Added schema validation

CMakeLists.txt

@@ -46,16 +46,23 @@ if (ENABLE_COVERAGE)
     add_link_options(--coverage)
 endif ()
 
+add_compile_options(-Wall -Wextra -Wpedantic)
+
 #####################################
 # External projects
 #####################################
-# simdjson
+
 FetchContent_Declare(
-        nlohmann_json
-        GIT_REPOSITORY https://github.com/nlohmann/json.git
-        GIT_TAG v3.12.0
+        jsoncons
+        GIT_REPOSITORY https://github.com/danielaparker/jsoncons.git
+        GIT_TAG v1.4.3
 )
-FetchContent_MakeAvailable(nlohmann_json)
+
+set(JSONCONS_BUILD_TESTS OFF CACHE BOOL "" FORCE)
+set(JSONCONS_BUILD_EXAMPLES OFF CACHE BOOL "" FORCE)
+set(JSONCONS_BUILD_FUZZERS OFF CACHE BOOL "" FORCE)
+
+FetchContent_MakeAvailable(jsoncons)
 
 if (BUILD_PYTHON_BINDINGS)
     FetchContent_Declare(
@@ -66,22 +73,57 @@ if (BUILD_PYTHON_BINDINGS)
     FetchContent_MakeAvailable(pybind11)
 endif ()
 
+#####################################
+# Encode CAPIO-CL JSON Schemas
+#####################################
+set(CAPIOCL_JSON_SCHEMAS_DIRECTORY "${CMAKE_BINARY_DIR}/schemas")
+set(OUTPUT_HEADER "${CAPIOCL_JSON_SCHEMAS_DIRECTORY}/capio_cl_json_schemas.hpp")
+file(GLOB SCHEMA_FILES "${CMAKE_SOURCE_DIR}/schema/*.json")
+
+set_property(DIRECTORY APPEND PROPERTY CMAKE_CONFIGURE_DEPENDS "${CMAKE_SOURCE_DIR}/schema*.json")
+
+file(WRITE ${OUTPUT_HEADER} "// Bundled CAPIO-CL encoded JSON schemas\n")
+file(APPEND ${OUTPUT_HEADER} "#pragma once\n\n")
+
+foreach (SCHEMA_FILE ${SCHEMA_FILES})
+    get_filename_component(SCHEMA_NAME ${SCHEMA_FILE} NAME_WE)
+    get_filename_component(SCHEMA_BASENAME ${SCHEMA_FILE} NAME)
+    message(STATUS "Bundling CAPIO-CL schema: ${SCHEMA_BASENAME}")
+    file(APPEND ${OUTPUT_HEADER} "// CAPIO-CL version: ${SCHEMA_NAME}\n")
+
+    execute_process(
+            COMMAND xxd -i ${SCHEMA_BASENAME}
+            WORKING_DIRECTORY ${CMAKE_SOURCE_DIR}/schema
+            OUTPUT_VARIABLE HEXDATA
+            RESULT_VARIABLE RES
+    )
+    if (NOT RES EQUAL 0)
+        message(FATAL_ERROR "xxd failed for ${SCHEMA_FILE}")
+    endif ()
+
+    file(APPEND ${OUTPUT_HEADER} "${HEXDATA}\n\n")
+endforeach ()
+
+message(STATUS "Generated header: ${OUTPUT_HEADER}")
 
 #####################################
 # Sources and headers
 #####################################
 file(GLOB_RECURSE CAPIO_SRC CONFIGURE_DEPENDS "src/*.cpp")
 set(CAPIO_CL_HEADERS capiocl.hpp)
 
-# Library target (for use with external projects)
+# Library target
 add_library(libcapio_cl STATIC ${CAPIO_SRC} ${CAPIO_CL_HEADERS})
 
 target_include_directories(libcapio_cl PUBLIC
         ${CMAKE_CURRENT_SOURCE_DIR}
         ${CMAKE_CURRENT_SOURCE_DIR}/src
-        ${nlohmann_json_SOURCE_DIR}/include
+        ${jsoncons_SOURCE_DIR}/include
+        ${CAPIOCL_JSON_SCHEMAS_DIRECTORY}
 )
 
+# jsoncons is header-only, no linking required
+target_link_libraries(libcapio_cl PUBLIC)
 
 #####################################
 # Install rules
@@ -90,22 +132,18 @@ install(TARGETS libcapio_cl
         LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
 )
 
-
 #####################################
 # Python bindings
 #####################################
 if (BUILD_PYTHON_BINDINGS)
-
     set(PYTHON_BIND_NAME _py_capio_cl)
 
     pybind11_add_module(${PYTHON_BIND_NAME}
             bindings/python_bindings.cpp
     )
 
-
-    # Make sure the binding sees the CAPIO CL headers & links against the core lib
-    target_link_libraries(${PYTHON_BIND_NAME}
-            PRIVATE libcapio_cl
+    target_link_libraries(${PYTHON_BIND_NAME} PRIVATE
+            libcapio_cl
     )
 
     target_include_directories(${PYTHON_BIND_NAME}
@@ -114,10 +152,8 @@ if (BUILD_PYTHON_BINDINGS)
     )
 
     install(TARGETS _py_capio_cl DESTINATION py_capio_cl)
-
 endif ()
 
-
 #####################################
 # Tests (only when built standalone)
 #####################################
@@ -161,14 +197,12 @@ if (CAPIO_CL_BUILD_TESTS)
     )
 
     #####################################
-    # Install rules
+    # Install rules for tests
     #####################################
-    # Optionally install tests
     install(TARGETS CAPIO_CL_tests
             RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
     )
 
-    # Copy tests jsons
     install(DIRECTORY ${TEST_JSON_DIR}
             DESTINATION ${CMAKE_INSTALL_BINDIR}/jsons
             FILES_MATCHING PATTERN "*.json"

README.md

@@ -42,7 +42,7 @@ At runtime, CAPIO-CL’s parser and engine components analyze, track, and manage
 ### Requirements & dependencies
 - C++17 or greater
 - Cmake 3.15 or newer
-- [nlohmann/json](https://github.com/nlohmann/json) to parse JSON config files
+- [danielaparker/jsoncons](https://github.com/danielaparker/jsoncons) to parse, serialize and validate CAPIO-CL JSON config files
 - [GoogleTest](https://github.com/google/googletest) for automated testing
 
 All dependencies are fetched automatically by CMake — no manual setup required.

capiocl.hpp

@@ -34,7 +34,7 @@ constexpr char UPDATE[]    = "update";
 namespace commit_rules {
 constexpr char ON_CLOSE[]       = "on_close";
 constexpr char ON_FILE[]        = "on_file";
-constexpr char N_FILES[]        = "n_files";
+constexpr char N_FILES[]        = "on_n_files";
 constexpr char ON_TERMINATION[] = "on_termination";
 } // namespace commit_rules
 
@@ -413,10 +413,6 @@ class Engine {
  *
  */
 class Parser {
-
-    static std::filesystem::path resolve(std::filesystem::path path,
-                                         const std::filesystem::path &prefix);
-
   public:
     /**
      * @brief Perform the parsing of the capio_server configuration file

schema/v1.json

@@ -0,0 +1,220 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "title": "CAPIO-CL Configuration Schema",
+  "type": "object",
+
+  "definitions": {
+    "streaming_name_rule": {
+      "type": "object",
+      "properties": {
+        "name": {
+          "type": "array",
+          "description": "Filenames to which the rule applies.",
+          "items": { "type": "string" }
+        },
+        "committed": {
+          "type": "string",
+          "description": "Commit rule associated with the file.",
+          "pattern": "^(on_close(:[0-9]+)?|on_file|on_termination)$"
+        },
+        "mode": {
+          "type": "string",
+          "description": "Firing rule associated with the files.",
+          "enum": ["update", "no_update"]
+        },
+        "file_deps": {
+          "type": "array",
+          "description": "List of dependent files required when committed = on_file.",
+          "items": { "type": "string" }
+        },
+        "n_files": {
+          "type": "integer",
+          "description": "Number of files expected when commit rule == n_files."
+        }
+      },
+      "required": ["name"],
+      "additionalProperties": false,
+
+      "if": {
+        "properties": {
+          "committed": { "pattern": "^on_file" }
+        },
+        "required": ["committed"]
+      },
+      "then": {
+        "required": ["file_deps"]
+      }
+    },
+
+    "streaming_dirname_rule": {
+      "type": "object",
+      "properties": {
+        "dirname": {
+          "type": "array",
+          "description": "Directory names to which the rule applies.",
+          "items": { "type": "string" }
+        },
+        "committed": {
+          "type": "string",
+          "description": "Commit rule associated with the directory.",
+          "pattern": "^(on_close(:[0-9]+)?|on_file|on_n_files|on_termination)$"
+        },
+        "mode": {
+          "type": "string",
+          "description": "Firing rule associated with the directory contents.",
+          "enum": ["update", "no_update"]
+        },
+        "file_deps": {
+          "type": "array",
+          "description": "List of dependent files required when committed = on_file.",
+          "items": { "type": "string" }
+        },
+        "n_files": {
+          "type": "integer",
+          "description": "Number of files expected when commit rule == n_files."
+        }
+      },
+      "required": ["dirname"],
+      "additionalProperties": false,
+
+      "allOf": [
+        {
+          "if": {
+            "properties": {
+              "committed": { "pattern": "^on_file" }
+            },
+            "required": ["committed"]
+          },
+          "then": {
+            "required": ["file_deps"]
+          }
+        },
+        {
+          "if": {
+            "properties": {
+              "committed": { "pattern": "^on_n_files" }
+            },
+            "required": ["committed"]
+          },
+          "then": {
+            "required": ["n_files"]
+          }
+        }
+      ]
+    }
+  },
+  "properties": {
+    "name": {
+      "type": "string",
+      "description": "Identifies the application workflow."
+    },
+
+    "aliases": {
+      "type": "array",
+      "description": "Groups files or directories under a convenient name.",
+      "items": {
+        "type": "object",
+        "properties": {
+          "group_name": { "type": "string" },
+          "files": {
+            "type": "array",
+            "items": { "type": "string" }
+          }
+        },
+        "required": ["group_name", "files"],
+        "additionalProperties": false
+      }
+    },
+
+    "IO_Graph": {
+      "type": "array",
+      "description": "Describes file data dependencies among application modules.",
+      "items": {
+        "type": "object",
+        "properties": {
+          "name": { "type": "string" },
+
+          "input_stream": {
+            "type": "array",
+            "items": { "type": "string" }
+          },
+
+          "output_stream": {
+            "type": "array",
+            "items": { "type": "string" }
+          },
+
+          "streaming": {
+            "type": "array",
+            "items": {
+              "oneOf": [
+                { "$ref": "#/definitions/streaming_name_rule" },
+                { "$ref": "#/definitions/streaming_dirname_rule" }
+              ]
+            }
+          }
+        },
+        "required": ["name", "input_stream", "output_stream"],
+        "additionalProperties": false
+      }
+    },
+
+    "permanent": {
+      "type": "array",
+      "items": { "type": "string" }
+    },
+
+    "exclude": {
+      "type": "array",
+      "items": { "type": "string" }
+    },
+
+    "home_node_policies": {
+      "type": "object",
+      "properties": {
+        "create": {
+          "type": "array",
+          "items": { "type": "string" }
+        },
+        "hashing": {
+          "type": "array",
+          "items": { "type": "string" }
+        },
+        "manual": {
+          "type": "array",
+          "items": {
+            "type": "object",
+            "properties": {
+              "name": {
+                "type": "array",
+                "items": { "type": "string" }
+              },
+              "app_node": { "type": "string" }
+            },
+            "required": ["name", "app_node"],
+            "additionalProperties": false
+          }
+        }
+      },
+      "additionalProperties": false
+    },
+
+    "storage": {
+      "type": "object",
+      "properties": {
+        "memory": {
+          "type": "array",
+          "items": { "type": "string" }
+        },
+        "fs": {
+          "type": "array",
+          "items": { "type": "string" }
+        }
+      },
+      "additionalProperties": false
+    }
+  },
+
+  "required": ["name", "IO_Graph"],
+  "additionalProperties": false
+}