diff options
author | Ingo Schwarze <schwarze@openbsd.org> | 2018-03-16 15:05:44 +0000 |
---|---|---|
committer | Ingo Schwarze <schwarze@openbsd.org> | 2018-03-16 15:05:44 +0000 |
commit | bbd260dcd38e0e50fc5c35b6d65bf2caf4459198 (patch) | |
tree | 176c4b9e670190db28d74ef1c446e7d5121837b4 /mdoc_validate.c | |
parent | 9fe450461bc85144434bb25f42f7c9f6bafa8da4 (diff) | |
download | mandoc-bbd260dcd38e0e50fc5c35b6d65bf2caf4459198.tar.gz mandoc-bbd260dcd38e0e50fc5c35b6d65bf2caf4459198.tar.zst mandoc-bbd260dcd38e0e50fc5c35b6d65bf2caf4459198.zip |
Style message about bad input encoding of em-dashes as -- instead of \(em.
Suggested by Thomas Klausner <wiz at NetBSD>; discussed with jmc@.
Diffstat (limited to 'mdoc_validate.c')
-rw-r--r-- | mdoc_validate.c | 75 |
1 files changed, 66 insertions, 9 deletions
diff --git a/mdoc_validate.c b/mdoc_validate.c index 642a670a..a6e3d5e7 100644 --- a/mdoc_validate.c +++ b/mdoc_validate.c @@ -1,7 +1,7 @@ -/* $Id: mdoc_validate.c,v 1.354 2018/02/06 16:29:57 schwarze Exp $ */ +/* $Id: mdoc_validate.c,v 1.355 2018/03/16 15:05:44 schwarze Exp $ */ /* * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv> - * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org> + * Copyright (c) 2010-2018 Ingo Schwarze <schwarze@openbsd.org> * Copyright (c) 2010 Joerg Sonnenberger <joerg@netbsd.org> * * Permission to use, copy, modify, and distribute this software for any @@ -53,10 +53,11 @@ enum check_ineq { typedef void (*v_post)(POST_ARGS); static int build_list(struct roff_man *, int); -static void check_text(struct roff_man *, int, int, char *); static void check_argv(struct roff_man *, struct roff_node *, struct mdoc_argv *); static void check_args(struct roff_man *, struct roff_node *); +static void check_text(struct roff_man *, int, int, char *); +static void check_text_em(struct roff_man *, int, int, char *); static void check_toptext(struct roff_man *, int, int, const char *); static int child_an(const struct roff_node *); static size_t macro2len(enum roff_tok); @@ -288,7 +289,7 @@ static const char * const secnames[SEC__MAX] = { void mdoc_node_validate(struct roff_man *mdoc) { - struct roff_node *n; + struct roff_node *n, *np; const v_post *p; n = mdoc->last; @@ -305,13 +306,18 @@ mdoc_node_validate(struct roff_man *mdoc) mdoc->next = ROFF_NEXT_SIBLING; switch (n->type) { case ROFFT_TEXT: + np = n->parent; if (n->sec != SEC_SYNOPSIS || - (n->parent->tok != MDOC_Cd && n->parent->tok != MDOC_Fd)) + (np->tok != MDOC_Cd && np->tok != MDOC_Fd)) check_text(mdoc, n->line, n->pos, n->string); - if (n->parent->tok == MDOC_It || - (n->parent->type == ROFFT_BODY && - (n->parent->tok == MDOC_Sh || - n->parent->tok == MDOC_Ss))) + if (np->tok != MDOC_Ql && np->tok != MDOC_Dl && + (np->tok != MDOC_Bd || + (mdoc->flags & MDOC_LITERAL) == 0) && + (np->tok != MDOC_It || np->type != ROFFT_HEAD || + np->parent->parent->norm->Bl.type != LIST_diag)) + check_text_em(mdoc, n->line, n->pos, n->string); + if (np->tok == MDOC_It || (np->type == ROFFT_BODY && + (np->tok == MDOC_Sh || np->tok == MDOC_Ss))) check_toptext(mdoc, n->line, n->pos, n->string); break; case ROFFT_EQN: @@ -395,6 +401,57 @@ check_text(struct roff_man *mdoc, int ln, int pos, char *p) } static void +check_text_em(struct roff_man *mdoc, int ln, int pos, char *p) +{ + const struct roff_node *np, *nn; + char *cp; + + np = mdoc->last->prev; + nn = mdoc->last->next; + + /* Look for em-dashes wrongly encoded as "--". */ + + for (cp = p; *cp != '\0'; cp++) { + if (*cp != '-' || *++cp != '-') + continue; + + /* Skip input sequences of more than two '-'. */ + + if (cp[1] == '-') { + while (cp[1] == '-') + cp++; + continue; + } + + /* Skip "--" directly attached to something else. */ + + if ((cp - p > 1 && cp[-2] != ' ') || + (cp[1] != '\0' && cp[1] != ' ')) + continue; + + /* Require a letter right before or right afterwards. */ + + if ((cp - p > 2 ? + isalpha((unsigned char)cp[-3]) : + np != NULL && + np->type == ROFFT_TEXT && + np->string != '\0' && + isalpha((unsigned char)np->string[ + strlen(np->string) - 1])) || + (cp[2] != '\0' ? + isalpha((unsigned char)cp[2]) : + nn != NULL && + nn->type == ROFFT_TEXT && + nn->string != '\0' && + isalpha((unsigned char)*nn->string))) { + mandoc_msg(MANDOCERR_DASHDASH, mdoc->parse, + ln, pos + (int)(cp - p) - 1, NULL); + break; + } + } +} + +static void check_toptext(struct roff_man *mdoc, int ln, int pos, const char *p) { const char *cp, *cpr; |