Autodetect configuration file format if is not .xml, .yml or .yaml

src/Common/Config/ConfigProcessor.cpp

@@ -138,9 +138,14 @@ static Node * getRootNode(Document * document)
     return XMLUtils::getRootNode(document);
 }
 
+static size_t firstNonWhitespacePos(const std::string & s)
+{
+    return s.find_first_not_of(" \t\n\r");
+}
+
 static bool allWhitespace(const std::string & s)
 {
-    return s.find_first_not_of(" \t\n\r") == std::string::npos;
+    return firstNonWhitespacePos(s) == std::string::npos;
 }
 
 static void deleteAttributesRecursive(Node * root)
@@ -622,6 +627,52 @@ ConfigProcessor::Files ConfigProcessor::getConfigMergeFiles(const std::string &
     return files;
 }
 
+XMLDocumentPtr ConfigProcessor::parseConfig(const std::string & config_path)
+{
+    fs::path p(config_path);
+    std::string extension = p.extension();
+    boost::algorithm::to_lower(extension);
+
+    if (extension == ".xml")
+        return dom_parser.parse(config_path);
+    else if (extension == ".yaml" || extension == ".yml")
+        return YAMLParser::parse(config_path);
+    else
+    {
+        /// Suppose non regular file parsed as XML, such as pipe: /dev/fd/X (regardless it has .xml extension or not)
+        if (!fs::is_regular_file(config_path))
+            return dom_parser.parse(config_path);
+
+        /// If the regular file begins with < it might be XML, otherwise it might be YAML.
+        bool maybe_xml = false;
+        {
+            std::ifstream file(config_path);
+            if (!file.is_open())
+                throw Exception(ErrorCodes::CANNOT_LOAD_CONFIG, "Unknown format of '{}' config", config_path);
+
+            std::string line;
+            while (std::getline(file, line))
+            {
+                const size_t pos = firstNonWhitespacePos(line);
+                if (pos == std::string::npos)
+                    continue;
+
+                if (pos < line.size() && '<' == line[pos])
+                {
+                    maybe_xml = true;
+                }
+
+                break;
+            }
+        }
+
+        if (maybe_xml)
+            return dom_parser.parse(config_path);
+        else
+            return YAMLParser::parse(config_path);
+    }
+}
+
 XMLDocumentPtr ConfigProcessor::processConfig(
     bool * has_zk_includes,
     zkutil::ZooKeeperNodeCache * zk_node_cache,
@@ -633,23 +684,7 @@ XMLDocumentPtr ConfigProcessor::processConfig(
 
     if (fs::exists(path))
     {
-        fs::path p(path);
+        config = parseConfig(path);
-
-        std::string extension = p.extension();
-        boost::algorithm::to_lower(extension);
-
-        if (extension == ".yaml" || extension == ".yml")
-        {
-            config = YAMLParser::parse(path);
-        }
-        else if (extension == ".xml" || extension == ".conf" || extension.empty())
-        {
-            config = dom_parser.parse(path);
-        }
-        else
-        {
-            throw Exception(ErrorCodes::CANNOT_LOAD_CONFIG, "Unknown format of '{}' config", path);
-        }
     }
     else
     {
@@ -673,20 +708,7 @@ XMLDocumentPtr ConfigProcessor::processConfig(
             LOG_DEBUG(log, "Merging configuration file '{}'.", merge_file);
 
             XMLDocumentPtr with;
-
+            with = parseConfig(merge_file);
-            fs::path p(merge_file);
-            std::string extension = p.extension();
-            boost::algorithm::to_lower(extension);
-
-            if (extension == ".yaml" || extension == ".yml")
-            {
-                with = YAMLParser::parse(merge_file);
-            }
-            else
-            {
-                with = dom_parser.parse(merge_file);
-            }
-
             if (!merge(config, with))
             {
                 LOG_DEBUG(log, "Merging bypassed - configuration file '{}' doesn't belong to configuration '{}' - merging root node name '{}' doesn't match '{}'",
@@ -730,19 +752,7 @@ XMLDocumentPtr ConfigProcessor::processConfig(
         {
             LOG_DEBUG(log, "Including configuration file '{}'.", include_from_path);
 
-            fs::path p(include_from_path);
+            include_from = parseConfig(include_from_path);
-            std::string extension = p.extension();
-            boost::algorithm::to_lower(extension);
-
-            if (extension == ".yaml" || extension == ".yml")
-            {
-                include_from = YAMLParser::parse(include_from_path);
-            }
-            else
-            {
-                include_from = dom_parser.parse(include_from_path);
-            }
-
             contributing_files.push_back(include_from_path);
         }
 

src/Common/Config/ConfigProcessor.h

@@ -65,6 +65,8 @@ public:
         zkutil::ZooKeeperNodeCache * zk_node_cache = nullptr,
         const zkutil::EventPtr & zk_changed_event = nullptr);
 
+    XMLDocumentPtr parseConfig(const std::string & config_path);
+
     /// These configurations will be used if there is no configuration file.
     static void registerEmbeddedConfig(std::string name, std::string_view content);
 

tests/queries/0_stateless/01889_clickhouse_client_config_format.reference

@@ -12,5 +12,9 @@ yml
 2
 yaml
 2
-ini
+autodetect xml (with leading whitespaces)
-Code: 347. Unknown format of '/config_default.ini' config. (CANNOT_LOAD_CONFIG)
+2
+autodetect xml (non leading whitespaces)
+2
+autodetect yaml
+2

tests/queries/0_stateless/01889_clickhouse_client_config_format.sh

@@ -12,7 +12,9 @@ XML_config=$CLICKHOUSE_TMP/config_$CLICKHOUSE_DATABASE.XML
 conf_config=$CLICKHOUSE_TMP/config_$CLICKHOUSE_DATABASE.conf
 yml_config=$CLICKHOUSE_TMP/config_$CLICKHOUSE_DATABASE.yml
 yaml_config=$CLICKHOUSE_TMP/config_$CLICKHOUSE_DATABASE.yaml
-ini_config=$CLICKHOUSE_TMP/config_$CLICKHOUSE_DATABASE.ini
+autodetect_xml_with_leading_whitespace_config=$CLICKHOUSE_TMP/config_$CLICKHOUSE_DATABASE.config
+autodetect_xml_non_leading_whitespace_config=$CLICKHOUSE_TMP/config_$CLICKHOUSE_DATABASE.cfg
+autodetect_yaml_config=$CLICKHOUSE_TMP/config_$CLICKHOUSE_DATABASE.properties
 
 function cleanup()
 {
@@ -22,7 +24,9 @@ function cleanup()
     rm "${conf_config:?}"
     rm "${yml_config:?}"
     rm "${yaml_config:?}"
-    rm "${ini_config:?}"
+    rm "${autodetect_xml_with_leading_whitespace_config:?}"
+    rm "${autodetect_xml_non_leading_whitespace_config:?}"
+    rm "${autodetect_yaml_config:?}"
 }
 trap cleanup EXIT
 
@@ -52,9 +56,19 @@ EOL
 cat > "$yaml_config" <<EOL
 max_threads: 2
 EOL
-cat > "$ini_config" <<EOL
+cat > "$autodetect_xml_with_leading_whitespace_config" <<EOL
-[config]
+
-max_threads=2
+    <config>
+        <max_threads>2</max_threads>
+    </config>
+EOL
+cat > "$autodetect_xml_non_leading_whitespace_config" <<EOL
+<config>
+    <max_threads>2</max_threads>
+</config>
+EOL
+cat > "$autodetect_yaml_config" <<EOL
+max_threads: 2
 EOL
 
 echo 'default'
@@ -74,5 +88,10 @@ echo 'yml'
 $CLICKHOUSE_CLIENT --config "$yml_config" -q "select getSetting('max_threads')"
 echo 'yaml'
 $CLICKHOUSE_CLIENT --config "$yaml_config" -q "select getSetting('max_threads')"
-echo 'ini'
+
-$CLICKHOUSE_CLIENT --config "$ini_config" -q "select getSetting('max_threads')" 2>&1 |& sed -e "s#$CLICKHOUSE_TMP##" -e "s#DB::Exception: ##"
+echo 'autodetect xml (with leading whitespaces)'
+$CLICKHOUSE_CLIENT --config "$autodetect_xml_with_leading_whitespace_config" -q "select getSetting('max_threads')"
+echo 'autodetect xml (non leading whitespaces)'
+$CLICKHOUSE_CLIENT --config "$autodetect_xml_non_leading_whitespace_config" -q "select getSetting('max_threads')"
+echo 'autodetect yaml'
+$CLICKHOUSE_CLIENT --config "$autodetect_yaml_config" -q "select getSetting('max_threads')"