expr/parser.go

// Copyright (c) 2024 Celestino Amoroso (celestino.amoroso@gmail.com).
// All rights reserved.

// parser.go
package expr

import (
	"errors"
	"fmt"
)

//-------- parser

type parser struct {
	ctx ExprContext
}

func NewParser(ctx ExprContext) (p *parser) {
	p = &parser{
		ctx: ctx,
	}
	return p
}

func (self *parser) parseFuncCall(scanner *scanner, allowVarRef bool, tk *Token) (tree *term, err error) {
	// name, _ := tk.Value.(string)
	// funcObj := self.ctx.GetFuncInfo(name)
	// if funcObj == nil {
	// 	err = fmt.Errorf("unknown function %s()", name)
	// 	return
	// }
	// maxArgs := funcObj.MaxArgs()
	// if maxArgs < 0 {
	// 	maxArgs = funcObj.MinArgs() + 10
	// }
	// args := make([]*term, 0, maxArgs)
	args := make([]*term, 0, 10)
	lastSym := SymUnknown
	for lastSym != SymClosedRound && lastSym != SymEos {
		var subTree *ast
		if subTree, err = self.parseItem(scanner, allowVarRef, SymComma, SymClosedRound); err == nil {
			if subTree.root != nil {
				args = append(args, subTree.root)
			}
		} else {
			break
		}
		lastSym = scanner.Previous().Sym
	}
	if err == nil {
		// TODO Check arguments
		if lastSym != SymClosedRound {
			err = errors.New("unterminate arguments list")
		} else {
			tree = newFuncCallTerm(tk, args)
		}
	}
	return
}

// func (self *parser) parseFuncDef(scanner *scanner) (tree *term, err error) {
// 	// Example: "add = func(x,y) {x+y}
// 	var body *ast
// 	args := make([]*term, 0)
// 	tk := scanner.Next()
// 	for tk.Sym != SymClosedRound && tk.Sym != SymEos {
// 		if tk.Sym == SymIdentifier {
// 			t := newTerm(tk, nil)
// 			args = append(args, t)
// 		} else {
// 			err = tk.Errorf("invalid param %q, variable identifier expected", tk.source)
// 			break
// 		}
// 		tk = scanner.Next()
// 	}
// 	if err == nil && tk.Sym != SymClosedRound {
// 		err = tk.Errorf("unterminate function params list")
// 	}
// 	if err == nil {
// 		tk = scanner.Next()
// 		if tk.Sym == SymOpenBrace {
// 			body, err = self.parseGeneral(scanner, true, true, SymClosedBrace)
// 		}
// 	}
// 	if err == nil {
// 		// TODO Check arguments
// 		if scanner.Previous().Sym != SymClosedBrace {
// 			err = scanner.Previous().Errorf("not properly terminated function body")
// 		} else {
// 			tk = scanner.makeValueToken(SymExpression, "", body)
// 			tree = newFuncDefTerm(tk, args)
// 		}
// 	}
// 	return
// }

func (self *parser) parseFuncDef(scanner *scanner) (tree *term, err error) {
	// Example: "add = func(x,y) {x+y}
	var body *ast
	args := make([]*term, 0)
	lastSym := SymUnknown
	itemExpected := false
	tk := scanner.Previous()
	for lastSym != SymClosedRound && lastSym != SymEos {
		var subTree *ast
		if subTree, err = self.parseItem(scanner, true, SymComma, SymClosedRound); err == nil {
			if subTree.root != nil {
				if subTree.root.symbol() == SymIdentifier {
					args = append(args, subTree.root)
				} else {
					err = tk.Errorf("exptected identifier, got %q", subTree.root)
				}
			} else if itemExpected {
				prev := scanner.Previous()
				err = prev.Errorf("expected function parameter, got %q", prev)
				break
			}
		} else {
			break
		}
		lastSym = scanner.Previous().Sym
		itemExpected = lastSym == SymComma
	}

	if err == nil && lastSym != SymClosedRound {
		err = tk.Errorf("unterminated function parameters list")
	}
	if err == nil {
		tk = scanner.Next()
		if tk.Sym == SymOpenBrace {
			body, err = self.parseGeneral(scanner, true, true, SymClosedBrace)
		}
	}
	if err == nil {
		// TODO Check arguments
		if scanner.Previous().Sym != SymClosedBrace {
			err = scanner.Previous().Errorf("not properly terminated function body")
		} else {
			tk = scanner.makeValueToken(SymExpression, "", body)
			tree = newFuncDefTerm(tk, args)
		}
	}
	return
}

func (self *parser) parseList(scanner *scanner, allowVarRef bool) (subtree *term, err error) {
	args := make([]*term, 0)
	lastSym := SymUnknown
	itemExpected := false
	for lastSym != SymClosedSquare && lastSym != SymEos {
		var subTree *ast
		if subTree, err = self.parseItem(scanner, allowVarRef, SymComma, SymClosedSquare); err == nil {
			if subTree.root != nil {
				args = append(args, subTree.root)
			} else if itemExpected {
				prev := scanner.Previous()
				err = prev.Errorf("expected list item, got %q", prev)
				break
			}
		} else {
			break
		}
		lastSym = scanner.Previous().Sym
		itemExpected = lastSym == SymComma
	}
	if err == nil {
		// TODO Check arguments
		if lastSym != SymClosedSquare {
			err = scanner.Previous().Errorf("unterminate items list")
		} else {
			subtree = newListTerm(args)
		}
	}
	return
}

func (self *parser) parseIterDef(scanner *scanner, allowVarRef bool) (subtree *term, err error) {
	var ds *term
	tk := scanner.Previous()
	args := make([]*term, 0)
	lastSym := SymUnknown
	dsExpected := true
	itemExpected := false
	for lastSym != SymClosedRound && lastSym != SymEos {
		var subTree *ast
		if subTree, err = self.parseItem(scanner, allowVarRef, SymComma, SymClosedRound); err == nil {
			if subTree.root != nil {
				if dsExpected {
					if sym := subTree.root.symbol(); sym == SymDict || sym == SymIdentifier {
						ds = subTree.root
					} else {
						err = subTree.root.Errorf("data-source dictionary expected, got %q", subTree.root.source())
					}
					dsExpected = false
				} else {
					args = append(args, subTree.root)
				}
			} else if itemExpected {
				prev := scanner.Previous()
				err = prev.Errorf("expected iterator argument, got %q", prev)
				break
			}
		} else {
			break
		}
		lastSym = scanner.Previous().Sym
		itemExpected = lastSym == SymComma
	}
	if err == nil {
		// TODO Check arguments
		if lastSym != SymClosedRound {
			err = scanner.Previous().Errorf("unterminate iterator param list")
		} else if ds != nil {
			subtree = newIteratorTerm(tk, ds, args)
		} else {
			tk.Errorf("missing data-source param")
		}
	}
	return
}

func (self *parser) parseDictKey(scanner *scanner, allowVarRef bool) (key any, err error) {
	tk := scanner.Next()
	if tk.Sym == SymError {
		err = tk.Error()
		return
	}
	if tk.Sym == SymClosedBrace || tk.Sym == SymEos {
		return
	}
	if tk.Sym == SymInteger || tk.Sym == SymString {
		tkSep := scanner.Next()
		if tkSep.Sym != SymColon {
			err = tkSep.Errorf("expected \":\", got %q", tkSep)
		} else {
			key = tk.Value
		}
	} else {
		err = tk.Errorf("expected dictionary key or closed brace, got %q", tk)
	}
	return
}

func (self *parser) parseDictionary(scanner *scanner, allowVarRef bool) (subtree *term, err error) {
	args := make(map[any]*term, 0)
	lastSym := SymUnknown
	itemExpected := false
	for lastSym != SymClosedBrace && lastSym != SymEos {
		var subTree *ast
		var key any
		if key, err = self.parseDictKey(scanner, allowVarRef); err != nil {
			break
		} else if key == nil {
			tk := scanner.Previous()
			lastSym = tk.Sym
			if itemExpected {
				err = tk.Errorf("expected dictionary key, got %q", tk)
			}
			break
		}
		if subTree, err = self.parseItem(scanner, allowVarRef, SymComma, SymClosedBrace); err == nil {
			if subTree.root != nil {
				args[key] = subTree.root
			} else if key != nil {
				prev := scanner.Previous()
				err = prev.Errorf("expected dictionary value, got %q", prev)
				break
			}
		} else {
			break
		}
		lastSym = scanner.Previous().Sym
		itemExpected = lastSym == SymComma
	}
	if err == nil {
		// TODO Check arguments
		if lastSym != SymClosedBrace {
			err = scanner.Previous().Errorf("unterminated dictionary")
		} else {
			subtree = newDictTerm(args)
		}
	}
	return
}

func (self *parser) parseSelectorCase(scanner *scanner, allowVarRef bool, defaultCase bool) (caseTerm *term, err error) {
	var filterList *term
	var caseExpr *ast
	tk := scanner.Next()
	startRow := tk.row
	startCol := tk.col
	if tk.Sym == SymOpenSquare {
		if defaultCase {
			err = tk.Errorf("case list in default clause")
			return
		}
		if filterList, err = self.parseList(scanner, allowVarRef); err != nil {
			return
		}
		tk = scanner.Next()
		startRow = tk.row
		startCol = tk.col
	} else if !defaultCase {
		filterList = newListTerm(make([]*term, 0))
	}

	if tk.Sym == SymOpenBrace {
		if caseExpr, err = self.parseGeneral(scanner, true, allowVarRef, SymClosedBrace); err != nil {
			return
		}
	} else {
		err = tk.Errorf("selector-case expected, got %q", tk.source)
	}

	if err == nil {
		caseTerm = newSelectorCaseTerm(startRow, startCol, filterList, caseExpr)
	}
	return
}

func addSelectorCase(selectorTerm, caseTerm *term) {
	if len(selectorTerm.children) < 2 {
		caseListTerm := newListTermA(caseTerm)
		selectorTerm.children = append(selectorTerm.children, caseListTerm)
	} else {
		caseListTerm := selectorTerm.children[1]
		caseList, _ := caseListTerm.value().([]*term)
		caseList = append(caseList, caseTerm)
		caseListTerm.tk.Value = caseList
	}
	caseTerm.parent = selectorTerm
}

func (self *parser) parseSelector(scanner *scanner, tree *ast, allowVarRef bool) (selectorTerm *term, err error) {
	var caseTerm *term
	tk := scanner.makeToken(SymSelector, '?')
	if selectorTerm, err = tree.addToken2(tk); err != nil {
		return
	}

	if caseTerm, err = self.parseSelectorCase(scanner, allowVarRef, false); err == nil {
		addSelectorCase(selectorTerm, caseTerm)
	}
	return
}

func (self *parser) parseItem(scanner *scanner, allowVarRef bool, termSymbols ...Symbol) (tree *ast, err error) {
	return self.parseGeneral(scanner, false, allowVarRef, termSymbols...)
}

func (self *parser) Parse(scanner *scanner, termSymbols ...Symbol) (tree *ast, err error) {
	return self.parseGeneral(scanner, true, false, termSymbols...)
}

func (self *parser) parseGeneral(scanner *scanner, allowForest bool, allowVarRef bool, termSymbols ...Symbol) (tree *ast, err error) {
	var selectorTerm *term = nil
	var currentTerm *term = nil
	var tk *Token
	tree = NewAst()
	firstToken := true
	lastSym := SymUnknown
	for tk = scanner.Next(); err == nil && tk != nil && !tk.IsTerm(termSymbols); tk = scanner.Next() {
		if tk.Sym == SymComment {
			continue
		}

		if tk.Sym == SymSemiColon {
			if allowForest {
				tree.ToForest()
				firstToken = true
				continue
			} else {
				err = tk.Errorf(`unexpected token %q, expected ",", "]", or ")"`, tk.source)
				break
			}
		}

		//fmt.Println("Token:", tk)
		if firstToken {
			if tk.Sym == SymMinus {
				tk.Sym = SymChangeSign
			} else if tk.Sym == SymPlus {
				tk.Sym = SymUnchangeSign
			}
			firstToken = false
		}

		switch tk.Sym {
		case SymOpenRound:
			var subTree *ast
			if subTree, err = self.parseGeneral(scanner, false, allowVarRef, SymClosedRound); err == nil {
				subTree.root.priority = priValue
				err = tree.addTerm(subTree.root)
				currentTerm = subTree.root
			}
		case SymFuncCall:
			var funcCallTerm *term
			if funcCallTerm, err = self.parseFuncCall(scanner, allowVarRef, tk); err == nil {
				err = tree.addTerm(funcCallTerm)
				currentTerm = funcCallTerm
			}
		case SymOpenSquare:
			var listTerm *term
			if listTerm, err = self.parseList(scanner, allowVarRef); err == nil {
				err = tree.addTerm(listTerm)
				currentTerm = listTerm
			}
		case SymOpenBrace:
			var mapTerm *term
			if mapTerm, err = self.parseDictionary(scanner, allowVarRef); err == nil {
				err = tree.addTerm(mapTerm)
				currentTerm = mapTerm
			}
		case SymEqual:
			if err = checkPrevSymbol(lastSym, SymIdentifier, tk); err == nil {
				currentTerm, err = tree.addToken2(tk)
			}
		case SymFuncDef:
			var funcDefTerm *term
			if funcDefTerm, err = self.parseFuncDef(scanner); err == nil {
				err = tree.addTerm(funcDefTerm)
				currentTerm = funcDefTerm
			}
		case SymDollarRound:
			var iterDefTerm *term
			if iterDefTerm, err = self.parseIterDef(scanner, allowVarRef); err == nil {
				err = tree.addTerm(iterDefTerm)
				currentTerm = iterDefTerm
			}
		case SymIdentifier:
			if tk.source[0] == '@' && !allowVarRef {
				err = tk.Errorf("variable references are not allowed in top level expressions: %q", tk.source)
			} else {
				currentTerm, err = tree.addToken2(tk)
			}
		case SymQuestion:
			if selectorTerm, err = self.parseSelector(scanner, tree, allowVarRef); err == nil {
				currentTerm = selectorTerm
			}
		case SymColon, SymDoubleColon:
			var caseTerm *term
			if selectorTerm == nil {
				err = tk.Errorf("selector-case outside of a selector context")
			} else if caseTerm, err = self.parseSelectorCase(scanner, allowVarRef, tk.Sym == SymDoubleColon); err == nil {
				addSelectorCase(selectorTerm, caseTerm)
				currentTerm = caseTerm
				if tk.Sym == SymDoubleColon {
					selectorTerm = nil
				}
			}
		default:
			currentTerm, err = tree.addToken2(tk)
		}

		if currentTerm != nil && currentTerm.tk.Sym != SymSelector && currentTerm.parent != nil && currentTerm.parent.tk.Sym != SymSelector {
			selectorTerm = nil

		}
		lastSym = tk.Sym
	}
	if err == nil {
		err = tk.Error()
	}
	return
}

func checkPrevSymbol(lastSym, wantedSym Symbol, tk *Token) (err error) {
	if lastSym != wantedSym {
		err = fmt.Errorf(`assign operator (%q) must be preceded by a variable`, tk.source)
	}
	return
}