feat: support parsing of system environment variables in yaml

In order to allow the user to supply environment variables in standard
ways performed in other applications the get_scalar function has been
extended to support defining an environment variable as either the
format $FOO or the format ${FOO}. As this handles some additional
complexity a unit test has been added to cover this new functionality

Signed-off-by: Daniel Wright <danielwright@bitgo.com>

unit_tests/falco/test_configuration.cpp

@@ -88,16 +88,88 @@ TEST(Configuration, modify_yaml_fields)
 {
 	std::string key = "base_value.subvalue.subvalue2.boolean";
 	yaml_helper conf;
-	
+
     /* Get original value */
     conf.load_from_string(sample_yaml);
 	ASSERT_EQ(conf.get_scalar<bool>(key, false), true);
-	
+
     /* Modify the original value */
     conf.set_scalar<bool>(key, false);
 	ASSERT_EQ(conf.get_scalar<bool>(key, true), false);
-	
+
     /* Modify it again */
     conf.set_scalar<bool>(key, true);
 	ASSERT_EQ(conf.get_scalar<bool>(key, false), true);
 }
+
+TEST(Configuration, configuration_environment_variables)
+{
+    // Set an environment variable for testing purposes
+    std::string env_var_value = "envVarValue";
+    std::string env_var_name = "ENV_VAR";
+    std::string default_value = "default";
+    setenv(env_var_name.c_str(), env_var_value.c_str(), 1);
+    yaml_helper conf;
+
+    std::string sample_yaml =
+        "base_value:\n"
+        "    id: $ENV_VAR\n"
+        "    name: '${ENV_VAR}'\n"
+        "    string: my_string\n"
+        "    invalid: $${ENV_VAR}\n"
+		"    invalid_env: $$ENV_VAR\n"
+        "    escaped: \"${ENV_VAR}\"\n"
+        "    subvalue:\n"
+        "        subvalue2:\n"
+        "            boolean: ${UNSED_XX_X_X_VAR}\n"
+        "base_value_2:\n"
+        "    sample_list:\n"
+        "        - ${ENV_VAR}\n"
+        "        - ' ${ENV_VAR}'\n"
+        "        - $UNSED_XX_X_X_VAR\n";
+    conf.load_from_string(sample_yaml);
+
+    /* Check if the base values are defined */
+    ASSERT_TRUE(conf.is_defined("base_value"));
+    ASSERT_TRUE(conf.is_defined("base_value_2"));
+    ASSERT_FALSE(conf.is_defined("unknown_base_value"));
+
+    /* Test fetching of a regular string without any environment variable */
+    std::string base_value_string = conf.get_scalar<std::string>("base_value.string", default_value);
+    ASSERT_EQ(base_value_string, "my_string");
+
+    /* Test fetching of escaped environment variable format. Should return the string as-is after stripping the leading `$` */
+    std::string base_value_invalid = conf.get_scalar<std::string>("base_value.invalid", default_value);
+    ASSERT_EQ(base_value_invalid, "${ENV_VAR}");
+
+	/* Test fetching of escaped environment variable ub ubvakud format. Should return the string as-is */
+    std::string base_value_invalid_env = conf.get_scalar<std::string>("base_value.invalid_env", default_value);
+    ASSERT_EQ(base_value_invalid_env, "$$ENV_VAR");
+
+    /* Test fetching of strings that contain environment variables */
+    std::string base_value_id = conf.get_scalar<std::string>("base_value.id", default_value);
+    ASSERT_EQ(base_value_id, "$ENV_VAR"); // Does not follow the `${VAR}` format, so it should be treated as a regular string
+
+    std::string base_value_name = conf.get_scalar<std::string>("base_value.name", default_value);
+    ASSERT_EQ(base_value_name, env_var_value); // Proper environment variable format
+
+    std::string base_value_escaped = conf.get_scalar<std::string>("base_value.escaped", default_value);
+    ASSERT_EQ(base_value_escaped, env_var_value); // Environment variable within quotes
+
+    /* Test fetching of an undefined environment variable. Expected to return the default value.*/
+    std::string unknown_boolean = conf.get_scalar<std::string>("base_value.subvalue.subvalue2.boolean", default_value);
+    ASSERT_EQ(unknown_boolean, default_value);
+
+    /* Test fetching of environment variables from a list */
+    std::string base_value_2_list_0 = conf.get_scalar<std::string>("base_value_2.sample_list[0]", default_value);
+    ASSERT_EQ(base_value_2_list_0, env_var_value); // Proper environment variable format
+
+    std::string base_value_2_list_1 = conf.get_scalar<std::string>("base_value_2.sample_list[1]", default_value);
+    ASSERT_EQ(base_value_2_list_1, " ${ENV_VAR}"); // Environment variable preceded by a space, hence treated as a regular string
+
+    std::string base_value_2_list_2 = conf.get_scalar<std::string>("base_value_2.sample_list[2]", default_value);
+    ASSERT_EQ(base_value_2_list_2, "$UNSED_XX_X_X_VAR"); // Does not follow the `${VAR}` format, so should be treated as a regular string
+
+    /* Clear the set environment variable after testing */
+    unsetenv(env_var_name.c_str());
+}

userspace/falco/yaml_helper.h

@@ -72,6 +72,64 @@ class yaml_helper
 		get_node(node, key);
 		if(node.IsDefined())
 		{
+			std::string value = node.as<std::string>();
+
+			// If the value starts with `$$`, check for a subsequent `{...}`
+			if (value.size() >= 3 && value[0] == '$' && value[1] == '$')
+			{
+				// If after stripping the first `$`, the string format is like `${VAR}`, treat it as a plain string and don't resolve.
+				if (value[2] == '{' && value[value.size() - 1] == '}')
+				{
+					value = value.substr(1);
+					std::stringstream ss(value);
+					T stripped_value;
+					if (ss >> stripped_value)
+					{
+						return stripped_value;
+					}
+					// If the conversion fails for some reason, we should return the default
+					return default_value;
+				}
+				else
+				{
+					// If it doesn't have the `${...}` format after `$$`, then it should return the value as-is without stripping.
+					std::stringstream ss(value);
+					T preserved_value;
+					if (ss >> preserved_value)
+					{
+						return preserved_value;
+					}
+					return default_value;
+				}
+			}
+
+			// Check if the value is an environment variable reference
+			if(!value.empty() && value[0] == '$' && value.size() > 2 && value[1] == '{' && value[value.size() - 1] == '}')
+			{
+				// Format: ${ENV_VAR_NAME}
+				std::string env_var = value.substr(2, value.size() - 3);
+
+				const char* env_value = std::getenv(env_var.c_str()); // Get the environment variable value
+				if(env_value)
+				{
+					// If the environment variable is found, parse and convert it to the desired type
+					std::stringstream ss(env_value);
+					T parsed_value;
+					if(ss >> parsed_value)
+					{
+						return parsed_value;
+					}
+					// Parsing failed, return the default value
+					return default_value;
+				}
+				else
+				{
+					// Environment variable declared in config not defined on system, return the default value
+					return default_value;
+				}
+			}
+
+			// If it's not an environment variable reference, return the value as is
 			return node.as<T>();
 		}
 
@@ -118,10 +176,10 @@ class yaml_helper
 	 * The provided key string can navigate the document in its
 	 * nested nodes, with arbitrary depth. The key string follows
 	 * this regular language:
-	 * 
+	 *
 	 * Key 		:= NodeKey ('.' NodeKey)*
 	 * NodeKey	:= (any)+ ('[' (integer)+ ']')*
-	 * 
+	 *
 	 * Some examples of accepted key strings:
 	 * - NodeName
 	 * - ListValue[3].subvalue
@@ -146,7 +204,7 @@ class yaml_helper
 					if (i > 0 && nodeKey.empty() && key[i - 1] != '.')
 					{
 						throw std::runtime_error(
-							"Parsing error: expected '.' character at pos " 
+							"Parsing error: expected '.' character at pos "
 							+ std::to_string(i - 1));
 					}
 					nodeKey += c;
@@ -157,7 +215,7 @@ class yaml_helper
 					if (nodeKey.empty())
 					{
 						throw std::runtime_error(
-							"Parsing error: unexpected character at pos " 
+							"Parsing error: unexpected character at pos "
 							+ std::to_string(i));
 					}
 					ret.reset(ret[nodeKey]);
@@ -181,7 +239,7 @@ class yaml_helper
 			throw std::runtime_error("Config error at key \"" + key + "\": " + std::string(e.what()));
 		}
 	}
-	
+
 	template<typename T>
 	void get_sequence_from_node(T& ret, const YAML::Node& node) const
 	{
@@ -250,7 +308,7 @@ namespace YAML {
 				default:
 					break;
 			}
-			
+
 			return true;
 		}
 	};