-
-/*
- * m_getfld.c -- read/parse a message
+/* m_getfld.c -- read/parse a message
*
* This code is Copyright (c) 2002, by the authors of nmh. See the
* COPYRIGHT file in the root directory of the nmh distribution for
#include <h/mh.h>
#include <h/mts.h>
#include <h/utils.h>
+#include <inttypes.h>
+
+/*
+ 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
+ ======
+ gstate: opaque 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
+ bufsz: number of characters loaded into buf
+ (return value): message parse state on return from function
+
+ Functions
+ =========
+ void m_getfld_state_destroy (m_getfld_state_t *gstate): destroys
+ the parse state pointed to by the gstate argument.
+
+ m_getfld_state_reset (m_getfld_state_t *gstate): resets the parse
+ state to FLD.
+
+ void m_unknown(FILE *iob): Determines the message delimiter string
+ for the maildrop. Called by inc and scan when reading from a
+ maildrop file.
+
+ State variables
+ ===============
+ m_getfld() retains state internally between calls in the
+ m_getfld_state_t variable. These are used for detecting the end of
+ each message when reading maildrops:
+
+ char **pat_map
+ char *fdelim
+ char *delimend
+ int fdelimlen
+ char *edelim
+ int edelimlen
+ char *msg_delim
+ int msg_style
+
+ Usage
+ =====
+ m_getfld_state_t gstate;
+
+ gstate = m_getfld_state_init(mailfp);
+ Perhaps m_getfld_track_filepos2(&gstate);
+ ...
+ state = m_getfld2(&gstate, ...);
+ ...Repeat until finished with mailfp.
+ m_getfld_state_destroy (&gstate);
+
+ The state is retained internally by gstate. To reset its state to FLD:
+ m_getfld_state_reset (&gstate);
+*/
+
+/* The following described the old implementation. The high-level
+ structure hasn't changed, but some of the details have. I'm
+ leaving this as-is, though, for posterity.
+ */
/* This module has a long and checkered history. First, it didn't burst
maildrops correctly because it considered two CTRL-A:s in a row to be
be parsed as well. Unfortunately the speed issue finally caught up with
us since this routine is at the very heart of MH.
- To speed things up considerably, the routine Eom() was made an auxilary
+ To speed things up considerably, the routine Eom() was made an auxiliary
function called by the macro eom(). Unless we are bursting a maildrop,
the eom() macro returns FALSE saying we aren't at the end of the
message.
names are typically short (~8 char) and the loop that extracts them
might terminate on a colon, newline or max width. I considered
using a Vax "scanc" to locate the end of the field followed by a
- "bcopy" but the routine call overhead on a Vax is too large for this
+ "memmove" but the routine call overhead on a Vax is too large for this
to work on short names. If Berkeley ever makes "inline" part of the
C optimiser (so things like "scanc" turn into inline instructions) a
change here would be worthwhile.
so message bodies average at least a few hundred characters.
Assuming your system uses reasonably sized stdio buffers (1K or
more), this routine should be able to remove the body in large
- (>500 byte) chunks. The makes the cost of a call to "bcopy"
+ (>500 byte) chunks. The makes the cost of a call to "memmove"
small but there is a premium on checking for the eom in packed
maildrops. The eom pattern is always a simple string so we can
construct an efficient pattern matcher for it (e.g., a Vax "matchc"
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
-======
-gstate: opaque 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
-bufsz: number of characters loaded into buf
-(return value): message parse state on return from function
-
-Functions
-=========
-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.
-
-State variables
-===============
-m_getfld() retains state internally between calls in the m_getfld_state_t
-variable. These are used for detecting the end of each message when reading
-maildrops:
-
- unsigned char **pat_map
- unsigned char *fdelim
- unsigned char *delimend
- int fdelimlen
- unsigned char *edelim
- int edelimlen
- char *msg_delim
- int msg_style
- int (*eom_action)(int)
-
-Usage
-=====
- m_getfld_state_t gstate;
- m_getfld_state_init (&gstate);
- int state = m_getfld (gstate, ...);
- ...
- m_getfld_state_destroy (&gstate);
-
-The state is retained internally by gstate. To reset its state to FLD:
-
- m_getfld_state_reset (&gstate);
-*/
-
/*
* static prototypes
*/
-struct m_getfld_state;
-static int m_Eom (m_getfld_state_t, int);
-static unsigned char *matchc(int, char *, int, char *);
+static void Ungetc(m_getfld_state_t s);
+static int m_Eom (m_getfld_state_t);
#define eom(c,s) (s->msg_style != MS_DEFAULT && \
- (((c) == *s->msg_delim && m_Eom(s,c)) || \
- (s->eom_action && (*s->eom_action)(c))))
+ ((c) == *s->msg_delim && m_Eom(s)))
+
+/*
+ * Maildrop styles
+ */
+#define MS_DEFAULT 0 /* default (one msg per file) */
+#define MS_UNKNOWN 1 /* type not known yet */
+#define MS_MBOX 2 /* Unix-style "from" lines */
+#define MS_MMDF 3 /* string MMDF_DELIM */
/* This replaces the old approach, with its direct access to stdio
* internals. It uses one fread() to load a buffer that we manage.
* separate messages in a maildrop, such as mbox "From ".
*
* Some of the tests in the test suite assume a MSG_INPUT_SIZE
- * of 8192. */
-#define MSG_INPUT_SIZE (BUFSIZ >= 1024 ? BUFSIZ : 1024)
+ * of 8192.
+ */
+#define MSG_INPUT_SIZE NMH_BUFSIZ
#define MAX_DELIMITER_SIZE 5
struct m_getfld_state {
- unsigned char msg_buf[2 * MSG_INPUT_SIZE + MAX_DELIMITER_SIZE];
- unsigned char *readpos;
- unsigned char *end; /* One past the last character read in. */
- /* The following support tracking of the read position in the
- input file stream so that callers can interleave m_getfld()
- calls with ftell() and fseek(). ytes_read replaces the old
- m_getfld() msg_count global. last_caller_pos is stored when
- leaving m_getfld()/m_unknown(), then checked on the next entry.
- last_internal_pos is used to remember the position used
- internally by m_getfld() (read_more(), actually). */
+ /* The file to read from; I/O block. Caller keeps passing it after
+ * initialisation due to historic interface so it keeps getting
+ * updated, presumably to the same value. */
+ FILE *iob;
+
+ /* Holds content of iob. */
+ char msg_buf[2 * MSG_INPUT_SIZE + MAX_DELIMITER_SIZE];
+ /* Points to the next byte to read from msg_buf. */
+ char *readpos;
+ /* Points to just after the last valid byte in msg_buf. If readpos
+ * equals end then msg_buf is empty. */
+ char *end;
+
+ /* Whether the caller intends to ftell(3)/fseek(3) iob's position,
+ * and thus whether m_getfld() needs to detect that and compensate. */
+ int track_filepos;
+ /* Position in iob given what's been consumed ready for returning to
+ * the caller. Further than this may have been read into msg_buf. */
+ off_t total_bytes_read;
+ /* Bytes of iob consumed during this call. */
off_t bytes_read;
- off_t total_bytes_read; /* by caller, not necessarily from input file */
+ /* What fseeko(3) tells us iob's position is having just explicitly
+ * set it to total_bytes_read. Surely always the same? */
off_t last_caller_pos;
+ /* Saved position in iob from filling msg_buf, prior to returning. */
off_t last_internal_pos;
- FILE *iob;
- unsigned char **pat_map;
+ /* One of the MS_* macros tracking the type of iob's content and
+ * thus if it's a single email, or several with delimeters. Default
+ * is MS_DEFAULT. */
int msg_style;
- /*
- * The "full" delimiter string for a packed maildrop consists
- * of a newline followed by the actual delimiter. E.g., the
- * full string for a Unix maildrop would be: "\n\nFrom ".
- * "Fdelim" points to the start of the full string and is used
- * in the BODY case of the main routine to search the buffer for
- * a possible eom. Msg_delim points to the first character of
- * the actual delim. string (i.e., fdelim+1). Edelim
- * points to the 2nd character of actual delimiter string. It
- * is used in m_Eom because the first character of the string
- * has been read and matched before m_Eom is called.
- */
+
+ /* The message delimeter if iob has multiple emails, else NULL. For
+ * MS_MBOX it's the string that separates two emails, "\nFrom ",
+ * i.e. the terminating blank line of the previous email, and the
+ * starting From_ line of the next, but for MS_MMDF it's
+ * "\001\001\001\001\n" that may start or terminate an email. */
char *msg_delim;
- unsigned char *fdelim;
- unsigned char *delimend;
+ /* The last non-NUL char of msg_delim. */
+ char *delimend;
+ /* When searching for msg_delim after an email, it's only of
+ * interest at the start of the line, i.e. when preceded by a
+ * linefeed. fdelim points to msg_delim[-1] that contains '\n' so
+ * it can be used as the needle. */
+ char *fdelim;
+ /* strlen(fdelim). */
int fdelimlen;
- unsigned char *edelim;
+ /* The second char of msg_delim. Used when the first char has
+ * already been matched to test the rest. */
+ char *edelim;
+ /* strlen(edelim). */
int edelimlen;
- int (*eom_action)(int);
+ /* The relationship between all of these pointers and lengths for
+ * the two possible msg_delim values.
+ *
+ * "\0\n\nFrom \0" 9 "\0\n\001\001\001\001\n\0" 8
+ * | || | | | | |
+ * | || s->delimend | | | s->delimend
+ * | || | | |
+ * | |s->edelim s->edelimlen=5 | | s->edelim s->edelimlen=4
+ * | | | |
+ * | s->msg_delim | s->msg_delim
+ * | |
+ * s->fdelim s->fdelimlen=7 s->fdelim s->fdelimlen=6
+ */
+
+ /* Maps all the bytes of msg_delim, apart from the last two,
+ * including the NUL, onto the last position in msg_delim where they
+ * occur. Bytes not present are NULL. */
+ char **pat_map;
+
+ /* The parser's current state. Also returned to the caller, amongst
+ * other possible values, to indicate the token consumed. One of
+ * FLD, FLDPLUS, BODY, or FILEEOF. */
int state;
};
+m_getfld_state_t m_getfld_state_init(FILE *iob)
+{
+ m_getfld_state_t s;
+
+ NEW(s);
+ s->readpos = s->end = s->msg_buf;
+ s->bytes_read = s->total_bytes_read = 0;
+ s->last_caller_pos = s->last_internal_pos = 0;
+ s->iob = iob;
+ s->pat_map = NULL;
+ s->msg_style = MS_DEFAULT;
+ s->msg_delim = "";
+ s->fdelim = s->delimend = s->edelim = NULL;
+ s->fdelimlen = s->edelimlen = 0;
+ s->state = FLD;
+ s->track_filepos = 0;
+
+ return s;
+}
+
+/* scan() needs to force an initial state of FLD for each message. */
void
-m_getfld_state_init (m_getfld_state_t *s) {
- *s = (m_getfld_state_t) mh_xmalloc(sizeof (struct m_getfld_state));
- (*s)->readpos = (*s)->end = (*s)->msg_buf;
- (*s)->bytes_read = (*s)->total_bytes_read = 0;
- (*s)->last_caller_pos = (*s)->last_internal_pos = 0;
- /* (*s)->iob gets loaded on every call to m_getfld()/m_unknown(). */
- (*s)->pat_map = NULL;
- (*s)->msg_style = MS_DEFAULT;
- (*s)->msg_delim = "";
- (*s)->fdelim = (*s)->delimend = (*s)->edelim = NULL;
- (*s)->fdelimlen = (*s)->edelimlen = 0;
- (*s)->eom_action = NULL;
- (*s)->state = FLD;
+m_getfld_state_reset (m_getfld_state_t *gstate) {
+ if (*gstate) {
+ (*gstate)->state = FLD;
+ }
}
-/* scan() needs to force a state an initial state of FLD for each message. */
+/* If the caller interleaves ftell*()/fseek*() calls with m_getfld()
+ calls, m_getfld() must keep track of the file position. The caller
+ must use this function to inform m_getfld(). */
void
-m_getfld_state_reset (m_getfld_state_t *s) {
- (*s)->state = FLD;
+m_getfld_track_filepos (m_getfld_state_t *gstate, FILE *iob) {
+ if (! *gstate) {
+ *gstate = m_getfld_state_init(iob);
+ }
+
+ (*gstate)->track_filepos = 1;
}
-void m_getfld_state_destroy (m_getfld_state_t *s) {
- if (*s) {
- if ((*s)->fdelim) free ((*s)->fdelim-1);
- free (*s);
- *s = 0;
+/* m_getfld_track_filepos() with the existing iob. */
+void m_getfld_track_filepos2(m_getfld_state_t *gstate)
+{
+ if (!*gstate)
+ adios(NULL, "m_getfld_track_filepos2 without gstate");
+
+ m_getfld_track_filepos(gstate, (*gstate)->iob);
+}
+
+void m_getfld_state_destroy (m_getfld_state_t *gstate) {
+ m_getfld_state_t s = *gstate;
+
+ if (s) {
+ if (s->fdelim) {
+ free (s->fdelim-1);
+ free (s->pat_map);
+ }
+ free (s);
+ *gstate = 0;
}
}
static void
-enter_getfld (m_getfld_state_t s, FILE *iob) {
- off_t pos = ftello (iob);
-
- /* Ugly. The parser opens the input file multiple times, so we
- have to always use the FILE * that's passed to m_getfld().
- Though this might not be necessary any more, as long as the
- parser inits a new m_getfld_state for each file. See comment
- below about the readpos shift code being currently unused. */
+enter_getfld (m_getfld_state_t *gstate, FILE *iob) {
+ m_getfld_state_t s;
+ off_t pos;
+ off_t pos_movement;
+
+ if (! *gstate) {
+ *gstate = m_getfld_state_init(iob);
+ }
+ s = *gstate;
+ s->bytes_read = 0;
+
+ /* This is ugly and no longer necessary, but is retained just in
+ case it's needed again. The parser used to open the input file
+ multiple times, so we had to always use the FILE * that's
+ passed to m_getfld(). Now the parser inits a new
+ m_getfld_state for each file. See comment below about the
+ readpos shift code being currently unused. */
s->iob = iob;
- if (pos != 0 || s->last_internal_pos != 0) {
- if (s->last_internal_pos == 0) {
- s->total_bytes_read = pos;
- } else {
- off_t pos_movement = pos - s->last_caller_pos; /* Can be < 0. */
-
- if (pos_movement == 0) {
- pos = s->last_internal_pos;
- } else {
- /* The current file stream position differs from the
- last one, so caller must have called ftell/o().
- Or, this is the first call and the file position
- was not at 0. */
-
- if (s->readpos + pos_movement >= s->msg_buf &&
- s->readpos + pos_movement < s->end) {
- /* This is currently unused. It could be used by
- parse_mime() if it was changed to use a global
- m_getfld_state. */
- /* We can shift readpos and remain within the
- bounds of msg_buf. */
- s->readpos += pos_movement;
- s->total_bytes_read += pos_movement;
- pos = s->last_internal_pos;
- } else {
- size_t num_read;
-
- /* This seek skips past an integral number of
- chunks of size MSG_INPUT_SIZE. */
- fseeko (iob, pos/MSG_INPUT_SIZE * MSG_INPUT_SIZE, SEEK_SET);
- num_read = fread (s->msg_buf, 1, MSG_INPUT_SIZE, iob);
- s->readpos = s->msg_buf + pos % MSG_INPUT_SIZE;
- s->end = s->msg_buf + num_read;
- s->total_bytes_read = pos;
- }
- }
+ if (!s->track_filepos)
+ return;
- fseeko (iob, pos, SEEK_SET);
- }
+ if ((pos = ftello(iob)) == -1)
+ adios("getfld's iob", "failed to get offset on entry");
+ if (pos == 0 && s->last_internal_pos == 0)
+ return;
+
+ if (s->last_internal_pos == 0) {
+ s->total_bytes_read = pos;
+ return;
}
- s->bytes_read = 0;
+ pos_movement = pos - s->last_caller_pos; /* Can be < 0. */
+ if (pos_movement == 0) {
+ pos = s->last_internal_pos;
+ } else {
+ /* The current file stream position differs from the
+ last one, so caller must have called ftell/o().
+ Or, this is the first call and the file position
+ was not at 0. */
+
+ if (s->readpos + pos_movement >= s->msg_buf &&
+ s->readpos + pos_movement < s->end) {
+ /* This is currently unused. It could be used by
+ parse_mime() if it was changed to use a global
+ m_getfld_state. */
+ /* We can shift readpos and remain within the
+ bounds of msg_buf. */
+ s->readpos += pos_movement;
+ s->total_bytes_read += pos_movement;
+ pos = s->last_internal_pos;
+ } else {
+ off_t off;
+ size_t num_read;
+
+ /* This seek skips past an integral number of
+ chunks of size MSG_INPUT_SIZE. */
+ off = pos / MSG_INPUT_SIZE * MSG_INPUT_SIZE;
+ if (fseeko(iob, off, SEEK_SET) == -1)
+ adios("getfld's iob", "failed to set offset to skip: "
+ "%" PRIdMAX, (intmax_t)off);
+ num_read = fread (s->msg_buf, 1, MSG_INPUT_SIZE, iob);
+ s->readpos = s->msg_buf + pos % MSG_INPUT_SIZE;
+ s->end = s->msg_buf + num_read;
+ s->total_bytes_read = pos;
+ }
+ }
+
+ if (fseeko(iob, pos, SEEK_SET) == -1)
+ adios("getfld's iob", "failed to set offset on entry: %" PRIdMAX,
+ (intmax_t)pos);
}
static void
leave_getfld (m_getfld_state_t s) {
- /* Save the internal file position that we use for the input buffer. */
- s->last_internal_pos = ftello (s->iob);
-
- /* Set file stream position so that callers can use ftell(). */
s->total_bytes_read += s->bytes_read;
- fseeko (s->iob, s->total_bytes_read, SEEK_SET);
- s->last_caller_pos = ftello (s->iob);
+
+ if (s->track_filepos) {
+ /* Save the internal file position that we use for the input buffer. */
+ if ((s->last_internal_pos = ftello(s->iob)) == -1)
+ adios("getfld's iob", "failed to get offset before seek");
+
+ /* Set file stream position so that callers can use ftell(). */
+ if (fseeko(s->iob, s->total_bytes_read, SEEK_SET) == -1)
+ adios("getfld's iob", "failed to set offset: %" PRIdMAX,
+ (intmax_t)s->total_bytes_read);
+
+ s->last_caller_pos = s->total_bytes_read;
+ }
}
static size_t
read_more (m_getfld_state_t s) {
- /* Retain at least edelimlen characters that have already been
- read so that we can back up to them in m_Eom(). */
- ssize_t retain = s->edelimlen;
+ /* Retain at least edelimlen characters that have already been read,
+ if at least edelimlen have been read, so that we can back up to them
+ in m_Eom(). */
+ ssize_t retain = s->end - s->msg_buf < s->edelimlen ? 0 : s->edelimlen;
size_t num_read;
- if (retain < s->end - s->readpos) retain = s->end - s->readpos;
- /* assert (retain <= s->readpos - s->msg_buf <= sizeof msg_buf); */
+ if (retain > 0) {
+ if (retain < s->end - s->readpos)
+ retain = s->end - s->readpos;
+ assert (retain <= s->readpos - s->msg_buf);
- /* Move what we want to retain at end of the buffer to the beginning. */
- memmove (s->msg_buf, s->readpos - retain, retain);
+ /* Move what we want to retain at end of the buffer to the beginning. */
+ memmove (s->msg_buf, s->readpos - retain, retain);
+ }
s->readpos = s->msg_buf + retain;
num_read = fread (s->readpos, 1, MSG_INPUT_SIZE, s->iob);
return num_read;
}
+/* Return the next character consumed from the input, fetching more of
+ * the input for the buffer if required, or EOF on end of file. */
static int
Getc (m_getfld_state_t s) {
- if (s->end - s->readpos < 1) {
- if (read_more (s) == 0) {
- /* Pretend that we read a character. That's what stdio does. */
- ++s->readpos;
- return EOF;
- }
- }
+ if ((s->end - s->readpos < 1 && read_more (s) == 0) ||
+ s->readpos >= s->end)
+ return EOF;
- ++s->bytes_read;
- return s->readpos < s->end ? *s->readpos++ : EOF;
+ s->bytes_read++;
+ return (unsigned char)*s->readpos++;
}
+/* Return the next character that Getc() would return, which may be EOF. */
static int
-Peek (m_getfld_state_t s) {
- if (s->end - s->readpos < 1) {
- if (read_more (s) == 0) {
- /* Pretend that we read a character. That's what stdio does. */
- ++s->readpos;
- return EOF;
- }
- }
+Peek (m_getfld_state_t s)
+{
+ int c;
+
+ c = Getc(s);
+ if (c != EOF)
+ Ungetc(s);
- return s->readpos < s->end ? *s->readpos : EOF;
+ return c;
}
-static int
-Ungetc (int c, m_getfld_state_t s) {
- if (s->readpos == s->msg_buf) {
- return EOF;
- } else {
- --s->bytes_read;
- return *--s->readpos = c;
- }
+/* If there's room, undo the consumption of one character from msg_buf,
+ * rewinding so it's read next, else die. */
+static void
+Ungetc(m_getfld_state_t s)
+{
+ if (s->readpos == s->msg_buf)
+ adios(NULL, "Ungetc() at start of message buffer.");
+
+ s->readpos--;
+ s->bytes_read--;
}
int
-m_getfld (m_getfld_state_t s, unsigned char name[NAMESZ], unsigned char *buf,
- int *bufsz, FILE *iob)
+m_getfld (m_getfld_state_t *gstate, char name[NAMESZ], char *buf, int *bufsz,
+ FILE *iob)
{
- register unsigned char *cp;
- register int max, n, c;
+ m_getfld_state_t s;
+ char *cp;
+ int max, n, c;
- enter_getfld (s, iob);
+ enter_getfld (gstate, iob);
+ s = *gstate;
- if ((c = Getc(s)) < 0) {
+ if ((c = Getc(s)) == EOF) {
*bufsz = *buf = 0;
leave_getfld (s);
return s->state = FILEEOF;
}
if (eom (c, s)) {
- if (! s->eom_action) {
- /* flush null messages */
- while ((c = Getc(s)) >= 0 && eom (c, s))
- ;
+ /* flush null messages */
+ while ((c = Getc(s)) != EOF && eom (c, s))
+ ;
- if (c >= 0)
- Ungetc(c, s);
- }
+ if (c != EOF)
+ Ungetc(s);
*bufsz = *buf = 0;
leave_getfld (s);
return s->state = FILEEOF;
case FLD:
if (c == '\n' || c == '-') {
/* we hit the header/body separator */
- while (c != '\n' && (c = Getc(s)) >= 0) continue;
-
- if (c < 0 || (c = Getc(s)) < 0 || eom (c, s)) {
- if (! s->eom_action) {
- /* flush null messages */
- while ((c = Getc(s)) >= 0 && eom (c, s))
- ;
- if (c >= 0)
- Ungetc(c, s);
- }
+ while (c != '\n' && (c = Getc(s)) != EOF)
+ ;
+
+ if (c == EOF || (c = Getc(s)) == EOF || eom (c, s)) {
+ /* flush null messages */
+ while ((c = Getc(s)) != EOF && eom (c, s))
+ ;
+ if (c != EOF)
+ Ungetc(s);
*bufsz = *buf = 0;
leave_getfld (s);
return s->state = FILEEOF;
int next_char;
if (c == EOF || (next_char = Peek (s)) == EOF) {
*bufsz = *cp = *buf = 0;
- advise (NULL, "eof encountered in field \"%s\"", name);
+ inform("eof encountered in field \"%s\"", name);
leave_getfld (s);
return s->state = FMTERR;
}
if (*bufsz < n + 1) {
/* No, it can't. Oh well, guess we'll blow up. */
*bufsz = *cp = *buf = 0;
- advise (NULL, "eol encountered in field \"%s\"", name);
+ inform("eol encountered in field \"%s\"", name);
s->state = FMTERR;
break;
}
memcpy (buf, name, n - 1);
buf[n - 1] = '\n';
buf[n] = '\0';
+ /* Indicate this wasn't a header field using a character
+ that can't appear in a header field. */
+ name[0] = ':';
/* The last character read was '\n'. s->bytes_read
(and n) include that, but it was not put into the
name array in the for loop above. So subtract 1. */
*bufsz = --s->bytes_read; /* == n - 1 */
leave_getfld (s);
return s->state = BODY;
- } else if (max <= n) {
+ }
+ if (max <= n) {
/* By design, the loop above discards the last character
it had read. It's in c, use it. */
*cp++ = c;
*bufsz = *cp = *buf = 0;
- advise (NULL, "field name \"%s\" exceeds %d bytes", name,
+ inform("field name \"%s\" exceeds %d bytes", name,
NAMESZ - 2);
s->state = LENERR;
break;
}
/* Trim any trailing spaces from the end of name. */
- while (isspace (*--cp) && cp >= name) continue;
+ while (isspace ((unsigned char) *--cp) && cp >= name) continue;
*++cp = 0;
/* readpos points to the first character of the field body. */
- /* fall through */
+ /* FALLTHRU */
case FLDPLUS: {
/*
n = 0;
for (finished = 0; ! finished; ) {
while (c != '\n' && c != EOF && n++ < max) {
- *cp++ = c = Getc (s);
+ if ((c = Getc (s)) != EOF)
+ *cp++ = c;
}
- if (c != EOF) c = Peek (s);
+ if (c != EOF)
+ c = Peek (s);
if (max < n) {
- /* the dest buffer is full */
+ /* The dest buffer is full. Need to back the read
+ pointer up by one because when m_getfld() is
+ reentered, it will read a character. Then
+ we'll jump right to the FLDPLUS handling code,
+ which will not store that character, but
+ instead move on to the next one. */
+ if (s->readpos > s->msg_buf) {
+ --s->readpos;
+ --s->bytes_read;
+ }
s->state = FLDPLUS;
finished = 1;
} else if (c != ' ' && c != '\t') {
* get the message body up to bufsz characters or the
* end of the message.
*/
- unsigned char *bp;
+ char *bp;
+ name[0] = '\0';
max = *bufsz-1;
/* Back up and store the current position. */
bp = --s->readpos;
- c = s->end - s->readpos < max ? s->end - s->readpos : max;
+ c = min(s->end - s->readpos, max);
if (s->msg_style != MS_DEFAULT && c > 1) {
/*
* packed maildrop - only take up to the (possible)
* algorithms vs. brute force.) Since I (currently)
* run MH on a vax, we use the matchc instruction. --vj
*/
- unsigned char *ep;
+ char *ep;
- if ((ep = matchc( s->fdelimlen, s->fdelim, c, bp )))
+ if ((ep = memmem(bp, c, s->fdelim, s->fdelimlen)))
+ /* Plus one to nab the '\n' that starts fdelim as
+ * that ends the previous line; it isn't part of
+ * msg_delim. */
c = ep - bp + 1;
else {
/*
* ends with one of the characters in the pattern
* (excluding the first and last), we do only one test.
*/
- unsigned char *sp;
+ char *sp;
- ep = bp + c - 1;
- if ((sp = s->pat_map[*ep])) {
+ ep = bp + c - 1; /* The last byte. */
+ if ((sp = s->pat_map[(unsigned char) *ep])) {
do {
/* This if() is true unless (a) the buffer is too
* small to contain this delimiter prefix, or
}
+/* m_getfld() with the existing iob. */
+int m_getfld2(m_getfld_state_t *gstate, char name[NAMESZ], char *buf, int *bufsz)
+{
+ if (!*gstate)
+ adios(NULL, "m_getfld2 without gstate");
+
+ return m_getfld(gstate, name, buf, bufsz, (*gstate)->iob);
+}
+
+
void
-m_unknown(m_getfld_state_t s, FILE *iob)
+m_unknown(m_getfld_state_t *gstate, FILE *iob)
{
- register int c;
+ m_getfld_state_t s;
+ int c;
char text[MAX_DELIMITER_SIZE];
char from[] = "From ";
- register char *cp;
- register char *delimstr;
+ char *cp;
+ char *delimstr;
unsigned int i;
- enter_getfld (s, iob);
+ enter_getfld (gstate, iob);
+ s = *gstate;
/*
- * Figure out what the message delimitter string is for this
+ * Figure out what the message delimiter string is for this
* maildrop. (This used to be part of m_Eom but I didn't like
* the idea of an "if" statement that could only succeed on the
* first call to m_Eom getting executed on each call, i.e., at
s->msg_style = MS_UNKNOWN;
for (i = 0, cp = text; i < sizeof text; ++i, ++cp) {
- if ((*cp = Getc (s)) == EOF) {
+ if ((c = Getc (s)) == EOF) {
+ *cp = '\0';
break;
}
+ *cp = c;
}
if (i == sizeof from-1 && strncmp (text, "From ", sizeof from-1) == 0) {
s->msg_style = MS_MBOX;
delimstr = "\nFrom ";
- while ((c = Getc (s)) != '\n' && c >= 0) continue;
+ while ((c = Getc(s)) != EOF && c != '\n')
+ ;
} else {
/* not a Unix style maildrop */
s->readpos -= s->bytes_read;
- delimstr = mmdlm2;
+ s->bytes_read = 0;
+ delimstr = MMDF_DELIM;
s->msg_style = MS_MMDF;
}
+
+ /* "\nFrom \0" 7 "\001\001\001\001\n\0" 6
+ * | |
+ * delimstr c=6 delimstr c=5
+ */
c = strlen (delimstr);
- s->fdelim = (unsigned char *) mh_xmalloc((size_t) (c + 3));
+ s->fdelim = mh_xmalloc (c + 3); /* \0, \n, delimstr, \0 */
*s->fdelim++ = '\0';
*s->fdelim = '\n';
- s->msg_delim = (char *)s->fdelim+1;
- s->edelim = (unsigned char *)s->msg_delim+1;
s->fdelimlen = c + 1;
- s->edelimlen = c - 1; /* == strlen (delimstr) */
+ s->msg_delim = s->fdelim+1;
strcpy (s->msg_delim, delimstr);
- s->delimend = (unsigned char *)s->msg_delim + s->edelimlen;
+ s->edelim = s->msg_delim+1;
+ s->edelimlen = c - 1;
+ s->delimend = s->msg_delim + s->edelimlen;
if (s->edelimlen <= 1)
adios (NULL, "maildrop delimiter must be at least 2 bytes");
+
/*
* build a Boyer-Moore end-position map for the matcher in m_getfld.
* N.B. - we don't match just the first char (since it's the newline
* separator) or the last char (since the matchc would have found it
* if it was a real delim).
*/
- s->pat_map = (unsigned char **) calloc (256, sizeof(unsigned char *));
+ s->pat_map = (char **) mh_xcalloc (256, sizeof(char *));
- for (cp = (char *) s->fdelim + 1; cp < (char *) s->delimend; cp++ )
- s->pat_map[(unsigned char)*cp] = (unsigned char *) cp;
+ for (cp = s->fdelim + 1; cp < s->delimend; cp++ )
+ s->pat_map[(unsigned char)*cp] = cp;
if (s->msg_style == MS_MMDF) {
/* flush extra msg hdrs */
- while ((c = Getc(s)) >= 0 && eom (c, s))
+ while ((c = Getc(s)) != EOF && eom (c, s))
;
- if (c >= 0)
- Ungetc(c, s);
+ if (c != EOF)
+ Ungetc(s);
}
leave_getfld (s);
}
-void
-m_eomsbr (m_getfld_state_t s, int (*action)(int))
+/* m_unknown() with the existing iob. */
+void m_unknown2(m_getfld_state_t *gstate)
{
- if ((s->eom_action = action)) {
- s->msg_style = MS_MSH;
- *s->msg_delim = 0;
- s->fdelimlen = 1;
- s->delimend = s->fdelim;
- } else {
- s->msg_style = MS_MMDF;
- s->msg_delim = (char *)s->fdelim + 1;
- s->fdelimlen = strlen((char *)s->fdelim);
- s->delimend = (unsigned char *)(s->msg_delim + s->edelimlen);
- }
+ if (!*gstate)
+ adios(NULL, "m_unknown2 without gstate");
+
+ m_unknown(gstate, (*gstate)->iob);
}
*/
static int
-m_Eom (m_getfld_state_t s, int c)
+m_Eom (m_getfld_state_t s)
{
- register int i;
+ int i;
char text[MAX_DELIMITER_SIZE];
char *cp;
+ int adjust = 1;
for (i = 0, cp = text; i < s->edelimlen; ++i, ++cp) {
- if ((*cp = Getc (s)) == EOF) {
+ int c2;
+
+ if ((c2 = Getc (s)) == EOF) {
+ *cp = '\0';
break;
}
+ *cp = c2;
}
if (i != s->edelimlen ||
strncmp (text, (char *)s->edelim, s->edelimlen)) {
- if (i == 0 && s->msg_style == MS_MBOX)
+ if (i == 0 && s->msg_style == MS_MBOX) {
/* the final newline in the (brain damaged) unix-format
- * maildrop is part of the delimitter - delete it.
+ * maildrop is part of the delimiter - delete it.
*/
return 1;
+ }
+
+ if (i <= 2 && s->msg_style == MS_MBOX &&
+ i != s->edelimlen && ! strncmp(text, s->fdelim, i)) {
+ /* If all or part of fdelim appeared at the end of the file,
+ back up even more so that the bytes are included in the
+ message. */
+ adjust = 2;
+ }
/* Did not find delimiter, so restore the read position.
Note that on input, a character had already been read
with Getc(). It will be unget by m_getfld () on return. */
- s->readpos -= s->bytes_read - 1;
+ s->readpos -= s->bytes_read - adjust;
+ s->bytes_read = adjust;
return 0;
}
if (s->msg_style == MS_MBOX) {
- while ((c = Getc (s)) != '\n')
- if (c < 0)
- break;
+ int c;
+ while ((c = Getc(s)) != EOF && c != '\n')
+ ;
}
return 1;
}
-
-
-static unsigned char *
-matchc(int patln, char *pat, int strln, char *str)
-{
- register char *es = str + strln - patln;
- register char *sp;
- register char *pp;
- register char *ep = pat + patln;
- register char pc = *pat++;
-
- for(;;) {
- while (pc != *str++)
- if (str > es)
- return 0;
- if (str > es+1)
- return 0;
- sp = str; pp = pat;
- while (pp < ep && *sp++ == *pp)
- pp++;
- if (pp >= ep)
- return ((unsigned char *)--str);
- }
-}
projects / nmh / blobdiff