From 4eaeda6604e2f8977728f14415fe92350d047970 Mon Sep 17 00:00:00 2001
From: Robert Hensing <robert@roberthensing.nl>
Date: Tue, 12 Dec 2023 17:43:54 +0100
Subject: [PATCH] isValidSchemeName: Use regex

As requested by Eelco Dolstra. I think it used to be simpler.
---
 src/libutil/url.cc        | 15 +++------------
 tests/unit/libutil/url.cc |  5 +++++
 2 files changed, 8 insertions(+), 12 deletions(-)

diff --git a/src/libutil/url.cc b/src/libutil/url.cc
index e9acd67d0..152c06d8e 100644
--- a/src/libutil/url.cc
+++ b/src/libutil/url.cc
@@ -3,7 +3,6 @@
 #include "util.hh"
 #include "split.hh"
 #include "canon-path.hh"
-#include "string.hh"
 
 namespace nix {
 
@@ -187,17 +186,9 @@ std::string fixGitURL(const std::string & url)
 // https://www.rfc-editor.org/rfc/rfc3986#section-3.1
 bool isValidSchemeName(std::string_view s)
 {
-    if (s.empty()) return false;
-    if (!isASCIIAlpha(s[0])) return false;
-    for (auto c : s.substr(1)) {
-        if (isASCIIAlpha(c)) continue;
-        if (isASCIIDigit(c)) continue;
-        if (c == '+') continue;
-        if (c == '-') continue;
-        if (c == '.') continue;
-        return false;
-    }
-    return true;
+    static std::regex regex(schemeNameRegex, std::regex::ECMAScript);
+
+    return std::regex_match(s.begin(), s.end(), regex, std::regex_constants::match_default);
 }
 
 }
diff --git a/tests/unit/libutil/url.cc b/tests/unit/libutil/url.cc
index 09fa4e218..7d08f467e 100644
--- a/tests/unit/libutil/url.cc
+++ b/tests/unit/libutil/url.cc
@@ -360,6 +360,11 @@ TEST(nix, isValidSchemeName) {
     ASSERT_FALSE(isValidSchemeName(".file"));
     ASSERT_FALSE(isValidSchemeName("-file"));
     ASSERT_FALSE(isValidSchemeName("1file"));
+    // regex ok?
+    ASSERT_FALSE(isValidSchemeName("\nhttp"));
+    ASSERT_FALSE(isValidSchemeName("\nhttp\n"));
+    ASSERT_FALSE(isValidSchemeName("http\n"));
+    ASSERT_FALSE(isValidSchemeName("http "));
 }
 
 }