Bash test for a dot terminatior in pop3 server LIST output

Howard_L · #1 28-Sep-2009, 20:50

Yes , an 0x2e dot.
I've been writing a script to talk to my pop email sever.
Here is an actual short manual session of what i'm trying to script so far:

Code:

  ~>  telnet mail.blahblah.org 110
Trying 192.94.73.19...
Connected to mail.blahblah.org (192.74.51.10).
Escape character is '^]'.
+OK Teapop [0.3.8] - Teaspoon stirs around again <1254161111.670F3C71@Llywellyn>
USER <my_email_username>
+OK Welcome, do you have any type of ID?
PASS <my_email_password>
+OK I'm ready to serve you, Master.
STAT
+OK 11 30687
LIST
+OK These are my measures.
1 686
2 657
3 1481
4 619
5 2410
6 5951
7 1932
8 6117
9 2227
10 4102
11 4505
.
QUIT
+OK It has been a pleasure serving you.
Connection closed by foreign host.
~>

I used this reference to learn those commands: :faqs.org/rfcs/rfc1939.html
I manually enter all the commands - USER , PASS etc.

Ok , so you can see there that after the LIST command is issued I
get a list of two-"word" strings each on a separate line terminated
by a single dot on a line.

In my script I use a while loop to 'read' each line that is output by LIST.
The condition of the while loop is that damn dot.

I can catch a dot I 'read' from my keyboard
but when I try it on the output from the server it's not caught.
I don't know if it's unicode or what but I'm stumped.

Here is a demo of catching the dot with me doing the input:

CPP / C++ / C Code:

#!/bin/sh
 
echo -e "----------\n\
Simulate pop server LIST output by entering <Msg# Size> strings.\n\
Terminate by typing a single . . eg:\n1 345\n2 654\n.\n";
 
while [ "$REPLY" != . ]; do     # a string containing only  dot (0x2e) 
 
  # read automatically assigns input to $REPLY unless another's specified
  read ;     # read tmp  will get the string into $tmp  
 
  msg=( $REPLY );  # this parses the string into "word" elements
 
  echo  "${REPLY[*]} , ${msg[0]} , ${msg[1]} , ${msg[2]}";
 
done;
 
echo -e "----------\n\
And if you could do a couple more, please:\n.\n";
 
msg="";
while [ "${msg[0]}" != . ] ;  do   # a word containing only one dot (0x2e)
 
  read   -a msg         # the -a automatically parses input into "word" elements
 
  echo  "${msg[*]} , ${msg[0]} , ${msg[1]} , ${msg[2]}";
 
  if [ "${msg[0]}" == . ]; then
    echo "Got dot";
  fi
 
done
 
echo "Loop done ...goodbye.";

And here is the script for the server where I'm not catching the dot.

When I run it the server list output stops , the dot is not detected
so the loop continues and I end up hung on a read and must use
ctrl-c to stop the process.

CPP / C++ / C Code:

 #!/bin/sh
server="mail.blahblah.org";   user="MYUSERNAME"; pass="MYPASSWORD";
 
exec  3<>/dev/tcp/$server/pop3 || exit 1;  # assign a tcp to port 110 connection
                                           # to fd 3
read  -u 3 ;               # read out the greeting (msg gets last line)
echo  "$REPLY";            # print the message to the screen
 
echo  "USER $user" 1>&3;   # send username to the server 
read  -u 3 ;               # read out the response into msg
echo  "$REPLY";            # print message to screen
 
echo  "PASS $pass" 1>&3;   # send password
read  -u 3 ;               # read out the response into msg
echo  "$REPLY";            # print message to screen
 
echo  "STAT";
echo  "STAT" 1>&3;         # stat show number of messages and total size
read  -u 3 ;
echo  "$REPLY";
 
echo  "LIST";
echo  "LIST" 1>&3;         # LIST shows each message and it's size 
read  -u 3  ;              # get and print the OK line
echo  "$REPLY";
 
 
# loop through the LISTings
 
while [ "$REPLY" != . ]; do
  
  read  -u 3  ;
 
  if [ "$REPLY" == . ]; then
    echo "Got dot";
  fi
  
  echo  "$REPLY";
 
done;
 
echo "LOOP DONE";
 
echo "quiting"
echo   "QUIT" 1>&3
read   -u 3
echo   $REPLY

I got some of the ideas use in the above from: linuxgazette.net/110/park1.html
Obviously I am not too experienced with bash...
HELP! , thanks, Howard();

davekw7x · #2 29-Sep-2009, 01:32

Quote:

Originally Posted by Howard_L

CPP / C++ / C Code:

.
.
# loop through the LISTings
 
while [ "$REPLY" != . ]; do
  
  read -u 3  ;
 
  if [ "$REPLY" == . ]; then
    echo "Got dot";
  fi
  
  echo  "$REPLY";
 
done;

What do you see if you change the loop output messages to be a little more informative:

Code:

.
.
.
# loop through the LISTings
 
while [ "$REPLY" != "." ] ; do
  read -u 3
  echo -n "REPLY = --->$REPLY<---"
  echo "Stuff after Reply"
 
  if [ "$REPLY" == "." ] ; then
    echo "Got dot"
  fi

done
 
echo "LOOP DONE"

Looking at your logic, if the last message is simply a dot, I would expect to see something like

Code:

REPLY = --->1 686<---Stuff after Reply
REPLY = --->2 657<---Stuff after Reply
REPLY = --->3 1481<---Stuff after Reply
REPLY = --->4 619<---Stuff after Reply
REPLY = --->5 2410<---Stuff after Reply
REPLY = --->6 5951<---Stuff after Reply
REPLY = --->7 1932<---Stuff after Reply
REPLY = --->8 6117<---Stuff after Reply
REPLY = --->9 2227<---Stuff after Reply
REPLY = --->10 4102<---Stuff after Reply
REPLY = --->11 4505<---Stuff after Reply
REPLY = --->.<---Stuff after Reply
Got dot
LOOP DONE

Since your script never showed "Got dot", maybe there is some extra character on the line after the dot (A '\r' comes to mind; maybe some other whitespace.) By surrounding the response with ---><--- you can (maybe) see if there is some extra little "gift." So, for example, if the server is sending "\r\n" as line termination characters you might see something like

Code:

<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply
<---Stuff after Reply

If that's the case, then try something like the following for your loop:

Code:

 #!/bin/bash
.
.
.
# loop through the LISTings
 
FIRST=${REPLY:0:1}
while [ "${FIRST}" != "." ] ; do
  read -u 3
  FIRST=${REPLY:0:1}
  echo ${REPLY}

  if [ "${FIRST}" = "." ] ; then
    echo "Got dot"
  fi
  
done
 
echo "LOOP DONE"

(This should work whether there is extra stuff after the dot or not.)

Regards,

Dave

Howard_L · #3 29-Sep-2009, 09:47

CPP / C++ / C Code:

${REPLY:0:1}

Wow now that's a different dereferencer! So you CAN get down to the chars?
No offense to Mr. Bourne et al but all these bash nuances are mind boggling...

You were right! There is a \r\n line ending.
The read command strips the \n.
I added a line to write each string to a file:

CPP / C++ / C Code:

echo -n "$REPLY" >> xfile; # -n supresses echo's default of adding a newline

...and check the file with hexdump:

Code:

...
LIST
  stat=+OK  :These are my measures.
1 686
2 657
3 1481
4 619
5 1932
6 2884
~>  hexdump -C xfile
00000000  31 20 36 38 36 0d 32 20  36 35 37 0d 33 20 31 34  |1 686.2 657.3 14|
00000010  38 31 0d 34 20 36 31 39  0d 35 20 31 39 33 32 0d  |81.4 619.5 1932.|
00000020  36 20 32 38 38 34 0d 2e  0d                       |6 2884...|
00000029

...you can see the 0x0d line ends.
hmm, the fact that there is no 0x0a in each $REPLY string may be why I can't get it to test as expected.

Code:

while [ "$REPLY" != . ]; do        # is . 2e 0a       ???
while [ "$REPLY" != "." ]; do      # is . 2e 0a       ???
while [ "$REPLY" != .\r  ]; do     # is .\r 2e 0d 0a  ???
while [ "$REPLY" != ".\r"  ]; do   # is .\r 2e 0d 0a  ???
while [ "$REPLY" != "\.\r"  ]; do  # who knows what that might evaluate to...

...anyhow none of them are 0e 0d which is what the last  $REPLY is in hexdump.
(btw, so where are the bash code tags Luciwiz?)

hmmm so last night I had resorted to getting the number of messages
and using a for loop. That turned out to be a "beginning math evaluation" learning session.
I think the for statement could be better but... it works for now.
I have now added a test for your ${REPLY:0:1} and it tests true at the dot.
So here is program as it stands now:

Code:

#!/bin/sh
server="mail.blahblah.org";   user="MYUSERNAME"; pass="MYPASSWORD";

exec  3<>/dev/tcp/$server/pop3 || exit 1;  # assign a tcp to port 110 connection
                                           # to fd 3
read  -u 3 ;               # read out the greeting (msg gets last line)
echo  "$REPLY";            # print the message to the screen

echo  "USER $user" 1>&3;   # send username to the server 
read  -u 3 ;               # read out the response into msg
echo  "$REPLY";            # print message to screen

echo  "PASS $pass" 1>&3;   # send password
read  -u 3 ;               # read out the response into msg
echo  "$REPLY";            # print message to screen

echo  "STAT";
echo  "STAT" 1>&3;         # stat show number of messages and total size
read  -u 3 stat num size ; # get the "words" into separate variables (nifty)
echo  "  stat=$stat num=$num  size=$size ";

 echo  "LIST";
echo  "LIST" 1>&3;         # LIST shows each message and it's size 
read  -u 3  stat trash ;              # get and print the OK line
echo  "  stat=$stat  :$trash";
if [ "$stat" != +OK ]; then
  echo "LIST failed";
else
  # loop through the LISTings
  # while [ "$REPLY" != . ]; do
  # while [ "$REPLY" != "." ]; do
  # while [ "$REPLY" != .\r  ]; do
  # while [ "$REPLY" != ".\r"  ]; do
  # while [ "$REPLY" != "\.\r"  ]; do

  i=0;
  echo -n "" | cat > xfile;     # start a fresh file
                                # -n supresses echo's \n line ending
  while [ $i -le $num ];  do

    read  -u 3  ;

    echo  -n "$REPLY" >> xfile;      # copy LIST output to file for hexdump

    FIRST=${REPLY:0:1}
    if [ "${FIRST}" = "." ] ; then
      echo "FIRST Got dot"
    fi

    if [ "$REPLY" == "\.\r" ]; then  # well, I still can't get this
      echo "Got dot";
    fi

    echo  "$REPLY";
    ((i++)) ;                  # increment,  (()) = "do math operation"
  done;
fi;
echo "LOOP DONE";

echo "quiting"
echo   "QUIT" 1>&3
read   -u 3
echo   $REPLY

...and here is the output

Code:

~>  ./popscript-3
+OK Teapop [0.3.8] - Teaspoon stirs around again <1254233090.1AFA9D1D@Llywellyn>
+OK Welcome, do you have any type of ID?
+OK I'm ready to serve you, Master.
STAT
  stat=+OK num=7  size=12823
LIST
  stat=+OK  :These are my measures.
1 686
2 657
3 1481
4 619
5 1932
6 2884
7 4564
FIRST Got dot
.
LOOP DONE
quiting
+OK It has been a pleasure serving you.
~>

I gotta go drive nails now but I am anxious to play with that new
indexing (I hate to say the word anymore) method later tonight...
Thanks for the help Dave. back later.
Howard();

davekw7x · #4 29-Sep-2009, 10:40

Quote:

Originally Posted by Howard_L

CPP / C++ / C Code:

${REPLY:0:1}

Wow now that's a different dereferencer! So you CAN get down to the chars?
No offense to Mr. Bourne et al but all these bash nuances are mind boggling...

You can't do that with Mr Bourne's original shell (traditionally named sh). That is one of the many useful features that was added to bash. (That's born-again sh---or is it bourne-again sh?)

Note that for some Linux distributions (RedHat/Centos, for example), /bin/sh is a symbolic link to /bin/bash, so a shell script with #!/bin/sh as its first line is actually runs in a bash shell.

With other distributions (Ubuntu, as I recall), /bin/sh is a shell that is very close to the original Bourne shell and doesn't have the Bash additions, so the first line of a script that uses bash-isms should be #!/bin/bash, not #!/bin/sh.

Since many (most?) Linux distributions have bash as the default shell, you may not need anything in particular in the first line of a script unless you have changed the default to some other shell. (Some people like the Korn shell, ksh, for example, and ksh doesn't have this notation to address characters within a string.)

A final note about variables in shell scripts: Old-timers who learned sh (maybe some not-so-old timers who learned from tutorials written by or copied from old-timers who learned sh) tend to use $xyz everywhere to get the value of a shell variable. This notation is still valid, but you can always use ${xyz} instead. The reverse is not true: There are a number of things that can be done with ${xyz} that do not compute if you try to use $xyz Some tutorials go into this in some detail, including examples where the newer notation works and the older wouldn't be possible (or in some cases could be used but would be more awkward.) Accessing substrings with the ${xyz:} notation is one example for which no such operation defined using the older notation.

Regards,

Dave

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Rate This Thread
Rate This Thread:

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Str_Misaligned in Double Link List	Peter_APIIT	C Programming Language	1	29-Feb-2008 21:50
[Include] Doubly-linked List	dsmith	C Programming Language	6	14-Apr-2006 14:12
fltk-2.0 cvs	Plumb	FLTK Forum	20	13-Nov-2004 08:10
Can't view pages from another machine on the Intranet	aevans	Apache Web Server Forum	9	14-May-2004 03:26

Bash test for a dot terminatior in pop3 server LIST output

Re: bash test for a dot terminatior in pop3 server LIST output

Re: Bash test for a dot terminatior in pop3 server LIST output

Re: Bash test for a dot terminatior in pop3 server LIST output