X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/2e768e3cc0ce768373d3320eca4cdd08b8cf646a..7711f3fc00259e55f630cfe6104eff3083dc9d77:/sbr/m_getfld.c diff --git a/sbr/m_getfld.c b/sbr/m_getfld.c index c6716284..b0960547 100644 --- a/sbr/m_getfld.c +++ b/sbr/m_getfld.c @@ -41,12 +41,9 @@ 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 + for the maildrop. Called by inc and scan 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 @@ -61,7 +58,6 @@ int edelimlen char *msg_delim int msg_style - int (*eom_action)(int) Usage ===== @@ -207,12 +203,11 @@ * static prototypes */ struct m_getfld_state; -static int m_Eom (m_getfld_state_t, int); +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. @@ -222,9 +217,9 @@ static 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 4096. + * of 8192. */ -#define MSG_INPUT_SIZE 4096 +#define MSG_INPUT_SIZE NMH_BUFSIZ #define MAX_DELIMITER_SIZE 5 struct m_getfld_state { @@ -264,7 +259,6 @@ struct m_getfld_state { int fdelimlen; char *edelim; int edelimlen; - int (*eom_action)(int); int state; int track_filepos; }; @@ -284,7 +278,6 @@ 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; } @@ -313,7 +306,10 @@ 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; } @@ -431,7 +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); @@ -501,14 +497,12 @@ m_getfld (m_getfld_state_t *gstate, char name[NAMESZ], char *buf, int *bufsz, 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; @@ -521,13 +515,11 @@ m_getfld (m_getfld_state_t *gstate, char name[NAMESZ], char *buf, int *bufsz, 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; @@ -627,12 +619,21 @@ m_getfld (m_getfld_state_t *gstate, char name[NAMESZ], char *buf, int *bufsz, 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') { @@ -787,8 +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; } } @@ -799,6 +803,7 @@ 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; } @@ -820,7 +825,7 @@ 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 = (char **) calloc (256, sizeof(char *)); + s->pat_map = (char **) mh_xcalloc (256, sizeof(char *)); for (cp = s->fdelim + 1; cp < s->delimend; cp++ ) s->pat_map[(unsigned char)*cp] = cp; @@ -837,37 +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 = s->fdelim + 1; - s->fdelimlen = strlen (s->fdelim); - s->delimend = 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; } } @@ -883,10 +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;