-/* $Id: man.c,v 1.54 2010/03/24 20:10:53 kristaps Exp $ */
+/* $Id: man.c,v 1.65 2010/05/12 16:46:28 kristaps Exp $ */
/*
* Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
*
"literal context already open", /* WOLITERAL */
"no literal context open", /* WNLITERAL */
"invalid nesting of roff declarations", /* WROFFNEST */
+ "scope in roff instructions broken", /* WROFFSCOPE */
+ "document title should be uppercase", /* WTITLECASE */
+ "deprecated comment style", /* WBADCOMMENT */
};
const char *const __man_macronames[MAN_MAX] = {
static int man_pmacro(struct man *, int, char *);
static void man_free1(struct man *);
static void man_alloc1(struct man *);
-static int pstring(struct man *, int, int,
- const char *, size_t);
static int macrowarn(struct man *, int, const char *);
man_parseln(struct man *m, int ln, char *buf)
{
- return('.' == *buf ?
+ return('.' == *buf || '\'' == *buf ?
man_pmacro(m, ln, buf) :
man_ptext(m, ln, buf));
}
}
-static int
-pstring(struct man *m, int line, int pos,
- const char *p, size_t len)
+int
+man_word_alloc(struct man *m, int line, int pos, const char *word)
{
struct man_node *n;
- size_t sv;
+ size_t sv, len;
+
+ len = strlen(word);
n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
n->string = mandoc_malloc(len + 1);
- sv = strlcpy(n->string, p, len + 1);
+ sv = strlcpy(n->string, word, len + 1);
/* Prohibit truncation. */
assert(sv < len + 1);
if ( ! man_node_append(m, n))
return(0);
+
m->next = MAN_NEXT_SIBLING;
return(1);
}
-int
-man_word_alloc(struct man *m, int line, int pos, const char *word)
-{
-
- return(pstring(m, line, pos, word, strlen(word)));
-}
-
-
/*
* Free all of the resources held by a node. This does NOT unlink a
* node from its context; for that, see man_node_unlink().
static int
man_ptext(struct man *m, int line, char *buf)
{
- int i, j;
- char sv;
+ int i;
+
+ /* Ignore bogus comments. */
+
+ if ('\\' == buf[0] && '.' == buf[1] && '\"' == buf[2])
+ return(man_pwarn(m, line, 0, WBADCOMMENT));
/* Literal free-form text whitespace is preserved. */
goto descope;
}
- /* First de-chunk and allocate words. */
+ /* Pump blank lines directly into the backend. */
for (i = 0; ' ' == buf[i]; i++)
/* Skip leading whitespace. */ ;
if ('\0' == buf[i]) {
- /* Trailing whitespace? */
- if (i && ' ' == buf[i - 1])
- if ( ! man_pwarn(m, line, i - 1, WTSPACE))
- return(0);
- if ( ! pstring(m, line, 0, &buf[i], 0))
+ /* Allocate a blank entry. */
+ if ( ! man_word_alloc(m, line, 0, ""))
return(0);
goto descope;
}
- for (j = i; buf[i]; i++) {
- if (' ' != buf[i])
- continue;
-
- /* Escaped whitespace. */
- if (i && ' ' == buf[i] && '\\' == buf[i - 1])
- continue;
-
- sv = buf[i];
- buf[i++] = '\0';
-
- if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
- return(0);
+ /*
+ * Warn if the last un-escaped character is whitespace. Then
+ * strip away the remaining spaces (tabs stay!).
+ */
- /* Trailing whitespace? Check at overwritten byte. */
+ i = (int)strlen(buf);
+ assert(i);
- if (' ' == sv && '\0' == buf[i])
+ if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
+ if (i > 1 && '\\' != buf[i - 2])
if ( ! man_pwarn(m, line, i - 1, WTSPACE))
return(0);
- for ( ; ' ' == buf[i]; i++)
- /* Skip trailing whitespace. */ ;
+ for (--i; i && ' ' == buf[i]; i--)
+ /* Spin back to non-space. */ ;
- j = i;
+ /* Jump ahead of escaped whitespace. */
+ i += '\\' == buf[i] ? 2 : 1;
- /* Trailing whitespace? */
+ buf[i] = '\0';
+ }
- if (' ' == buf[i - 1] && '\0' == buf[i])
- if ( ! man_pwarn(m, line, i - 1, WTSPACE))
- return(0);
+ if ( ! man_word_alloc(m, line, 0, buf))
+ return(0);
- if ('\0' == buf[i])
+ /*
+ * End-of-sentence check. If the last character is an unescaped
+ * EOS character, then flag the node as being the end of a
+ * sentence. The front-end will know how to interpret this.
+ */
+
+ assert(i);
+
+ switch (buf[i - 1]) {
+ case ('.'):
+ if (i > 1 && '\\' == buf[i - 2])
break;
- }
+ /* FALLTHROUGH */
+ case ('!'):
+ /* FALLTHROUGH */
+ case ('?'):
+ m->last->flags |= MAN_EOS;
+ break;
+ default:
+ break;
- if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
- return(0);
+ }
descope:
-
/*
* Co-ordinate what happens with having a next-line scope open:
* first close out the element scope (if applicable), then close
if (MAN_ELINE & m->flags) {
m->flags &= ~MAN_ELINE;
- if ( ! man_unscope(m, m->last->parent))
+ if ( ! man_unscope(m, m->last->parent, WERRMAX))
return(0);
}
return(1);
m->flags &= ~MAN_BLINE;
- if ( ! man_unscope(m, m->last->parent))
+ if ( ! man_unscope(m, m->last->parent, WERRMAX))
return(0);
return(man_body_alloc(m, line, 0, m->last->tok));
}
macrowarn(struct man *m, int ln, const char *buf)
{
if ( ! (MAN_IGN_MACRO & m->pflags))
- return(man_verr(m, ln, 0,
- "unknown macro: %s%s",
+ return(man_verr(m, ln, 0, "unknown macro: %s%s",
buf, strlen(buf) > 3 ? "..." : ""));
return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
buf, strlen(buf) > 3 ? "..." : ""));
int
man_pmacro(struct man *m, int ln, char *buf)
{
- int i, j, ppos, fl;
+ int i, j, ppos;
enum mant tok;
char mac[5];
struct man_node *n;
/* Comments and empties are quickly ignored. */
- fl = m->flags;
-
if ('\0' == buf[1])
return(1);
i = 1;
- if (' ' == buf[i]) {
+ /*
+ * Skip whitespace between the control character and initial
+ * text. "Whitespace" is both spaces and tabs.
+ */
+
+ if (' ' == buf[i] || '\t' == buf[i]) {
i++;
- while (buf[i] && ' ' == buf[i])
+ while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
i++;
if ('\0' == buf[i])
goto out;
while (buf[i] && ' ' == buf[i])
i++;
- /* Trailing whitespace? */
+ /*
+ * Trailing whitespace. Note that tabs are allowed to be passed
+ * into the parser as "text", so we only warn about spaces here.
+ */
if ('\0' == buf[i] && ' ' == buf[i - 1])
if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
* Remove prior ELINE macro, as it's being clobbering by a new
* macro. Note that NSCOPED macros do not close out ELINE
* macros---they don't print text---so we let those slip by.
+ * NOTE: we don't allow roff blocks (NOCLOSE) to be embedded
+ * here because that would stipulate blocks as children of
+ * elements!
*/
if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
m->flags &= ~MAN_ELINE;
}
- /* Begin recursive parse sequence. */
+ /*
+ * Save the fact that we're in the next-line for a block. In
+ * this way, embedded roff instructions can "remember" state
+ * when they exit.
+ */
+
+ if (MAN_BLINE & m->flags)
+ m->flags |= MAN_BPLINE;
- assert(man_macros[tok].fp);
+ /* Call to handler... */
+ assert(man_macros[tok].fp);
if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
goto err;
/*
* We weren't in a block-line scope when entering the
* above-parsed macro, so return.
- *
- * FIXME: this prohibits the nesting of blocks (e.g., `de' and
- * family) within BLINE or ELINE systems. This is annoying.
*/
- if ( ! (MAN_BLINE & fl)) {
+ if ( ! (MAN_BPLINE & m->flags)) {
m->flags &= ~MAN_ILINE;
return(1);
}
+ m->flags &= ~MAN_BPLINE;
/*
* If we're in a block scope, then allow this macro to slip by
assert(MAN_BLINE & m->flags);
m->flags &= ~MAN_BLINE;
- if ( ! man_unscope(m, m->last->parent))
+ if ( ! man_unscope(m, m->last->parent, WERRMAX))
return(0);
return(man_body_alloc(m, ln, 0, m->last->tok));
git.ckatri.com / mandoc.git / blobdiff