Started implementing backreferences (octal values should now be prefaced with \0)

regex/postfixNode.go

@@ -1,6 +1,8 @@
 package regex
 
-import "fmt"
+import (
+	"fmt"
+)
 
 type nodeType int
 
@@ -20,6 +22,7 @@ const (
 	assertionNode
 	lparenNode
 	rparenNode
+	backreferenceNode
 )
 
 // Helper constants for lookarounds
@@ -31,15 +34,16 @@ const lookbehind = -1
 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
-	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
-	allChars       bool          // Whether or not the current node represents all characters (eg. dot metacharacter)
-	except         []postfixNode // For inverted character classes, we match every unicode character _except_ a few. In this case, allChars is true and the exceptions are placed here.
-	lookaroundSign int           // ONLY USED WHEN nodetype == ASSERTION. Whether we have a positive or negative lookaround.
-	lookaroundDir  int           // Lookbehind or lookahead
-	nodeContents   []postfixNode // ONLY USED WHEN nodetype == CHARCLASS. Holds all the nodes inside the given CHARCLASS node.
+	nodetype        nodeType
+	contents        []rune        // Contents of the node
+	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
+	allChars        bool          // Whether or not the current node represents all characters (eg. dot metacharacter)
+	except          []postfixNode // For inverted character classes, we match every unicode character _except_ a few. In this case, allChars is true and the exceptions are placed here.
+	lookaroundSign  int           // ONLY USED WHEN nodetype == ASSERTION. Whether we have a positive or negative lookaround.
+	lookaroundDir   int           // Lookbehind or lookahead
+	nodeContents    []postfixNode // ONLY USED WHEN nodetype == CHARCLASS. Holds all the nodes inside the given CHARCLASS node.
+	referencedGroup int           // ONLY USED WHEN nodetype == backreferenceNode. Holds the group which this one refers to. After parsing is done, the expression will be rewritten eg. (a)\1 will become (a)(a). So the return value of ShuntingYard() shouldn't contain a backreferenceNode.
 }
 
 // Converts the given list of postfixNodes to one node of type CHARCLASS.
@@ -208,3 +212,44 @@ func newPostfixCharNode(contents ...rune) postfixNode {
 	toReturn.contents = append(toReturn.contents, contents...)
 	return toReturn
 }
+
+// newPostfixBackreferenceNode creates and returns a backreference node, referring to the given group
+func newPostfixBackreferenceNode(referred int) postfixNode {
+	toReturn := postfixNode{}
+	toReturn.startReps = 1
+	toReturn.endReps = 1
+	toReturn.nodetype = backreferenceNode
+	toReturn.referencedGroup = referred
+	return toReturn
+}
+
+// rewriteBackreferences rewrites any backreferences in the given postfixNode slice, into their respective groups.
+// It stores the relation in a map, and returns it as the second return value.
+// It uses parenIndices to determine where a group starts and ends in nodes.
+// For example, \1(a) will be rewritten into (a)(a), and 1 -> 2 will be the hashmap value.
+// It returns an error if a backreference points to an invalid group.
+// func rewriteBackreferences(nodes []postfixNode, parenIndices []Group) ([]postfixNode, map[int]int, error) {
+// 	rtv := make([]postfixNode, 0)
+// 	referMap := make(map[int]int)
+// 	numGroups := 0
+// 	groupIncrement := 0 // If we have a backreference before the group its referring to, then the group its referring to will have its group number incremented.
+// 	for i, node := range nodes {
+// 		if node.nodetype == backreferenceNode {
+// 			if node.referencedGroup >= len(parenIndices) {
+// 				return nil, nil, fmt.Errorf("invalid backreference")
+// 			}
+// 			rtv = slices.Concat(rtv, nodes[parenIndices[node.referencedGroup].StartIdx:parenIndices[node.referencedGroup].EndIdx+1]) // Add all the nodes in the group to rtv
+// 			numGroups += 1
+// 			if i < parenIndices[node.referencedGroup].StartIdx {
+// 				groupIncrement += 1
+// 			}
+// 			referMap[numGroups] = node.referencedGroup + groupIncrement
+// 		} else {
+// 			rtv = append(rtv, node)
+// 			if node.nodetype == lparenNode {
+// 				numGroups += 1
+// 			}
+// 		}
+// 	}
+// 	return rtv, referMap, nil
+// }