/*
* m_getfld.c -- read/parse a message
*
- * $Id$
- *
* This code is Copyright (c) 2002, by the authors of nmh. See the
* COPYRIGHT file in the root directory of the nmh distribution for
* complete copyright information.
there is data in "name" or "buf").
*/
+/*
+Purpose
+=======
+Reads an Internet message (RFC 5322), or one or more messages stored in a
+maildrop in mbox (RFC 4155) or MMDF format, from a file stream. Each call
+to m_getfld() reads one header field, or a portion of the body, in sequence.
+
+Inputs
+======
+state: message parse state
+bufsz: maximum number of characters to load into buf
+iob: input file stream
+
+Outputs
+=======
+name: header field name (array of size NAMESZ=999)
+buf: either a header field body or message body
+(return value): message parse state on return from function
+(global) int msg_count: number of characters loaded into buf
+
+Functions (part of Inputs, really)
+=========
+void m_unknown(FILE *iob): Determines the message delimiter string for the
+ maildrop. Called by inc, scan, and msh when reading from a maildrop file.
+
+void m_eomsbr (int (*action)(int)): Sets the hook to check for end of
+ message in a maildrop. Called only by msh.
+
+Those functions save state in the State variables listed below.
+
+Definitions
+===========
+state is one of:
+ FLD // Field returned
+ FLDPLUS // Field returned with more to come
+ FLDEOF // Field returned ending at eom
+ BODY // Body returned with more to come
+ BODYEOF // Body returned ending at eom
+ FILEEOF // Reached end of input file
+ FMTERR // Message Format error
+ LENERR // Name too long error from getfld
+
+msg_style is maildrop style, one of:
+ MS_UNKNOWN // type not known yet
+ MS_DEFAULT // default (one msg per file)
+ MS_MBOX // Unix-style "from" lines
+ MS_MMDF // string mmdlm2
+ MS_MSH // whacko msh
+
+State variables (part of Outputs)
+===============
+m_getfld() retains state internally between calls in some state variables.
+
+These two variables are global, but only used internally by m_getfld.c:
+int msg_style
+char *msg_delim
+
+These are used for the end-of-message matcher when reading maildrops:
+static unsigned char **pat_map
+static unsigned char *fdelim
+static unsigned char *delimend
+static int fdelimlen
+static unsigned char *edelim
+static int edelimlen
+
+Restriction
+===========
+m_getfld() is restricted to operate on one file stream at a time because of
+the retained state (see "State variables" above).
+
+Current usage
+=============
+The first call to m_getfld() on a file stream is with a state of FLD.
+Subsequent calls provide the state returned by the previous call.
+
+Along the way, I thought of these possible interface changes that we
+might want to consider before rototilling the internals:
+
+1) To remove globals that don't need to be:
+ Change msg_style and msg_delim to be file static.
+
+2) To remove a global:
+ Change bufsz to be in-out instead of in, and therefore int * instead of
+ int, and use that instead of global msg_count. There are only 3 call
+ sites that use msg_count so it wouldn't take much effort to remove use of
+ it. Of course, all call sites would have to change to provide an int *
+ instead of an int. Some now pass constants.
+
+3) To remove the state argument from the signature:
+ Given the Current usage and Restriction above, the state variable could
+ be removed from the signature and just retained internally.
+
+4) To remove the Restriction above:
+ One approach would be for m_getfld() to retain multiple copies of that
+ state, one per iob that it sees. Another approach would be for the
+ caller to store it in an opaque struct, the address of which is passed
+ through the interface.
+*/
/*
* static prototypes
# define _cnt _r /* Retch */
# define _filbuf __srget /* Puke */
# define DEFINED__FILBUF_TO_SOMETHING_SPECIFIC
-#endif
-#ifdef SCO_5_STDIO
-# define _ptr __ptr
-# define _cnt __cnt
-# define _base __base
-# define _filbuf(fp) ((fp)->__cnt = 0, __filbuf(fp))
-# define DEFINED__FILBUF_TO_SOMETHING_SPECIFIC
+# if defined __CYGWIN__
+ /* Cygwin's stdio.h does not declare __srget(). */
+ int __srget(FILE *);
+# endif /* __CYGWIN__ */
#endif
#ifndef DEFINED__FILBUF_TO_SOMETHING_SPECIFIC
int
-m_getfld (int state, unsigned char *name, unsigned char *buf,
+m_getfld (int state, unsigned char name[NAMESZ], unsigned char *buf,
int bufsz, FILE *iob)
{
register unsigned char *bp, *cp, *ep, *sp;
* . hit the end of the buffer. (loop)
*/
if (c == '\n') {
- *cp = *buf = 0;
- advise (NULL, "eol encountered in field \"%s\"", name);
- state = FMTERR;
- goto finish;
+ /* We hit the end of the line without seeing ':' to
+ * terminate the field name. This is usually (always?)
+ * spam. But, blowing up is lame, especially when
+ * scan(1)ing a folder with such messages. Pretend such
+ * lines are the first of the body (at least mutt also
+ * handles it this way). */
+
+ /* See if buf can hold this line, since we were assuming
+ * we had a buffer of NAMESZ, not bufsz. */
+ /* + 1 for the newline */
+ if (bufsz < j + 1) {
+ /* No, it can't. Oh well, guess we'll blow up. */
+ *cp = *buf = 0;
+ advise (NULL, "eol encountered in field \"%s\"", name);
+ state = FMTERR;
+ goto finish;
+ }
+ memcpy (buf, name, j - 1);
+ buf[j - 1] = '\n';
+ buf[j] = '\0';
+ /* mhparse.c:get_content wants to find the position of the
+ * body start, but it thinks there's a blank line between
+ * the header and the body (naturally!), so seek back so
+ * that things line up even though we don't have that
+ * blank line in this case. Simpler parsers (e.g. mhl)
+ * get extra newlines, but that should be harmless enough,
+ * right? This is a corrupt message anyway. */
+ fseek (iob, ftell (iob) - 2, SEEK_SET);
+ return BODY;
}
if ((i -= j) <= 0) {
*cp = *buf = 0;
- advise (NULL, "field name \"%s\" exceeds %d bytes", name, NAMESZ - 1);
+ advise (NULL, "field name \"%s\" exceeds %d bytes", name, NAMESZ - 2);
state = LENERR;
goto finish;
}
ep = bp + c - 1;
if ((sp = pat_map[*ep])) {
do {
- cp = sp;
- while (*--ep == *--cp)
- ;
- if (cp < fdelim) {
- if (ep >= bp)
- /*
- * ep < bp means that all the buffer
- * contains is a prefix of delim.
- * If this prefix is really a delim, the
- * m_eom call at entry should have found
- * it. Thus it's not a delim and we can
- * take all of it.
+ /* This if() is true unless (a) the buffer is too
+ * small to contain this delimiter prefix, or
+ * (b) it contains exactly enough chars for the
+ * delimiter prefix.
+ * For case (a) obviously we aren't going to match.
+ * For case (b), if the buffer really contained exactly
+ * a delim prefix, then the m_eom call at entry
+ * should have found it. Thus it's not a delim
+ * and we know we won't get a match.
+ */
+ if (((sp - fdelim) + 2) <= c) {
+ cp = sp;
+ /* Unfortunately although fdelim has a preceding NUL
+ * we can't use this as a sentinel in case the buffer
+ * contains a NUL in exactly the wrong place (this
+ * would cause us to run off the front of fdelim).
+ */
+ while (*--ep == *--cp)
+ if (cp < fdelim)
+ break;
+ if (cp < fdelim) {
+ /* we matched the entire delim prefix,
+ * so only take the buffer up to there.
+ * we know ep >= bp -- check above prevents underrun
*/
c = (ep - bp) + 2;
- break;
- }
+ break;
+ }
+ }
/* try matching one less char of delim string */
ep = bp + c - 1;
} while (--sp > fdelim);
}
-#ifdef RPATHS
-static char unixbuf[BUFSIZ] = "";
-#endif /* RPATHS */
-
void
m_unknown(FILE *iob)
{
&& strncmp (text, "From ", 5) == 0) {
msg_style = MS_MBOX;
delimstr = "\nFrom ";
-#ifndef RPATHS
while ((c = getc (iob)) != '\n' && c >= 0)
;
-#else /* RPATHS */
- cp = unixbuf;
- while ((c = getc (iob)) != '\n' && cp - unixbuf < BUFSIZ - 1)
- *cp++ = c;
- *cp = 0;
-#endif /* RPATHS */
} else {
/* not a Unix style maildrop */
fseek (iob, pos, SEEK_SET);
register long pos = 0L;
register int i;
char text[10];
-#ifdef RPATHS
- register char *cp;
-#endif /* RPATHS */
pos = ftell (iob);
if ((i = fread (text, sizeof *text, edelimlen, iob)) != edelimlen
}
if (msg_style == MS_MBOX) {
-#ifndef RPATHS
while ((c = getc (iob)) != '\n')
if (c < 0)
break;
-#else /* RPATHS */
- cp = unixbuf;
- while ((c = getc (iob)) != '\n' && c >= 0 && cp - unixbuf < BUFSIZ - 1)
- *cp++ = c;
- *cp = 0;
-#endif /* RPATHS */
}
return 1;
}
-#ifdef RPATHS
-/*
- * Return the Return-Path and Delivery-Date
- * header information.
- *
- * Currently, I'm assuming that the "From " line
- * takes one of the following forms.
- *
- * From sender date remote from host (for UUCP delivery)
- * From sender@host date (for sendmail delivery)
- */
-
-int
-get_returnpath (char *rp, int rplen, char *dd, int ddlen)
-{
- char *ap, *bp, *cp, *dp;
-
- ap = unixbuf;
- if (!(bp = cp = strchr(ap, ' ')))
- return 0;
-
- /*
- * Check for "remote from" in envelope to see
- * if this message uses UUCP style addressing
- */
- while ((cp = strchr(++cp, 'r'))) {
- if (strncmp (cp, "remote from", 11) == 0) {
- cp = strrchr (cp, ' ');
- break;
- }
- }
-
- /*
- * Get the Return-Path information from
- * the "From " envelope.
- */
- if (cp) {
- /* return path for UUCP style addressing */
- dp = strchr (++cp, '\n');
- snprintf (rp, rplen, "%.*s!%.*s\n", (int)(dp - cp), cp, (int)(bp - ap), ap);
- } else {
- /* return path for standard domain addressing */
- snprintf (rp, rplen, "%.*s\n", (int)(bp - ap), ap);
- }
-
- /*
- * advance over the spaces to get to
- * delivery date on envelope
- */
- while (*bp == ' ')
- bp++;
-
- /* Now get delivery date from envelope */
- snprintf (dd, ddlen, "%.*s\n", 24, bp);
-
- unixbuf[0] = 0;
- return 1;
-}
-#endif /* RPATHS */
-
-
static unsigned char *
matchc(int patln, char *pat, int strln, char *str)
{
projects / mmh / blobdiff