regex/postfixNode.go

package main

import (
	"slices"
)

type NodeType int

// This is a list of the possible node types
const (
	CHARACTER NodeType = iota
	PIPE
	CONCATENATE
	KLEENE
	QUESTION
	PLUS
	ASSERTION
	LPAREN
	RPAREN
)

var INFINITE_REPS int = -1 // Represents infinite reps eg. the end range in {5,}
// This represents a node in the postfix representation of the expression
type postfixNode struct {
	nodetype  NodeType
	contents  []rune // Contents of the node - the length of this would only be >1 if the node represents a character class
	startReps int    // Minimum number of times the node should be repeated - used with numeric specifiers
	endReps   int    // Maximum number of times the node should be repeated - used with numeric specifiers
	isDot     bool   // Whether or not the current node represents a 'dot' metacharacter
}

// Creates a new escaped node - the given character is assumed to have been preceded by a backslash
func newEscapedNode(c rune) postfixNode {
	toReturn := postfixNode{}
	toReturn.startReps = 1
	toReturn.endReps = 1
	switch c {
	case 's': // Whitespace
		toReturn.nodetype = CHARACTER
		toReturn.contents = append(toReturn.contents, whitespaceChars...)
	case 'S': // Non-whitespace - I am doing this in a fancy way, generating all dot characters, then removing whitespace characters from it
		toReturn.nodetype = CHARACTER
		toReturn.contents = append(toReturn.contents, slices.DeleteFunc(dotChars(), func(r rune) bool {
			return slices.Contains(whitespaceChars, r)
		})...)
	case 'd': // Digits
		toReturn.nodetype = CHARACTER
		toReturn.contents = append(toReturn.contents, digitChars...)
	case 'D': // Non-digits - same fancy way as 'S'
		toReturn.nodetype = CHARACTER
		toReturn.contents = append(toReturn.contents, slices.DeleteFunc(dotChars(), func(r rune) bool {
			return slices.Contains(digitChars, r)
		})...)
	case 'w': // word character
		toReturn.nodetype = CHARACTER
		toReturn.contents = append(toReturn.contents, wordChars...)
	case 'W': // Non-word character - same fancy way as 'S' and 'D'
		toReturn.nodetype = CHARACTER
		toReturn.contents = append(toReturn.contents, slices.DeleteFunc(dotChars(), func(r rune) bool {
			return slices.Contains(wordChars, r)
		})...)
	case 'b', 'B':
		toReturn.nodetype = ASSERTION
		toReturn.contents = append(toReturn.contents, c)
	case 'n': // Newline character
		toReturn.nodetype = CHARACTER
		toReturn.contents = append(toReturn.contents, '\n')
	default: // None of the above - append it as a regular character
		toReturn.nodetype = CHARACTER
		toReturn.contents = append(toReturn.contents, c)
	}
	return toReturn
}

// Creates and returns a postfixNode based on the given contents
func newPostfixNode(contents ...rune) postfixNode {
	if len(contents) < 1 {
		panic("Empty node.")
	}
	to_return := postfixNode{}
	to_return.startReps = 1
	to_return.endReps = 1
	if len(contents) > 1 { // If the node has more than element, it must be a character class - the type must be CHARACTER
		to_return.nodetype = CHARACTER
		to_return.contents = contents
	} else { // Node has one element, could be anything
		switch contents[0] {
		case '+':
			to_return.nodetype = PLUS
		case '?':
			to_return.nodetype = QUESTION
		case '*':
			to_return.nodetype = KLEENE
		case '|':
			to_return.nodetype = PIPE
		case CONCAT:
			to_return.nodetype = CONCATENATE
		case '^', '$':
			to_return.nodetype = ASSERTION
		case '(':
			to_return.nodetype = LPAREN
		case ')':
			to_return.nodetype = RPAREN
		default:
			to_return.nodetype = CHARACTER
		}
		to_return.contents = append(to_return.contents, contents...)
	}
	return to_return
}

// Creates and returns a postfixNode representing the 'dot' metacharacter.
func newPostfixDotNode() postfixNode {
	toReturn := postfixNode{}
	toReturn.startReps = 1
	toReturn.endReps = 1
	toReturn.nodetype = CHARACTER
	toReturn.isDot = true
	toReturn.contents = []rune{ANY_CHAR}
	return toReturn
}

// Creates a character node, regardless of the contents
func newPostfixCharNode(contents ...rune) postfixNode {
	toReturn := postfixNode{}
	toReturn.startReps = 1
	toReturn.endReps = 1
	toReturn.nodetype = CHARACTER
	toReturn.contents = append(toReturn.contents, contents...)
	return toReturn
}