X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/adc0a3232c43033729dbc036f0757bddfba463bd..2d8a2d23b4c17ffb277d008f4e011fcc2bb2fa7e:/sbr/m_getfld.c?ds=inline diff --git a/sbr/m_getfld.c b/sbr/m_getfld.c index 59b49677..f84d2933 100644 --- a/sbr/m_getfld.c +++ b/sbr/m_getfld.c @@ -11,6 +11,75 @@ #include #include +/* + 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, 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: + + char **pat_map + char *fdelim + char *delimend + int fdelimlen + char *edelim + int edelimlen + char *msg_delim + int msg_style + int (*eom_action)(int) + + Usage + ===== + m_getfld_state_t gstate = 0; + ... + 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); +*/ + +/* 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 an inter-message delimiter. It really is four CTRL-A:s followed by a @@ -134,69 +203,12 @@ 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 char *matchc(int, char *, int, char *); #define eom(c,s) (s->msg_style != MS_DEFAULT && \ (((c) == *s->msg_delim && m_Eom(s,c)) || \ @@ -207,17 +219,18 @@ static unsigned char *matchc(int, char *, int, char *); * * MSG_INPUT_SIZE is the size of the buffer. * MAX_DELIMITER_SIZE is the maximum size of the delimiter used to - * separate messages in a maildrop, such as mbox "\nFrom ". + * 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) -#define MAX_DELIMITER_SIZE 32 + * of 4096. + */ +#define MSG_INPUT_SIZE 4096 +#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. */ + char msg_buf[2 * MSG_INPUT_SIZE + MAX_DELIMITER_SIZE]; + char *readpos; + 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 @@ -231,7 +244,7 @@ struct m_getfld_state { off_t last_internal_pos; FILE *iob; - unsigned char **pat_map; + char **pat_map; int msg_style; /* * The "full" delimiter string for a packed maildrop consists @@ -246,41 +259,66 @@ struct m_getfld_state { * has been read and matched before m_Eom is called. */ char *msg_delim; - unsigned char *fdelim; - unsigned char *delimend; + char *fdelim; + char *delimend; int fdelimlen; - unsigned char *edelim; + char *edelim; int edelimlen; int (*eom_action)(int); int state; + int track_filepos; }; +static 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_init (m_getfld_state_t *gstate, FILE *iob) { + m_getfld_state_t s; + + s = *gstate = (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 = 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->eom_action = NULL; + s->state = FLD; + s->track_filepos = 0; } /* scan() needs to force a state an initial state of FLD for each message. */ void -m_getfld_state_reset (m_getfld_state_t *s) { - (*s)->state = FLD; +m_getfld_state_reset (m_getfld_state_t *gstate) { + if (*gstate) { + (*gstate)->state = FLD; + } } -void m_getfld_state_destroy (m_getfld_state_t *s) { - if (*s) { - free (*s); - *s = 0; +/* 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_track_filepos (m_getfld_state_t *gstate, FILE *iob) { + if (! *gstate) { + m_getfld_state_init (gstate, iob); + } + + (*gstate)->track_filepos = 1; +} + +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; } } @@ -314,60 +352,78 @@ void m_getfld_state_destroy (m_getfld_state_t *s) { static void -enter_getfld (m_getfld_state_t s, FILE *iob) { - off_t pos; +enter_getfld (m_getfld_state_t *gstate, FILE *iob) { + m_getfld_state_t s; + 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(). */ - s->iob = iob; - - pos = ftello (iob); + if (! *gstate) { + m_getfld_state_init (gstate, iob); + } + s = *gstate; + s->bytes_read = 0; - if (pos != 0) { - off_t pos_movement = pos - s->last_caller_pos; /* Can be < 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_movement == 0) { - pos = s->last_internal_pos; + if (s->track_filepos && (pos != 0 || s->last_internal_pos != 0)) { + if (s->last_internal_pos == 0) { + s->total_bytes_read = 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) { - /* We can shift readpos and remain within the bounds of - msg_buf. */ - s->readpos += pos_movement; - s->total_bytes_read += pos_movement; + off_t pos_movement = pos - s->last_caller_pos; /* Can be < 0. */ + + if (pos_movement == 0) { 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; + /* 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->last_internal_pos != 0) fseeko (iob, pos, SEEK_SET); + fseeko (iob, pos, SEEK_SET); + } } - - s->bytes_read = 0; } 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. */ + s->last_internal_pos = ftello (s->iob); + + /* Set file stream position so that callers can use ftell(). */ + fseeko (s->iob, s->total_bytes_read, SEEK_SET); + s->last_caller_pos = ftello (s->iob); + } } static size_t @@ -378,7 +434,7 @@ read_more (m_getfld_state_t s) { 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); */ + 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); @@ -390,6 +446,9 @@ read_more (m_getfld_state_t s) { return num_read; } +/* The return values of the following functions are a bit + subtle. They can return 0x00 - 0xff as a valid character, + but EOF is typically 0xffffffff. */ static int Getc (m_getfld_state_t s) { if (s->end - s->readpos < 1) { @@ -401,16 +460,20 @@ Getc (m_getfld_state_t s) { } ++s->bytes_read; - return s->readpos < s->end ? *s->readpos++ : EOF; + return s->readpos < s->end ? (unsigned char) *s->readpos++ : EOF; } static int Peek (m_getfld_state_t s) { - int next_char = Getc (s); - --s->readpos; - --s->bytes_read; + 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; + } + } - return next_char; + return s->readpos < s->end ? (unsigned char) *s->readpos : EOF; } static int @@ -419,19 +482,21 @@ Ungetc (int c, m_getfld_state_t s) { return EOF; } else { --s->bytes_read; - return *--s->readpos = c; + return *--s->readpos = (unsigned char) c; } } 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; + m_getfld_state_t s; + register char *cp; register int max, n, c; - enter_getfld (s, iob); + enter_getfld (gstate, iob); + s = *gstate; if ((c = Getc(s)) < 0) { *bufsz = *buf = 0; @@ -547,7 +612,7 @@ m_getfld (m_getfld_state_t s, unsigned char name[NAMESZ], unsigned char *buf, } /* 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 */ @@ -594,10 +659,10 @@ m_getfld (m_getfld_state_t s, unsigned char name[NAMESZ], unsigned char *buf, * get the message body up to bufsz characters or the * end of the message. */ - unsigned char *bp; + char *bp; max = *bufsz-1; - /* Back up and store the current position and update cnt. */ + /* Back up and store the current position. */ bp = --s->readpos; c = s->end - s->readpos < max ? s->end - s->readpos : max; if (s->msg_style != MS_DEFAULT && c > 1) { @@ -613,7 +678,7 @@ m_getfld (m_getfld_state_t s, unsigned char name[NAMESZ], unsigned char *buf, * 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 ))) c = ep - bp + 1; @@ -631,10 +696,10 @@ m_getfld (m_getfld_state_t s, unsigned char name[NAMESZ], unsigned char *buf, * 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])) { + 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 @@ -696,14 +761,18 @@ m_getfld (m_getfld_state_t s, unsigned char name[NAMESZ], unsigned char *buf, void -m_unknown(m_getfld_state_t s, FILE *iob) +m_unknown(m_getfld_state_t *gstate, FILE *iob) { + m_getfld_state_t s; register int c; char text[MAX_DELIMITER_SIZE]; + char from[] = "From "; register char *cp; register 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 @@ -720,37 +789,33 @@ m_unknown(m_getfld_state_t s, FILE *iob) s->msg_style = MS_UNKNOWN; - for (c = 0, cp = text; c < 5; ++c, ++cp) { + for (i = 0, cp = text; i < sizeof text; ++i, ++cp) { if ((*cp = Getc (s)) == EOF) { break; } } - if (c == 5 && strncmp (text, "From ", 5) == 0) { + 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; - /* m_unknown is only called on maildrop files, and they are only - read using m_getfld (). The caller musn't try to read from - the stream directly because the file position indicator was - not advanced based on bytes_read, but instead on whatever - was read into the message buffer. */ } else { /* not a Unix style maildrop */ s->readpos -= s->bytes_read; + s->bytes_read = 0; delimstr = mmdlm2; s->msg_style = MS_MMDF; } c = strlen (delimstr); - s->fdelim = (unsigned char *) mh_xmalloc((size_t) (c + 3)); + s->fdelim = mh_xmalloc (c + 3); *s->fdelim++ = '\0'; *s->fdelim = '\n'; - s->msg_delim = (char *)s->fdelim+1; - s->edelim = (unsigned char *)s->msg_delim+1; + s->msg_delim = s->fdelim+1; + s->edelim = s->msg_delim+1; s->fdelimlen = c + 1; s->edelimlen = c - 1; /* == strlen (delimstr) */ strcpy (s->msg_delim, delimstr); - s->delimend = (unsigned char *)s->msg_delim + s->edelimlen; + s->delimend = s->msg_delim + s->edelimlen; if (s->edelimlen <= 1) adios (NULL, "maildrop delimiter must be at least 2 bytes"); /* @@ -759,10 +824,10 @@ m_unknown(m_getfld_state_t s, FILE *iob) * 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 **) calloc (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 */ @@ -786,9 +851,9 @@ m_eomsbr (m_getfld_state_t s, int (*action)(int)) 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); + s->msg_delim = s->fdelim + 1; + s->fdelimlen = strlen (s->fdelim); + s->delimend = s->msg_delim + s->edelimlen; } } @@ -822,6 +887,7 @@ m_Eom (m_getfld_state_t s, int c) 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->bytes_read = 1; return 0; } @@ -835,7 +901,7 @@ m_Eom (m_getfld_state_t s, int c) } -static unsigned char * +static char * matchc(int patln, char *pat, int strln, char *str) { register char *es = str + strln - patln; @@ -854,6 +920,6 @@ matchc(int patln, char *pat, int strln, char *str) while (pp < ep && *sp++ == *pp) pp++; if (pp >= ep) - return ((unsigned char *)--str); + return --str; } }