👻 Config changes

.github/workflows/build-and-push-binaries.yml

@@ -189,23 +189,36 @@ jobs:
           cp ./dist/kai-rpc-server ./example/analysis/
           cp ./kai_analyzer_rpc/kai-analyzer-rpc ./example/analysis/
 
-          cat << EOF > ./example/config.toml
+          rm -f ./example/initialize.toml
+
+          cat << EOF > ./example/initialize.toml
+          root_path = "./coolstore"
+          analyzer_lsp_java_bundle_path = "./analysis/bundle.jar"
+          analyzer_lsp_lsp_path = "./analysis/jdtls/bin/jdtls"
+          analyzer_lsp_rpc_path = "./analysis/kai-analyzer-rpc"
+          analyzer_lsp_rules_path = "./analysis/rulesets/default/generated"
+          analyzer_lsp_dep_labels_path = "./analysis/maven.default.index"
+
+          demo_mode = true
+          enable_reflection = false
+
+          [log_config]
           log_level = "debug"
           file_log_level = "debug"
-          log_dir = "/home/runner/work/kai/kai/kai/../../logs/"
-          demo_mode = true
-          [models]
+          log_dir_path = "/home/runner/work/kai/kai/kai/../../logs/"
+
+          [model_provider]
           provider = "${{ matrix.models.provider }}"
-          [models.args]
+          [model_provider.args]
           model_id = "${{ matrix.models.model_id }}"
           EOF
 
           if [[ -n "${{ matrix.models.max_new_tokens }}" ]]; then
-          cat << EOF >> ./example/config.toml
+          cat << EOF >> ./example/initialize.toml
           parameters.max_new_tokens = ${{ matrix.models.max_new_tokens }}
           EOF
           fi
-          cat ./example/config.toml
+          cat ./example/initialize.toml
 
           which python
           cd example

.trunk/trunk.yaml

@@ -57,6 +57,7 @@ lint:
         - kai_solution_server/samples/macos/**
         - example/analysis/**
         - example/default_rules/**
+        - kai/data/**
         # This file is from  https://github.com/rh-aiservices-bu/llm-on-openshift
         # It is included here only for convenience
         - notebooks/jms_to_smallrye_reactive/caikit_tgis_langchain.py

Makefile

@@ -60,7 +60,7 @@ run-demo:
 
 run-debug-driver:
 	python kai/reactive_codeplanner/main.py \
-		--kai-config example/config.toml \
+		--kai-config example/initialize.toml \
 		--source-directory example/coolstore \
 		--rules-directory example/analysis/rulesets/default/generated \
 		--analyzer-lsp-server-binary example/analysis/kai-analyzer-rpc \

example/initialize.toml

@@ -0,0 +1,23 @@
+# This file will transformed into a json-rpc request object and sent to the
+# server to its `initialize` request endpoint
+
+root_path = "./coolstore"
+analyzer_lsp_java_bundle_path = "./analysis/bundle.jar"
+analyzer_lsp_lsp_path = "./analysis/jdtls/bin/jdtls"
+analyzer_lsp_rpc_path = "./analysis/kai-analyzer-rpc"
+analyzer_lsp_rules_path = "./analysis/rulesets/default/generated"
+analyzer_lsp_dep_labels_path = "./analysis/maven.default.index"
+
+demo_mode = true
+enable_reflection = false
+
+[log_config]
+log_level = "INFO"
+file_log_level = "DEBUG"
+log_dir_path = "./logs"
+
+[model_provider]
+provider = "ChatIBMGenAI"
+
+[model_provider.args]
+model_id = "meta-llama/llama-3-1-70b-instruct"

example/run_demo.py

@@ -7,7 +7,7 @@
 import time
 from io import BufferedReader, BufferedWriter
 from pathlib import Path
-from typing import Generator, Optional, cast
+from typing import Generator, cast
 
 from opentelemetry import trace
 from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
@@ -22,8 +22,7 @@
 from kai.jsonrpc.core import JsonRpcServer
 from kai.jsonrpc.models import JsonRpcError, JsonRpcResponse
 from kai.jsonrpc.streams import LspStyleStream
-from kai.kai_config import KaiConfig
-from kai.logging.logging import get_logger, init_logging_from_config
+from kai.logging.logging import get_logger, init_logging_from_log_config
 from kai.rpc_server.server import (
     GetCodeplanAgentSolutionParams,
     KaiRpcApplication,
@@ -38,14 +37,9 @@ def get_binary_path(path: str) -> Path:
 
 
 SERVER_URL = "http://0.0.0.0:8080"
-APP_NAME = "coolstore"
 SAMPLE_APP_DIR = Path("coolstore")
-ANALYSIS_BUNDLE_PATH = Path(".", "analysis", "bundle.jar")
-ANALYSIS_LSP_PATH = Path(".", "analysis", "jdtls", "bin", "jdtls")
 ANALYSIS_RPC_PATH = get_binary_path("./analysis/kai-analyzer-rpc")
 RPC_BINARY_PATH = get_binary_path("./analysis/kai-rpc-server")
-ANALYSIS_RULES_PATH = Path(".", "analysis", "rulesets", "default", "generated")
-ANALYSIS_DEP_LABELS_FILE = Path(".", "analysis", "maven.default.index")
 TRACING_ENABLED = "ENABLE_TRACING"
 
 KAI_LOG = get_logger("run_demo")
@@ -57,11 +51,6 @@ def get_binary_path(path: str) -> Path:
 
 def pre_flight_checks() -> None:
     for path in [
-        SAMPLE_APP_DIR,
-        ANALYSIS_BUNDLE_PATH,
-        ANALYSIS_LSP_PATH,
-        ANALYSIS_RPC_PATH,
-        ANALYSIS_DEP_LABELS_FILE,
         RPC_BINARY_PATH,
     ]:
         if not path.exists():
@@ -73,32 +62,29 @@ def pre_flight_checks() -> None:
 
 @contextlib.contextmanager
 def initialize_rpc_server(
-    kai_config: KaiConfig,
+    config: KaiRpcApplicationConfig,
 ) -> Generator[JsonRpcServer, None, None]:
-
-    cache_dir: Optional[Path] = None
-    if kai_config.cache_dir is not None:
-        cache_dir = Path(kai_config.cache_dir)
+    # NOTE: This is a hack. Config should probably be globally accessible in
+    # this script.
+    global SAMPLE_APP_DIR
+    SAMPLE_APP_DIR = config.root_path
 
     log = get_logger("client")
-    config = KaiRpcApplicationConfig(
-        process_id=None,
-        root_path=SAMPLE_APP_DIR,
-        kai_backend_url=SERVER_URL,
-        log_dir_path=Path("./logs"),
-        model_provider=kai_config.models,
-        demo_mode=True,
-        cache_dir=cache_dir,
-        analyzer_lsp_java_bundle_path=ANALYSIS_BUNDLE_PATH,
-        analyzer_lsp_lsp_path=ANALYSIS_LSP_PATH,
-        analyzer_lsp_rpc_path=ANALYSIS_RPC_PATH,
-        analyzer_lsp_rules_path=ANALYSIS_RULES_PATH,
-        analyzer_lsp_dep_labels_path=ANALYSIS_DEP_LABELS_FILE,
-        enable_reflection=False,
-    )
 
     rpc_subprocess = subprocess.Popen(  # trunk-ignore(bandit/B603)
-        [RPC_BINARY_PATH, "-c", "config.toml"],
+        [
+            RPC_BINARY_PATH,
+            "--log-level",
+            str(config.log_config.log_level),
+            "--stderr-log-level",
+            str(config.log_config.stderr_log_level),
+            "--file-log-level",
+            str(config.log_config.file_log_level),
+            "--log-dir-path",
+            config.log_config.log_dir_path,
+            "--log-file-name",
+            config.log_config.log_file_name,
+        ],
         stdin=subprocess.PIPE,
         stdout=subprocess.PIPE,
         env=os.environ,
@@ -132,13 +118,21 @@ def initialize_rpc_server(
             raise Exception(
                 f"Failed to initialize RPC server - {response.code} {response.message}"
             )
+        elif response.error is not None:
+            if isinstance(response.error, str):
+                raise Exception(f"Failed to initialize RPC server - {response.error}")
+            else:
+                raise Exception(
+                    f"Failed to initialize RPC server - {response.error.code} {response.error.message}"
+                )
+
         yield rpc_server
     except Exception as e:
         log.error("Failed to initialize the server:", e)
     finally:
         # send shutdown
         response = rpc_server.send_request("shutdown", params={})
-        log.debug(f"shutdown resposne -- {response}")
+        log.debug(f"shutdown response -- {response}")
         log.info("Stopping RPC Server")
         rpc_subprocess.wait()
         log.info("Stopped RPC Server")
@@ -237,8 +231,8 @@ def run_demo(report: Report, server: JsonRpcServer) -> None:
 
 
 def main() -> None:
-    kai_config = KaiConfig.model_validate_filepath("config.toml")
-    init_logging_from_config(kai_config)
+    kai_config = KaiRpcApplicationConfig.model_validate_filepath("initialize.toml")
+    init_logging_from_log_config(kai_config.log_config)
     start = time.time()
 
     tracer_provider: TracerProvider | None = None
@@ -254,7 +248,7 @@ def main() -> None:
     report = Report.load_report_from_file(coolstore_analysis_dir)
     try:
         pre_flight_checks()
-        with initialize_rpc_server(kai_config=kai_config) as server:
+        with initialize_rpc_server(kai_config) as server:
             run_demo(report, server)
         KAI_LOG.info(
             f"Total time to process '{coolstore_analysis_dir}' was {time.time()-start}s"

kai/analyzer.py

@@ -17,12 +17,12 @@
 CONST_KAI_ANALYZER_LOG_FILE = "kai-analyzer-server.log"
 
 
-def get_logfile_dir() -> str:
+def get_logfile_dir() -> Path:
     if not log:
         return PATH_KAI
     for h in log.handlers:
         if isinstance(h, logging.FileHandler):
-            return os.path.dirname(h.baseFilename)
+            return Path(os.path.dirname(h.baseFilename))
     return PATH_KAI
 
 
@@ -41,7 +41,7 @@ def __init__(
         analyzer_java_bundle_path: Path,
         dep_open_source_labels_path: Optional[Path],
     ) -> None:
-        """This will start and analyzer-lsp jsonrpc server"""
+        """This will start an analyzer-lsp jsonrpc server"""
         # trunk-ignore-begin(bandit/B603)
         args: list[str] = [
             str(analyzer_lsp_server_binary),

kai/constants.py

@@ -1,38 +1,40 @@
 import os
-import pathlib
 import sys
+from pathlib import Path
 from typing import Any
 
 """
 This file exists because we need to define some constants - specifically file
 paths - that are used in multiple places in the codebase. There might be a more
-robust solution, but for now, this should suffice
+robust solution, but for now, this should suffice.
+
+Note: We sometimes use `os.path.abspath` as opposed to `Path.resolve()` because
+the former will resolve relative paths, but not symlinks. This is in line with
+what Go does. 
 """
 
-PATH_KAI = os.path.dirname(os.path.abspath(__file__))
+PATH_KAI = Path(os.path.dirname(os.path.abspath(__file__)))
 
 # pyinstaller sets sys attributes to help determine when program runs in bin
 if getattr(sys, "frozen", False) and hasattr(sys, "_MEIPASS"):
-    PATH_KAI = sys._MEIPASS
+    PATH_KAI = Path(sys._MEIPASS)
 
-PATH_GIT_ROOT = os.path.join(PATH_KAI, "..")
+PATH_GIT_ROOT = Path(os.path.abspath(os.path.join(PATH_KAI, "..")))
 
-PATH_SOLUTION_SERVER_ROOT = os.path.join(PATH_GIT_ROOT, "kai_solution_server")
+PATH_SOLUTION_SERVER_ROOT = PATH_GIT_ROOT / "kai_solution_server"
 
-PATH_DATA = os.path.join(PATH_SOLUTION_SERVER_ROOT, "data")
+PATH_DATA = PATH_SOLUTION_SERVER_ROOT / "data"
 
-PATH_LLM_CACHE = os.path.join(PATH_KAI, "data", "llm_cache")
-PATH_BENCHMARKS = os.path.join(PATH_DATA, "benchmarks")
-PATH_MISC = os.path.join(PATH_DATA, "misc")
-PATH_SQL = os.path.join(PATH_DATA, "sql")
-PATH_TEMPLATES = os.path.join(PATH_DATA, "templates")
+PATH_LLM_CACHE = PATH_KAI / "data" / "llm_cache"
+PATH_BENCHMARKS = PATH_DATA / "benchmarks"
+PATH_MISC = PATH_DATA / "misc"
+PATH_SQL = PATH_DATA / "sql"
+PATH_TEMPLATES = PATH_DATA / "templates"
 
-PATH_LOCAL_REPO = os.path.join(
-    PATH_GIT_ROOT, "kai_solution_server/samples/sample_repos"
-)
+PATH_LOCAL_REPO = PATH_GIT_ROOT / "kai_solution_server" / "samples" / "sample_repos"
 
-PATH_TESTS = os.path.join(PATH_GIT_ROOT, "tests")
-PATH_TEST_DATA = pathlib.Path(os.path.join(PATH_GIT_ROOT, "tests/test_data"))
+PATH_TESTS = PATH_GIT_ROOT / "tests"
+PATH_TEST_DATA = PATH_GIT_ROOT / "tests" / "test_data"
 
 
 def __clean_env() -> dict[str, Any]:

kai/jsonrpc/util.py

@@ -1,7 +1,9 @@
 import logging
-from typing import Any
+import os
+from pathlib import Path
+from typing import Annotated, Any
 
-from pydantic import AliasChoices, AliasGenerator, BaseModel, ConfigDict
+from pydantic import AfterValidator, AliasChoices, AliasGenerator, BaseModel, ConfigDict
 from pydantic.alias_generators import to_camel
 
 
@@ -28,7 +30,58 @@ def log_record_to_dict(record: logging.LogRecord) -> dict[str, Any]:
     }
 
 
+AutoAbsPath = Annotated[Path, AfterValidator(lambda x: Path(os.path.abspath(x)))]
+"""
+`AutoAbsPath` is a type that can be used with Pydantic models to automatically
+convert an inputted relative path to an absolute path. This is different than
+`.resolve()` because it will resolve relative paths, but not symlinks. For
+example:
+
+```python
+class TheModel(BaseModel):
+    the_path: AutoAbsPath
+
+the_model = TheModel(the_path="build/build.spec")
+print(the_model.the_path)  #/path/to/build/build.spec
+```
+"""
+
+
+def auto_abs_path_exists_validator(x: Path) -> Path:
+    x = Path(os.path.abspath(x))
+    if not x.exists():
+        raise Exception(f"Path `{x}` does not exist!")
+    return x
+
+
+AutoAbsPathExists = Annotated[
+    Path,
+    AfterValidator(auto_abs_path_exists_validator),
+]
+
+AutoUpperStr = Annotated[str, AfterValidator(lambda x: x.upper())]
+"""
+`AutoUpperStr` is a type that can be used with Pydantic models to automatically
+convert an inputted string to uppercase.
+"""
+
+
 class CamelCaseBaseModel(BaseModel):
+    """
+    This class will accept both camelCase and snake_case keys when creating an
+    instance of the model. When serializing, it will produce camelCase keys.
+    For example:
+
+    ```python
+    class TheModel(CamelCaseBaseModel):
+        the_thing: str
+
+    a = TheModel.model_validate({"theThing": "hello"})  # Works!
+    b = TheModel.model_validate({"the_thing": "hello"})  # Works!
+    c = TheModel(the_thing="hello").model_dump()  # {"theThing": "hello"}
+    ```
+    """
+
     model_config = ConfigDict(
         alias_generator=AliasGenerator(
             validation_alias=lambda field_name: AliasChoices(