1. Chapter4 Syntax-Directed Translation Introduction ： 1.In the lexical analysis step, each token has its attribute ， e.g., the attribute of an id is a pointer to the symbol-table entry for that id, the attribute of a num is the value of the num ； 2.thinking ： In the syntax analysis step, should each grammar symbol has its attribute? tip1 ： Complier not only gives an answer whether the string is a sentence, but also need to do a lot of work, such as the generation of object code, error management and so on. tip2 ： Many compiler is based on syntax analysis stage as the core, around syntax analysis to complete semantic analysis, intermediate code generation, code optimization and so on 。 tip3 ： The attribute of grammar symbol is very board, it can be generation of code, calculation of the value, information stored in the symbol table and so on.

2. 4.1 Syntax-Directed Definitions e.g.Syntax-directed definition of a simple desk calculator PRODUCTION SEMANTIC RULES L  E n L  E n print (E.val) print (E.val) E  E 1 + T E  E 1 + T E.val := E 1.val + T.val E.val := E 1.val + T.val E  T E  T E.val := T.val E.val := T.val T  T 1 * F T  T 1 * F T.val := T 1.val * F.val T.val := T 1.val * F.val T  F T  F T.val := F.val T.val := F.val F  (E) F  (E) F.val := E.val F.val := E.val F  digit F  digit F.val := digit.lexval F.val := digit.lexval

3. 4.1 Syntax-Directed Definitions 4.1.1 the form of Syntax-Directed Definitions Basic grammar Attributes are associated with grammar symbol Each grammar production A   has a set of semantic rules like b := f(c 1, c 2, …, c k ), where f is the function, b and c 1, c 2, …, c k are the attribute of the grammar symbol. Synthesized Attributes ： if b is the attribute of A ， c 1, c 2, …, c k are the attribute of right side symbol in the grammar production or other attribute of A. Inherited Attributes ： if b is the attribute of a right side sumbol of x. Attribute is defined at the children or it self Attribute is defined at its parent, it self and its siblings 15/36

4. 4.1 Syntax-Directed Definitions 4.1.2 Synthesized Attributes S-Attributed Definitions ： every attribute is synthesized PRODUCTION SEMANTIC RULES L  E n L  E n print (E.val) print (E.val) E  E 1 + T E  E 1 + T E.val := E 1.val + T.val E.val := E 1.val + T.val E  T E  T E.val := T.val E.val := T.val T  T 1 * F T  T 1 * F T.val := T 1.val * F.val T.val := T 1.val * F.val T  F T  F T.val := F.val T.val := F.val F  (E) F  (E) F.val := E.val F.val := E.val F  digit F  digit F.val := digit.lexval F.val := digit.lexval

5. 4.1 Syntax-Directed Definitions The annotated parse tree for 8+5*2 n digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 Each node’s attribute value is annotated According to the annotated parse tree, let’s study the calculaton order of S-attributed definition

6. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

7. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

8. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n 20/36

9. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

10. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

11. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

12. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

13. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n 25/36

14. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

15. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

16. 4.1 Syntax-Directed Definitions The calculation of attribute value of nodes of the annotated tree can be done from bottom to up digit.lexval = 2 L E.val = 18 n T.val = 10 E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

17. 4.1 Syntax-Directed Definitions The annotated parse tree :Each node’s attribute value is annotated digit.lexval = 2 L E.val = 18 n T.val = 10E.val = 8 T.val = 8 F.val = 8 digit.lexval = 8 T.val = 5 + * F.val = 5 F.val = 2 digit.lexval = 5 8+5*2 n

18. 4.1 Syntax-Directed Definitions 4.1.3 Inherited Attributes int id, id, id PRODUCTION SEMANTIC RULES D  TL D  TL L.in := T.type L.in := T.type T  int T  int T. type := integer T. type := integer T  real T  real T. type := real T. type := real L  L 1, id L  L 1, id L 1.in := L.in; L 1.in := L.in; addtype (id.entry, L.in ) addtype (id.entry, L.in ) L  id L  id addtype (id.entry, L.in ) addtype (id.entry, L.in ) 30/36

19. 4.1 Syntax-Directed Definitions Annotated parse tree for int id 1, id 2, id 3 Dint T.type = integer, id 3 L.in = integer id 2 id 1,

20. 4.1 Syntax-Directed Definitions 4.1.4 Dependency Graphs Dependency graph for the annotated parse tree of int id 1, id 2, id 3 D  TL L.in := T.type D  TL L.in := T.typeD int int T, id 3 L L L id 2 id 1, 1 entry 10 2 entry 3 entry in 9 8 in 7 6 in 5 4 type

21. 4.1 Syntax-Directed Definitions 4.1.4 Dependency Graphs Dependency graph for the annotated parse tree of int id 1, id 2, id 3 L  L 1, id L 1.in := L.in; addtype (id.entry, L.in ) addtype (id.entry, L.in )D int int T, id 3 L L L id 2 id 1, 1 entry 10 2 entry 3 entry in 9 8 in 7 6 in 5 4 type

22. 4.1 Syntax-Directed Definitions 4.1.5 Ordering the Evaluation of Attributes Topological sort ： if there is an edge of the dependency graph from Ni to Nj,then i < j e.g. ： 1 ， 2 ， 3 ， 4 ， 5 ， 6 ， 7 ， 8 ， 9 ， 10 D int int T, id 3 L L L id 2 id 1, 1 entry 10 2 entry 3 entry in 9 8 in 7 6 in 5 4 type

23. 4.1 Syntax-Directed Definitions Ordering the Evaluation of Attributes 1 、 Structure input parse tree ， 2 、 Structure dependency graph ， 3 、 sort nodes by topological sort, 4 、 calculate the attribute by the order of topological sort 。 35/36

24. Supplement The understanding of attribute nonterminal  analysis process （ function ） synthesized attributes  the return value of process inherited attributes  the parameters of process

25. e.g. int id, id, id PRODUCTION SEMANTIC RULES D  TL D  TL L.in := T.type L.in := T.type T  int T  int T. type := integer T. type := integer T  real T  real T. type := real T. type := real L  id, L 1 L  id, L 1 L 1.in := L.in; L 1.in := L.in; addtype (id.entry, L.in ) addtype (id.entry, L.in ) L  id L  id addtype (id.entry, L.in ) addtype (id.entry, L.in ) Review

26. Review e.g. void D() { T_temp = T(); L_in = T_temp; L(L_in); L(L_in); return ; } int T() { switch lookahead { case INT: return INTEGER; { case INT: return INTEGER; case REAL: return REAL; case REAL: return REAL; default: error; default: error;}} void L(int L_in) {match(id); match( ‘, ’ ); match( ‘, ’ ); addtype(id.entry, L_in); L(L_in); L(L_in);} Void L(int L_in) { match (id); addtype(id.entry, L.in); } PRODUCT ION SEMANTIC RULES D  TL D  TL L.in := T.type T  int T  int T. type := integer T  real T  real T. type := real L  id, L 1 L  id, L 1 L 1.in := L.in; addtype (id.entry, L.in ) L  id L  id addtype (id.entry, L.in )

27. Exercise 4.1 According to the syntax-directed definition of table 4.1 ， please structure the annotated parse tree for expression5 * (4*3+2) 4.1 According to the syntax-directed definition of table 4.1 ， please structure the annotated parse tree for expression5 * (4*3+2) 。