-/* $Id: read.c,v 1.49 2014/06/20 17:24:00 schwarze Exp $ */
+/* $Id: read.c,v 1.122 2015/02/04 18:03:48 schwarze Exp $ */
/*
* Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
- * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
* Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
*
* Permission to use, copy, modify, and distribute this software for any
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
-#ifdef HAVE_CONFIG_H
#include "config.h"
-#endif
-#ifdef HAVE_MMAP
-# include <sys/stat.h>
-# include <sys/mman.h>
+#include <sys/types.h>
+#if HAVE_MMAP
+#include <sys/mman.h>
+#include <sys/stat.h>
#endif
+#include <sys/wait.h>
#include <assert.h>
#include <ctype.h>
#include "libmandoc.h"
#include "mdoc.h"
#include "man.h"
-#include "main.h"
#define REPARSE_LIMIT 1000
-struct buf {
- char *buf; /* binary input buffer */
- size_t sz; /* size of binary buffer */
-};
-
struct mparse {
- enum mandoclevel file_status; /* status of current parse */
- enum mandoclevel wlevel; /* ignore messages below this */
- int line; /* line number in the file */
- int options; /* parser options */
struct man *pman; /* persistent man parser */
struct mdoc *pmdoc; /* persistent mdoc parser */
struct man *man; /* man parser */
struct mdoc *mdoc; /* mdoc parser */
struct roff *roff; /* roff parser (!NULL) */
+ const struct mchars *mchars; /* character table */
char *sodest; /* filename pointed to by .so */
- int reparse_count; /* finite interp. stack */
+ const char *file; /* filename of current input file */
+ struct buf *primary; /* buffer currently being parsed */
+ struct buf *secondary; /* preprocessed copy of input */
+ const char *defos; /* default operating system */
mandocmsg mmsg; /* warning/error message handler */
- const char *file;
- struct buf *secondary;
- char *defos; /* default operating system */
+ enum mandoclevel file_status; /* status of current parse */
+ enum mandoclevel wlevel; /* ignore messages below this */
+ int options; /* parser options */
+ int filenc; /* encoding of the current file */
+ int reparse_count; /* finite interp. stack */
+ int line; /* line number in the file */
+ pid_t child; /* the gunzip(1) process */
};
+static void choose_parser(struct mparse *);
static void resize_buf(struct buf *, size_t);
-static void mparse_buf_r(struct mparse *, struct buf, int);
-static void pset(const char *, int, struct mparse *);
+static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
static int read_whole_file(struct mparse *, const char *, int,
struct buf *, int *);
static void mparse_end(struct mparse *);
MANDOCERR_WARNING,
MANDOCERR_WARNING,
MANDOCERR_ERROR,
- MANDOCERR_FATAL,
+ MANDOCERR_UNSUPP,
MANDOCERR_MAX,
MANDOCERR_MAX
};
"generic warning",
/* related to the prologue */
- "no TH macro in document",
- "document title should be all caps",
+ "missing manual title, using UNTITLED",
+ "missing manual title, using \"\"",
+ "lower case character in document title",
+ "missing manual section, using \"\"",
"unknown manual section",
- "unknown manual volume or arch",
- "date missing, using today's date",
+ "missing date, using today's date",
"cannot parse date, using it verbatim",
- "prologue macros out of order",
+ "missing Os macro, using \"\"",
"duplicate prologue macro",
- "macro not allowed in prologue",
- "macro not allowed in body",
+ "late prologue macro",
+ "skipping late title macro",
+ "prologue macros out of order",
/* related to document structure */
".so is fragile, better use ln(1)",
- "NAME section must come first",
+ "no document body",
+ "content before first section header",
+ "first section is not \"NAME\"",
"bad NAME section contents",
+ "missing description line, using \"\"",
"sections out of conventional order",
- "duplicate section name",
- "section header suited to sections 2, 3, and 9 only",
+ "duplicate section title",
+ "unexpected section",
+ "unusual Xr order",
+ "unusual Xr punctuation",
+ "AUTHORS section without An macro",
/* related to macros and nesting */
- "skipping obsolete macro",
+ "obsolete macro",
+ "macro neither callable nor escaped",
"skipping paragraph macro",
"moving paragraph macro out of list",
"skipping no-space macro",
"blocks badly nested",
- "child violates parent syntax",
"nested displays are not portable",
- "already in literal mode",
+ "moving content out of list",
+ ".Vt block has child macro",
+ "fill mode already enabled, skipping",
+ "fill mode already disabled, skipping",
"line scope broken",
/* related to missing macro arguments */
+ "skipping empty request",
+ "conditional request controls empty scope",
"skipping empty macro",
+ "empty argument, using 0n",
"argument count wrong",
- "missing display type",
- "list type must come first",
- "tag lists require a width argument",
- "missing font type",
- "skipping end of block that is not open",
+ "missing display type, using -ragged",
+ "list type is not the first argument",
+ "missing -width in -tag list, using 8n",
+ "missing utility name, using \"\"",
+ "empty head in list item",
+ "empty list item",
+ "missing font type, using \\fR",
+ "unknown font type, using \\fR",
+ "nothing follows prefix",
+ "empty reference block",
+ "missing -std argument, adding it",
+ "missing eqn box, using \"\"",
/* related to bad macro arguments */
- "skipping argument",
+ "unterminated quoted argument",
"duplicate argument",
- "duplicate display type",
- "duplicate list type",
+ "skipping duplicate argument",
+ "skipping duplicate display type",
+ "skipping duplicate list type",
+ "skipping -width argument",
"unknown AT&T UNIX version",
- "bad Boolean value",
- "unknown font",
- "unknown standard specifier",
- "bad width argument",
+ "comma in function argument",
+ "parenthesis in function name",
+ "invalid content in Rs block",
+ "invalid Boolean argument",
+ "unknown font, skipping request",
/* related to plain text */
- "blank line in non-literal context",
- "tab in non-literal context",
- "end of line whitespace",
+ "blank line in fill mode, using .sp",
+ "tab in filled text",
+ "whitespace at end of input line",
"bad comment style",
- "bad escape sequence",
- "unterminated quoted string",
+ "invalid escape sequence",
+ "undefined string, using \"\"",
- /* related to equations */
- "unexpected literal in equation",
+ /* related to tables */
+ "tbl line starts with span",
+ "tbl column starts with span",
+ "skipping vertical bar in tbl layout",
"generic error",
- /* related to equations */
- "unexpected equation scope closure",
- "equation scope open on exit",
- "overlapping equation scopes",
- "unexpected end of equation",
- "equation syntax error",
-
/* related to tables */
- "bad table syntax",
- "bad table option",
- "bad table layout",
- "no table layout cells specified",
- "no table data cells specified",
- "ignore data in cell",
- "data block still open",
- "ignoring extra data cells",
-
+ "non-alphabetic character in tbl options",
+ "skipping unknown tbl option",
+ "missing tbl option argument",
+ "wrong tbl option argument size",
+ "empty tbl layout",
+ "invalid character in tbl layout",
+ "unmatched parenthesis in tbl layout",
+ "tbl without any data cells",
+ "ignoring data in spanned tbl cell",
+ "ignoring extra tbl data cells",
+ "data block open at end of tbl",
+
+ /* related to document structure and macros */
+ NULL,
"input stack limit exceeded, infinite loop?",
"skipping bad character",
- "escaped character not allowed in a name",
- "manual name not yet set",
- "skipping text before the first section header",
"skipping unknown macro",
- "NOT IMPLEMENTED, please use groff: skipping request",
- "argument count wrong",
+ "skipping insecure request",
+ "skipping item outside list",
"skipping column outside column list",
"skipping end of block that is not open",
- "missing end of block",
- "scope open on exit",
- "uname(3) system call failed",
- "macro requires line argument(s)",
- "macro requires body argument(s)",
- "macro requires argument(s)",
- "request requires a numeric argument",
- "missing list type",
- "line argument(s) will be lost",
- "body argument(s) will be lost",
-
- "generic fatal error",
+ "fewer RS blocks open, skipping",
+ "inserting missing end of block",
+ "appending missing end of block",
- "input too large",
- "not a manual",
- "column syntax is inconsistent",
- "NOT IMPLEMENTED: .Bd -file",
- "argument count wrong, violates syntax",
- "child violates parent syntax",
- "argument count wrong, violates syntax",
+ /* related to request and macro arguments */
+ "escaped character not allowed in a name",
+ "argument count wrong",
+ "NOT IMPLEMENTED: Bd -file",
+ "missing list type, using -item",
+ "missing manual name, using \"\"",
+ "uname(3) system call failed, using UNKNOWN",
+ "unknown standard specifier",
+ "skipping request without numeric argument",
"NOT IMPLEMENTED: .so with absolute path or \"..\"",
- "no document body",
- "no document prologue",
- "static buffer exhausted",
+ ".so request failed",
+ "skipping all arguments",
+ "skipping excess arguments",
+ "divide by zero",
- /* system errors */
- "cannot open file",
- "cannot stat file",
- "cannot read file",
+ "unsupported feature",
+ "input too large",
+ "unsupported control character",
+ "unsupported roff request",
+ "eqn delim option in tbl",
+ "unsupported tbl layout modifier",
+ "ignoring macro in table",
};
static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
"RESERVED",
"WARNING",
"ERROR",
- "FATAL",
+ "UNSUPP",
"BADARG",
"SYSERR"
};
}
static void
-pset(const char *buf, int pos, struct mparse *curp)
+choose_parser(struct mparse *curp)
{
- int i;
+ char *cp, *ep;
+ int format;
/*
- * Try to intuit which kind of manual parser should be used. If
- * passed in by command-line (-man, -mdoc), then use that
- * explicitly. If passed as -mandoc, then try to guess from the
- * line: either skip dot-lines, use -mdoc when finding `.Dt', or
- * default to -man, which is more lenient.
- *
- * Separate out pmdoc/pman from mdoc/man: the first persists
- * through all parsers, while the latter is used per-parse.
+ * If neither command line arguments -mdoc or -man select
+ * a parser nor the roff parser found a .Dd or .TH macro
+ * yet, look ahead in the main input buffer.
*/
- if ('.' == buf[0] || '\'' == buf[0]) {
- for (i = 1; buf[i]; i++)
- if (' ' != buf[i] && '\t' != buf[i])
+ if ((format = roff_getformat(curp->roff)) == 0) {
+ cp = curp->primary->buf;
+ ep = cp + curp->primary->sz;
+ while (cp < ep) {
+ if (*cp == '.' || *cp == '\'') {
+ cp++;
+ if (cp[0] == 'D' && cp[1] == 'd') {
+ format = MPARSE_MDOC;
+ break;
+ }
+ if (cp[0] == 'T' && cp[1] == 'H') {
+ format = MPARSE_MAN;
+ break;
+ }
+ }
+ cp = memchr(cp, '\n', ep - cp);
+ if (cp == NULL)
break;
- if ('\0' == buf[i])
- return;
+ cp++;
+ }
}
- if (MPARSE_MDOC & curp->options) {
+ if (format == MPARSE_MDOC) {
if (NULL == curp->pmdoc)
curp->pmdoc = mdoc_alloc(
curp->roff, curp, curp->defos,
assert(curp->pmdoc);
curp->mdoc = curp->pmdoc;
return;
- } else if (MPARSE_MAN & curp->options) {
- if (NULL == curp->pman)
- curp->pman = man_alloc(curp->roff, curp,
- MPARSE_QUICK & curp->options ? 1 : 0);
- assert(curp->pman);
- curp->man = curp->pman;
- return;
}
- if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
- if (NULL == curp->pmdoc)
- curp->pmdoc = mdoc_alloc(
- curp->roff, curp, curp->defos,
- MPARSE_QUICK & curp->options ? 1 : 0);
- assert(curp->pmdoc);
- curp->mdoc = curp->pmdoc;
- return;
- }
+ /* Fall back to man(7) as a last resort. */
if (NULL == curp->pman)
- curp->pman = man_alloc(curp->roff, curp,
+ curp->pman = man_alloc(
+ curp->roff, curp, curp->defos,
MPARSE_QUICK & curp->options ? 1 : 0);
assert(curp->pman);
curp->man = curp->pman;
}
/*
- * Main parse routine for an opened file. This is called for each
- * opened file and simply loops around the full input file, possibly
- * nesting (i.e., with `so').
+ * Main parse routine for a buffer.
+ * It assumes encoding and line numbering are already set up.
+ * It can recurse directly (for invocations of user-defined
+ * macros, inline equations, and input line traps)
+ * and indirectly (for .so file inclusion).
*/
static void
-mparse_buf_r(struct mparse *curp, struct buf blk, int start)
+mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
{
const struct tbl_span *span;
struct buf ln;
+ const char *save_file;
+ char *cp;
+ size_t pos; /* byte number in the ln buffer */
enum rofferr rr;
- int i, of, rc;
- int pos; /* byte number in the ln buffer */
+ int of;
int lnn; /* line number in the real file */
+ int fd;
+ pid_t save_child;
unsigned char c;
- memset(&ln, 0, sizeof(struct buf));
+ memset(&ln, 0, sizeof(ln));
lnn = curp->line;
pos = 0;
- for (i = 0; i < (int)blk.sz; ) {
+ while (i < blk.sz) {
if (0 == pos && '\0' == blk.buf[i])
break;
if (start) {
curp->line = lnn;
curp->reparse_count = 0;
+
+ if (lnn < 3 &&
+ curp->filenc & MPARSE_UTF8 &&
+ curp->filenc & MPARSE_LATIN1)
+ curp->filenc = preconv_cue(&blk, i);
}
- while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
+ while (i < blk.sz && (start || blk.buf[i] != '\0')) {
/*
* When finding an unescaped newline character,
* Skip a preceding carriage return, if any.
*/
- if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
+ if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
'\n' == blk.buf[i + 1])
++i;
if ('\n' == blk.buf[i]) {
}
/*
- * Make sure we have space for at least
- * one backslash and one other character
- * and the trailing NUL byte.
+ * Make sure we have space for the worst
+ * case of 11 bytes: "\\[u10ffff]\0"
*/
- if (pos + 2 >= (int)ln.sz)
+ if (pos + 11 > ln.sz)
resize_buf(&ln, 256);
/*
- * Warn about bogus characters. If you're using
- * non-ASCII encoding, you're screwing your
- * readers. Since I'd rather this not happen,
- * I'll be helpful and replace these characters
- * with "?", so we don't display gibberish.
- * Note to manual writers: use special characters.
+ * Encode 8-bit input.
*/
- c = (unsigned char) blk.buf[i];
+ c = blk.buf[i];
+ if (c & 0x80) {
+ if ( ! (curp->filenc && preconv_encode(
+ &blk, &i, &ln, &pos, &curp->filenc))) {
+ mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
+ curp->line, pos, "0x%x", c);
+ ln.buf[pos++] = '?';
+ i++;
+ }
+ continue;
+ }
- if ( ! (isascii(c) &&
- (isgraph(c) || isblank(c)))) {
- mandoc_msg(MANDOCERR_BADCHAR, curp,
- curp->line, pos, NULL);
+ /*
+ * Exclude control characters.
+ */
+
+ if (c == 0x7f || (c < 0x20 && c != 0x09)) {
+ mandoc_vmsg(c == 0x00 || c == 0x04 ||
+ c > 0x0a ? MANDOCERR_CHAR_BAD :
+ MANDOCERR_CHAR_UNSUPP,
+ curp, curp->line, pos, "0x%x", c);
i++;
ln.buf[pos++] = '?';
continue;
/* Trailing backslash = a plain char. */
- if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
+ if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
ln.buf[pos++] = blk.buf[i++];
continue;
}
* skip that one as well.
*/
- if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
+ if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
'\n' == blk.buf[i + 2])
++i;
if ('\n' == blk.buf[i + 1]) {
if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
i += 2;
/* Comment, skip to end of line */
- for (; i < (int)blk.sz; ++i) {
+ for (; i < blk.sz; ++i) {
if ('\n' == blk.buf[i]) {
++i;
++lnn;
if ( ! (isascii(c) &&
(isgraph(c) || isblank(c)))) {
- mandoc_msg(MANDOCERR_BADCHAR, curp,
- curp->line, pos, NULL);
+ mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
+ curp->line, pos, "0x%x", c);
i += 2;
ln.buf[pos++] = '?';
continue;
ln.buf[pos++] = blk.buf[i++];
}
- if (pos >= (int)ln.sz)
+ if (pos >= ln.sz)
resize_buf(&ln, 256);
ln.buf[pos] = '\0';
[curp->secondary->sz] = '\0';
}
rerun:
- rr = roff_parseln(curp->roff, curp->line,
- &ln.buf, &ln.sz, of, &of);
+ rr = roff_parseln(curp->roff, curp->line, &ln, &of);
switch (rr) {
case ROFF_REPARSE:
if (REPARSE_LIMIT >= ++curp->reparse_count)
- mparse_buf_r(curp, ln, 0);
+ mparse_buf_r(curp, ln, of, 0);
else
mandoc_msg(MANDOCERR_ROFFLOOP, curp,
curp->line, pos, NULL);
pos = 0;
continue;
case ROFF_APPEND:
- pos = (int)strlen(ln.buf);
+ pos = strlen(ln.buf);
continue;
case ROFF_RERUN:
goto rerun;
case ROFF_IGN:
pos = 0;
continue;
- case ROFF_ERR:
- assert(MANDOCLEVEL_FATAL <= curp->file_status);
- break;
case ROFF_SO:
- if (0 == (MPARSE_SO & curp->options) &&
- (i >= (int)blk.sz || '\0' == blk.buf[i])) {
+ if ( ! (curp->options & MPARSE_SO) &&
+ (i >= blk.sz || blk.buf[i] == '\0')) {
curp->sodest = mandoc_strdup(ln.buf + of);
free(ln.buf);
return;
*/
if (curp->secondary)
curp->secondary->sz -= pos + 1;
- mparse_readfd(curp, -1, ln.buf + of);
- if (MANDOCLEVEL_FATAL <= curp->file_status)
- break;
+ save_file = curp->file;
+ save_child = curp->child;
+ if (mparse_open(curp, &fd, ln.buf + of) ==
+ MANDOCLEVEL_OK) {
+ mparse_readfd(curp, fd, ln.buf + of);
+ curp->file = save_file;
+ } else {
+ curp->file = save_file;
+ mandoc_vmsg(MANDOCERR_SO_FAIL,
+ curp, curp->line, pos,
+ ".so %s", ln.buf + of);
+ ln.sz = mandoc_asprintf(&cp,
+ ".sp\nSee the file %s.\n.sp",
+ ln.buf + of);
+ free(ln.buf);
+ ln.buf = cp;
+ of = 0;
+ mparse_buf_r(curp, ln, of, 0);
+ }
+ curp->child = save_child;
pos = 0;
continue;
default:
break;
}
- /*
- * If we encounter errors in the recursive parse, make
- * sure we don't continue parsing.
- */
-
- if (MANDOCLEVEL_FATAL <= curp->file_status)
- break;
-
/*
* If input parsers have not been allocated, do so now.
* We keep these instanced between parsers, but set them
*/
if ( ! (curp->man || curp->mdoc))
- pset(ln.buf + of, pos - of, curp);
+ choose_parser(curp);
/*
- * Lastly, push down into the parsers themselves. One
- * of these will have already been set in the pset()
- * routine.
+ * Lastly, push down into the parsers themselves.
* If libroff returns ROFF_TBL, then add it to the
* currently open parse. Since we only get here if
* there does exist data (see tbl_data.c), we're
* Do the same for ROFF_EQN.
*/
- rc = -1;
-
- if (ROFF_TBL == rr)
- while (NULL != (span = roff_span(curp->roff))) {
- rc = curp->man ?
- man_addspan(curp->man, span) :
- mdoc_addspan(curp->mdoc, span);
- if (0 == rc)
- break;
- }
- else if (ROFF_EQN == rr)
- rc = curp->mdoc ?
- mdoc_addeqn(curp->mdoc,
- roff_eqn(curp->roff)) :
- man_addeqn(curp->man,
- roff_eqn(curp->roff));
- else if (curp->man || curp->mdoc)
- rc = curp->man ?
- man_parseln(curp->man,
- curp->line, ln.buf, of) :
- mdoc_parseln(curp->mdoc,
- curp->line, ln.buf, of);
-
- if (0 == rc) {
- assert(MANDOCLEVEL_FATAL <= curp->file_status);
- break;
- } else if (2 == rc)
- break;
+ if (rr == ROFF_TBL) {
+ while ((span = roff_span(curp->roff)) != NULL)
+ if (curp->man == NULL)
+ mdoc_addspan(curp->mdoc, span);
+ else
+ man_addspan(curp->man, span);
+ } else if (rr == ROFF_EQN) {
+ if (curp->man == NULL)
+ mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
+ else
+ man_addeqn(curp->man, roff_eqn(curp->roff));
+ } else if ((curp->man == NULL ?
+ mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
+ man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
+ break;
/* Temporary buffers typically are not full. */
size_t off;
ssize_t ssz;
-#ifdef HAVE_MMAP
+#if HAVE_MMAP
struct stat st;
if (-1 == fstat(fd, &st)) {
- curp->file_status = MANDOCLEVEL_SYSERR;
- if (curp->mmsg)
- (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
- file, 0, 0, strerror(errno));
- return(0);
+ perror(file);
+ exit((int)MANDOCLEVEL_SYSERR);
}
/*
if (S_ISREG(st.st_mode)) {
if (st.st_size >= (1U << 31)) {
- curp->file_status = MANDOCLEVEL_FATAL;
- if (curp->mmsg)
- (*curp->mmsg)(MANDOCERR_TOOLARGE,
- curp->file_status, file, 0, 0, NULL);
+ mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
return(0);
}
*with_mmap = 1;
for (;;) {
if (off == fb->sz) {
if (fb->sz == (1U << 31)) {
- curp->file_status = MANDOCLEVEL_FATAL;
- if (curp->mmsg)
- (*curp->mmsg)(MANDOCERR_TOOLARGE,
- curp->file_status,
- file, 0, 0, NULL);
+ mandoc_msg(MANDOCERR_TOOLARGE, curp,
+ 0, 0, NULL);
break;
}
resize_buf(fb, 65536);
return(1);
}
if (ssz == -1) {
- curp->file_status = MANDOCLEVEL_SYSERR;
- if (curp->mmsg)
- (*curp->mmsg)(MANDOCERR_SYSREAD,
- curp->file_status, file, 0, 0,
- strerror(errno));
- break;
+ perror(file);
+ exit((int)MANDOCLEVEL_SYSERR);
}
off += (size_t)ssz;
}
mparse_end(struct mparse *curp)
{
- if (MANDOCLEVEL_FATAL <= curp->file_status)
- return;
-
- if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
- assert(MANDOCLEVEL_FATAL <= curp->file_status);
- return;
- }
-
- if (curp->man && ! man_endparse(curp->man)) {
- assert(MANDOCLEVEL_FATAL <= curp->file_status);
- return;
- }
-
- if ( ! (curp->mdoc || curp->man || curp->sodest)) {
- mandoc_msg(MANDOCERR_NOTMANUAL, curp, 1, 0, NULL);
- curp->file_status = MANDOCLEVEL_FATAL;
- return;
+ if (curp->mdoc == NULL &&
+ curp->man == NULL &&
+ curp->sodest == NULL) {
+ if (curp->options & MPARSE_MDOC)
+ curp->mdoc = curp->pmdoc;
+ else {
+ if (curp->pman == NULL)
+ curp->pman = man_alloc(
+ curp->roff, curp, curp->defos,
+ curp->options & MPARSE_QUICK ? 1 : 0);
+ curp->man = curp->pman;
+ }
}
-
+ if (curp->mdoc)
+ mdoc_endparse(curp->mdoc);
+ if (curp->man)
+ man_endparse(curp->man);
roff_endparse(curp->roff);
}
static void
mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
{
+ struct buf *svprimary;
const char *svfile;
+ size_t offset;
static int recursion_depth;
if (64 < recursion_depth) {
/* Line number is per-file. */
svfile = curp->file;
curp->file = file;
+ svprimary = curp->primary;
+ curp->primary = &blk;
curp->line = 1;
recursion_depth++;
- mparse_buf_r(curp, blk, 1);
+ /* Skip an UTF-8 byte order mark. */
+ if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
+ (unsigned char)blk.buf[0] == 0xef &&
+ (unsigned char)blk.buf[1] == 0xbb &&
+ (unsigned char)blk.buf[2] == 0xbf) {
+ offset = 3;
+ curp->filenc &= ~MPARSE_LATIN1;
+ } else
+ offset = 0;
- if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
+ mparse_buf_r(curp, blk, offset, 1);
+
+ if (--recursion_depth == 0)
mparse_end(curp);
+ curp->primary = svprimary;
curp->file = svfile;
}
enum mandoclevel
-mparse_readmem(struct mparse *curp, const void *buf, size_t len,
+mparse_readmem(struct mparse *curp, void *buf, size_t len,
const char *file)
{
struct buf blk;
- blk.buf = UNCONST(buf);
+ blk.buf = buf;
blk.sz = len;
mparse_parse_buffer(curp, blk, file);
return(curp->file_status);
}
+/*
+ * Read the whole file into memory and call the parsers.
+ * Called recursively when an .so request is encountered.
+ */
enum mandoclevel
mparse_readfd(struct mparse *curp, int fd, const char *file)
{
struct buf blk;
int with_mmap;
+ int save_filenc;
+
+ if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
+ save_filenc = curp->filenc;
+ curp->filenc = curp->options &
+ (MPARSE_UTF8 | MPARSE_LATIN1);
+ mparse_parse_buffer(curp, blk, file);
+ curp->filenc = save_filenc;
+#if HAVE_MMAP
+ if (with_mmap)
+ munmap(blk.buf, blk.sz);
+ else
+#endif
+ free(blk.buf);
+ }
+
+ if (fd != STDIN_FILENO && close(fd) == -1)
+ perror(file);
- if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) {
- curp->file_status = MANDOCLEVEL_SYSERR;
- if (curp->mmsg)
- (*curp->mmsg)(MANDOCERR_SYSOPEN,
- curp->file_status,
- file, 0, 0, strerror(errno));
- goto out;
+ mparse_wait(curp);
+ return(curp->file_status);
+}
+
+enum mandoclevel
+mparse_open(struct mparse *curp, int *fd, const char *file)
+{
+ int pfd[2];
+ int save_errno;
+ char *cp;
+
+ curp->file = file;
+
+ /* Unless zipped, try to just open the file. */
+
+ if ((cp = strrchr(file, '.')) == NULL ||
+ strcmp(cp + 1, "gz")) {
+ curp->child = 0;
+ if ((*fd = open(file, O_RDONLY)) != -1)
+ return(MANDOCLEVEL_OK);
+
+ /* Open failed; try to append ".gz". */
+
+ mandoc_asprintf(&cp, "%s.gz", file);
+ file = cp;
+ } else
+ cp = NULL;
+
+ /* Before forking, make sure the file can be read. */
+
+ save_errno = errno;
+ if (access(file, R_OK) == -1) {
+ if (cp != NULL)
+ errno = save_errno;
+ free(cp);
+ *fd = -1;
+ curp->child = 0;
+ mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
+ return(MANDOCLEVEL_ERROR);
}
- /*
- * Run for each opened file; may be called more than once for
- * each full parse sequence if the opened file is nested (i.e.,
- * from `so'). Simply sucks in the whole file and moves into
- * the parse phase for the file.
- */
+ /* Run gunzip(1). */
- if ( ! read_whole_file(curp, file, fd, &blk, &with_mmap))
- goto out;
+ if (pipe(pfd) == -1) {
+ perror("pipe");
+ exit((int)MANDOCLEVEL_SYSERR);
+ }
- mparse_parse_buffer(curp, blk, file);
+ switch (curp->child = fork()) {
+ case -1:
+ perror("fork");
+ exit((int)MANDOCLEVEL_SYSERR);
+ case 0:
+ close(pfd[0]);
+ if (dup2(pfd[1], STDOUT_FILENO) == -1) {
+ perror("dup");
+ exit((int)MANDOCLEVEL_SYSERR);
+ }
+ execlp("gunzip", "gunzip", "-c", file, NULL);
+ perror("exec");
+ exit((int)MANDOCLEVEL_SYSERR);
+ default:
+ close(pfd[1]);
+ *fd = pfd[0];
+ return(MANDOCLEVEL_OK);
+ }
+}
-#ifdef HAVE_MMAP
- if (with_mmap)
- munmap(blk.buf, blk.sz);
- else
-#endif
- free(blk.buf);
+enum mandoclevel
+mparse_wait(struct mparse *curp)
+{
+ int status;
- if (STDIN_FILENO != fd && -1 == close(fd))
- perror(file);
-out:
- return(curp->file_status);
+ if (curp->child == 0)
+ return(MANDOCLEVEL_OK);
+
+ if (waitpid(curp->child, &status, 0) == -1) {
+ perror("wait");
+ exit((int)MANDOCLEVEL_SYSERR);
+ }
+ if (WIFSIGNALED(status)) {
+ mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
+ "gunzip died from signal %d", WTERMSIG(status));
+ return(MANDOCLEVEL_ERROR);
+ }
+ if (WEXITSTATUS(status)) {
+ mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
+ "gunzip failed with code %d", WEXITSTATUS(status));
+ return(MANDOCLEVEL_ERROR);
+ }
+ return(MANDOCLEVEL_OK);
}
struct mparse *
-mparse_alloc(int options, enum mandoclevel wlevel,
- mandocmsg mmsg, char *defos)
+mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
+ const struct mchars *mchars, const char *defos)
{
struct mparse *curp;
- assert(wlevel <= MANDOCLEVEL_FATAL);
-
curp = mandoc_calloc(1, sizeof(struct mparse));
curp->options = options;
curp->mmsg = mmsg;
curp->defos = defos;
- curp->roff = roff_alloc(curp, options);
+ curp->mchars = mchars;
+ curp->roff = roff_alloc(curp, curp->mchars, options);
+ if (curp->options & MPARSE_MDOC)
+ curp->pmdoc = mdoc_alloc(
+ curp->roff, curp, curp->defos,
+ curp->options & MPARSE_QUICK ? 1 : 0);
+ if (curp->options & MPARSE_MAN)
+ curp->pman = man_alloc(
+ curp->roff, curp, curp->defos,
+ curp->options & MPARSE_QUICK ? 1 : 0);
+
return(curp);
}
{
enum mandoclevel level;
- level = MANDOCLEVEL_FATAL;
+ level = MANDOCLEVEL_UNSUPP;
while (er < mandoclimits[level])
level--;
- if (level < m->wlevel)
+ if (level < m->wlevel && er != MANDOCERR_FILE)
return;
if (m->mmsg)
git.ckatri.com / mandoc.git / blobdiff