dns: Add is_hostname() for RFC 1123 §2.1 Internet hostname validation (#2346)

Signed-off-by: Tushar Verma <tusharmyself06@gmail.com>

Author: vtushar06

.github/workflows/website-build.yml

@@ -29,6 +29,7 @@ jobs:
           -DSOURCEMETA_CORE_CRYPTO:BOOL=OFF
           -DSOURCEMETA_CORE_REGEX:BOOL=OFF
           -DSOURCEMETA_CORE_IP:BOOL=OFF
+          -DSOURCEMETA_CORE_DNS:BOOL=OFF
           -DSOURCEMETA_CORE_URI:BOOL=OFF
           -DSOURCEMETA_CORE_URITEMPLATE:BOOL=OFF
           -DSOURCEMETA_CORE_JSON:BOOL=OFF

.github/workflows/website-deploy.yml

@@ -39,6 +39,7 @@ jobs:
           -DSOURCEMETA_CORE_CRYPTO:BOOL=OFF
           -DSOURCEMETA_CORE_REGEX:BOOL=OFF
           -DSOURCEMETA_CORE_IP:BOOL=OFF
+          -DSOURCEMETA_CORE_DNS:BOOL=OFF
           -DSOURCEMETA_CORE_URI:BOOL=OFF
           -DSOURCEMETA_CORE_URITEMPLATE:BOOL=OFF
           -DSOURCEMETA_CORE_JSON:BOOL=OFF

CMakeLists.txt

@@ -17,6 +17,7 @@ option(SOURCEMETA_CORE_CRYPTO "Build the Sourcemeta Core Crypto library" ON)
 option(SOURCEMETA_CORE_CRYPTO_USE_SYSTEM_OPENSSL "Use system OpenSSL for the Sourcemeta Core Crypto library" OFF)
 option(SOURCEMETA_CORE_REGEX "Build the Sourcemeta Core Regex library" ON)
 option(SOURCEMETA_CORE_IP "Build the Sourcemeta Core IP library" ON)
+option(SOURCEMETA_CORE_DNS "Build the Sourcemeta Core DNS library" ON)
 option(SOURCEMETA_CORE_URI "Build the Sourcemeta Core URI library" ON)
 option(SOURCEMETA_CORE_URITEMPLATE "Build the Sourcemeta Core URI Template library" ON)
 option(SOURCEMETA_CORE_JSON "Build the Sourcemeta Core JSON library" ON)
@@ -118,6 +119,10 @@ if(SOURCEMETA_CORE_IP)
   add_subdirectory(src/core/ip)
 endif()
 
+if(SOURCEMETA_CORE_DNS)
+  add_subdirectory(src/core/dns)
+endif()
+
 if(SOURCEMETA_CORE_URI)
   add_subdirectory(src/core/uri)
 endif()
@@ -237,6 +242,10 @@ if(SOURCEMETA_CORE_TESTS)
     add_subdirectory(test/ip)
   endif()
 
+  if(SOURCEMETA_CORE_DNS)
+    add_subdirectory(test/dns)
+  endif()
+
   if(SOURCEMETA_CORE_URI)
     add_subdirectory(test/uri)
   endif()

config.cmake.in

@@ -15,6 +15,7 @@ if(NOT SOURCEMETA_CORE_COMPONENTS)
   list(APPEND SOURCEMETA_CORE_COMPONENTS crypto)
   list(APPEND SOURCEMETA_CORE_COMPONENTS regex)
   list(APPEND SOURCEMETA_CORE_COMPONENTS ip)
+  list(APPEND SOURCEMETA_CORE_COMPONENTS dns)
   list(APPEND SOURCEMETA_CORE_COMPONENTS uri)
   list(APPEND SOURCEMETA_CORE_COMPONENTS uritemplate)
   list(APPEND SOURCEMETA_CORE_COMPONENTS json)
@@ -61,6 +62,8 @@ foreach(component ${SOURCEMETA_CORE_COMPONENTS})
     include("${CMAKE_CURRENT_LIST_DIR}/sourcemeta_core_regex.cmake")
   elseif(component STREQUAL "ip")
     include("${CMAKE_CURRENT_LIST_DIR}/sourcemeta_core_ip.cmake")
+  elseif(component STREQUAL "dns")
+    include("${CMAKE_CURRENT_LIST_DIR}/sourcemeta_core_dns.cmake")
   elseif(component STREQUAL "uri")
     include("${CMAKE_CURRENT_LIST_DIR}/sourcemeta_core_ip.cmake")
     include("${CMAKE_CURRENT_LIST_DIR}/sourcemeta_core_uri.cmake")

src/core/dns/CMakeLists.txt

@@ -0,0 +1,6 @@
+sourcemeta_library(NAMESPACE sourcemeta PROJECT core NAME dns
+  SOURCES hostname.cc)
+
+if(SOURCEMETA_CORE_INSTALL)
+  sourcemeta_library_install(NAMESPACE sourcemeta PROJECT core NAME dns)
+endif()

src/core/dns/hostname.cc

@@ -0,0 +1,74 @@
+#include <sourcemeta/core/dns.h>
+
+namespace sourcemeta::core {
+
+// RFC 952 §B: let-dig = ALPHA / DIGIT
+// RFC 1123 §2.1: first character of a label is letter or digit
+static constexpr auto is_let_dig(const char character) -> bool {
+  return (character >= 'A' && character <= 'Z') ||
+         (character >= 'a' && character <= 'z') ||
+         (character >= '0' && character <= '9');
+}
+
+// RFC 952 §B: let-dig-hyp = ALPHA / DIGIT / "-"
+static constexpr auto is_let_dig_hyp(const char character) -> bool {
+  return is_let_dig(character) || character == '-';
+}
+
+auto is_hostname(const std::string_view value) -> bool {
+  // RFC 952 §B: <hname> requires at least one <name>
+  if (value.empty()) {
+    return false;
+  }
+
+  // RFC 1123 §2.1: SHOULD handle host names of up to 255 characters
+  if (value.size() > 255) {
+    return false;
+  }
+
+  std::string_view::size_type position{0};
+
+  while (position < value.size()) {
+    const auto label_start{position};
+
+    // RFC 1123 §2.1: first character is letter or digit
+    if (!is_let_dig(value[position])) {
+      return false;
+    }
+    position += 1;
+
+    while (position < value.size() && value[position] != '.') {
+      // RFC 952 §B: interior characters are let-dig-hyp
+      if (!is_let_dig_hyp(value[position])) {
+        return false;
+      }
+      position += 1;
+    }
+
+    const auto label_length{position - label_start};
+
+    // RFC 1123 §2.1: MUST handle host names of up to 63 characters (per label)
+    if (label_length > 63) {
+      return false;
+    }
+
+    // RFC 952 §B + ASSUMPTIONS: last character must not be a minus sign
+    if (value[position - 1] == '-') {
+      return false;
+    }
+
+    // If we stopped on a dot, there must be another label following it
+    if (position < value.size()) {
+      // value[position] == '.'
+      position += 1;
+      // Trailing dot: JSON Schema test suite requires rejection (TS d7+ #15)
+      if (position >= value.size()) {
+        return false;
+      }
+    }
+  }
+
+  return true;
+}
+
+} // namespace sourcemeta::core

src/core/dns/include/sourcemeta/core/dns.h

@@ -0,0 +1,46 @@
+#ifndef SOURCEMETA_CORE_DNS_H_
+#define SOURCEMETA_CORE_DNS_H_
+
+#ifndef SOURCEMETA_CORE_DNS_EXPORT
+#include <sourcemeta/core/dns_export.h>
+#endif
+
+#include <string_view> // std::string_view
+
+/// @defgroup dns DNS
+/// @brief DNS and hostname validation utilities.
+///
+/// This functionality is included as follows:
+///
+/// ```cpp
+/// #include <sourcemeta/core/dns.h>
+/// ```
+
+namespace sourcemeta::core {
+
+/// @ingroup dns
+/// Check whether the given string is a valid Internet host name per
+/// RFC 1123 Section 2.1, which relaxes the first-character rule of
+/// RFC 952 to allow either a letter or a digit. This matches the
+/// definition used by the JSON Schema `hostname` format. For example:
+///
+/// ```cpp
+/// #include <sourcemeta/core/dns.h>
+///
+/// #include <cassert>
+///
+/// assert(sourcemeta::core::is_hostname("www.example.com"));
+/// assert(sourcemeta::core::is_hostname("1host"));
+/// assert(!sourcemeta::core::is_hostname("-bad"));
+/// assert(!sourcemeta::core::is_hostname("example."));
+/// ```
+///
+/// This function implements RFC 1123 §2.1 (ASCII only). It does not
+/// perform A-label or Punycode decoding. Those belong to the separate
+/// `idn-hostname` format.
+SOURCEMETA_CORE_DNS_EXPORT
+auto is_hostname(const std::string_view value) -> bool;
+
+} // namespace sourcemeta::core
+
+#endif

test/dns/CMakeLists.txt

@@ -0,0 +1,5 @@
+sourcemeta_googletest(NAMESPACE sourcemeta PROJECT core NAME dns
+  SOURCES hostname_test.cc)
+
+target_link_libraries(sourcemeta_core_dns_unit
+  PRIVATE sourcemeta::core::dns)