Minimized message files reads in m_getfld() when the caller

interleaves calls to fseek().  Added check of handling of
long header field name without a colon to test-header.

diff --git a/sbr/m_getfld.c b/sbr/m_getfld.c
index 2b5ef094214bd3815d9554b7ba6de467887f535b..05b9e953178aedd23baebe9e36f3c03dd6805d25 100644 (file)
--- a/sbr/m_getfld.c
+++ b/sbr/m_getfld.c
@@ -246,16 +246,45 @@ static struct m_getfld_buffer {
     /* 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
     /* 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_file_pos is stored when
+       m_getfld() msg_count global.  last_caller_pos is stored when

        leaving m_getfld()/m_unkown(), then checked on the next entry.
        last_internal_pos is used to remember the position used
        internally by m_getfld() (read_more(), actually). */
     off_t bytes_read;
        leaving m_getfld()/m_unkown(), then checked on the next entry.
        last_internal_pos is used to remember the position used
        internally by m_getfld() (read_more(), actually). */
     off_t bytes_read;

-    off_t total_bytes_read;
-    off_t last_file_pos;
+    off_t total_bytes_read; /* by caller, not necessarily from input file */
+    off_t last_caller_pos;

     off_t last_internal_pos;
 } m;
 
     off_t last_internal_pos;
 } m;
 

+/*
+  Summary of file and message input buffer positions:
+
+  input file      -------------------------------------------EOF
+                                 |              |
+                          last_caller_pos  last_internal_pos
+
+
+  msg_buf                   --------------------EOF
+                            |         |         |
+                         msg_buf   readpos     end
+
+                            |<>|=retained characters, difference
+                                 between last_internal_pos and
+                                 first readpos value after reading
+                                 in new chunk in read_more()
+
+  When returning from m_getfld()/m_unknown():
+  1) Save the internal file position in last_internal_pos.  That's the
+     m_getfld() position reference in the input file.
+  2) Set file stream position so that callers can use ftell().
+
+  When entering m_getfld()/m_unknown():
+  Check to see if the call had changed the file position.  If so,
+  adjust the internal position reference accordingly.  If not, restore
+  the internal file position from last_internal_pos.
+*/
+
+

 static void
 enter_getfld (FILE *iob, struct m_getfld_buffer *m) {
     off_t pos = ftello (iob);
 static void
 enter_getfld (FILE *iob, struct m_getfld_buffer *m) {
     off_t pos = ftello (iob);


@@ -266,30 +295,33 @@ enter_getfld (FILE *iob, struct m_getfld_buffer *m) {
        /* A new file stream, so reset the buffer state. */
        m->readpos = m->end = m->msg_buf;
        m->total_bytes_read = 0;
        /* A new file stream, so reset the buffer state. */
        m->readpos = m->end = m->msg_buf;
        m->total_bytes_read = 0;

-       m->last_file_pos = m->last_internal_pos = ftello (iob);
+       m->last_caller_pos = m->last_internal_pos = ftello (iob);

     } else {
     } else {

-       /* If the current file stream position differs from the last one,
-          then caller must have called ftell(), so adjust. */
-       if (pos != m->last_file_pos) {
-           size_t num_read;
-
-           /* Opportunity for optimization here:  don't reread if the
-              new position had already been read into the buffer,
-              just move m->readpos to it. */
-           fseeko (iob, 0, SEEK_SET);
-           do {
-               num_read = fread (m->msg_buf, 1, MSG_INPUT_SIZE, iob);
-               pos -= num_read;
-           } while (pos > 0);
-           pos += num_read;
-           /* assert (ftello (iob) == pos); */
-           m->readpos = m->msg_buf + pos;
-           m->end = m->msg_buf + num_read;
-           m->last_internal_pos = m->last_file_pos = pos;
-           m->total_bytes_read = pos;
-       } else {
-           /* Restore the file position that we use for the input buffer. */
+       off_t pos_movement = pos - m->last_caller_pos; /* Can be < 0. */
+
+       if (pos_movement == 0) {

            pos = m->last_internal_pos;
            pos = m->last_internal_pos;

+       } else {
+           /* The current file stream position differs from the last one, so
+              caller must have called ftell/o().  Adjust accordingly. */
+           if (m->readpos + pos_movement >= m->msg_buf  &&
+               m->readpos + pos_movement < m->end) {
+               /* We can shift readpos and remain within the bounds of
+                  msg_buf. */
+               m->readpos += pos_movement;
+               m->total_bytes_read += pos_movement;
+               pos = m->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 (m->msg_buf, 1, MSG_INPUT_SIZE, iob);
+               m->readpos = m->msg_buf  +  pos % MSG_INPUT_SIZE;
+               m->end = m->msg_buf + num_read;
+               m->total_bytes_read = pos;
+           }

        }
 
        fseeko (iob, pos, SEEK_SET);
        }
 
        fseeko (iob, pos, SEEK_SET);


@@ -303,10 +335,10 @@ leave_getfld (struct m_getfld_buffer *m, FILE *iob) {
     /* Save the internal file position that we use for the input buffer. */
     m->last_internal_pos = ftello (iob);
 
     /* Save the internal file position that we use for the input buffer. */
     m->last_internal_pos = ftello (iob);
 

-    /* Set file stream position so that callers can use ftell (). */
+    /* Set file stream position so that callers can use ftell(). */

     m->total_bytes_read += m->bytes_read;
     fseeko (iob, m->total_bytes_read, SEEK_SET);
     m->total_bytes_read += m->bytes_read;
     fseeko (iob, m->total_bytes_read, SEEK_SET);

-    m->last_file_pos = ftello (iob);
+    m->last_caller_pos = ftello (iob);

 }
 
 static size_t
 }
 
 static size_t


@@ -433,7 +465,7 @@ m_getfld (int state, unsigned char name[NAMESZ], unsigned char *buf,
                *cp++ = c;
            }
 
                *cp++ = c;
            }
 

-           /* Skip next character, which is either the space after
+           /* Check for next character, which is either the space after

               the ':' or the first folded whitespace. */
            {
                int next_char;
               the ':' or the first folded whitespace. */
            {
                int next_char;


@@ -538,13 +570,11 @@ m_getfld (int state, unsigned char name[NAMESZ], unsigned char *buf,
             * end of the message.
             */
            unsigned char *bp;
             * end of the message.
             */
            unsigned char *bp;

-           register int cnt;

 
            max = *bufsz-1;
            /* Back up and store the current position and update cnt. */
            bp = --m.readpos;
 
            max = *bufsz-1;
            /* Back up and store the current position and update cnt. */
            bp = --m.readpos;

-           cnt = m.end - m.readpos;
-           c = cnt < max ? cnt : max;
+           c = m.end - m.readpos < max  ?  m.end - m.readpos  :  max;

            if (msg_style != MS_DEFAULT && c > 1) {
                /*
                 * packed maildrop - only take up to the (possible)
            if (msg_style != MS_DEFAULT && c > 1) {
                /*
                 * packed maildrop - only take up to the (possible)

diff --git a/test/bad-input/test-header b/test/bad-input/test-header
index 2cef784eb8e09d9b0b3272f64a0f8cb1203a436d..0e210dbab32bf61f568b4fe0b42f0d79d60bc6b3 100755 (executable)
--- a/test/bad-input/test-header
+++ b/test/bad-input/test-header
@@ -14,16 +14,14 @@ fi
 
 setup_test
 
 
 setup_test
 

-thisdir="$srcdir/test/bad-input"

 expected="$MH_TEST_DIR/$$.expected"
 expected_err="$MH_TEST_DIR/$$.expected_err"
 actual="$MH_TEST_DIR/$$.actual"
 actual_err="$MH_TEST_DIR/$$.actual_err"
 
 # Write message with bogus header field (missing blank line, really).
 expected="$MH_TEST_DIR/$$.expected"
 expected_err="$MH_TEST_DIR/$$.expected_err"
 actual="$MH_TEST_DIR/$$.actual"
 actual_err="$MH_TEST_DIR/$$.actual_err"
 
 # Write message with bogus header field (missing blank line, really).

-msgfile=`mhpath new`
-msgnum=`basename $msgfile`
-cat > "$msgfile" <<EOF
+msgfile="$MH_TEST_DIR/Mail/inbox/11"
+cat >"$msgfile" <<EOF

 Date: Sun, 18 Dec 2005 00:52:39 +0100
 From: foo@example.edu
 To: bar@example.edu
 Date: Sun, 18 Dec 2005 00:52:39 +0100
 From: foo@example.edu
 To: bar@example.edu


@@ -34,14 +32,14 @@ I am a stupid spammer.
 EOF
 
 # check scan
 EOF
 
 # check scan

-cat > "$expected" <<EOF
+cat >"$expected" <<EOF

   11  12/18 foo@example.edu    test<<This is a multi-part message in MIME forma
 EOF
   11  12/18 foo@example.edu    test<<This is a multi-part message in MIME forma
 EOF

-scan -width 80 $msgnum > "$actual" 2>&1
+scan -width 80 last > "$actual" 2>&1

 check "$expected" "$actual"
 
 # check show (mhl)
 check "$expected" "$actual"
 
 # check show (mhl)

-cat > "$expected" <<EOF
+cat >"$expected" <<EOF

 (Message inbox:11)
 
 Date:    Sun, 18 Dec 2005 00:52:39 +0100
 (Message inbox:11)
 
 Date:    Sun, 18 Dec 2005 00:52:39 +0100


@@ -54,11 +52,11 @@ This is a multi-part message in MIME format.
 
 I am a stupid spammer.
 EOF
 
 I am a stupid spammer.
 EOF

-show $msgnum > "$actual" 2>&1
+show last > "$actual" 2>&1

 check "$expected" "$actual"
 
 # check mhshow
 check "$expected" "$actual"
 
 # check mhshow

-cat > "$expected" <<EOF
+cat >"$expected" <<EOF

 Date:    Sun, 18 Dec 2005 00:52:39 +0100
 To:      bar@example.edu
 From:    foo@example.edu
 Date:    Sun, 18 Dec 2005 00:52:39 +0100
 To:      bar@example.edu
 From:    foo@example.edu


@@ -71,41 +69,63 @@ This is a multi-part message in MIME format.
 
 I am a stupid spammer.
 EOF
 
 I am a stupid spammer.
 EOF

-mhshow -nopause $msgnum > "$actual" 2>&1
+mhshow -nopause last > "$actual" 2>&1

 check "$expected" "$actual"
 
 # check m_getfld() handling of empty header field
 check "$expected" "$actual"
 
 # check m_getfld() handling of empty header field

-printf 'Date: Sat, 12 Jan 2013 09:07:01 -0600\nReceived:' >"$MH_TEST_DIR/mess"
+msgfile="$MH_TEST_DIR/Mail/inbox/12"
+printf 'Date: Sat, 12 Jan 2013 09:07:01 -0600\nReceived:' >"$msgfile"

 cat >"$expected" <<EOF
 cat >"$expected" <<EOF

-   1  01/12 
+  12  01/12 

 EOF
 cat >"$expected_err" <<EOF
 scan: eof encountered in field "Received"
 EOF
 cat >"$expected_err" <<EOF
 scan: eof encountered in field "Received"

-??Format error (message -1) in component 2
+??Format error (message 12) in component 2

 EOF
 
 EOF
 

-scan -width 13 -file "$MH_TEST_DIR/mess" >"$actual" 2>"$actual_err"
+scan -width 13 last >"$actual" 2>"$actual_err"

 check "$expected" "$actual"
 check "$expected_err" "$actual_err"
 check "$expected" "$actual"
 check "$expected_err" "$actual_err"

-rm -f "$MH_TEST_DIR/mess"

 
 # check m_getfld() handling of excessively long header field name
 
 # check m_getfld() handling of excessively long header field name

-cat >"$MH_TEST_DIR/mess" <<EOF
+msgfile="$MH_TEST_DIR/Mail/inbox/13"
+cat >"$msgfile" <<EOF

 Date: Tue, 15 Jan 2013 21:13:12 -0600
 ThisHeaderFieldNameIsWaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaayTooLong: OK
 EOF
 cat >"$expected" <<EOF
 Date: Tue, 15 Jan 2013 21:13:12 -0600
 ThisHeaderFieldNameIsWaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaayTooLong: OK
 EOF
 cat >"$expected" <<EOF

-   1  01/15 
+  13  01/15 

 EOF
 cat >"$expected_err" <<EOF
 scan: field name "ThisHeaderFieldNameIsWaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaayTooLong" exceeds 997 bytes
 EOF
 cat >"$expected_err" <<EOF
 scan: field name "ThisHeaderFieldNameIsWaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaayTooLong" exceeds 997 bytes

-??Format error (message -1) in component 2
+??Format error (message 13) in component 2
+EOF
+
+scan -width 13 last >"$actual" 2>"$actual_err"
+check "$expected" "$actual"
+check "$expected_err" "$actual_err"
+
+# check m_getfld() handling of long header field name without a colon
+msgfile="$MH_TEST_DIR/Mail/inbox/14"
+cat >"$msgfile" <<EOF
+Date: Thu, 17 Jan 2013 19:33:46 -0600
+If a header field name has at least 512 characters without a newline or colon, it will raise a format error in m_getfld().  Here is a test of that.  01234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901
+
+Test
+
+
+EOF
+cat >"$expected" <<EOF
+  14  01/17 
+EOF
+cat >"$expected_err" <<EOF
+scan: eol encountered in field "If a header field name has at least 512 characters without a newline or colon, it will raise a format error in m_getfld().  Here is a test of that.  01234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901"
+??Format error (message 14) in component 2

 EOF
 
 EOF
 

-scan -width 13 -file "$MH_TEST_DIR/mess" >"$actual" 2>"$actual_err"
+scan -width 13 last >"$actual" 2>"$actual_err"

 check "$expected" "$actual"
 check "$expected_err" "$actual_err"
 check "$expected" "$actual"
 check "$expected_err" "$actual_err"

-rm -f "$MH_TEST_DIR/mess"

 
 
 exit $failed
 
 
 exit $failed