QTERM chat scripts
==================

A chat script is a means for getting QTERM to automatically send and
receive text, this can be used to auto-dial, connect to remote systems,
log in to them, and do whatever else is wanted. In addition, chat scripts
have a number of commands available, to do such things as protocol sends
and receives, transfer of text files, and many other things.

There are two ways of invoking a chat script. Firstly when QTERM is
executed from CP/M, a chat script and parameters can be provided there:

	A>QTERM SCRIPT 1200

would be an example, alternatively the ^\ X command will prompt for a
filename, the response can be exactly the same:

	Filename: SCRIPT 1200

will have the same effect.

QTERM will look in several places to try to find the script. The first
thing it will do is to take the filename as given (subject to the
current default drive/user, which may have been changed by the !n or
^\ N commands). If this is not successful, QTERM then searches the
drive/user area that was active when it first started. It should be
noted that if the entry subroutine includes BDOS calls to change either
the drive or user, then the values rememberd by QTERM will be those on
return from the entry subroutine. This provides a mechanism for setting
up a default script area, a place where QTERM will always try to find
scripts. In addition, if it can't find the script as a file in the
default script area, QTERM will look for a .LBR file /QTERM.LBR and
see if this library contains the script. The reason behind this is that
scripts tend be fairly small, and it is far more efficient to keep them
all together in one .LBR, since this saves disk space. It goes without
saying that QTERM cannot deal with squeezed or crunched scripts, they
must be saved in the .LBR as uncompressed ASCII text files.

When a chat script is running it can be terminated prematurely by typing
^X on the keyboard: this will return to normal terminal mode.

There are two types of lines in a chat script: send/expect lines, and
command lines. Command lines are always started with a '!' character,
any other character starts a send/expect line.

Looking first at send/expect lines, they can contain up to six fields,
and the first two must be provided, even if they are empty. An example
of such a line is:

	-AT\r-OK\r\n-3-2-3-0-

In this example the '-' (first character) is the delimiter used to
separate fields. Any character can be used except for '!', but whatever
character is chosen cannot appear in the strings. Also note that
chosing a delimiter from the characters above 'z' in the ASCII character
set (i.e. '{', '|', '}', and '~') has a special effect, which is explained
below. Taking the fields in order they are SEND, EXPECT, TIME, TRIES,
SUCCESS, and FAIL. SEND is a string that is transmitted by QTERM, so in
the example above QTERM would transmit 'AT<carriage return>'. As was noted
above, delimiters above 'z' have a special effect: they cause the SEND
string to be written out slowly: there is a tenth of a second delay
after each character. EXPECT is a string that QTERM is looking for in
response to it's SEND string: so in the above example, QTERM would be
looking for the 'OK<carriage return><linefeed>' that a Hayes compatible
modem would respond with, when presented with 'AT<return>'.

The remining four fields are all decimal numbers, and can be omitted
as QTERM will provide default values. TIME is the number of seconds
to wait before assuming failure, if not given it defaults to 15. TRIES
is the number of times to retry on failure, so taking our first example,
TRIES is 2. If QTERM matched the EXPECT string on the first sending of
SEND, all is well, but on the first failure it would resend the SEND string
and look for the EXPECT string a second time. If it failed on this second
attempt, only then would it consider this line to have failed. SUCCESS
specifies the line number to transfer to in the chat script if it matched
the EXPECT string. The default for this is the line following the current
line. FAIL is the line to transfer to if the EXPECT string is not
matched. This can be a line in the chat script, or as shown above 0 is
allowed, which terminates the script immediately.

In the example above, the success and fail values are given as simple
line numbers, it is also possible to use labels in chat scripts, see !:
below for an explanation of how to define a label. If a label is being
used, the line might look like this:

	-ATDT5551234\r-CONNECT-30--`connect-`fail-

In this case, the `connect and `fail are label usages, and cause transfer
to wherever the corresponding label is. Label useage is introduced with the
backquote character `, which must be followed by the label itself, with no
intervening white space: -` connect- will not work. Using an undefined label
does not directly cause an error, but the substitution of the non-existant
label will usually create a line that cannot be parsed, thus flagging the
error.

In another example, if the first line were:

	-AT\r-OK\r\n--5-

since TIME is empty, it defaults to 15, but as TRIES is 5, this line
would try five times before giving up. Note also from this example
that there are two ways of causing QTERM to default a value: an empty
field (TIME) or end of the string (SUCCESS and FAIL). Note that the
closing '-' after the 5 for TRIES is necessary. On the basis of this,
the absulute minimum line is:

	-send-expect-

This uses all four defaults: 15 seconds timeout, 1 try, success goes to
the next line, failure terminates the script. The idea behind these
defaults is that a collection of simple send/expect lines like the above
allow a "conversation" to be held with the remote system.

It is possible that either of SEND or EXPECT can be empty: an empty SEND
causes nothing to be sent, but the EXPECT must be matched to continue;
an empty EXPECT automatically matches. Note that if both are empty then
the chat script will terminate when it reaches that line, so a line like:

	---

will serve as a means to terminate a chat script, returning to terminal
mode.

Command lines in chat scripts start with '!', and following the '!' is
a command letter. If input is needed (e.g. for a '!b' or '!s' line)
it should be placed after the command letter:

	!b 1200 8n1 -5 30 500 +\x13\x11

As is shown in the above example, spaces are permitted after the command
letter, but not before.

Several of the ! commands correspond to ^\ commands available from terminal
mode: the !b above would set the baud rate etc., just like the corresponding
^\ B command would.

Commands available in this group are:

!,	hangup
!.	break
!B	set baud rate
!E	set local echo
!H	set half duplex
!J	toggle junking of control characters
!L	set linefeed send for 'P'
!M	set bit 7 mask
!N	select new drive/user
!O	toggle output to the printer
!V	toggle VT100 emulation
!W	toggle split window mode
!K	program function key
!P	print file to remote
!U	invoke user function
!R	protocol receive
!S	protocol send
!C	open catch file
!Y	put catch file on hold
!Z	close catch file
!X	activate chat script
!Q	exit QTERM

Note also that the toggles ('!E', '!H', '!J', '!L', '!M', '!O', '!V' and '!W')
behave a little differently. Since the state of these toggles is not defined
when a chat script starts, with one exception (!W) there are three ways of
invoking these. Using '!H' as an example:

	!h

behaves as would an <escape> 'H' in normal operation, i.e. it toggles
the half duplex switch. However, if the following is given:

	!h 1

the trailing '1' forces half duplex to be enabled, irrespective of
it's original state, and:

	!h 0

guarantees to turn half duplex off. The other toggles work in the same
manner: a trailing '0' always disables, and a trailing '1' always
enables. !W is a little different, in that there are three possibilities:
window mode off, window mode on with big receive, and on with small. Also
allowing a pure toggle could have undefined results, since if window mode
were toggled on, there would be no indication what size was wanted. As
a result of this, there are three forms for the !W command in a script:

	!w 0

forces window mode off,

	!w b

forces it on with a big receive window, and:

	!w s

forces it on, but with a small window. With all these toggles (!h etc. and
!w) the options above are guaranteed, using any other option letters will
have undefined results.

The 'X' command to activate a chat script can be used to chain scripts
together: when an 'X' is encountered the specified chat script is
invoked, however the current script is lost: it is overwritten by the
new one.

There are other commands that are not normally available are as follows:


!: - Define a label. Label usage was described above: the '`' character
introduces a label usage. NOTE that this is the BACKQUOTE character, not
the usual single quote character. To define a label, simply include a
line of the form:

!: connect

in the script. A few comments may make labels easier to use, firstly
they cannot be longer than seven characters, and where they are defined
there should be no trailing blanks. When a label is used, it is done by
means of a simple text substitution: after seeing a '`' character, QTERM
tries to match the following text with a label in the script, and it
stops at the first match. So if you have two labels one of which is a
prefix of the other, the results can be unpredictable. As a byproduct
of this, undefined labels do not generate an error (they just become
line zero), but the text substitution doesn't remove the label, so the
resulting line usually generates an error. In the event that a '`'
character is needed as part of a send or expect string, it can be
escaped by preceeding it with a $, so the line:

	.send.exp$`ect.

will look for exp`ect, whereas:

	.send.exp`ect.

will not work, it would try to look for and substitute the label 'ect'.
Since '$' is used to escape '`', it must also be used to escape itself,
so to match the string XXX$YYY, the script line would need to be:

	.send.XXX$$YYY.

Labels should always be used to transfer control in a script: when a
script is being read in, comments and blank lines are stripped very
early in the process. This means that a given line in the script file
may have a different line number as far as QTERM is concerned. Labels
bypass this problem, so their usage is highly recommended.


The commands !@ and !# can be used for variable manipulation. Their
main purpose is to prevent infinite loops in chat scripts. In the
following example:

	!: reset
	.AT\r.OK\r\n.5.5.
	.ATDT5551212\r.CONNECT.30..`connect.`reset.
	!: connect
	. ........

if the system being called is off line and not answering, QTERM will
loop here for ever. The !@ and !# provide the ability to keep count and
terminate the loop after some specified number of tries.

	!@ var term +/- term

is the form of an @ line. var is a single letter variable (there are 26
available: a through z), and term is either a number or a variable. This
is very simplistic, in that two terms must be present: to set a variable
simply say something like:

	!@ a 5 + 0

the operator can be either + or - and they act as you would expect. so:

	!@ a a - 1

will subtract 1 from a, or:

	!@ a a + b

will add b to a, etc. etc. Note that variables are recognised in either
upper or lower case:

	!@ A a + B

would have exactly the same effect as the line above. Note that these are
single bytes, so there is some risk of working with values above 255.

!# tests variables: the general syntax is:

	!# var operator term line

where var is a variable letter, term is a variable or a number, and the
operator can be '=' to test for equality, '#' to test for inequality, '<'
to check for less than and '>' to test for greater than. line is simply
the line number in the script to go to is the test succedes. Note that this
also provides a goto capability:

	!# a = a `doit

will always go to doit, since a is always equal to itself.

All variables are initialized to zero when the first script in a series is
invoked, but values are retained when a !x command chains from one script
to another.


In addition to !@ and !# which set and test numeric variables, there are two
corresponding commands which set and test string variables. !$ and !% will
set and test strings. A line like:

	!$ a This is a string

will put 'This is a string' into string variable a. Note that when a script
is started, or chained to with a !x command, the first nine parameters
become strings $1 through $9. These can be assigned to only if they are not
set when the script is invoked. The reason behind this is that if the
command line parameters are set with default values at the top of the script,
then missing parameters can be dealt with. An example might be a script
where the first parameter provides the baud rate, if it's invoked as:

	SCRIPT 1200

then $1 will contain '1200', however if the following line appears in the
script:

	!$ 1 2400

this will not disturb the 1200 in the case above, however if the script
is invoked with no parameters, then $1 will be empty initially, and the
assignment will put 2400 into it. String variables are used simply by
naming them, so to use $1 in this case, a line like:

	!b $1 8n1

would set the baud rate as needed. In a similar manner, numeric variables
can be used simply by naming them:

	-Send @a\r-expect-

would have the @a expanded to whatever the current value in variable a is.

As is done with labels, $a string use and @a variable use is handled by
simple text substitution, so a little care will make their use easier.
As was described above, '$' is used to escape '`' characters in strings,
it is also used to escape itself, and it can also be used to escape an
'@' character in a string. So, to actually place a '$' character in a
string say '$$', so that:

	.send.exp$$ect.

would look for the string 'exp$ect'. There are only three things that
can follow a '$' sign: either '$' or '`' for escaping purposes,
and '1' through '9' and 'a' through 'z' for string substitution.
Placing any other character after a '$' will have an undefined result.

To test a string variable the !% line can be used: there are three forms
this line takes:

	!% = .string1.string2. `label
	!% _ .string1.string2. `label
	!% # .string1.string2. `label

The first one will jump to label if the two strings are equal, so to test
a string, a line like:

	!% = .$a.exit. `doexit

would jump to label doexit if string variable a contained 'exit'. The second
case is identical, except that the test is done ignoring case, so if _ were
used instead of = in the example above, 'exit', 'Exit', 'EXIT', and 'eXiT'
would all test as equal. The last case jumps to the label if the two strings
are different.


Two commands exist to manipulate the appearance of chat scripts:

	!> This is a line of text\r\n

!> simply prints the text, after processing '\' escapes. Note that
leading and trailing spaces are ignored, so the above case would start
with the 'T' of 'This'. In order to start or end with a with a space,
\x20 can be used.

	!&

This command is actually three different commands rolled into one:
!& o manipulates the echoing of characters received from the modem
while the script is running:

	!& o 1

forces modem echo on,

	!& o 0

forces it off, and:

	!& o

simply switches state. In the same manner, !& m controls printout of
the 'Match: OK' messages that are printed when QTERM matches the expect
string in a send/expect line, and the 'Fail' and 'Retry' messages as
well. These two are initially on when a script starts, the third case
is !% l which controls printout of 'Looking for: ' messages. When this
option is turned on, these messages are printed when QTERM starts
looking for the expect string.


As a complement to !>, the !< command can be used to take keyboard input,
and make decisions based on what happens. This includes two subcommands
altogether:

	!< - variable

The '-' causes QTERM to prompt for a line of input using CP/M's BDOS
buffered command. The line is then in the named variable, and can later
be tested with !% lines.

This allows for such things as multiple choice:

	!> \r\nSelect system to call\r\n
	!> 1. System 1 ..... \r\n
	!> 2. System 2 ..... \r\n
	!> 3. System 3 ..... \r\n
	!: prompt
	!> Enter 1, 2 or 3:\x20
	!< - a
	!% = .$a.1. `sys1
	!% = .$a.2. `sys2
	!% = .$s.3. `sys3
	!> Error, invalid input\r\n
	!# a = a `prompt

Where the first 4 lines print a menu, the next line defines a label. Then
comes a prompt, followed by an input command. After this, the line is
checked against 1, 2 and 3, and a jump is made to the appropriate label.
If there is no match an error message is printed, and the !# a = a line
is used as a goto, since a is always equal to a.

In a similar manner, '!< . a' will provide a "hot key" input - this reads
a single keystroke from the keyboard, and saves the numeric value of the
key pressed in the named variable, where it can be tested with a !# line.


The ![ command provides a similar function to the !< command, but it
works on text coming from the modem. There are several different
sub-commands available:

	![ -

reads text from the modem. In this instance there are two ways that
reading can end:

	![ - 15

would simply read text for 15 seconds. This numeric timeout must be
provided, but in addition up to four extra strings can be provided:

	![ - 5 .string1.string2.string3.string4.

in which case input will terminate when 5 seconds have elapsed, or one
of the four strings is read. In this line, the '.' following the 5 is a
delimiter, this serves to separate the strings exactly like the delimiter
in a send / expect line. Not all four strings need to be there:

	![ - 5 .OK.ERROR.

is acceptable, but the trailing delimiter must be there after the last
string. In addition to scanning for the four lines, QTERM keeps the last
sixty-four characters seen in a buffer, these can then be inspected with
![ =, ![ +, and ![ _ lines.

	![ : 20

watches the data arriving from the modem, and only stops when 20 seconds
of silence are detected. This can prove useful when !C has opened a file,
and you want to capture the incoming text until a period of silence.
In addition, the last 64 characters are saved, just like with ![ - and
they can be tested with the same three commands.

The three commands that test are:

	![ = string `label
	![ + string `label
	![ _ string `label

These are all very similar in that they all jump to label if the string is
found in the "last 64 character buffer", however the exact nature of the
test varies. Also note that during the search for the expect string in a
normal send/expect line, the last 64 characters seen are also saved in
this buffer, and can be tested using these commands.

	![ = string `label

searches the saved text for the string, which in this case can include
\ escapes. If string occured anwhere in the last 64 characters, control
goes to label. ![ + looks exactly the same, but the difference is whether
the test is done on seven or eight bit data: ![ = just compares the least
significant seven bits, ignoring the parity bit, whereas ![ + compares
all eight. Note also that the strings provided in the ![ - line are only
checked in seven bit mode. ![ _ also does a seven bit test, but in addition
it ignores case.

Since spaces are used as the delimiter on a ![ = line, it is necessary to
escape a space: to match the 'CONNECT 1200' string that might be received
from a modem, it would be necessary to do something like this:

	![ = CONNECT\x201200 `con1200

where the \x20 is an encoded space. Consult QTERM.DOC for a discourse on
\ escapes in strings.

Finally:

	![ ^ a XY

allows a string to be extracted from the 64 character buffer and copied
into a variable. a is the variable name in this case (only letter variables
are allowed), and XY are two characters that delimit the string. If they
are left off, they both default to zero (NUL character, not '0' which is
ascii 0x30), which matches any white space. X and Y can be \ escapes, and
note that \0 (i.e. a NUL) has the special effect mentioned. The exact
algorithm used to determine the text to be saved is:

1. Scan backwards from the end of the buffer (last characters received), till
a 'Y' is seen, then skip over any adjacent duplicates. Then scan back to the
next 'X', and take all characters between, not including the X or Y. So if
the tail of the buffer holds:

	..... YYY XXThis is a stringYY hello X world X

the string that would be extracted is 'This is a string'. As always, a little
care and attention will help in chosing the correct way of getting a buffer
of 64 characters, and then some care in chosing the delimiters will ensure
the correct string is extracted into the variable.


The !~ command has been added for rudimentary file manipulation
during chat script operation. Four options exist:

	!~ - file

will erase a file. Note that QTERM will silently ignore this command
if the file doesn't exist, or if it is write protected. As with all
filenames, a drive/user can be given:

	!~ - d15:foo.bar

does like you'd expect.

	!~ = newname = oldname

renames a file: note that if a drive/user is given on oldname, it will
be ignored: newname completely defines where the action will happen.
This will fail silently if newname already exists, or if old name doesn't,
or if oldname does exist but is write protected.

	!~ + newname = oldname

copies a file. In this case a file can be copied to a different
drive / user, so if needed a drive / user spec should be attached to
oldname. This will fail silently if newname exists or if oldname doesn't.
These can be used to good effect when QTERM is sending text files as
messages to a BBS, after sending the file with a !P command, a !~ - will
erase it, or files can be erased after uploading, or a file might be
renamed after a batch download.

The last options don't change any files, but allow QTERM to conditionally
alter script execution depending on whether a file exists or not:

	!~ Y filename line
	!~ N filename line

This simply transfers control to line if filename exists. As always line can
be a label. If needed, filename can be a wildcard:

	!~ Y A17:*.TXT `sendtxt

will go to label sendtxt only if there are any files on A17: matching the
*.TXT wildcard.


Two commands exist to allow strings to be saved during a script run, for
possible reloading during a later script run. The commands are:

	!( r a 7

to read, and:

	!( w b 3

to write. The r or w specifies read or write, following this is a single
letter that tells which string variable to read or write, and finally a
file position to use. QTERM uses the file /QTERM.STR in the default chat
drive/user area as the file that holds thee saved strings, the last parameter
specifies which record will be used. '0' uses the first record, and up to a
certain point, the file can be made as large as is needed: ten records
would provide '0' through '9', however the file can be made larger to hold
records 'A' through 'Z', or even 'a' through 'z' if needed. To create a
record with a given index, simply use the !( w command to write to it,
so if a file containing just ten empty is desired, the following ten line
script would create it:

	!( a 0
	!( a 1
	!( a 2
	!( a 3
	!( a 4
	!( a 5
	!( a 6
	!( a 7
	!( a 8
	!( a 9

This would set all the saved strings to empty, since string a will be
initially empty when the script starts. Note that while QTERM can add
records to an already existing file, the file must exist in order for
QTERM to access it. Several means exist for creating it:

	A>SAVE 0 /QTERM.STR

under CP/M 2.2 will work, or using a text editor to create an empty
file, or even using the <ESCAPE> C command under QTERM to open a catch
file, and then immediately closing it with an <ESCAPE> Z command.

This function can be used to good effect when a script is used to
call a BBS, and you want to use a catch file to capture new messages.
If the BBS can't do this for you, then the ![ - and ![ : lines can
be used to catch the high message number in the 64 character buffer,
then the ![ ^ line can be used to pluck it out into a string, from
where it can be saved in the file. Then next time the script is
running, it can read the previous high message number back into a
string, and use it in a line line:

	.read new $a\r..

to send it back to the BBS as the first message number required.


As a final note, any other command character is silently ignored,
this can be put to use to introduce comments. At this stage, !; is
not in use, and this is the official comment entry, it is guaranteed
that !; will never be used for a command line function in a QTERM
chat script. Further, when a script is being read in by QTERM, blank
lines and lines starting with !; are discarded fairly early. This has
two side effects, it means that comments can be used freely without
using up any of the 4K space available for saving scripts, but it
also means that line numbers should not be used in scripts as the
targets of jumps (i.e. the last two fields of a send / expect line)
since the target line number may well change. Use labels instead.