From 35b25224ec197b45f887e5d71ed546ee2d382721 Mon Sep 17 00:00:00 2001
From: Juan Pablo Tosso <jptosso@gmail.com>
Date: Tue, 5 Nov 2024 11:14:43 +0000
Subject: [PATCH] feat(transaction): data type prediction for lazy predictions

---
 internal/corazarules/rule_match.go |  2 +
 types/value_metadata.go            | 84 ++++++++++++++++++++++++++++++
 2 files changed, 86 insertions(+)
 create mode 100644 types/value_metadata.go

diff --git a/internal/corazarules/rule_match.go b/internal/corazarules/rule_match.go
index 67f8f7dc1..2ccf5b988 100644
--- a/internal/corazarules/rule_match.go
+++ b/internal/corazarules/rule_match.go
@@ -29,6 +29,8 @@ type MatchData struct {
 	// Keeps track of the chain depth in which the data matched.
 	// Multiphase specific field
 	ChainLevel_ int
+	// Metadata of the matched data
+	Metadata_ types.DataMetadataList
 }
 
 var _ types.MatchData = (*MatchData)(nil)
diff --git a/types/value_metadata.go b/types/value_metadata.go
new file mode 100644
index 000000000..b02a74cb8
--- /dev/null
+++ b/types/value_metadata.go
@@ -0,0 +1,84 @@
+// Copyright 2023 Juan Pablo Tosso and the OWASP Coraza contributors
+// SPDX-License-Identifier: Apache-2.0
+package types
+
+import "unicode"
+
+// ValueMetadata is the type of metadata that a value can have.
+type ValueMetadata int
+
+const (
+	// ValueMetadataAlphanumeric represents an alphanumeric value.
+	ValueMetadataAlphanumeric ValueMetadata = iota
+	// ValueMetadataAscii represents an ASCII value.
+	ValueMetadataAscii
+	// ValueMetadataBase64 represents a base64 value.
+	ValueMetadataBase64
+	// ValueMetadataURI represents a URI value.
+	ValueMetadataURI
+	// ValueMetadataDomain represents a domain value.
+	ValueMetadataDomain
+	// ValueMetadataNumeric represents a numeric value, either integer or float.
+	ValueMetadataNumeric
+	// ValueMetadataBoolean represents a boolean value.
+	ValueMetadataBoolean
+	// ValueMetadataUnicode represents a unicode value.
+	ValueMetadataUnicode
+)
+
+// NewValueMetadata returns a new ValueMetadata from a string.
+func NewValueMetadata(metadata string) (ValueMetadata, bool) {
+	switch metadata {
+	case "alphanumeric":
+		return ValueMetadataAlphanumeric, true
+	case "ascii":
+		return ValueMetadataAscii, true
+	case "base64":
+		return ValueMetadataBase64, true
+	case "uri":
+		return ValueMetadataURI, true
+	case "domain":
+		return ValueMetadataDomain, true
+	case "numeric":
+		return ValueMetadataNumeric, true
+	case "boolean":
+		return ValueMetadataBoolean, true
+	case "unicode":
+		return ValueMetadataUnicode, true
+	}
+	return 0, false
+}
+
+// DataMetadataList is a list of ValueMetadata.
+type DataMetadataList struct {
+	metadata    map[ValueMetadata]bool
+	testedTypes []ValueMetadata
+}
+
+func (v *DataMetadataList) Test(data string, metadataType ValueMetadata) bool {
+	result, ok := v.metadata[metadataType]
+	if !ok {
+		// we do the analysis only once
+		switch metadataType {
+		case ValueMetadataAlphanumeric:
+			return v.testAlphanumeric(data)
+		default:
+			// this should not happen
+			return false
+		}
+	}
+	return result
+
+}
+
+func (v *DataMetadataList) testAlphanumeric(data string) bool {
+	res := true
+	for _, c := range data {
+		if !unicode.IsLetter(c) && !unicode.IsNumber(c) {
+			res = false
+			break
+		}
+	}
+	v.metadata[ValueMetadataAlphanumeric] = res
+	return res
+}