jgs Compilers Lecture 15: Error Handling III Dr. Javier Gonzalez-Sanchez jgonzalezs@up.edu.mx

jgs Error Handling Version 2

Dr. Javier Gonzalez-Sanchez | Compilers | 3 jgs Error Handling a) Panic One error, then Stop a) Simple Either ignore or increase the current token. a) First and Follow Each rule is responsible for validating the start and end.

Dr. Javier Gonzalez-Sanchez | Compilers | 4 jgs Error Recovery At this point, Error recovery is about ignoring tokens. § How do we know which tokens should be ignored and how many?

Dr. Javier Gonzalez-Sanchez | Compilers | 5 jgs Parser | Error Recovery Rule FIRST set FOLLOW set PROGRAM { EOF BODY FIRST (PRINT) U FIRST (ASIGNMENT) U FIRST(VARIABLE) U FIRST (WHILE) U FIRST(IF) U FIRST (RETURN) } PRINT print ; ASSIGNMENT identifier ; VARIABLE int, float, boolean, void, char, string ; WHILE while } U FIRST(BODY) IF if } U FIRST(BODY) RETURN return ; EXPRESSION FIRST(X) ), ; X FIRST(Y) | U FOLLOW(EXPRESSION) Y ! U FIRST(R) & U FOLLOW(X) R FIRST(E) FOLLOW(Y) E FIRST (A) !=, ==, >, < U FOLLOW(R) A FIRST (B) -, + U FOLLOW(E) B - U FIRST (C) *, /, U FOLLOW(A) C integer, octal, hexadecimal, binary, true, false, string, char, float, identifier, ( FOLLOW(B)

jgs Calculating the Follow Set

Dr. Javier Gonzalez-Sanchez | Compilers | 9 jgs FOLLOW set Define FOLLOW (BODY) FIRST(BODY) = }

Dr. Javier Gonzalez-Sanchez | Compilers | 11 jgs FOLLOW set → → → d → a → ab → ε → c → d → e → → f → ε

Dr. Javier Gonzalez-Sanchez | Compilers | 12 jgs Calculate the FOLLOW set 1.First put $ (the end of input marker) in Follow(S) (S is the start symbol) 2.If there is a production A → aBb, (where a can be a whole string) then everything in FIRST(b) except for ε is placed in FOLLOW(B). (apply the rule 4 in calculate FIRST set) 3.If there is a production A → aB, then add FOLLOW(A) to FOLLOW(B) 4.If there is a production A → aBb, where FIRST(b) contains ε, then add FOLLOW(A) to FOLLOW(B)

Dr. Javier Gonzalez-Sanchez | Compilers | 13 jgs Calculate the FOLLOW set

Dr. Javier Gonzalez-Sanchez | Compilers | 14 jgs FOLLOW set S → ABC S → F A → EFd A → a B → aBb B → ε C → cC C → d E → eE E → F F → Ff F → ε rule FOLLOW set - evolution S {eof} A {a} {a, c, d} B {c, d} {c, d, b} C {eof} E {f} {f, d} F {eof} {eof, d} {eof, d, f} FIRST sets: S={a,ε,e,f,d} A={a, e, f, d} B={a, ε} C= {c, d} E={e, ε, f} F={ε,f}

Dr. Javier Gonzalez-Sanchez | Compilers | 15 jgs Another Example → {+} → {*} → () | integer FIRST (E) = {(, integer} FIRST (T) = {(, integer} FIRST (F) = {(, integer} FOLLOW(E) = {$, )} FOLLOW(T) = {$, ),+ } FOLLOW(F) = {$, ),+, * }

jgs Prediction Rules

Dr. Javier Gonzalez-Sanchez | Compilers | 17 jgs Prediction Rules Rule 1. It should always be possible to choose among several alternatives in a grammar rule. FIRST(R1 ) FIRST(R2 ) FIRST(R3 )... FIRST(Rn ) = Ø BODY

Dr. Javier Gonzalez-Sanchez | Compilers | 18 jgs Prediction Rules Rule 1.1 The FIRST sets of any two choices in one rule must not have tokens in common in order to implement a single-symbol look ahead predictive parser.

Dr. Javier Gonzalez-Sanchez | Compilers | 19 jgs Prediction Rules Rule 2. For any optional part, no token beginning the optional part can also come after the optional part. FIRST(RULE) != FOLLOW(RULE) BODY PROGRAM

jgs Rules

Dr. Javier Gonzalez-Sanchez | Compilers | 21 jgs Parser | Error Recovery PROGRAM Line N: expected { Line N: expected } currentToken++; Searching for FIRST(BODY) or }

Dr. Javier Gonzalez-Sanchez | Compilers | 22 jgs Parser | Error Recovery Line N: expected ; Line N: expected identifier or keyword currentToken++; Searching for FIRST(BODY) or FOLLOW(BODY) BODY

Dr. Javier Gonzalez-Sanchez | Compilers | 23 jgs Parser | Error Recovery ASSIGNMENT Line N: expected = currentToken++; Searching for FIRST(EXPRESSION) or FOLLOW(EXPRESSION)

Dr. Javier Gonzalez-Sanchez | Compilers | 24 jgs Parser | Error Recovery VARIABLE Line N: expected identifier

Dr. Javier Gonzalez-Sanchez | Compilers | 25 jgs Parser | Error Recovery WHILE Line N: expected ( Line N: expected ) currentToken++; Searching for FIRST(EXPRESSION) or ) currentToken++; Searching for FIRST(PROGRAM) or FOLLOW(PROGRAM)

Dr. Javier Gonzalez-Sanchez | Compilers | 26 jgs Parser | Error Recovery IF Line N: expected ( Line N: expected ) currentToken++; Searching for FIRST(EXPRESSION) or ) currentToken++; Searching for FIRST(PROGRAM) or FOLLOW(PROGRAM)

Dr. Javier Gonzalez-Sanchez | Compilers | 27 jgs Parser | Error Recovery RETURN

Dr. Javier Gonzalez-Sanchez | Compilers | 28 jgs Parser | Error Recovery PRINT Line N: expected ( Line N: expected ) currentToken++; Searching for FIRST(EXPRESSION) or )

Dr. Javier Gonzalez-Sanchez | Compilers | 29 jgs Parser | Error Recovery EXPRESSION X Y R E A B

Dr. Javier Gonzalez-Sanchez | Compilers | 30 jgs Parser | Error Recovery C Line N: expected value, identifier or ( Line N: expected )

Dr. Javier Gonzalez-Sanchez | Compilers | 31 jgs Error Recovery PROGRAM BODY ASSIGNMENT VARIABLE WHILE IF RETURN PRINT C EXPRESSION X Y R E A B

Dr. Javier Gonzalez-Sanchez | Compilers | 32 jgs Homework Submit Parser 1 Submit Parser 2 Grading Work on Parser 3

Dr. Javier Gonzalez-Sanchez | Compilers | 33 jgs Homework Let us do an Exam Lexical + Syntax When is a good date?

jgs Test Yourselves

Dr. Javier Gonzalez-Sanchez | Compilers | 35 jgs Error Recovery { x = a; x = 1 + ( x = (y;) if (a < b + ) {} else {} if (a < b) { if (a < b) { } else { } } Input: We will not allow multi- line expressions; line 3 and 4 should not be considered as follow: x= 1 + ( x = ( y;)

Dr. Javier Gonzalez-Sanchez | Compilers | 36 jgs Error Recovery Line 3: expected value, identifier, ( Line 3: expected ) Line 3: expected ; // move to the next line Line 4: expected ) Line 4: expected identifier or keyword // infinite loop or end Line 5: expected value, identifier, ( // simple Line 12: expected } // reported by program Output:

Dr. Javier Gonzalez-Sanchez | Compilers | 37 jgs Questions

jgs Compilers Javier Gonzalez-Sanchez, Ph.D. jgonzalezs@up.edu.mx Spring 2024 Copyright. These slides can only be used as study material for the Compilers course at Universidad Panamericana. They cannot be distributed or used for another purpose.