(some text drawn from Brent Welch’s book, "Practical Programming in Tcl/Tk")

Aidan Low

Student Information Processing Board

MIT

January 1999

http://www.mit.edu/iap/tk

Tcl

Background

Variables – Set, Unset

Comments

Basic Input/Output

Procedures

Upvar

Arithmatic Operations

String Operations

List Operations

Control Flow

Global variables

Array Operations

More Input/Output

File Operations

System Operations

Tcl is an intepreted high-level programming language designed to be easily understandable and easily customizable. It was developed by John Ousterhout when he wanted to give his students an editor that they could alter and extend themselves. One of tcl’s nicest features is that it’s very easy to write and very easy to read. Tcl code is simple and straightforward, for the most part, and as such is easy to modify.

Tk is an associated graphical toolkit which provides the same power to users in the graphical domain. Now complex GUIs can be written in a few dozen lines of code, rather than the pages it requires in more complicated arenas.

By the end of the night, this (highly inefficient) zsig randomizer should make sense to all of you.

This course will cover Tcl and Tk at a basic level, and looking at features of the language one by one along with examples of the code at work. In the end, we’ll see that a basic word processing program can be written very simply using Tcl and Tk. We’ll close by briefly looking at a number of advanced topics which are beyond the scope of this class, but that you can learn about on your own if you like.

In the first class, we’ll concentrate on Tcl, and in the second, we’ll look at how Tk can extend Tcl to allow the creation of graphical interfaces very simply.

Text in Courier font

represents code exactly as typed into the interpreter.

Text in italics represents an abstract thing to be filled in. This text

means that the command takes a filename in that space. (i.e.

Text wrapped in ? ? means that it is optional.

Finally, … means that there are multiple things there.

This will bring up the tcl interpreter in the window you run it in.

This will bring up the Tcl interpreter in the window you run it in, but this tcl interpreter can run tk commands as well. A Tk window will also be created.

To execute a file of tcl or tk code, there are two ways to do it. One is to make the file executable:

and to put

or #!/mit/tcl/bin/wish

at the start of the file.

The other way is to run

in the interpreter once started.

I encourage all of you to work together as you’re learning Tcl and Tk, and a zephyr instance is a good way to do that. I’ll be subbed to this whenever I can, and you can ask your questions about Tcl and Tk on the instance where everyone can see them and benefit from the answers.

To subscribe for this login session only:

To unsubscribe for this login session only:

To subscribe forever:

To unsubscribe forever:

To send a message

To read the logs of the instance

more tcltk (or emacs tcltk, or whatever)

The core of the tcl language is variables. Unlike other languages, every variable in tcl is a string. The string "123" is the number one hundred twenty-three, but it is also the character string made of the characters ‘1’,’2’ and ‘3’. It depends on the context in which the variable appears.

Assigning a variable a value is done with the set command

set a 3 <- This sets the variable a to be the string "3"

To get the value of a variable, the set command is used with just the name of the variable.

set a <- This is equal to 3

One variable can be set to the value of another. A command can be run on its own line, but to nest commands, we use the [ ] braces.

set d [set a] <- This sets d to be equal to the value of a

Note that [set a] is the value of the variable, and once done, the data has nothing to do with the variable. If the variable a changes later, d will not change.

The $ symbol is a shorthand which gets the value of a variable.

A variable can be unset, removing it from the scope of the program, with the unset command.

set f $e <- this will generate an error, since e is no longer defined

To find out if a variable is defined, you can use info exists, which returns 1 if the variable is defined, and 0 if not.

set b [info exists a] <- b is now 1

set c [info exists gobbledygook] <- c is now 0

These commands can be arbitrarily nested, either in a string

or even in the name of a variable

set h 12 <- variable "h" is set to value "12"

set i$h 14 <- variable "i12" is set to the value "14"

Input and output in tcl is pretty simple. Puts and gets are the output and input commands.

set a [gets stdin] <- this reads in a value from the console and stores it in a

puts stdout $a <- this command echos back the value to the console

The default location for puts is stdout, but gets has no default location. Go figure.

By default, puts puts a carriage return at the end of the string it prints, but you can disable this with the nonewline flag.

You can define procedures in tcl and call them.

A procedure can take arguments.

A procedure can return a single value.

Procedure arguments can have default values

set b [foo2] <- b is assigned 12 + 3 = 15

set c [foo2 7] <- c is assigned 7 + 3 = 10

set d [foo2 1 8]<- d is assigned 1 + 8 = 9

Note that the arguments with default values must be at the end of the list of arguments, and no arguments with default values may precede arguments without default values.

Procedures can even take any number of parameters, if the last parameter is named "args". Here, calling this procedure with any number of arguments will pass in the arguments as a list (more on lists later) bound to the identifier args.

This somewhat contrived example takes in values from the console and echos them back.

In the arguments we have seen before, arguments are passed "by value". This means that in the code

The interpreter first computes the value of variable f (11), the value of variable g (22), and the value of variable h (33), and then send them (11,22,33) to the procedure foo. Foo knows nothing about where those values came from, and from foo’s perspective, the call might well have been

The other method of passing arguments is called "call by name". In this calling convention, a function can access and modify variables from the caller. The upvar command works to allow you to do this.

Upvar takes in the name of a variable from the caller and the name of a variable to "bind" that variable to. After the upvar command, any reference to the second variable actually refers to the named variable in the caller. To pass a variable by name, you send in the name of the variable, rather than its value. For example:

Now, the caller calls foo with the names of its three variables (f, g, and h). Foo uses upvar to say that after the 3 upvar commands, any reference to "a" really means a reference to the variable "f" as defined in the scope of the caller. Now, in the puts command when the value of a is printed out, the value of f in the caller is really the thing that gets printed.

Upvar can also allow you to modify variables in the caller.

Comments are text within a program that is ignored when running the program, but used to leave reminders and notes for other people who read the source code later.

Comments in Tcl are specified by the # character at the beginning of the line. This comments all the text to the end of the line. Note that you can only use # where the Tcl interpreter expects a Tcl expression, so you are somewhat restricted in where you can use it. In general, comments should be on their own lines to be safe.

The simplest arithmatic operations are simply incrementing and decrementing an integer. This is done with the incr and decr commands, which take in the name of the variable to increment. Note that this is a call by name command, so you send in the variable name, not the value of the variable.

(i.e. incr a, not incr $a)

incr a <- a is now 4

decr a <- a is now 3 again

Arithmatic operations in Tcl are executed with the expr command

A number of more complex operations can be done with expr as well:

set c [expr 7 << 3] <- c is assigned 7 left shifted by 3, i.e. 7 * 2³ = 56

set d [expr 7 >> 3] <- d is assigned 7 right shifted by 3, i.e. 7 / 2³

Note that the "type" of the input determines the operation in some cases

set e [expr 7 / 3] <- e is assigned 7 integer-divided by 3, or 2

set f [expr 7.0 / 3.0] <- f is assigned 7 float-divided by 3, or 2.3333

Other math operations have a slightly different notation

set g [expr pow(3,2)] <- g is assigned 3 raised to the second power

set h [expr acos(.8)] <- h is assigned the arc-cosine of .8

Tcl supports many math operations:

acos(x)

asin(x)

atan(x)

atan2(y,x) Rectangular (x,y) to polar (r,th), atan2 gives th

ceil(x)

cos(x)

cosh(x)

exp(x) e^x

floor(x)

fmod(x,y) Floating point remainder of x/y

hypot(x,y) Returns sqrt(x*x + y*y) (r part of polar expresion)

log(x) Natural log

log10(x) Log base 10

pow(x,y) x^y

sin(x)

sinh(x)

sqrt(x)

tan(x)

tanh(x)

abs(x)

double(x) Turns x into floating point number

int(x) Turns x into an integer (truncates)

round(x) Turns x into an integer (rounds off)

rand() Return a random floating point value between 0.0 and 1.0

srand(x) Seed the random number generator

This procedures takes in inputs and adds them up until they sum to more than 100.

Since all Tcl values are stores as strings, it makes sense to have a number of string operations that can be done. Just as expr preceded arithmatic expressions, string precedes these commands.

set a [string length "Hello world"] <- a gets the string length of the string, which is 11

set b [string index "Hello world" 3] <- b gets the 3^rd character in Hello World

Note that strings are 0-indexed, so the 0^th character is the first one in the string.

Other string operations:

string compare str1 str2 Returns 0 if equal, -1 if str1 sorts before str2

string first sub str Returns the index of the first character of the first

string last sub str Returns the index of the first character of the last

string match pattern str Returns 1 if str matches the pattern in pattern.

string range str I j Returns the range of characters from index i to index j.

string tolower str

string toupper str

string trim str ?chars? Trims the characters in the string chars from the front and

end of str. chars defaults to whitespace if it is not present.

string trimleft str ?chars?

string trimright str ?chars?

string wordend str ix Return the index in str of the character after the word

string wordstart str ix Return the index in str of the first character in the word

Another useful string command is append. This appends one or more arguments to a variable. Note that the first argument is the NAME of the variable, not [set variable] or $variable.

puts $a <- this will print "abc is some lettersthat I like"

Note: append is efficient, much more so than doing something like

Format is like the C printf function.

set a [format "%d %f %s" 10 4.3 "hello"] <- a is assigned "10 4.3 hello"

Scan is like the C scanf function.

scan "10 4.3 hello" "%d %f %s" x y z <- assigns x to be 10, assigns y to be 4.3, assigns z to be hello

This example reads in a string and then prints out some information about it

The most common data structure in Tcl is a list.

Lists can be created in a number of ways.

set a [list "a" "b" c"]

<- a is assigned to be a list with three elements, "a", "b", and "c"

Curly braces {} can be used as shorthand for the list command.

set a {"a" "b" "c"}

<- a is assigned to be the same list as earlier

Lists can be nested:

A number of list operators are very useful.

lindex list i <- returns the ith element of the list

llength list <- returns the number of elements in the list

lappend listName arg … arg <- appends some number of arguments to a list

lrange list i j <- return the sublist of the list found between

linsert list i arg … arg <- insert some number of arguments into a list before the

lreplace list i j arg … arg <- replace the elements i thru j of the list with the args.

returns the new list.

lsearch ?mode? list value <- returns the first index of an element in the list that

concat list … list <- concatenates multiple lists together into one big list.

lsort ?switches? list <- Sort the list as per the switches passed in. –ascii,

join list joinString <- join the elements of a list into a single string, each

element separated by joinString

split string splitChars <- split a string into a list by dividing the string up with

all the characters in splitChars as the delimiters.

Control flow in Tcl works like in many other languages.

One construct that’s cool is foreach. Foreach is like for, except that a loop is run once for each element in a list rather than for each value between two boundaries. This example would print out all the colors in the list.

Foreach can have more than one list variable, in that case the list would be taken in pairs. This example would print out the colors two at a time.

The continue and break commands allow control to jump out from the inside of a loop. Break breaks out of the loop and runs the next command, and continue jumps back to the start of the loop.

In some sense, break stops the loop totally and continue skips to the next run through the loop.

Switch allows control to jump to a number of expressions depending on what a value is.

The default is to do exact matching, but flags can be put after switch to change this.

or switch –glob $name

or switch –regexp $name

Put -- before $name if there’s any chance that $name might begin with a -, or else $name will be interpreted as a flag. It’s best to put -- before $name anyway.

If a - is written as the expression for a match, then control "falls through" to the next non-"-" expression.

Tcl supports error handling. If you use the catch command, you can try execute a block of Tcl expressions and trap any errors that result.

catch returns true (1) if the command failed, or false (0) otherwise. The resultVar, if present, is given the return value of the command, or an error message if an error occurred.

catch can catch return, break, and continue commands as well. See the documentation for information on this.

The error command is the counterpart to catch. It signals an error that will trigger an enclosing catch or halt the program with an error message if there is no enclosing catch.

The return command returns from a procedure, returning a value.

So far we’ve looked only at local variables, variables in the same procedure where they’re used. Variables can also be global, and thus accessible anywhere in the program. Global variables are somewhat dangerous to use, but they’re really quite useful, especially in small programs that are easily understood.

To declare a variable global, you simply use the global command

If you declare a variable outside of any procedures, it is implicitly global.

Another construct Tcl provides is an array indexed by string values. This is essentially a hash table, and allows efficient association of keys and values.

To create an array, simply set a variable along with an index.

set myArray("redKey") "redValue" <- this will create an array called myArray if it

To get the value out again, just use set like we did before.

or

You can test whether a key is present in an array by using info exists, just like for regular variables.

The array command allows a number of operations on arrays as well:

array exists arr <- returns 1 is arr is the name of an array variable

array get arr ?pattern? <- turns an array into a list of alternating keys and values

pattern selects the entries to put into the list, if present

(beats the hell out of me how the pattern works, though)

array names arr ?pattern? <- returns a list of the names of the keys in the array

applies the pattern to filter, if present

array set arr list <- create arr from the data in list, which is in the same form

as returned from get

array size arr <- return the number of keys in the array arr

You can iterate through an array as well. Look at the documentation for information on startsearch, nextelement, anymore, and donesearch.

Note that passing arrays between procedures requires call by name, using upvar.

File input/output is reasonably straightforward.

access can be:

r Open for reading, file must exist

r+ Open for reading and writing, file must exist

w Open for writing, replace if exists, create if does not

w+ Open for reading and writing, replace or create as needed

a Open for writing, data is appended to file

a+ Open for reading and writing, data is appended

Permissions are the standard numbers used in chmod, but you need a leading 0 to get an octal number.

Once open, a file can be written to

or read from

To close an open file (you must close a file before the Tcl program ends to store your changes), simply use close on the fileID

You can also pass –nonewline to puts, if you don’t want to put an end-of-line after the string.

Tcl provides a number of operators on files that can be quite useful.

file atime name <- returns file access time as a decimal string

file attributes name ?option? ?value? … <- query or set file attributes

file copy ?-force? source destination <- copy files or directories

file delete ?-force? name

file dirname name <- return parent directory of file name

file executable name <- return 1 if executable, else 0

file extension name <- return the extension (.txt) of the file

file isdirectory name <- returns 1 or 0

file isfile name <- return 1 if name is not a directory, symlink, or device, else 0

file join path path … <- join pathname components into a single pathname

file lstat name var <- place attributes of the link name into var

file mkdir name

file nativename name <- return the platform-native version of name

file owned name <- return 1 if current user owns the file name, else 0

file pathtype name <- relative, absolute, or driverelative

file readable name <- return 1 if readable, else 0

file readlink name <- return the contents of symlink name

file rename ?-force? old new

file rootname name <- return all but the extension in name (i.e. strip off .txt or whatever)

file size name <- return file size of name in bytes

file split name <- split name into its pathname components

file stat name var <- place attributes of name into array var

file tail name <- return the last pathname component of name

file type name <- return type identifier, which is either file, directory, characterSpecial, blockSpecial, fifo, link, or socket

file writeable name <- return 1 if writeable, else 0

Tcl provides the functionality to execute a number of system calls directly.

The exec program allows you to execute programs from your Tcl script. The standard output of the program will be returned.

If the program writes to the standard error channel or exits with a non-zero exit code, you’ll need to use catch to get the information you want.

Exec has a lot of weird specifics about how it works, so look at the documentation for specifics.

http://www.scriptics.com/man/tcl8.0/TclCmd/exec.htm

The exit command terminates your Tcl program. An integer argument to exit sets the exit status of the process

The pid command returns the process ID of the current process.

You can get access to environment variables with the global variable env. This loop prints out all the environment variables and their values.

The Tcl Platform Company

http://www.scriptics.com/

Brent Welch’s book (excellent)

http://www.beedub.com/book/

Tcl/Tk Man pages

http://www.scriptics.com/man/tcl8.0/contents.htm

Tcl/Tk Plugin for Netscape, Internet Explorer

http://www.scriptics.com:8123/plugin/