Validate and clean input yaml

Code.v05-00/src/YamlInputReader/YamlInputReader.cpp

@@ -5,7 +5,11 @@
 #include <algorithm> // std::equal
 #include <cctype>    // std::tolower
 #include <iostream>
+#include <stdexcept>
 #include <string_view> // std::string_view
+#include <set>
+#include <string>
+#include <vector>
 
 
 // Read default configuration from CMake-generated include file.
@@ -25,79 +29,122 @@ bool iequals(std::string_view lhs, std::string_view rhs) {
 
 
 namespace YamlInputReader{
+    // Helper function to get all keys from a YAML Map node
+    std::set<std::string> getYamlKeys(const YAML::Node& node) {
+        std::set<std::string> keys;
+        if (!node.IsMap()) {
+            return keys;
+        }
+        for (const auto& it : node) {
+            keys.insert(it.first.as<std::string>());
+        }
+        return keys;
+    }
+
+    void validateYamlKeys(const YAML::Node& defaultNode, const YAML::Node& userNode, const std::string& currentPath = "") {
+        if (!userNode.IsMap()) {
+            // If the user node is not a map, we don't need to check its keys.
+            return;
+        }
+
+        if (!defaultNode.IsMap()) {
+            // If the user node is a map but the default is not, it's an error
+            // because the user is trying to add a structure that doesn't exist.
+            throw std::runtime_error("Invalid key: '" + currentPath + "' is a map in provided YAML but not in the default input.yaml (should be a value).");
+        }
+
+        auto defaultKeys = getYamlKeys(defaultNode);
+        auto userKeys = getYamlKeys(userNode);
+
+        for (const auto& key : userKeys) {
+            if (!defaultKeys.contains(key)) {
+                // The key from the user's YAML does not exist in the default YAML.
+                std::string errorPath = currentPath.empty() ? key : currentPath + " -> " + key;
+                throw std::runtime_error("Unknown key found: '" + errorPath + "'");
+            }
+
+            // Recurse to check nested maps
+            const YAML::Node nextUserNode = userNode[key];
+            const YAML::Node nextDefaultNode = defaultNode[key];
+
+            if (nextUserNode.IsMap()) {
+                std::string nextPath = currentPath.empty() ? key : currentPath + " -> " + key;
+                validateYamlKeys(nextDefaultNode, nextUserNode, nextPath);
+            }
+        }
+    }
+
     void readYamlInputFiles(OptInput& input, const vector<string> &filenames){
-        YAML::Node data = YAML::Load(default_input);
+        YAML::Node defaultData = YAML::Load(default_input);
+        YAML::Node mergedData = YAML::Load(default_input);
+
         for (auto filename: filenames) {
+            YAML::Node userData = YAML::LoadFile(filename);
+
+            // Validate the user's YAML file against the default structure
+            try {
+                validateYamlKeys(defaultData, userData);
+            } catch (const std::runtime_error& e) {
+                throw std::runtime_error("Invalid field in YAML input file '" + filename + "': " + e.what());
+            }
             INPUT_FILE_PATH = std::filesystem::path(filename);
-            data = mergeYamlNodes(data, YAML::LoadFile(filename));
+            mergedData = mergeYamlNodes(mergedData, userData);
         }
 
         try {
-            readSimMenu(input, data["SIMULATION MENU"]);
+            readSimMenu(input, mergedData["SIMULATION MENU"]);
         }
-        catch (...) {
-            std::cout << "Something went wrong in reading the SIMULATION MENU! Please double-check your input file with the reference in SampleRunDir!";
-            exit(1);
+        catch (const std::exception& e) {
+            throw std::runtime_error("Something went wrong in reading the SIMULATION MENU! Please double-check your input file with the reference in Code.v05-00/defaults/input.yaml\n  Exception: " + std::string(e.what()));
         }
 
         try {
-            readParamMenu(input, data["PARAMETER MENU"]);
+            readParamMenu(input, mergedData["PARAMETER MENU"]);
         }
-        catch (...) {
-            std::cout << "Something went wrong in reading the PARAMETER MENU! Please double-check your input file with the reference in SampleRunDir!";
-            exit(1);
+        catch (const std::exception& e) {
+            throw std::runtime_error("Something went wrong in reading the PARAMETER MENU! Please double-check your input file with the reference in Code.v05-00/defaults/input.yaml\n  Exception: " + std::string(e.what()));
         }
 
         try {
-            readTransportMenu(input, data["TRANSPORT MENU"]);
+            readTransportMenu(input, mergedData["TRANSPORT MENU"]);
         }
-        catch (...) {
-            std::cout << "Something went wrong in reading the TRANSPORT MENU! Please double-check your input file with the reference in SampleRunDir!";
-            exit(1);
+        catch (const std::exception& e) {
+            throw std::runtime_error("Something went wrong in reading the TRANSPORT MENU! Please double-check your input file with the reference in Code.v05-00/defaults/input.yaml\n  Exception: " + std::string(e.what()));
         }
 
         try {
-            readChemMenu(input, data["CHEMISTRY MENU"]);
+            readChemMenu(input, mergedData["CHEMISTRY MENU"]);
         }
-        catch (...) {
-            std::cout << "Something went wrong in reading the CHEMISTRY MENU! Please double-check your input file with the reference in SampleRunDir!";
-            exit(1);
+        catch (const std::exception& e) {
+            throw std::runtime_error("Something went wrong in reading the CHEMISTRY MENU! Please double-check your input file with the reference in Code.v05-00/defaults/input.yaml\n  Exception: " + std::string(e.what()));
         }
 
         try {
-            readAeroMenu(input, data["AEROSOL MENU"]);  
+            readAeroMenu(input, mergedData["AEROSOL MENU"]);  
         }
-        catch (...) {
-            std::cout << "Something went wrong in reading the AEROSOL MENU! Please double-check your input file with the reference in SampleRunDir!";
-            exit(1);
+        catch (const std::exception& e) {
+            throw std::runtime_error("Something went wrong in reading the AEROSOL MENU! Please double-check your input file with the reference in Code.v05-00/defaults/input.yaml\n  Exception: " + std::string(e.what()));
         }
 
         try {
-            readMetMenu(input, data["METEOROLOGY MENU"]);
-        }
-        catch (const std::invalid_argument& e) {
-            std::cerr << "ERROR: " << e.what() << std::endl;
-            exit(1);
+            readMetMenu(input, mergedData["METEOROLOGY MENU"]);
         }
-        catch (...) {
-            std::cout << "Something went wrong in reading the METEOROLOGY MENU! Please double-check your input file with the reference in SampleRunDir!";
-            exit(1);
+        catch (const std::exception& e) {
+            throw std::runtime_error("Something went wrong in reading the METEOROLOGY MENU! Please double-check your input file with the reference in Code.v05-00/defaults/input.yaml\n  Exception: " + std::string(e.what()));
         }
 
         try {
-            readDiagMenu(input, data["DIAGNOSTIC MENU"]);
+            readDiagMenu(input, mergedData["DIAGNOSTIC MENU"]);
         }
-        catch (...) {
-            std::cout << "Something went wrong in reading the DIAGNOSTIC MENU! Please double-check your input file with the reference in SampleRunDir!";
-            exit(1);
+        catch (const std::exception& e) {
+            throw std::runtime_error("Something went wrong in reading the DIAGNOSTIC MENU! Please double-check your input file with the reference in Code.v05-00/defaults/input.yaml\n  Exception: " + std::string(e.what()));
         }
 
         try {
-            readAdvancedMenu(input, data["ADVANCED OPTIONS MENU"]);
+            readAdvancedMenu(input, mergedData["ADVANCED OPTIONS MENU"]);
         }
-        catch (...) {
-            std::cout << "Something went wrong in reading the ADVANCED OPTIONS MENU! Please double-check your input file with the reference in SampleRunDir!";
-            exit(1);
+        catch (const std::exception& e) {
+            throw std::runtime_error("Something went wrong in reading the ADVANCED OPTIONS MENU! Please double-check your input file with the reference in Code.v05-00/defaults/input.yaml\n  Exception: " + std::string(e.what()));
         }
     }
     void readSimMenu(OptInput& input, const YAML::Node& simNode){

Code.v05-00/tests/test1.yaml

@@ -100,7 +100,6 @@ METEOROLOGY MENU:
     Init wind shear from met. (T/F): T
     Wind shear time series input (T/F): T
     Interpolate shear met. data (T/F): T
-    Init vertical velocity from met. data: T
     Init vert. veloc. from met. data (T/F): T
     Vert. veloc. time series input (T/F): T
     Interpolate vert. veloc. met. data (T/F): T 

Code.v05-00/tests/test3.yaml

@@ -0,0 +1,3 @@
+PARAMETER MENU:
+  Plume Process [hr] (double): 14
+  INVALID YAML INPUT: 0

Code.v05-00/tests/test4.yaml

@@ -0,0 +1,9 @@
+PARAMETER MENU:
+  Plume Process [hr] (double): 14
+
+METEOROLOGY MENU:
+  METEOROLOGICAL INPUT SUBMENU:
+    Use met. input (T/F): F
+    Met input file path (string): /path/to/met/input
+    INVALID YAML KEY:
+      BAD FIELD: 0

Code.v05-00/tests/test5.yaml

@@ -0,0 +1,7 @@
+METEOROLOGY MENU:
+  METEOROLOGICAL INPUT SUBMENU:
+    Use met. input (T/F): F
+    # This is a valid value in the default yaml
+    # but it is not a valid key! Should throw
+    Met input file path (string):
+      BAD VALUE: 0

Code.v05-00/tests/test_yamlreader.cpp

@@ -1,4 +1,5 @@
 #include <catch2/catch_test_macros.hpp>
+#include <catch2/matchers/catch_matchers_string.hpp>
 #include <YamlInputReader/YamlInputReader.hpp>
 #include <Core/Input.hpp>
 #include "APCEMM.h"
@@ -458,3 +459,45 @@ TEST_CASE("Merge Input Files"){
     REQUIRE(caseInput.coreExitTemp() == 547.3);
     REQUIRE(caseInput.bypassArea() == 1.804);
 }
+
+TEST_CASE("Validate Input Files"){
+    OptInput input;
+    string validFile = string(APCEMM_TESTS_DIR)+"/test1.yaml";
+
+    string filename1 = string(APCEMM_TESTS_DIR)+"/test3.yaml";
+    string filename2 = string(APCEMM_TESTS_DIR)+"/test4.yaml";
+    string filename3 = string(APCEMM_TESTS_DIR)+"/test5.yaml";
+
+    SECTION("Invalid key (scalar) at the root level") {
+        // Check that it detects the invalid key, that it points to the correct file and that
+        // it prints out the name of the invalid key (here a scalar)
+        string invalid_file = string(APCEMM_TESTS_DIR) + "/test3.yaml";
+
+        REQUIRE_THROWS_WITH(
+            YamlInputReader::readYamlInputFiles(input, {validFile, filename1}), 
+            Catch::Matchers::ContainsSubstring("Unknown key found") &&
+            Catch::Matchers::ContainsSubstring("test3.yaml") &&
+            Catch::Matchers::ContainsSubstring("INVALID YAML INPUT")
+    );
+    }
+
+    SECTION("Invalid key (map) in a submenu"){
+        // Check that it detects the invalid key and that it prints out the name
+        // of the invalid key (here a map)
+        REQUIRE_THROWS_WITH(
+            YamlInputReader::readYamlInputFiles(input, {filename2}),
+            Catch::Matchers::ContainsSubstring("Unknown key found") &&
+            Catch::Matchers::ContainsSubstring("INVALID YAML KEY")
+        );
+    }
+
+    SECTION("Valid key but wrong type (map instead of scalar)"){
+        // Here we have a key that is supposed to be a scalar but instead is a map
+        // Check that if detects this and prints the name correctly
+        REQUIRE_THROWS_WITH(
+            YamlInputReader::readYamlInputFiles(input, {filename3}),
+            Catch::Matchers::ContainsSubstring("is a map in provided YAML but not in the default input.yaml") &&
+            Catch::Matchers::ContainsSubstring("Met input file path (string)")
+        );
+    }
+}