5. Control Statements

Transcription

1 5. Control Statements This section of the course will introduce you to the major control statements in C++. These control statements are used to specify the branching in an algorithm/recipe. Control statements are also used to specify the looping of an algorithm and the conditions upon which the loop should be ended. Readings: [Savitch99] Section 2.4 and Chapter Control Statements Basic Flow of Control The C++ If Statement Boolean Expressions and Operators Boolean Variables and Literals The C++ Switch Statement What Kind of Loop? While Loops Do Loops Loops With Mid-Body Exits For Loops Copyright 1998 by R. Tront 5-1 Copyright 1998 by R. Tront 5-2

2 5.1 Basic Flow of Control A computer program is like a recipe. It is a list of instructions which described the steps of an algorithm. Most useful or very interesting algorithms are composed of more than just a linear list of steps. Such algorithms may have alternative sub-algorithms, only one of which you want executed under any particular conditions (for instance, only on Wednesdays). Other algorithms may be iterative in nature, repeating the same basic steps over and over again until some final desired result is obtained. Most introductory computer courses will tell you that any computer program can be composed with just three basic types of flow. Sequence Alternation Iteration There is a very visual technique called flow charting that I will use to introduce these three different types of flow of control. Yes Statement1 Statement2 Statement3 Is it Wednesday? Sequence No Statement4 Alternation Note 1: The flow chart branch symbol is often just a diamond, but a flattened hexagon provides more room inside to write the branch condition. Note 2: There is a variation of Alternation where something is done on Wednesdays and nothing if it is not Wednesday. Copyright 1998 by R. Tront 5-3 Copyright 1998 by R. Tront 5-4

3 Finished Iterating Yet? No Statement 7; Statement 8; Statement 9; Yes Iteration Most computer languages provide a variety of control statements. Only three different ones are required. And providing too many different control statements makes a program hard to read. Most modern computer languages have about the right number of kinds of control statements. Copyright 1998 by R. Tront The C++ If Statement Sequence in C++ is simple provided by programming one statement after another, usually separated by a semi-colon ( ; ). Alternation in C++ is provided by the C++ if and switch statements. The condition used in deciding which branch of the alteratives to take must be a boolean expression, or convertible to a boolean expression. short int temperature = 25; if (temperature < 15) cout << Temperature is low ; else { cout << Temp is high, << endl; cout << so turn on computer fan. ; } //The next statement goes here, //so this is where the two //alternative paths rejoin. This example of a C++ if statement shows the decision being made on a boolean expression. Boolean expressions are often formed by comparing the size of two numbers with relational (also called comparison) operators like: < - which at execution time determines if its first operand is less than the second. The result is a boolean value, either true or false. This is used by the if to Copyright 1998 by R. Tront 5-6

4 determine whether the then or else alterative is taken. (Note that contrary to the PASCAL programming language, C++ never requires you, and in fact doesn t allow you to write the word then ). <= - is an relational operator which determines if the first operand is less than or equal to the second. == - This is two equal sign characters, one right after the other. It is used to compare whether two operands are equal. If they are, the resulting boolean value used by the if statement is a true, otherwise the result is false which causes the else alternative to be executed. A single = should NOT be used in this case as that is the assignment operator, not a comparison operator! The two alternatives of an if statement are called the then and else clauses. In their simplest form, each may only contain one statement. But like compound statements in Pascal, you can replace the one statement with multiple statements suitably delimited by opening and closing brace characters {}. This is similar to Pascal s BEGIN and END statements. I have illustrated this in the else clause in the above example. Generally you do not need to put a semi-colon after a closing brace, as the closing brace is interpreted by the compiler to mean the end of a compound statement. There is important exceptions to this rule which will be pointed out later. Copyright 1998 by R. Tront 5-7 Sometimes you want to do something if the value in the temperature variable is less than 15, and simply skip doing that something otherwise and carry on with the rest of the algorithm. In that case, you can leave the else and the else clause out of your program! Note that it is common to indent the then and else clauses to improve readability. I will insist on this in Cmpt 101/104. A fast way to consistently do this is to use the tab key. Generally, the body of a function and an iterative loop are also indented from their first line. Note that when a then or else clause itself contains another if statement, you will want to indent further this entire second if statement. If that one contains yet another if statement, you should probably abandon indenting and go to a form like the following. This is particularly true if the first few ifs involved have nothing in their else clause but the next if. if (condition1) statement1; else if (condition2) statement2; else if (condition3) statement3; else otherwisestatement; You would use actual statements and conditionals when programming rather than this skeleton. Copyright 1998 by R. Tront 5-8

5 An (inner) if statement within an (outer) if statement is called a nested if. As discussed in the textbook readings for this course, there is a question of which if the final else applies to. You should know that it applies to the last if, unless you somehow use braces {} around one of the inner ifs to override this rule. Some other languages have an elsif or elseif keyword to make it very clear that the last else normally applies to the last if. Copyright 1998 by R. Tront Boolean Expressions and Operators Sometimes, you might nest ifs unnecessarily. It is sometimes possible to put the two ifs together with Boolean operators to form a compound boolean expression. e.g. if((temp < 15) && (speed > MACH1))... //instead of: if (temp < 15) if (speed > MACH1)... The && operator is the logical AND operator. This conditional can be read if temp is less than 15 and speed is greater than MACH1. The OR operator is written (i.e. one vertical bar immediately after another; you might have to search hard to find a vertical bar character on your keyboard). Boolean operators are used for more than just combining some special cases of single ifs; they are widely used for expressing and storing all kinds of logical conditions. Note that && is a boolean operator that works on boolean operands on both sides of it. i.e. bool boolvariable = true && false It is very important to examine the truth tables in your textbook to completely understand the operation of these operators. In the above case, boolvariable would be assigned the value false. In the if statement at the top of this section, the two boolean values being Copyright 1998 by R. Tront 5-10

6 ANDed together are the results of individual numerical comparisons. There is another boolean operator, in fact a unary boolean operator, called NOT. It is kind of like a unary negation - operator, but for boolean rather than numerical variables. It is! rather than the - used for arithmetic negation. Copyright 1998 by R. Tront Boolean Variables and Literals Boolean variables have just recently been added to C++, and many compilers more than a year old do not have this container type! An Appendix of [Savitch99] discusses how you might compensate for this. Boolean variables are containers to store boolean values. They usually only take up one byte of space, though in theory they could be reduced to one bit! Boolean variables are good if you need to store the result of a relational expression such as: bool b = (variable1 < variable2); b = (variable1 < PI); b = variable == PI; Notice that you should use lots of parentheses in boolean expressions and assignments. The bottom one just above has none and is very hard to read. In addition, it begs the question of which has the higher precedence and is thus done first: the assignment, or the comparison? If you use lots of parentheses, you will rarely have to concern yourself with the ugly details in your textbook regarding this. The only two literal values proper C++ boolean variables can have are false and true, spelled in Copyright 1998 by R. Tront 5-12

7 lower case with no quotes. But you can, for backward compatibility with C, assign a zero to a boolean and it will be automatically converted to a false and stored. Any non-zero value will be converted into and stored as a true. An aside: There are ways to convert a boolean to a number. This is part of an advanced topic called casting. To maintain some backward compatibility with old C, assigning a boolean to an integer will convert false into zero and true into a one. Sometimes this happens automatically and the results can surprise you. Also, cout << false annoyingly will print 0 rather than false. Copyright 1998 by R. Tront The C++ Switch Statement In the case of a long series of if, else if, else if,... there is often a better way to code this. It is called in Pascal a CASE statement and in C++ a switch statement. It works like this. int myint; const int FOUR = 4; cin >> myint; switch (myint * 2) { case 2: cout << first case ; cout << endl; break; case 4: cout << second case ; break; case 6: case 8: cout << third or fourth ; break; case 20/FOUR: cout << five ; break; default: cout << all others ; } //no semi-colon needed after }. A switch statement basically is a multi-way branch. The a variable or expression in parentheses controls which branch is taken. The flow of control jumps to Copyright 1998 by R. Tront 5-14

8 whatever corresponding case clause matches. That clause is executed. Note that several statements are allowed in each clause without needing braces {} to join them into one compound statement. The break keyword encountered at the end of a case clause indicates that execution is to again jump, this time to the first statement following the closing } of the switch statement. Don t forget the break at the end of each clause as otherwise one clause will execute right into the next! If none of the cases match the switch expression, then the optional default clause is executed. It is always a good idea to provide a default clause at the end of any switch statement. Note the default clause does not need a break because it is already at the bottom. Question: What happens when there is no default clause, and no matching case? Why don t you write such a program and use the debugger to single step through and watch what happens! The switch expression must be a variable or expression of some type that can be converted to an integer (e.g. short, long, char, bool. Recall that characters are stored as 1 byte long integers!). Note that A (including the quotes) is a usable integer constant. The case values must be integral literals or constant expressions. Though you can probably use a constant variable in a switch case, you cannot put regular variables there. This is because the series of comparisons done against each case is constructed by the compiler at compile time. If you must compare some expression against a variety of values including against values that might change at run time because they are stored in variables, then instead of a switch you simply have to use a series of if statements instead. Finally note that I have shown that you may use two cases in front of one case clause. Copyright 1998 by R. Tront 5-15 Copyright 1998 by R. Tront 5-16

9 5.6 What Kind of Loop? There are three C++ statements commonly used for iterative calculations that require a sequence of steps to be repeated over and over again. They are the: while statement do statement for statement Before you decide which one to use, you should draw a flow chart. Of particular concern is whether the decision to break out of the loop is at the top, bottom, or middle of the loop. Finished Iterating Yet? No Statement 7; Statement 8; Statement 9; Yes Loop with exit at top Top exit loops are nice because you can check first whether to exit the loop body at all! It allows that some days when you run the program with some data, the body of the loop will be skipped. Generally the body of the loop modifies some variables that are used in the boolean exit condition so as to definitely cause you to eventually exit. Copyright 1998 by R. Tront 5-17 Copyright 1998 by R. Tront 5-18

10 Mid-Exit Bottom Exit Done Not Done Not Done Done Sometimes, as illustrated above, you must execute some of the loop before deciding to exit. A good example is when attempting to read input data from a file. First you must attempt to read. It you were unsuccessful because you are at the end of the file and the read failed, you should exit the loop. But if you were successful, you likely need to do something in the bottom box with the data that was read in, before attempting to read yet another data value from the file. Note that this logic applies whether you just opened and attempted to read an empty file, or whether you have read 50,000 values and finally reach the end. Copyright 1998 by R. Tront 5-19 The above flowchart illustrates a situation where you need to execute the body of the loop before trying to decide whether to exit the loop. This is because the statements you execute in the body of the loop determine the value of some variable(s) that are used in the boolean expression coded to decide whether to exit the loop. To which form of loop to use in a program, it is best to draw a flow chart of what you want to do because humans are very visual and are less likely to make the wrong loop choice with visual reasoning. To help you with this, it is important to write pseudo-code in the boxes. Pseudo-code is neither legal or acceptable C++, Copyright 1998 by R. Tront 5-20

11 but rather an English-like set of statements that describe the steps of the algorithm you want to implement (you can translate into C++ later; first and most importantly, get the algorithm laid out correct by using visual aids). When designing loops, you MUST consider very carefully: what you need to do in preparation for entering the loop (e.g. define some variables the loop might need, and/or set some variables to suitable starting values), what do you want to do within the loop, do you always want to execute the body of the loop at least once? Perhaps then the exit decision should be at the bottom. Or should it be in the middle? Or if sometimes you don t want to execute the body of the loop at all, perhaps the decision should be made at the top so as to, if appropriate, allow for sometimes not executing the body of the loop! on which iteration do you want to leave the loop (9th iteration, 10th, or when myvar becomes greater or equal to 25), lastly, make sure that your exit condition will ALWAYS eventually come true, otherwise your program could loop forever. This often soaks up the processor s entire attention and there is sometimes no way to stop the program except to turn off the computer (which may cause you to loose data or your most recent program changes). Copyright 1998 by R. Tront 5-21 Copyright 1998 by R. Tront 5-22

12 5.7 While Loops Here is a loop that adds the numbers from one to n, where n is an integer that is read in. int n; cout << Type an integer: ; cin >> n; cout << endl << The sum of 1 to << n << is ; int runningtotal = 0; while( n >= 1 ){ } runningtotal = runningtotal + n; n = n-1; cout << runningtotal; Notice that: Note that most C++ loops require you to write a boolean expression saying when it is appropriate to continue looping rather than when to exit. If I had an continuation condition of n>1 or n=0, the wrong answer would be calculated. But n>0 would work fine. if I had reversed the order of the two statements in the body of the loop, the displayed answer would be wrong because the value the user entered would never have been added to the runningtotal. The first time through the loop, one less than the value entered would be initially assigned to the total! if I had instead printed out n after the loop, it would have printed out 0 rather than the value that the user entered. If I had really wanted to delay printing out n until after the loop, I would have had to copy n to another variable that counted down (or with suitable modifications, up), instead of modifying n. Or I could have decremented n as long as I first copied it to the other variable for safe storage, then print that other variable, which at the end, would still contain the value that the user typed in. So you can see that there are many different ways to screw up. Programming requires great care! Also, you see that there are often several different ways to write a program. This is what design is: choosing wisely between alternatives. It is not just doing it the first way that comes into your head. If you are a beginner programmer, you should try writing all the different versions mentioned in the last bulleted paragraph! In addition, also try running each within the debugger in single step mode and watching how each variable changes as the program executes! Copyright 1998 by R. Tront 5-23 Copyright 1998 by R. Tront 5-24

13 This will be extremely valuable to you. See the TAs if you need help with any of this. And yes some of you real C and C++ hackers, I know that you can replace the lines: runningtotal = runningtotal + n; n = n-1; with: runningtotal += n; n--; 5.8 Do Loops If you want your exit decision to be performed at the end of the loop body, you need a do... while(); style loop. Here is a familiar program changes to instead use a do loop: int n; cout << Type an integer: ; cin >> n; cout << endl << The sum of 1 to << n << is ; int runningtotal = 0; do{ runningtotal = runningtotal + n; n = n-1; }while (n >= 0); cout << runningtotal; Notice the structure of a do loop. There is only the keyword do at the beginning (plus a brace if there is more than one statement in the body of the loop). You indicate whether the loop will continue iterating by placing the while keyword and parenthesized boolean Copyright 1998 by R. Tront 5-25 Copyright 1998 by R. Tront 5-26

14 continuation expression after the loop body. Don t forget the semi-colon after the closing parenthesis. How does the compiler know that this while is not the beginning of some other while loop that comes next? Because it is placed after a do. After any do, the first while encountered by the compiler is not considered the start of a while style loop. This is what is meant by the term syntax analysis, which is done very carefully by the compiler. If I had written the continuation condition as n>=1 (like I did for the while style loop several pages back), 80% of you would not have noticed. In fact, many of you, if writing the do style version of the program, would have possibly used n>=1. But your program would have printed the wrong total! It is very easy to make such exit condition mistakes! I will say that again. It is very easy to make exit condition mistakes! Exercise: Would n > -1 be an acceptable exit condition? Note: There is usually a reason to prefer either while versus do style loops. But for this particular algorithm there is none. When an algorithm can be written either way, though, the exit condition is usually different for the two different styles! Copyright 1998 by R. Tront Loops With Mid-Body Exits In some languages like Modula-2, there are special loops for this. In C++ though, there is just a kludge that will break you out of the middle. char ch; while (true){ fin >> ch; if ( fin.eof() ) break; cout << ch; } What you do is purposely program an infinite loop! The while condition above is always true (it is the boolean literal true ). Then you use an if statement where desired in the middle of the loop to test an exit (not continuation) boolean expression; in the above case, to test whether your last read from a file encountered the end of file (abbreviated in computer science eof). The eof() function of the fin object returns a boolean value which is used by the if. Finally, the keyword break is used as the then clause of the if. You can put the break to the right of the if, but in this class I will insist that it be put right under the Copyright 1998 by R. Tront 5-28

15 w of the while, so you can quickly spot in a long loop body where the loop exit is and what condition caused the exit. Copyright 1998 by R. Tront For Loops All computer languages generally have a for style loops. They are mainly used to have a loop body executed several times with one variable linearly increasing (or decreasing) by one each iteration. C++ s for loop though is quite a bit more flexible than that. In fact, the so called control variable can increase or decrease by any amount. In fact there needn t even be a control variable as long as there is some kind of control expression. Here is an example of a for loop: int n; cout << Type an integer: ; cin >> n; cout << endl << The sum of 1 to << n << is ; int runningtotal = 0; for(int i = n; i>=1; i=i-1){ //note no ; just before the ) runningtotal = runningtotal + i; } cout << runningtotal; This says: for integer variable i starting at n, while i is still greater or equal to 1, execute the body of the loop, and decrement i before starting the loop again. Copyright 1998 by R. Tront 5-30

16 A for loop is a top exit loop. If the exit condition (which may be any boolean expression like x>y) is not satisfied when the for statement is first encountered, the loop body will not be executed even once. Also note the update action i=i-1 is NOT done before the first iteration. We are glad of this because we want the first value added to running total to be n, not one less! Given what I have just said you should be able to see that a for loop is just a different, slightly more compact way of saying the same thing as: int i = n; while (i>=1){ runningtotal = runningtotal + i; i = i-1; } The initializing action is placed before the while, the continuation expression is placed in the while condition, and the update action is put as the last statement of the while loop. Warning #1: In many languages it is forbidden to mess with the control variable in the body of the for loop. This is not the case in C++. But, nonetheless any for loop that messes with the control variable both within the update action and also in the loop body has a much higher likelihood of a mistake that causes the loop to exit at the wrong time or not at all! Warning #2: For all styles of loops, it is wiser to use an inequality in your continuation expression rather than an equality. This is because if you are iterating up your controlvariable from zero by two each time through the loop, and you want to keep going if controlvariable hasn t reached 13 yet (i.e. controlvariable!= 13) your loop will never end. Why? This problem is particularly common when comparing real numbers rather than integer numbers, because round off error can cause two numbers to not quite be the same. e.g. (1/3)*3 is not generally exactly equal to 1. The expression 1/3*3 is , not 1. Warning #4: Though it has always been legal C++ to define a variable right after the ( of while and for loops, only in the 1998 compilers is the scope of this variable just that of the loop body. In older compilers, the variable is also valid outside the loop and will conflict with another variable with the same name which is defined outside the loop. Copyright 1998 by R. Tront 5-31 Copyright 1998 by R. Tront 5-32