libutil: add simple NUL escaping/unescaping functions

Change-Id: Ia4cc7b8f1058439f312066422eebfaff1c2c0c6c

lix/libutil/strings.cc

@@ -366,4 +366,27 @@ std::string showBytes(uint64_t bytes)
     return fmt("%.2f MiB", bytes / (1024.0 * 1024.0));
 }
 
+std::string escapeNul(const std::string & in)
+{
+    using std::operator""sv;
+    return replaceStrings(replaceStrings(std::move(in), R"(\)", R"(\\)"), "\0"sv, R"(\0)");
+}
+
+std::string unescapeNul(const std::string & in)
+{
+    std::string result;
+    result.reserve(in.size());
+    bool inSequence = false;
+    for (auto c : in) {
+        if (inSequence) {
+            result += c == '0' ? '\0' : c;
+            inSequence = false;
+        } else if (c == '\\') {
+            inSequence = true;
+        } else {
+            result += c;
+        }
+    }
+    return result;
+}
 }

lix/libutil/strings.hh

@@ -141,7 +141,15 @@ inline std::string rewriteStrings(std::string s, const StringMap & rewrites)
     return Rewriter(rewrites)(s);
 }
 
-
+/**
+ * Escape NUL bytes and `\` in a string by replacing them with `\0` and `\\` respectively.
+ */
+std::string escapeNul(const std::string & in);
+/**
+ * Undo replacements done by `escapeNul`. Other escape sequences (eg `\n`) are replaced by
+ * the second character in the sequence (eg `n`), `\` at the end of the string is dropped.
+ */
+std::string unescapeNul(const std::string & in);
 
 /**
  * Parse a string into an integer.

tests/unit/libutil/tests.cc

@@ -817,4 +817,21 @@ namespace nix {
         ASSERT_TRUE(e.is<std::invalid_argument>());
         ASSERT_NE(e.as<std::invalid_argument>(), nullptr);
     }
+
+    TEST(NulEscaping, function)
+    {
+        for (int c1 = CHAR_MIN; c1 < CHAR_MAX; c1++) {
+            for (int c2 = CHAR_MIN; c2 < CHAR_MAX; c2++) {
+                auto sequence = std::format("{}{}", char(c1), char(c2));
+                for (auto str : {sequence, "a" + sequence + "b"}) {
+                    ASSERT_EQ(unescapeNul(escapeNul(str)), str);
+                }
+            }
+        }
+
+        // unknown escapes are character identity for simplicity of use
+        ASSERT_EQ(unescapeNul("\\a"), "a");
+        // escape of end-of-string is dropped silently
+        ASSERT_EQ(unescapeNul("a\\"), "a");
+    }
 }