From 9845e1cd9df09640a99ff3f42cf0539c6f21d1eb Mon Sep 17 00:00:00 2001
From: Mihaela Mihaljevic <mihaelamj@me.com>
Date: Mon, 8 Jun 2026 17:38:58 +0200
Subject: [PATCH] feat(relaxng): fold RNC ~ string concatenation

A string literal split with ~ ("a" ~ "b") is the single literal "ab",
the way long values and namespace URIs are wrapped across lines. The fold
applies wherever a string literal is read: value literals and the URIs of
namespace/datatypes/include/external declarations.
---
 Sources/Schema/RelaxNGCompactHelpers.swift | 15 +++++++++-
 Sources/Schema/RelaxNGCompactParser.swift  |  4 +--
 Tests/RelaxNGCompactConcatTests.swift      | 33 ++++++++++++++++++++++
 3 files changed, 49 insertions(+), 3 deletions(-)
 create mode 100644 Tests/RelaxNGCompactConcatTests.swift

diff --git a/Sources/Schema/RelaxNGCompactHelpers.swift b/Sources/Schema/RelaxNGCompactHelpers.swift
index 3d9c2e3..878d820 100644
--- a/Sources/Schema/RelaxNGCompactHelpers.swift
+++ b/Sources/Schema/RelaxNGCompactHelpers.swift
@@ -98,11 +98,24 @@ extension RNCParser {
 
     func expectString() -> String {
         if case let .string(value) = peek() { advance()
-            return value
+            return concatenated(value)
         }
         return ""
     }
 
+    /// Folds RNC string concatenation: `"a" ~ "b"` is the single literal `"ab"`,
+    /// the way long strings are split across lines. The leading string has been
+    /// consumed; this appends each following `~ "…"`.
+    func concatenated(_ value: String) -> String {
+        var result = value
+        while peek() == .symbol("~"), case let .string(next)? = peek(1) {
+            advance()
+            advance()
+            result += next
+        }
+        return result
+    }
+
     func strip(_ qualified: String) -> String {
         qualified.split(separator: ":").last.map(String.init) ?? qualified
     }
diff --git a/Sources/Schema/RelaxNGCompactParser.swift b/Sources/Schema/RelaxNGCompactParser.swift
index 0bd1f61..6d66d52 100644
--- a/Sources/Schema/RelaxNGCompactParser.swift
+++ b/Sources/Schema/RelaxNGCompactParser.swift
@@ -250,7 +250,7 @@ extension RNCParser {
         switch peek() {
         case let .string(value): advance()
             // A bare string literal is shorthand for a `token`-typed value.
-            return .value(PureXML.Schema.SimpleType(base: .token), value)
+            return .value(PureXML.Schema.SimpleType(base: .token), concatenated(value))
         case .symbol("("): advance()
             let pattern = parsePattern()
             expectSymbol(")")
@@ -269,7 +269,7 @@ extension RNCParser {
         var facets = PureXML.Schema.Facets()
         if peek() == .symbol("{") { parseParams(into: &facets) }
         if case let .string(value) = peek() { advance()
-            return .value(PureXML.Schema.SimpleType(base: type), value)
+            return .value(PureXML.Schema.SimpleType(base: type), concatenated(value))
         }
         return .data(PureXML.Schema.SimpleType(base: type, facets: facets))
     }
diff --git a/Tests/RelaxNGCompactConcatTests.swift b/Tests/RelaxNGCompactConcatTests.swift
new file mode 100644
index 0000000..b2e78f6
--- /dev/null
+++ b/Tests/RelaxNGCompactConcatTests.swift
@@ -0,0 +1,33 @@
+@testable import PureXML
+import Testing
+
+@Suite("RELAX NG compact string concatenation")
+struct RelaxNGCompactConcatTests {
+    private func validRNC(_ rnc: String, _ xml: String) throws -> Bool {
+        try PureXML.Schema.RelaxNG(compact: rnc).validate(xml)
+    }
+
+    @Test("A value literal split with ~ is concatenated")
+    func test_valueConcatenation() throws {
+        let rnc = "element greeting { \"hello \" ~ \"world\" }"
+        #expect(try validRNC(rnc, "<greeting>hello world</greeting>"))
+        #expect(try !validRNC(rnc, "<greeting>hello</greeting>"))
+    }
+
+    @Test("More than two segments concatenate left to right")
+    func test_multiSegment() throws {
+        let rnc = "element n { \"a\" ~ \"b\" ~ \"c\" }"
+        #expect(try validRNC(rnc, "<n>abc</n>"))
+        #expect(try !validRNC(rnc, "<n>ab</n>"))
+    }
+
+    @Test("A namespace URI may be assembled with ~")
+    func test_namespaceConcatenation() throws {
+        let rnc = """
+        namespace p = "urn:" ~ "example"
+        element root { element p:item { empty } }
+        """
+        #expect(try validRNC(rnc, "<root xmlns:p=\"urn:example\"><p:item/></root>"))
+        #expect(try !validRNC(rnc, "<root xmlns:p=\"urn:other\"><p:item/></root>"))
+    }
+}