X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/2fc2c991583f130d972aab63e3e56340d780dcb3..8699f1cc:/sbr/m_getfld.c?ds=sidebyside diff --git a/sbr/m_getfld.c b/sbr/m_getfld.c index 1ca0bee3..f5fa9a4e 100644 --- a/sbr/m_getfld.c +++ b/sbr/m_getfld.c @@ -8,6 +8,7 @@ #include #include #include +#include /* Purpose @@ -59,10 +60,13 @@ Usage ===== - m_getfld_state_t gstate = 0; - ... - int state = m_getfld (&gstate, ...); - ... + 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: @@ -90,7 +94,7 @@ 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 + the eom() macro returns false saying we aren't at the end of the message. The next thing to do is to read the mts.conf file and initialize @@ -200,6 +204,7 @@ /* * static prototypes */ +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 && \ @@ -227,6 +232,11 @@ static int m_Eom (m_getfld_state_t); #define MAX_DELIMITER_SIZE 5 struct m_getfld_state { + /* 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. */ @@ -235,41 +245,24 @@ struct m_getfld_state { * equals end then msg_buf is empty. */ char *end; - /* Bytes of iob consumed during this call. */ - off_t bytes_read; + /* 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; /* 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; - /* 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; - /* 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; /* 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 ", @@ -277,13 +270,13 @@ struct m_getfld_state { * 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; + /* 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; - /* The last non-NUL char of msg_delim. */ - char *delimend; /* strlen(fdelim). */ int fdelimlen; /* The second char of msg_delim. Used when the first char has @@ -305,22 +298,22 @@ struct m_getfld_state { * 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; - /* 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; }; -static -void -m_getfld_state_init (m_getfld_state_t *gstate, FILE *iob) { +m_getfld_state_t m_getfld_state_init(FILE *iob) +{ m_getfld_state_t s; NEW(s); - *gstate = 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; @@ -332,6 +325,8 @@ m_getfld_state_init (m_getfld_state_t *gstate, FILE *iob) { 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. */ @@ -348,12 +343,21 @@ m_getfld_state_reset (m_getfld_state_t *gstate) { void m_getfld_track_filepos (m_getfld_state_t *gstate, FILE *iob) { if (! *gstate) { - m_getfld_state_init (gstate, iob); + *gstate = m_getfld_state_init(iob); } (*gstate)->track_filepos = 1; } +/* 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; @@ -403,7 +407,7 @@ enter_getfld (m_getfld_state_t *gstate, FILE *iob) { off_t pos_movement; if (! *gstate) { - m_getfld_state_init (gstate, iob); + *gstate = m_getfld_state_init(iob); } s = *gstate; s->bytes_read = 0; @@ -419,7 +423,8 @@ enter_getfld (m_getfld_state_t *gstate, FILE *iob) { if (!s->track_filepos) return; - pos = ftello(iob); + if ((pos = ftello(iob)) == -1) + adios("getfld's iob", "failed to get offset on entry"); if (pos == 0 && s->last_internal_pos == 0) return; @@ -448,11 +453,15 @@ enter_getfld (m_getfld_state_t *gstate, FILE *iob) { 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. */ - fseeko (iob, pos/MSG_INPUT_SIZE * MSG_INPUT_SIZE, SEEK_SET); + 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; @@ -460,7 +469,9 @@ enter_getfld (m_getfld_state_t *gstate, FILE *iob) { } } - fseeko (iob, pos, SEEK_SET); + if (fseeko(iob, pos, SEEK_SET) == -1) + adios("getfld's iob", "failed to set offset on entry: %" PRIdMAX, + (intmax_t)pos); } static void @@ -469,11 +480,15 @@ leave_getfld (m_getfld_state_t s) { if (s->track_filepos) { /* Save the internal file position that we use for the input buffer. */ - s->last_internal_pos = ftello (s->iob); + 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(). */ - fseeko (s->iob, s->total_bytes_read, SEEK_SET); - s->last_caller_pos = ftello (s->iob); + 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; } } @@ -513,27 +528,29 @@ Getc (m_getfld_state_t s) { return (unsigned char)*s->readpos++; } -/* Return the next character that would be read by Getc() without - * consuming it, fetching more of the input for the buffer if required, - * or EOF on end of file. */ +/* 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 && read_more (s) == 0) { - return EOF; - } - return s->readpos < s->end ? (unsigned char) *s->readpos : EOF; +Peek (m_getfld_state_t s) +{ + int c; + + c = Getc(s); + if (c != EOF) + Ungetc(s); + + return c; } -/* If there's room, put non-EOF c back into msg_buf and rewind so it's - * read next. c need not be the value already in the buffer. If there - * isn't room then return EOF, else return c. */ -static int -Ungetc (int c, m_getfld_state_t s) { - if (s->readpos == s->msg_buf) { - return EOF; - } - --s->bytes_read; - return *--s->readpos = (unsigned char) 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--; } @@ -559,7 +576,7 @@ m_getfld (m_getfld_state_t *gstate, char name[NAMESZ], char *buf, int *bufsz, ; if (c != EOF) - Ungetc(c, s); + Ungetc(s); *bufsz = *buf = 0; leave_getfld (s); return s->state = FILEEOF; @@ -577,7 +594,7 @@ m_getfld (m_getfld_state_t *gstate, char name[NAMESZ], char *buf, int *bufsz, while ((c = Getc(s)) != EOF && eom (c, s)) ; if (c != EOF) - Ungetc(c, s); + Ungetc(s); *bufsz = *buf = 0; leave_getfld (s); return s->state = FILEEOF; @@ -826,6 +843,16 @@ m_getfld (m_getfld_state_t *gstate, char name[NAMESZ], char *buf, int *bufsz, } +/* 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 *gstate, FILE *iob) { @@ -909,13 +936,23 @@ m_unknown(m_getfld_state_t *gstate, FILE *iob) while ((c = Getc(s)) != EOF && eom (c, s)) ; if (c != EOF) - Ungetc(c, s); + Ungetc(s); } leave_getfld (s); } +/* m_unknown() with the existing iob. */ +void m_unknown2(m_getfld_state_t *gstate) +{ + if (!*gstate) + adios(NULL, "m_unknown2 without gstate"); + + m_unknown(gstate, (*gstate)->iob); +} + + /* * test for msg delimiter string */