X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/d02aabc331c325d8206a862a43438f4a472a5306..58eaf2c2:/sbr/m_getfld.c?ds=sidebyside diff --git a/sbr/m_getfld.c b/sbr/m_getfld.c index ba135aa5..77dbeda1 100644 --- a/sbr/m_getfld.c +++ b/sbr/m_getfld.c @@ -11,6 +11,71 @@ #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 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 = 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 @@ -134,73 +199,15 @@ 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 = 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); -*/ - /* * static prototypes */ struct m_getfld_state; -static int m_Eom (m_getfld_state_t, int); -static unsigned char *matchc(int, char *, int, char *); +static int m_Eom (m_getfld_state_t); +static char *matchc(int, char *, int, char *); #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))) /* This replaces the old approach, with its direct access to stdio * internals. It uses one fread() to load a buffer that we manage. @@ -210,14 +217,15 @@ static unsigned char *matchc(int, char *, int, char *); * 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 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 +239,7 @@ 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 @@ -246,13 +254,13 @@ 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 @@ -270,8 +278,8 @@ m_getfld_state_init (m_getfld_state_t *gstate, FILE *iob) { 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. */ @@ -282,11 +290,26 @@ m_getfld_state_reset (m_getfld_state_t *gstate) { } } +/* 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); + if (s->fdelim) { + free (s->fdelim-1); + free (s->pat_map); + } free (s); *gstate = 0; } @@ -340,7 +363,7 @@ enter_getfld (m_getfld_state_t *gstate, FILE *iob) { readpos shift code being currently unused. */ s->iob = iob; - if (pos != 0 || s->last_internal_pos != 0) { + if (s->track_filepos && (pos != 0 || s->last_internal_pos != 0)) { if (s->last_internal_pos == 0) { s->total_bytes_read = pos; } else { @@ -384,13 +407,16 @@ enter_getfld (m_getfld_state_t *gstate, FILE *iob) { 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 @@ -401,7 +427,7 @@ 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); @@ -413,6 +439,9 @@ 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) { @@ -424,7 +453,7 @@ 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 @@ -437,7 +466,7 @@ Peek (m_getfld_state_t s) { } } - return s->readpos < s->end ? *s->readpos : EOF; + return s->readpos < s->end ? (unsigned char) *s->readpos : EOF; } static int @@ -446,17 +475,17 @@ 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 *gstate, 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) { m_getfld_state_t s; - register unsigned char *cp; + register char *cp; register int max, n, c; enter_getfld (gstate, iob); @@ -468,14 +497,12 @@ m_getfld (m_getfld_state_t *gstate, unsigned char name[NAMESZ], 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)) >= 0 && eom (c, s)) + ; - if (c >= 0) - Ungetc(c, s); - } + if (c >= 0) + Ungetc(c, s); *bufsz = *buf = 0; leave_getfld (s); return s->state = FILEEOF; @@ -488,13 +515,11 @@ m_getfld (m_getfld_state_t *gstate, unsigned char name[NAMESZ], 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); - } + /* flush null messages */ + while ((c = Getc(s)) >= 0 && eom (c, s)) + ; + if (c >= 0) + Ungetc(c, s); *bufsz = *buf = 0; leave_getfld (s); return s->state = FILEEOF; @@ -576,7 +601,7 @@ m_getfld (m_getfld_state_t *gstate, unsigned char name[NAMESZ], } /* 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 */ @@ -594,12 +619,21 @@ m_getfld (m_getfld_state_t *gstate, unsigned char name[NAMESZ], 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 (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') { @@ -623,7 +657,7 @@ m_getfld (m_getfld_state_t *gstate, unsigned char name[NAMESZ], * 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. */ @@ -642,7 +676,7 @@ m_getfld (m_getfld_state_t *gstate, unsigned char name[NAMESZ], * 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; @@ -660,10 +694,10 @@ m_getfld (m_getfld_state_t *gstate, unsigned char name[NAMESZ], * 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 @@ -754,8 +788,11 @@ m_unknown(m_getfld_state_t *gstate, FILE *iob) 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; + } else { + *cp = c; } } @@ -766,19 +803,20 @@ m_unknown(m_getfld_state_t *gstate, FILE *iob) } 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"); /* @@ -787,10 +825,10 @@ m_unknown(m_getfld_state_t *gstate, 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 */ @@ -804,37 +842,25 @@ m_unknown(m_getfld_state_t *gstate, FILE *iob) } -void -m_eomsbr (m_getfld_state_t s, int (*action)(int)) -{ - 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); - } -} - - /* * test for msg delimiter string */ static int -m_Eom (m_getfld_state_t s, int c) +m_Eom (m_getfld_state_t s) { register int i; char text[MAX_DELIMITER_SIZE]; char *cp; 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; + } else { + *cp = c2; } } @@ -850,10 +876,12 @@ 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; } if (s->msg_style == MS_MBOX) { + int c; while ((c = Getc (s)) != '\n') if (c < 0) break; @@ -863,7 +891,7 @@ 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; @@ -882,6 +910,6 @@ matchc(int patln, char *pat, int strln, char *str) while (pp < ep && *sp++ == *pp) pp++; if (pp >= ep) - return ((unsigned char *)--str); + return --str; } }