC_BODY == node->type) - return(0); - break; - default: - break; - } - - return(1); + return(0); } -/* ARGSUSED */ -static void -termp_it_post(DECL_ARGS) +static int +term_isopendelim(const char *p, int len) { - int type; - if (MDOC_BODY != node->type && MDOC_HEAD != node->type) - return; + if (1 != len) + return(0); - type = arg_listtype(node->parent->parent->parent); - - switch (type) { - case (MDOC_Diag): + switch (*p) { + case('('): /* FALLTHROUGH */ - case (MDOC_Item): + case('['): /* FALLTHROUGH */ - case (MDOC_Inset): - if (MDOC_BODY == node->type) - term_flushln(p); - break; - case (MDOC_Column): - if (MDOC_HEAD == node->type) - term_flushln(p); - break; + case('{'): + return(1); default: - term_flushln(p); break; } - p->offset = pair->offset; - p->rmargin = pair->rmargin; - p->flags = pair->flag; -} - - -/* ARGSUSED */ -static int -termp_nm_pre(DECL_ARGS) -{ - - if (SEC_SYNOPSIS == node->sec) - term_newln(p); - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_PROG]); - if (NULL == node->child) - term_word(p, meta->name); - - return(1); -} - - -/* ARGSUSED */ -static int -termp_fl_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_CMD_FLAG]); - term_word(p, "\\-"); - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ -static int -termp_ar_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_CMD_ARG]); - return(1); -} - - -/* ARGSUSED */ -static int -termp_ns_pre(DECL_ARGS) -{ - - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ -static int -termp_pp_pre(DECL_ARGS) -{ - - term_vspace(p); - return(1); -} - - -/* ARGSUSED */ -static int -termp_st_pre(DECL_ARGS) -{ - const char *cp; - - if (node->child && (cp = mdoc_a2st(node->child->string))) - term_word(p, cp); return(0); } -/* ARGSUSED */ -static int -termp_rs_pre(DECL_ARGS) +/* + * Flush a line of text. A "line" is loosely defined as being something + * that should be followed by a newline, regardless of whether it's + * broken apart by newlines getting there. A line can also be a + * fragment of a columnar list. + * + * Specifically, a line is whatever's in p->buf of length p->col, which + * is zeroed after this function returns. + * + * The variables TERMP_NOLPAD, TERMP_LITERAL and TERMP_NOBREAK are of + * critical importance here. Their behaviour follows: + * + * - TERMP_NOLPAD: when beginning to write the line, don't left-pad the + * offset value. This is useful when doing columnar lists where the + * prior column has right-padded. + * + * - TERMP_NOBREAK: this is the most important and is used when making + * columns. In short: don't print a newline and instead pad to the + * right margin. Used in conjunction with TERMP_NOLPAD. + * + * - TERMP_NONOBREAK: don't newline when TERMP_NOBREAK is specified. + * + * In-line line breaking: + * + * If TERMP_NOBREAK is specified and the line overruns the right + * margin, it will break and pad-right to the right margin after + * writing. If maxrmargin is violated, it will break and continue + * writing from the right-margin, which will lead to the above + * scenario upon exit. + * + * Otherwise, the line will break at the right margin. Extremely long + * lines will cause the system to emit a warning (TODO: hyphenate, if + * possible). + */ +void +term_flushln(struct termp *p) { + int i, j; + size_t vsz, vis, maxvis, mmax, bp; - if (MDOC_BLOCK == node->type && node->prev) - term_vspace(p); - return(1); -} + /* + * First, establish the maximum columns of "visible" content. + * This is usually the difference between the right-margin and + * an indentation, but can be, for tagged lists or columns, a + * small set of values. + */ + assert(p->offset < p->rmargin); + maxvis = p->rmargin - p->offset; + mmax = p->maxrmargin - p->offset; + bp = TERMP_NOBREAK & p->flags ? mmax : maxvis; + vis = 0; -/* ARGSUSED */ -static int -termp_rv_pre(DECL_ARGS) -{ - int i; - - if (-1 == (i = arg_getattr(MDOC_Std, node))) - errx(1, "expected -std argument"); - if (1 != node->args->argv[i].sz) - errx(1, "expected -std argument"); - - term_newln(p); - term_word(p, "The"); - - p->flags |= ttypes[TTYPE_FUNC_NAME]; - term_word(p, *node->args->argv[i].value); - p->flags &= ~ttypes[TTYPE_FUNC_NAME]; - p->flags |= TERMP_NOSPACE; - - term_word(p, "() function returns the value 0 if successful;"); - term_word(p, "otherwise the value -1 is returned and the"); - term_word(p, "global variable"); - - p->flags |= ttypes[TTYPE_VAR_DECL]; - term_word(p, "errno"); - p->flags &= ~ttypes[TTYPE_VAR_DECL]; - - term_word(p, "is set to indicate the error."); - - return(1); -} - - -/* ARGSUSED */ -static int -termp_ex_pre(DECL_ARGS) -{ - int i; - - if (-1 == (i = arg_getattr(MDOC_Std, node))) - errx(1, "expected -std argument"); - if (1 != node->args->argv[i].sz) - errx(1, "expected -std argument"); - - term_word(p, "The"); - p->flags |= ttypes[TTYPE_PROG]; - term_word(p, *node->args->argv[i].value); - p->flags &= ~ttypes[TTYPE_PROG]; - term_word(p, "utility exits 0 on success, and >0 if an error occurs."); - - return(1); -} - - -/* ARGSUSED */ -static int -termp_nd_pre(DECL_ARGS) -{ - - term_word(p, "\\-"); - return(1); -} - - -/* ARGSUSED */ -static void -termp_bl_post(DECL_ARGS) -{ - - if (MDOC_BLOCK == node->type) - term_newln(p); -} - - -/* ARGSUSED */ -static void -termp_op_post(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return; - p->flags |= TERMP_NOSPACE; - term_word(p, "\\(rB"); -} - - -/* ARGSUSED */ -static int -termp_xr_pre(DECL_ARGS) -{ - const struct mdoc_node *n; - - if (NULL == (n = node->child)) - errx(1, "expected text line argument"); - term_word(p, n->string); - if (NULL == (n = n->next)) - return(0); - p->flags |= TERMP_NOSPACE; - term_word(p, "("); - p->flags |= TERMP_NOSPACE; - term_word(p, n->string); - p->flags |= TERMP_NOSPACE; - term_word(p, ")"); - return(0); -} - - -/* ARGSUSED */ -static int -termp_vt_pre(DECL_ARGS) -{ - - /* FIXME: this can be "type name". */ - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_VAR_DECL]); - return(1); -} - - -/* ARGSUSED */ -static void -termp_vt_post(DECL_ARGS) -{ - - if (node->sec == SEC_SYNOPSIS) - term_vspace(p); -} - - -/* ARGSUSED */ -static int -termp_fd_pre(DECL_ARGS) -{ - - /* - * FIXME: this naming is bad. This value is used, in general, - * for the #include header or other preprocessor statement. + /* + * If in the standard case (left-justified), then begin with our + * indentation, otherwise (columns, etc.) just start spitting + * out text. */ - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_FUNC_DECL]); - return(1); -} + if ( ! (p->flags & TERMP_NOLPAD)) + /* LINTED */ + for (j = 0; j < (int)p->offset; j++) + putchar(' '); -/* ARGSUSED */ -static void -termp_fd_post(DECL_ARGS) -{ + for (i = 0; i < (int)p->col; i++) { + /* + * Count up visible word characters. Control sequences + * (starting with the CSI) aren't counted. A space + * generates a non-printing word, which is valid (the + * space is printed according to regular spacing rules). + */ - if (node->sec != SEC_SYNOPSIS) - return; - term_newln(p); - if (node->next && MDOC_Fd != node->next->tok) - term_vspace(p); -} + /* LINTED */ + for (j = i, vsz = 0; j < (int)p->col; j++) { + if (' ' == p->buf[j]) + break; + else if (8 == p->buf[j]) + j += 1; + else + vsz++; + } + /* + * Do line-breaking. If we're greater than our + * break-point and already in-line, break to the next + * line and start writing. If we're at the line start, + * then write out the word (TODO: hyphenate) and break + * in a subsequent loop invocation. + */ -/* ARGSUSED */ -static int -termp_sh_pre(DECL_ARGS) -{ + if ( ! (TERMP_NOBREAK & p->flags)) { + if (vis && vis + vsz > bp) { + putchar('\n'); + for (j = 0; j < (int)p->offset; j++) + putchar(' '); + vis = 0; + } + } else if (vis && vis + vsz > bp) { + putchar('\n'); + for (j = 0; j < (int)p->rmargin; j++) + putchar(' '); + vis = p->rmargin - p->offset; + } - switch (node->type) { - case (MDOC_HEAD): - term_vspace(p); - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_SECTION]); - break; - case (MDOC_BODY): - p->offset = INDENT; - break; - default: - break; - } - return(1); -} + /* + * Write out the word and a trailing space. Omit the + * space if we're the last word in the line or beyond + * our breakpoint. + */ - -/* ARGSUSED */ -static void -termp_sh_post(DECL_ARGS) -{ - - switch (node->type) { - case (MDOC_HEAD): - term_newln(p); - break; - case (MDOC_BODY): - term_newln(p); - p->offset = 0; - break; - default: - break; + for ( ; i < (int)p->col; i++) { + if (' ' == p->buf[i]) + break; + putchar(p->buf[i]); + } + vis += vsz; + if (i < (int)p->col && vis <= bp) { + putchar(' '); + vis++; + } } -} - -/* ARGSUSED */ -static int -termp_op_pre(DECL_ARGS) -{ - - switch (node->type) { - case (MDOC_BODY): - term_word(p, "\\(lB"); - p->flags |= TERMP_NOSPACE; - break; - default: - break; - } - return(1); -} - - -/* ARGSUSED */ -static int -termp_bt_pre(DECL_ARGS) -{ - - term_word(p, "is currently in beta test."); - return(1); -} - - -/* ARGSUSED */ -static int -termp_lb_pre(DECL_ARGS) -{ - const char *lb; - - if (NULL == node->child) - errx(1, "expected text line argument"); - if ((lb = mdoc_a2lib(node->child->string))) { - term_word(p, lb); - return(0); - } - term_word(p, "library"); - return(1); -} - - -/* ARGSUSED */ -static void -termp_lb_post(DECL_ARGS) -{ - - term_newln(p); -} - - -/* ARGSUSED */ -static int -termp_ud_pre(DECL_ARGS) -{ - - term_word(p, "currently under development."); - return(1); -} - - -/* ARGSUSED */ -static int -termp_d1_pre(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return(1); - term_newln(p); - p->offset += (pair->offset = INDENT); - return(1); -} - - -/* ARGSUSED */ -static void -termp_d1_post(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return; - term_newln(p); - p->offset -= pair->offset; -} - - -/* ARGSUSED */ -static int -termp_aq_pre(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return(1); - term_word(p, "\\(la"); - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ -static void -termp_aq_post(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return; - p->flags |= TERMP_NOSPACE; - term_word(p, "\\(ra"); -} - - -/* ARGSUSED */ -static int -termp_ft_pre(DECL_ARGS) -{ - - if (SEC_SYNOPSIS == node->sec) - if (node->prev && MDOC_Fo == node->prev->tok) - term_vspace(p); - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_FUNC_TYPE]); - return(1); -} - - -/* ARGSUSED */ -static void -termp_ft_post(DECL_ARGS) -{ - - if (SEC_SYNOPSIS == node->sec) - term_newln(p); -} - - -/* ARGSUSED */ -static int -termp_fn_pre(DECL_ARGS) -{ - const struct mdoc_node *n; - - if (NULL == node->child) - errx(1, "expected text line arguments"); - - /* FIXME: can be "type funcname" "type varname"... */ - - p->flags |= ttypes[TTYPE_FUNC_NAME]; - term_word(p, node->child->string); - p->flags &= ~ttypes[TTYPE_FUNC_NAME]; - - p->flags |= TERMP_NOSPACE; - term_word(p, "("); - - for (n = node->child->next; n; n = n->next) { - p->flags |= ttypes[TTYPE_FUNC_ARG]; - term_word(p, n->string); - p->flags &= ~ttypes[TTYPE_FUNC_ARG]; - if (n->next) - term_word(p, ","); - } - - term_word(p, ")"); - - if (SEC_SYNOPSIS == node->sec) - term_word(p, ";"); - - return(0); -} - - -/* ARGSUSED */ -static void -termp_fn_post(DECL_ARGS) -{ - - if (node->sec == SEC_SYNOPSIS && node->next) - term_vspace(p); - -} - - -/* ARGSUSED */ -static int -termp_sx_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_LINK]); - return(1); -} - - -/* ARGSUSED */ -static int -termp_fa_pre(DECL_ARGS) -{ - struct mdoc_node *n; - - if (node->parent->tok != MDOC_Fo) { - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_FUNC_ARG]); - return(1); - } - - for (n = node->child; n; n = n->next) { - p->flags |= ttypes[TTYPE_FUNC_ARG]; - term_word(p, n->string); - p->flags &= ~ttypes[TTYPE_FUNC_ARG]; - if (n->next) - term_word(p, ","); - } - - if (node->child && node->next && node->next->tok == MDOC_Fa) - term_word(p, ","); - - return(0); -} - - -/* ARGSUSED */ -static int -termp_va_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_VAR_DECL]); - return(1); -} - - -/* ARGSUSED */ -static int -termp_bd_pre(DECL_ARGS) -{ - int i, type, ln; - /* - * This is fairly tricky due primarily to crappy documentation. - * If -ragged or -filled are specified, the block does nothing - * but change the indentation. - * - * If, on the other hand, -unfilled or -literal are specified, - * then the game changes. Text is printed exactly as entered in - * the display: if a macro line, a newline is appended to the - * line. Blank lines are allowed. + * If we've overstepped our maximum visible no-break space, then + * cause a newline and offset at the right margin. */ - if (MDOC_BLOCK == node->type) - return(fmt_block_vspace(p, node, node)); - else if (MDOC_BODY != node->type) - return(1); - - if (NULL == node->parent->args) - errx(1, "missing display type"); - - pair->offset = p->offset; - - for (type = -1, i = 0; - i < (int)node->parent->args->argc; i++) { - switch (node->parent->args->argv[i].arg) { - case (MDOC_Ragged): - /* FALLTHROUGH */ - case (MDOC_Filled): - /* FALLTHROUGH */ - case (MDOC_Unfilled): - /* FALLTHROUGH */ - case (MDOC_Literal): - type = node->parent->args->argv[i].arg; - i = (int)node->parent->args->argc; - break; - default: - break; + if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) { + if ( ! (TERMP_NONOBREAK & p->flags)) { + putchar('\n'); + for (i = 0; i < (int)p->rmargin; i++) + putchar(' '); } + p->col = 0; + return; } - if (NULL == node->parent->args) - errx(1, "missing display type"); - - i = arg_getattr(MDOC_Offset, node->parent); - if (-1 != i) { - if (1 != node->parent->args->argv[i].sz) - errx(1, "expected single value"); - p->offset += arg_offset(&node->parent->args->argv[i]); - } - - switch (type) { - case (MDOC_Literal): - /* FALLTHROUGH */ - case (MDOC_Unfilled): - break; - default: - return(1); - } - /* - * Tricky. Iterate through all children. If we're on a - * different parse line, append a newline and then the contents. - * Ew. + * If we're not to right-marginalise it (newline), then instead + * pad to the right margin and stay off. */ - p->flags |= TERMP_LITERAL; - ln = node->child ? node->child->line : 0; + if (p->flags & TERMP_NOBREAK) { + if ( ! (TERMP_NONOBREAK & p->flags)) + for ( ; vis < maxvis; vis++) + putchar(' '); + } else + putchar('\n'); - for (node = node->child; node; node = node->next) { - if (ln < node->line) { - term_flushln(p); - p->flags |= TERMP_NOSPACE; - } - ln = node->line; - term_node(p, pair, meta, node); - } - - return(0); + p->col = 0; } -/* ARGSUSED */ -static void -termp_bd_post(DECL_ARGS) +/* + * A newline only breaks an existing line; it won't assert vertical + * space. All data in the output buffer is flushed prior to the newline + * assertion. + */ +void +term_newln(struct termp *p) { - if (MDOC_BODY != node->type) + p->flags |= TERMP_NOSPACE; + if (0 == p->col) { + p->flags &= ~TERMP_NOLPAD; return; - + } term_flushln(p); - p->flags &= ~TERMP_LITERAL; - p->offset = pair->offset; - p->flags |= TERMP_NOSPACE; + p->flags &= ~TERMP_NOLPAD; } -/* ARGSUSED */ -static int -termp_qq_pre(DECL_ARGS) +/* + * Asserts a vertical space (a full, empty line-break between lines). + * Note that if used twice, this will cause two blank spaces and so on. + * All data in the output buffer is flushed prior to the newline + * assertion. + */ +void +term_vspace(struct termp *p) { - if (MDOC_BODY != node->type) - return(1); - term_word(p, "\""); - p->flags |= TERMP_NOSPACE; - return(1); + term_newln(p); + putchar('\n'); } -/* ARGSUSED */ -static void -termp_qq_post(DECL_ARGS) +/* + * Break apart a word into "pwords" (partial-words, usually from + * breaking up a phrase into individual words) and, eventually, put them + * into the output buffer. If we're a literal word, then don't break up + * the word and put it verbatim into the output buffer. + */ +void +term_word(struct termp *p, const char *word) { + int i, j, len; - if (MDOC_BODY != node->type) - return; - p->flags |= TERMP_NOSPACE; - term_word(p, "\""); -} + len = (int)strlen(word); - -/* ARGSUSED */ -static int -termp_bsx_pre(DECL_ARGS) -{ - - term_word(p, "BSDI BSD/OS"); - return(1); -} - - -/* ARGSUSED */ -static void -termp_bx_post(DECL_ARGS) -{ - - if (node->child) - p->flags |= TERMP_NOSPACE; - term_word(p, "BSD"); -} - - -/* ARGSUSED */ -static int -termp_ox_pre(DECL_ARGS) -{ - - term_word(p, "OpenBSD"); - return(1); -} - - -/* ARGSUSED */ -static int -termp_dx_pre(DECL_ARGS) -{ - - term_word(p, "DragonFly"); - return(1); -} - - -/* ARGSUSED */ -static int -termp_ux_pre(DECL_ARGS) -{ - - term_word(p, "UNIX"); - return(1); -} - - -/* ARGSUSED */ -static int -termp_fx_pre(DECL_ARGS) -{ - - term_word(p, "FreeBSD"); - return(1); -} - - -/* ARGSUSED */ -static int -termp_nx_pre(DECL_ARGS) -{ - - term_word(p, "NetBSD"); - return(1); -} - - -/* ARGSUSED */ -static int -termp_sq_pre(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return(1); - term_word(p, "\\(oq"); - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ -static void -termp_sq_post(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) + if (p->flags & TERMP_LITERAL) { + term_pword(p, word, len); return; - p->flags |= TERMP_NOSPACE; - term_word(p, "\\(aq"); -} - - -/* ARGSUSED */ -static int -termp_pf_pre(DECL_ARGS) -{ - - p->flags |= TERMP_IGNDELIM; - return(1); -} - - -/* ARGSUSED */ -static void -termp_pf_post(DECL_ARGS) -{ - - p->flags &= ~TERMP_IGNDELIM; - p->flags |= TERMP_NOSPACE; -} - - -/* ARGSUSED */ -static int -termp_ss_pre(DECL_ARGS) -{ - - switch (node->type) { - case (MDOC_BLOCK): - term_newln(p); - if (node->prev) - term_vspace(p); - break; - case (MDOC_HEAD): - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_SSECTION]); - p->offset = INDENT / 2; - break; - default: - break; } - return(1); -} + /* LINTED */ + for (j = i = 0; i < len; i++) { + if (' ' != word[i]) { + j++; + continue; + } + + /* Escaped spaces don't delimit... */ + if (i && ' ' == word[i] && '\\' == word[i - 1]) { + j++; + continue; + } - -/* ARGSUSED */ -static void -termp_ss_post(DECL_ARGS) -{ - - switch (node->type) { - case (MDOC_HEAD): - term_newln(p); - p->offset = INDENT; - break; - default: - break; + if (0 == j) + continue; + assert(i >= j); + term_pword(p, &word[i - j], j); + j = 0; } + if (j > 0) { + assert(i >= j); + term_pword(p, &word[i - j], j); + } } -/* ARGSUSED */ -static int -termp_pa_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_FILE]); - return(1); -} - - -/* ARGSUSED */ -static int -termp_em_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_EMPH]); - return(1); -} - - -/* ARGSUSED */ -static int -termp_cd_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_CONFIG]); - term_newln(p); - return(1); -} - - -/* ARGSUSED */ -static int -termp_cm_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_CMD_FLAG]); - return(1); -} - - -/* ARGSUSED */ -static int -termp_ic_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_CMD]); - return(1); -} - - -/* ARGSUSED */ -static int -termp_in_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_INCLUDE]); - term_word(p, "#include"); - term_word(p, "<"); - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ +/* + * Determine the symbol indicated by an escape sequences, that is, one + * starting with a backslash. Once done, we pass this value into the + * output buffer by way of the symbol table. + */ static void -termp_in_post(DECL_ARGS) +term_nescape(struct termp *p, const char *word, size_t len) { + const char *rhs; + size_t sz; - p->flags |= TERMP_NOSPACE; - term_word(p, ">"); - - term_newln(p); - if (SEC_SYNOPSIS != node->sec) + if (NULL == (rhs = term_a2ascii(p->symtab, word, len, &sz))) return; - if (node->next && MDOC_In != node->next->tok) - term_vspace(p); + term_stringa(p, rhs, sz); } -/* ARGSUSED */ -static int -termp_at_pre(DECL_ARGS) -{ - const char *att; - - att = NULL; - - if (node->child) - att = mdoc_a2att(node->child->string); - if (NULL == att) - att = "AT&T UNIX"; - - term_word(p, att); - return(0); -} - - -/* ARGSUSED */ -static int -termp_brq_pre(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return(1); - term_word(p, "\\(lC"); - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ +/* + * Handle an escape sequence: determine its length and pass it to the + * escape-symbol look table. Note that we assume mdoc(3) has validated + * the escape sequence (we assert upon badly-formed escape sequences). + */ static void -termp_brq_post(DECL_ARGS) +term_pescape(struct termp *p, const char *word, int *i, int len) { + int j; - if (MDOC_BODY != node->type) + if (++(*i) >= len) return; - p->flags |= TERMP_NOSPACE; - term_word(p, "\\(rC"); -} + if ('(' == word[*i]) { + (*i)++; + if (*i + 1 >= len) + return; -/* ARGSUSED */ -static int -termp_bq_pre(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) - return(1); - term_word(p, "\\(lB"); - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ -static void -termp_bq_post(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) + term_nescape(p, &word[*i], 2); + (*i)++; return; - p->flags |= TERMP_NOSPACE; - term_word(p, "\\(rB"); -} + } else if ('*' == word[*i]) { + (*i)++; + if (*i >= len) + return; -/* ARGSUSED */ -static int -termp_pq_pre(DECL_ARGS) -{ + switch (word[*i]) { + case ('('): + (*i)++; + if (*i + 1 >= len) + return; - if (MDOC_BODY != node->type) - return(1); - term_word(p, "\\&("); - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ -static void -termp_pq_post(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) + term_nescape(p, &word[*i], 2); + (*i)++; + return; + case ('['): + break; + default: + term_nescape(p, &word[*i], 1); + return; + } + + } else if ('f' == word[*i]) { + (*i)++; + if (*i >= len) + return; + switch (word[*i]) { + case ('B'): + p->flags |= TERMP_BOLD; + break; + case ('I'): + p->flags |= TERMP_UNDER; + break; + case ('P'): + /* FALLTHROUGH */ + case ('R'): + p->flags &= ~TERMP_STYLE; + break; + default: + break; + } return; - term_word(p, ")"); -} - -/* ARGSUSED */ -static int -termp_fo_pre(DECL_ARGS) -{ - const struct mdoc_node *n; - - if (MDOC_BODY == node->type) { - term_word(p, "("); - p->flags |= TERMP_NOSPACE; - return(1); - } else if (MDOC_HEAD != node->type) - return(1); - - /* XXX - groff shows only first parameter */ - - p->flags |= ttypes[TTYPE_FUNC_NAME]; - for (n = node->child; n; n = n->next) { - if (MDOC_TEXT != n->type) - errx(1, "expected text line argument"); - term_word(p, n->string); + } else if ('[' != word[*i]) { + term_nescape(p, &word[*i], 1); + return; } - p->flags &= ~ttypes[TTYPE_FUNC_NAME]; - return(0); -} + (*i)++; + for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++) + /* Loop... */ ; - -/* ARGSUSED */ -static void -termp_fo_post(DECL_ARGS) -{ - - if (MDOC_BODY != node->type) + if (0 == word[*i]) return; - p->flags |= TERMP_NOSPACE; - term_word(p, ")"); - p->flags |= TERMP_NOSPACE; - term_word(p, ";"); - term_newln(p); -} - -/* ARGSUSED */ -static int -termp_bf_pre(DECL_ARGS) -{ - const struct mdoc_node *n; - - if (MDOC_HEAD == node->type) { - return(0); - } else if (MDOC_BLOCK != node->type) - return(1); - - if (NULL == (n = node->head->child)) { - if (arg_hasattr(MDOC_Emphasis, node)) - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_EMPH]); - else if (arg_hasattr(MDOC_Symbolic, node)) - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_SYMB]); - - return(1); - } - - if (MDOC_TEXT != n->type) - errx(1, "expected text line arguments"); - - if (0 == strcmp("Em", n->string)) - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_EMPH]); - else if (0 == strcmp("Sy", n->string)) - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_EMPH]); - - return(1); + term_nescape(p, &word[*i - j], (size_t)j); } -/* ARGSUSED */ -static int -termp_sy_pre(DECL_ARGS) +/* + * Handle pwords, partial words, which may be either a single word or a + * phrase that cannot be broken down (such as a literal string). This + * handles word styling. + */ +static void +term_pword(struct termp *p, const char *word, int len) { + int i; - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_SYMB]); - return(1); -} + if (term_isclosedelim(word, len)) + if ( ! (TERMP_IGNDELIM & p->flags)) + p->flags |= TERMP_NOSPACE; + if ( ! (TERMP_NOSPACE & p->flags)) + term_chara(p, ' '); -/* ARGSUSED */ -static int -termp_ms_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_SYMBOL]); - return(1); -} - - - -/* ARGSUSED */ -static int -termp_sm_pre(DECL_ARGS) -{ - - if (NULL == node->child || MDOC_TEXT != node->child->type) - errx(1, "expected boolean line argument"); - - if (0 == strcmp("on", node->child->string)) { - p->flags &= ~TERMP_NONOSPACE; + if ( ! (p->flags & TERMP_NONOSPACE)) p->flags &= ~TERMP_NOSPACE; - } else - p->flags |= TERMP_NONOSPACE; - return(0); -} + /* + * If ANSI (word-length styling), then apply our style now, + * before the word. + */ + for (i = 0; i < len; i++) { + if ('\\' == word[i]) { + term_pescape(p, word, &i, len); + continue; + } -/* ARGSUSED */ -static int -termp_ap_pre(DECL_ARGS) -{ + if (TERMP_STYLE & p->flags) { + if (TERMP_BOLD & p->flags) { + term_chara(p, word[i]); + term_chara(p, 8); + } + if (TERMP_UNDER & p->flags) { + term_chara(p, '_'); + term_chara(p, 8); + } + } - p->flags |= TERMP_NOSPACE; - term_word(p, "\\(aq"); - p->flags |= TERMP_NOSPACE; - return(1); -} + term_chara(p, word[i]); + } - -/* ARGSUSED */ -static int -termp__j_pre(DECL_ARGS) -{ - - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_REF_JOURNAL]); - return(1); + if (term_isopendelim(word, len)) + p->flags |= TERMP_NOSPACE; } -/* ARGSUSED */ -static int -termp__t_pre(DECL_ARGS) -{ - - term_word(p, "\""); - p->flags |= TERMP_NOSPACE; - return(1); -} - - -/* ARGSUSED */ +/* + * Like term_chara() but for arbitrary-length buffers. Resize the + * buffer by a factor of two (if the buffer is less than that) or the + * buffer's size. + */ static void -termp__t_post(DECL_ARGS) +term_stringa(struct termp *p, const char *c, size_t sz) { + size_t s; - p->flags |= TERMP_NOSPACE; - term_word(p, "\""); - termp____post(p, pair, meta, node); -} + if (0 == sz) + return; - -/* ARGSUSED */ -static void -termp____post(DECL_ARGS) -{ - - p->flags |= TERMP_NOSPACE; - term_word(p, node->next ? "," : "."); -} - - -/* ARGSUSED */ -static int -termp_lk_pre(DECL_ARGS) -{ - const struct mdoc_node *n; - - if (NULL == (n = node->child)) - errx(1, "expected line argument"); - - p->flags |= ttypes[TTYPE_LINK_ANCHOR]; - term_word(p, n->string); - p->flags &= ~ttypes[TTYPE_LINK_ANCHOR]; - p->flags |= TERMP_NOSPACE; - term_word(p, ":"); - - p->flags |= ttypes[TTYPE_LINK_TEXT]; - for ( ; n; n = n->next) { - term_word(p, n->string); + assert(c); + if (p->col + sz >= p->maxcols) { + if (0 == p->maxcols) + p->maxcols = 256; + s = sz > p->maxcols * 2 ? sz : p->maxcols * 2; + p->buf = realloc(p->buf, s); + if (NULL == p->buf) + err(1, "realloc"); + p->maxcols = s; } - p->flags &= ~ttypes[TTYPE_LINK_TEXT]; - return(0); + (void)memcpy(&p->buf[(int)p->col], c, sz); + p->col += sz; } -/* ARGSUSED */ -static int -termp_mt_pre(DECL_ARGS) +/* + * Insert a single character into the line-buffer. If the buffer's + * space is exceeded, then allocate more space by doubling the buffer + * size. + */ +static void +term_chara(struct termp *p, char c) { + size_t s; - TERMPAIR_SETFLAG(p, pair, ttypes[TTYPE_LINK_ANCHOR]); - return(1); + if (p->col + 1 >= p->maxcols) { + if (0 == p->maxcols) + p->maxcols = 256; + s = p->maxcols * 2; + p->buf = realloc(p->buf, s); + if (NULL == p->buf) + err(1, "realloc"); + p->maxcols = s; + } + p->buf[(int)(p->col)++] = c; }