1. Discussion #71/13 Discussion #7 Recursive Descent Parsing

2. Discussion #72/13 Topics Recursive Descent Parsing –Use tables to build recursive descent parsers –Parse tree construction Project #2  Parsing

3. Discussion #73/13 Recursive Descent Parsing Consider an arbitrary production S  xAySz and assume x is the current (top symbol), i.e.: S x (xAySz, …)

4. Discussion #74/13 Recursive Descent Parsing (continued…) Make a method for S (indeed, for every non-terminal) as follows: –For S  xAySz Attempt to read an x from the input. If success, call method A. If success, attempt to read a y from the input. If success call method S. If success attempt to read a z from the input. If success, method S reports success! –If any of the above attempts fail, report failure.

5. Discussion #75/13 Recursive Descent Parsing (continued…) Make a method for S (indeed, for every non-terminal) as follows: –For S  xAySz | q If x is the current input character then –call method A. –If success, attempt to read a y from the input. –If success call method S. –If success attempt to read a z from the input. –If success, method S reports success! Else if q is the current input character then, report success Else report error

6. Discussion #76/13 Recursive Descent Parsing (continued…) Output from each syntactical class is a left- child, right-sibling parse tree. E  OEE produces: E OEE

7. Discussion #77/13 Recursive Descent Parsing Example Consider our prefix grammar: E  N | OEE O  + |  | * | / N  0 | 1 | … | 9 Design a series of recursive methods: E() to process N or O, E, E O() to process +, , *, / N() to process numbers 0 thru 9

8. Discussion #78/13 Data Structure for Parse Tree class parseTree { … char value parseTree leftChild parseTree rightSibling … };

9. Discussion #79/13 Initialization: Call to Start Symbol parseTree buildTree() // build parse tree { … nextChar = readChar() // read 1st character ptree = E() // start syntactic class if (ptree == error) return error if (nextChar) return error // check for string finished return ptree // return the full parse tree }

10. parseTree E() // syntactic category E { parseTree ptree,sibling1,sibling2 if (isCharInString(nextChar, "+-*/")) // E -> OEE {FIRST(OEE)} { ptree = O() // Try to recognize O if (ptree == error) return error sibling1 = E() // Try to recognize E if (sibling1 == error) return error sibling2 = E() // Try to recognize E if (sibling2 == error) return error ptree.rightSibling = sibling1 // Success, link O->E->E ptree.rightSibling.rightSibling = sibling2 } else if (isCharInString(nextChar, "0123456789")) // E -> N { ptree = N() // Try to recognize N if (ptree == error) return error } else return error return new parseTree('E', ptree, null) } Discussion #710/13 Method for E

11. Discussion #711/13 Methods for N and O parseTree N() // syntactic category N { ptree = new parseTree(nextChar, null, null) nextChar = readChar() return new parseTree('N', ptree, null) } parseTree O() // syntactic category O { ptree = new parseTree(nextChar, null, null) nextChar = readChar() return new parseTree('O', ptree, null) }

12. Discussion #712/13 Recursive Descent Parser Execution for +*-748/92 E O + E O * E O - E N 7 E N 4 E N 8 E O / E N 9 E N 2 + * - 7 4 8 / 9 2 E OEE *NN 23

13. Discussion #713/13 Constructing Recursive Descent Parsers Can be applied to almost all grammars. Required property: By looking at the look-ahead symbol, we know which production to process next. Grammars with this single-symbol look-ahead property are called single-symbol look-ahead grammars. Alternatives: –Backtracking –LR(k)  k symbol look ahead –Make grammar have single-symbol look ahead.