Merge pull request #887 from ATheorell/naming_suggestions

Naming suggestions

gpt_engineer/applications/cli/cli_agent.py

@@ -1,4 +1,4 @@
-from gpt_engineer.core.code import Code
+from gpt_engineer.core.files_dict import FilesDict
 
 # from gpt_engineer.core.default.git_version_manager import GitVersionManager
 from gpt_engineer.core.ai import AI
@@ -8,22 +8,26 @@
     execute_entrypoint,
     improve,
 )
-from gpt_engineer.core.repository import Repository
-from gpt_engineer.core.default.on_disk_repository import OnDiskRepository
-from gpt_engineer.core.execution_env import ExecutionEnv
-from gpt_engineer.core.default.on_disk_execution_env import OnDiskExecutionEnv
+from gpt_engineer.core.base_memory import BaseMemory
+from gpt_engineer.core.default.disk_memory import DiskMemory
+from gpt_engineer.core.base_execution_env import BaseExecutionEnv
+from gpt_engineer.core.default.disk_execution_env import DiskExecutionEnv
 from gpt_engineer.core.default.paths import memory_path, ENTRYPOINT_FILE, PREPROMPTS_PATH
-from gpt_engineer.core.agent import Agent
+from gpt_engineer.core.base_agent import BaseAgent
 from gpt_engineer.core.preprompts_holder import PrepromptsHolder
 from typing import TypeVar, Callable, Union
 from pathlib import Path
 
-CodeGenType = TypeVar("CodeGenType", bound=Callable[[AI, str, Repository], Code])
-CodeProcessor = TypeVar("CodeProcessor", bound=Callable[[AI, ExecutionEnv, Code], Code])
-ImproveType = TypeVar("ImproveType", bound=Callable[[AI, str, Code, Repository], Code])
+CodeGenType = TypeVar("CodeGenType", bound=Callable[[AI, str, BaseMemory], FilesDict])
+CodeProcessor = TypeVar(
+    "CodeProcessor", bound=Callable[[AI, BaseExecutionEnv, FilesDict], FilesDict]
+)
+ImproveType = TypeVar(
+    "ImproveType", bound=Callable[[AI, str, FilesDict, BaseMemory], FilesDict]
+)
 
 
-class CliAgent(Agent):
+class CliAgent(BaseAgent):
     """
     The `Agent` class is responsible for managing the lifecycle of code generation and improvement.
 
@@ -51,7 +55,7 @@ class CliAgent(Agent):
                 prompt (str): A string prompt that guides the code generation process.
 
             Returns:
-                Code: An instance of the `Code` class containing the generated code.
+                FilesDict: An instance of the `Code` class containing the generated code.
 
         improve(self, prompt: str) -> Code:
             Improves an existing piece of code using the AI and step bundle based on the provided prompt.
@@ -61,13 +65,13 @@ class CliAgent(Agent):
                 prompt (str): A string prompt that guides the code improvement process.
 
             Returns:
-                Code: An instance of the `Code` class containing the improved code.
+                FilesDict: An instance of the `Code` class containing the improved code.
     """
 
     def __init__(
         self,
-        memory: Repository,
-        execution_env: ExecutionEnv,
+        memory: BaseMemory,
+        execution_env: BaseExecutionEnv,
         ai: AI = None,
         code_gen_fn: CodeGenType = gen_code,
         improve_fn: ImproveType = improve,
@@ -85,8 +89,8 @@ def __init__(
     @classmethod
     def with_default_config(
         cls,
-        memory: OnDiskRepository,
-        execution_env: OnDiskExecutionEnv,
+        memory: DiskMemory,
+        execution_env: DiskExecutionEnv,
         ai: AI = None,
         code_gen_fn: CodeGenType = gen_code,
         improve_fn: ImproveType = improve,
@@ -103,25 +107,37 @@ def with_default_config(
             preprompts_holder=preprompts_holder or PrepromptsHolder(PREPROMPTS_PATH),
         )
 
-    def init(self, prompt: str) -> Code:
-        code = self.code_gen_fn(self.ai, prompt, self.memory, self.preprompts_holder)
-        entrypoint = gen_entrypoint(self.ai, code, self.memory, self.preprompts_holder)
-        code = Code(code | entrypoint)
-        code = self.process_code_fn(
-            self.ai, self.execution_env, code, preprompts_holder=self.preprompts_holder
+    def init(self, prompt: str) -> FilesDict:
+        files_dict = self.code_gen_fn(
+            self.ai, prompt, self.memory, self.preprompts_holder
+        )
+        entrypoint = gen_entrypoint(
+            self.ai, files_dict, self.memory, self.preprompts_holder
         )
-        return code
+        files_dict = FilesDict(files_dict | entrypoint)
+        files_dict = self.process_code_fn(
+            self.ai,
+            self.execution_env,
+            files_dict,
+            preprompts_holder=self.preprompts_holder,
+        )
+        return files_dict
 
     def improve(
-        self, code: Code, prompt: str, execution_command: str | None = None
-    ) -> Code:
-        code = self.improve_fn(self.ai, prompt, code, self.memory, self.preprompts_holder)
-        if not execution_command and ENTRYPOINT_FILE not in code:
+        self, files_dict: FilesDict, prompt: str, execution_command: str | None = None
+    ) -> FilesDict:
+        files_dict = self.improve_fn(
+            self.ai, prompt, files_dict, self.memory, self.preprompts_holder
+        )
+        if not execution_command and ENTRYPOINT_FILE not in files_dict:
             entrypoint = gen_entrypoint(
-                self.ai, code, self.memory, self.preprompts_holder
+                self.ai, files_dict, self.memory, self.preprompts_holder
             )
-            code = Code(code | entrypoint)
-        code = self.process_code_fn(
-            self.ai, self.execution_env, code, preprompts_holder=self.preprompts_holder
+            files_dict = FilesDict(files_dict | entrypoint)
+        files_dict = self.process_code_fn(
+            self.ai,
+            self.execution_env,
+            files_dict,
+            preprompts_holder=self.preprompts_holder,
         )
-        return code
+        return files_dict

gpt_engineer/applications/cli/collect.py

@@ -27,7 +27,7 @@
 
 from typing import List, Tuple
 
-from gpt_engineer.core.default.on_disk_repository import OnDiskRepository
+from gpt_engineer.core.default.disk_memory import DiskMemory
 from gpt_engineer.applications.cli.learning import (
     Learning,
     extract_learning,
@@ -68,7 +68,7 @@ def collect_learnings(
     model: str,
     temperature: float,
     config: any,
-    memory: OnDiskRepository,
+    memory: DiskMemory,
     review: Review,
 ):
     """
@@ -134,7 +134,7 @@ def collect_and_send_human_review(
     model: str,
     temperature: float,
     config: Tuple[str, ...],
-    memory: OnDiskRepository,
+    memory: DiskMemory,
 ):
     """
     Collects human feedback on the code and stores it in memory.

gpt_engineer/applications/cli/file_selector.py

@@ -47,9 +47,9 @@
 from pathlib import Path
 from typing import List, Union
 
-from gpt_engineer.core.default.on_disk_repository import OnDiskRepository
+from gpt_engineer.core.default.disk_memory import DiskMemory
 from gpt_engineer.core.default.paths import metadata_path
-from gpt_engineer.core.code import Code
+from gpt_engineer.core.files_dict import FilesDict
 
 IGNORE_FOLDERS = {"site-packages", "node_modules", "venv"}
 FILE_LIST_NAME = "file_list.txt"
@@ -326,15 +326,15 @@ def is_in_ignoring_extensions(path: Path) -> bool:
     return is_hidden and is_pycache
 
 
-def ask_for_files(project_path: Union[str, Path]) -> Code:
+def ask_for_files(project_path: Union[str, Path]) -> FilesDict:
     """
     Ask user to select files to improve.
     It can be done by terminal, gui, or using the old selection.
 
     Returns:
         dict[str, str]: Dictionary where key = file name and value = file path
     """
-    metadata_db = OnDiskRepository(metadata_path(project_path))
+    metadata_db = DiskMemory(metadata_path(project_path))
     if FILE_LIST_NAME in metadata_db:
         print(
             f"File list detected at {metadata_db.path / FILE_LIST_NAME}. "
@@ -399,10 +399,10 @@ def ask_for_files(project_path: Union[str, Path]) -> Code:
         for file in file_list:
             with open(os.path.join(project_path, file.strip()), "r") as content:
                 content_dict[file.strip()] = content.read()
-    return Code(content_dict)
+    return FilesDict(content_dict)
 
 
-def get_all_code(project_path: str) -> Code:
+def get_all_code(project_path: str) -> FilesDict:
     file_selection = terminal_file_selector(project_path, all=True)
     # ToDO: Replace this hack that makes all file_path relative to the right thing
     file_selection = [
@@ -414,7 +414,7 @@ def get_all_code(project_path: str) -> Code:
     for file in file_selection:
         with open(os.path.join(project_path, file.strip()), "r") as content:
             content_dict[file.strip()] = content.read()
-    return Code(content_dict)
+    return FilesDict(content_dict)
 
 
 def gui_file_selector(input_path: str) -> List[str]:

gpt_engineer/applications/cli/learning.py

@@ -52,7 +52,7 @@
 from dataclasses_json import dataclass_json
 from termcolor import colored
 
-from gpt_engineer.core.default.on_disk_repository import OnDiskRepository
+from gpt_engineer.core.default.disk_memory import DiskMemory
 
 
 @dataclass_json
@@ -188,7 +188,7 @@ def extract_learning(
     model: str,
     temperature: float,
     config: Tuple[str, ...],
-    memory: OnDiskRepository,
+    memory: DiskMemory,
     review: Review,
 ) -> Learning:
     """

gpt_engineer/applications/cli/main.py

@@ -32,8 +32,8 @@
 import typer
 from dotenv import load_dotenv
 
-from gpt_engineer.core.default.disk_store import FileStore
-from gpt_engineer.core.default.on_disk_repository import OnDiskRepository
+from gpt_engineer.core.default.file_store import FileStore
+from gpt_engineer.core.default.disk_memory import DiskMemory
 from gpt_engineer.core.ai import AI
 from gpt_engineer.core.default.paths import PREPROMPTS_PATH, memory_path
 from gpt_engineer.applications.cli.file_selector import ask_for_files, get_all_code
@@ -49,7 +49,7 @@
 from gpt_engineer.applications.cli.cli_agent import CliAgent
 from gpt_engineer.applications.cli.collect import collect_and_send_human_review
 from gpt_engineer.core.preprompts_holder import PrepromptsHolder
-from gpt_engineer.core.default.on_disk_execution_env import OnDiskExecutionEnv
+from gpt_engineer.core.default.disk_execution_env import DiskExecutionEnv
 import logging
 
 app = typer.Typer()  # creates a CLI app
@@ -64,7 +64,7 @@ def load_env_if_needed():
     openai.api_key = os.getenv("OPENAI_API_KEY")
 
 
-def load_prompt(input_repo: OnDiskRepository, improve_mode):
+def load_prompt(input_repo: DiskMemory, improve_mode):
     if input_repo.get("prompt"):
         return input_repo.get("prompt")
 
@@ -101,12 +101,12 @@ def main(
         False,
         "--improve",
         "-i",
-        help="Improve code from existing project.",
+        help="Improve files_dict from existing project.",
     ),
     improve_all_mode: bool = typer.Option(
         False,
         "--improve_all_experimental",
-        help="Improve code from existing project, without manually choosing which files to improve, using vector store (EXPERIMENTAL).",
+        help="Improve files_dict from existing project, without manually choosing which files to improve, using vector store (EXPERIMENTAL).",
     ),
     lite_mode: bool = typer.Option(
         False,
@@ -171,7 +171,7 @@ def main(
     # )  # resolve the string to a valid path (eg "a/b/../c" to "a/c")
     path = Path(project_path)  # .absolute()
     print("Running gpt-engineer in", path.absolute(), "\n")
-    prompt = load_prompt(OnDiskRepository(path), improve_mode)
+    prompt = load_prompt(DiskMemory(path), improve_mode)
     # configure generation function
     if clarify_mode:
         code_gen_fn = clarified_gen
@@ -192,8 +192,8 @@ def main(
 
     preprompts_path = get_preprompts_path(use_custom_preprompts, Path(project_path))
     preprompts_holder = PrepromptsHolder(preprompts_path)
-    memory = OnDiskRepository(memory_path(project_path))
-    execution_env = OnDiskExecutionEnv()
+    memory = DiskMemory(memory_path(project_path))
+    execution_env = DiskExecutionEnv()
     agent = CliAgent.with_default_config(
         memory,
         execution_env,
@@ -207,14 +207,14 @@ def main(
     store = FileStore(project_path)
     if improve_mode:
         if improve_all_mode:
-            code = store.download()
+            files_dict = store.download()
         else:
-            code = ask_for_files(project_path)
-        code = agent.improve(code, prompt)
+            files_dict = ask_for_files(project_path)
+        files_dict = agent.improve(files_dict, prompt)
     else:
-        code = agent.init(prompt)
+        files_dict = agent.init(prompt)
 
-    store.upload(code)
+    store.upload(files_dict)
 
     # collect user feedback if user consents
     config = (code_gen_fn.__name__, execution_fn.__name__)

gpt_engineer/benchmark/benchmarks/gpteng/eval_tools.py

@@ -17,7 +17,7 @@
 
 from tabulate import tabulate
 
-from gpt_engineer.core.code import Code
+from gpt_engineer.core.files_dict import FilesDict
 
 EVAL_LIST_NAME = "evaluations"  # the top level list in the YAML file
 
@@ -27,27 +27,27 @@ def check_language(eval_d: dict) -> None:
         raise Exception(f"Language: {eval_d['language']} is not supported.")
 
 
-def assert_exists_in_source_code(eval_d: dict, files: Code) -> bool:
+def assert_exists_in_source_code(eval_d: dict, files_dict: FilesDict) -> bool:
     """Checks of some text exists in the source code."""
-    source_body = files[eval_d["source_file"]]
+    source_body = files_dict[eval_d["source_file"]]
     return source_body.find(eval_d["existing_string"]) > -1
 
 
-def run_code_class_has_property(eval_d: dict, files: Code) -> bool:
+def run_code_class_has_property(eval_d: dict, files_dict: FilesDict) -> bool:
     """Will execute code, then check if the code has the desired proprty."""
     check_language(eval_d)
-    source_body = files[eval_d["source_file"]]
+    source_body = files_dict[eval_d["source_file"]]
     exec(source_body)
 
     class_ref = locals().get(eval_d["class_name"])
     ob = class_ref()
     return hasattr(ob, eval_d["property_name"])
 
 
-def run_code_class_has_property_w_value(eval_d: dict, files: Code) -> bool:
+def run_code_class_has_property_w_value(eval_d: dict, files_dict: FilesDict) -> bool:
     """Will execute code, then check if the code has the desired proprty."""
     check_language(eval_d)
-    source_body = files[eval_d["source_file"]]
+    source_body = files_dict[eval_d["source_file"]]
     exec(source_body)
 
     class_ref = locals().get(eval_d["class_name"])
@@ -58,28 +58,28 @@ def run_code_class_has_property_w_value(eval_d: dict, files: Code) -> bool:
     return getattr(ob, eval_d["property_name"]) == eval_d["expected_value"]
 
 
-def run_code_eval_function(eval_d: dict, files: Code) -> bool:
+def run_code_eval_function(eval_d: dict, files_dict: FilesDict) -> bool:
     """Similar to run_code_class_has_property() except is evaluates a function call."""
     check_language(eval_d)
-    source_body = files[eval_d["source_file"]]
+    source_body = files_dict[eval_d["source_file"]]
     exec(source_body)
     function_ref = globals().get(eval_d["function_name"])
 
     # TODO: add the ability to have function arguments
     return function_ref() == eval_d["expected_value"]
 
 
-def check_evaluation_component(eval_d: dict, files: Code) -> bool:
+def check_evaluation_component(eval_d: dict, files_dict: FilesDict) -> bool:
     """Switch on evaluation components"""
     test_type = eval_d.get("type")
     if test_type == "assert_exists_in_source_code":
-        return assert_exists_in_source_code(eval_d, files)
+        return assert_exists_in_source_code(eval_d, files_dict)
     elif test_type == "run_code_class_has_property":
-        return run_code_class_has_property(eval_d, files)
+        return run_code_class_has_property(eval_d, files_dict)
     elif test_type == "run_code_class_has_property_w_value":
-        return run_code_class_has_property_w_value(eval_d, files)
+        return run_code_class_has_property_w_value(eval_d, files_dict)
     elif test_type == "run_code_eval_function":
-        return run_code_eval_function(eval_d, files)
+        return run_code_eval_function(eval_d, files_dict)
     # The following are for new code
     else:
         raise Exception(f"Test type '{test_type}' is not recognized.")