wcgw 2.6.2__py3-none-any.whl → 2.7.0__py3-none-any.whl

wcgw/client/file_ops/search_replace.py

@@ -0,0 +1,119 @@
+import re
+from typing import Callable
+
+from .diff_edit import FileEditInput, FileEditOutput
+
+
+def search_replace_edit(
+    lines: list[str], original_content: str, logger: Callable[[str], object]
+) -> tuple[str, str]:
+    if not lines:
+        raise Exception("Error: No input to search replace edit")
+    original_lines = original_content.split("\n")
+    n_lines = len(lines)
+    i = 0
+    search_replace_blocks = list[tuple[list[str], list[str]]]()
+    while i < n_lines:
+        if re.match(r"^<<<<<<+\s*SEARCH\s*$", lines[i]):
+            search_block = []
+            i += 1
+            while i < n_lines and not re.match(r"^======*\s*$", lines[i]):
+                search_block.append(lines[i])
+                i += 1
+            i += 1
+            if not search_block:
+                raise Exception("SEARCH block can not be empty")
+            replace_block = []
+            while i < n_lines and not re.match(r"^>>>>>>+\s*REPLACE\s*$", lines[i]):
+                replace_block.append(lines[i])
+                i += 1
+            i += 1
+
+            for line in search_block:
+                logger("> " + line)
+            logger("=======")
+            for line in replace_block:
+                logger("< " + line)
+            logger("\n\n\n\n")
+
+            search_replace_blocks.append((search_block, replace_block))
+        else:
+            i += 1
+
+    if not search_replace_blocks:
+        raise Exception(
+            "No valid search replace blocks found, ensure your SEARCH/REPLACE blocks are formatted correctly"
+        )
+
+    edited_content, comments_ = greedy_context_replace(
+        original_lines, [[x] for x in search_replace_blocks], original_lines, set(), 0
+    )
+    edited_file = "\n".join(edited_content)
+    if not comments_:
+        comments = "Edited successfully"
+    else:
+        comments = (
+            "Edited successfully. However, following warnings were generated while matching search blocks.\n"
+            + "\n".join(comments_)
+        )
+    return edited_file, comments
+
+
+def greedy_context_replace(
+    original_lines: list[str],
+    search_replace_blocks: list[list[tuple[list[str], list[str]]]],
+    running_lines: list[str],
+    running_comments: set[str],
+    current_block_offset: int,
+) -> tuple[list[str], set[str]]:
+    if current_block_offset >= len(search_replace_blocks):
+        return running_lines, running_comments
+    current_blocks = search_replace_blocks[current_block_offset]
+
+    outputs = FileEditInput(running_lines, 0, current_blocks, 0).edit_file()
+    best_matches, is_error = FileEditOutput.get_best_match(outputs)
+
+    if is_error:
+        best_matches[0].replace_or_throw(3)
+        raise Exception("Shouldn't happen")
+
+    if len(best_matches) > 1:
+        # Duplicate found, try to ground using previous blocks.
+        if current_block_offset == 0:
+            raise Exception(f"""
+    The following block matched more than once:
+    ---
+    ```
+    {'\n'.join(current_blocks[-1][0])}
+    ```
+    """)
+
+        else:
+            search_replace_blocks = (
+                search_replace_blocks[: current_block_offset - 1]
+                + [search_replace_blocks[current_block_offset - 1] + current_blocks]
+                + search_replace_blocks[current_block_offset + 1 :]
+            )
+            try:
+                return greedy_context_replace(
+                    original_lines, search_replace_blocks, original_lines, set(), 0
+                )
+            except Exception:
+                raise Exception(f"""
+        The following block matched more than once:
+        ---
+        ```
+        {'\n'.join(current_blocks[-1][0])}
+        ```
+        """)
+
+    best_match = best_matches[0]
+    running_lines, comments = best_match.replace_or_throw(3)
+    running_comments = running_comments | comments
+    return greedy_context_replace(
+        original_lines,
+        search_replace_blocks,
+        running_lines,
+        running_comments,
+        current_block_offset + 1,
+    )

wcgw/client/mcp_server/server.py

@@ -82,6 +82,7 @@ async def handle_list_tools() -> list[types.Tool]:
 - If the user has mentioned a folder or file with unclear project root, use the file or folder as `any_workspace_path`.
 - If user has mentioned any files use `initial_files_to_read` to read, use absolute paths only.
 - If `any_workspace_path` is provided, a tree structure of the workspace will be shown.
+- Leave `any_workspace_path` as empty if no file or folder is mentioned.
 """,
         ),
         ToolParam(
@@ -152,6 +153,24 @@ async def handle_list_tools() -> list[types.Tool]:
 """
             + diffinstructions,
         ),
+        #         ToolParam(
+        #             inputSchema=KnowledgeTransfer.model_json_schema(),
+        #             name="KnowledgeTransfer",
+        #             description="""
+        # Write detailed description in order to do a KT, if the user asks for it.
+        # Save all information necessary for a person to understand the task and the problems.
+        # - `all_user_instructions` should contain all instructions user shared in the conversation.
+        # - `current_status_of_the_task` should contain only what is already achieved, not what's remaining.
+        # - `all_issues_snippets` should only contain snippets of error, traceback, file snippets, commands, etc., no comments or solutions (important!).
+        # - Be very verbose in `all_issues_snippets` providing as much error context as possible.
+        # - Provide an id if the user hasn't provided one.
+        # - This tool will return a text file path where the information is saved.
+        # - After the tool completes succesfully, tell the user the task id and the generate file path. (important!)
+        # - Leave arguments as empty string if they aren't relevant.
+        # - This tool marks end of your conversation, do not run any further tools after calling this.
+        # - Provide absolute file paths only in `relevant_file_paths` containing all relevant files.
+        # """,
+        #         ),
     ]
     if COMPUTER_USE_ON_DOCKER_ENABLED:
         tools += [

wcgw/client/memory.py

@@ -0,0 +1,52 @@
+import os
+
+from ..types_ import KnowledgeTransfer
+
+
+def get_app_dir_xdg() -> str:
+    xdg_data_dir = os.environ.get("XDG_DATA_HOME", os.path.expanduser("~/.local/share"))
+    return os.path.join(xdg_data_dir, "wcgw")
+
+
+def format_memory(task_memory: KnowledgeTransfer, relevant_files: str) -> str:
+    memory_data = f"""# Goal: {task_memory.objective}\n\n
+# Instructions:\n{task_memory.all_user_instructions}\n\n
+# Current Status:\n{task_memory.current_status_of_the_task}\n\n
+# Pending Issues:\n{task_memory.all_issues_snippets}\n\n
+# Build Instructions:\n{task_memory.build_and_development_instructions}\n"""
+
+    memory_data += "\n# Relevant Files:\n" + relevant_files
+
+    return memory_data
+
+
+def save_memory(task_memory: KnowledgeTransfer, relevant_files: str) -> str:
+    app_dir = get_app_dir_xdg()
+    memory_dir = os.path.join(app_dir, "memory")
+    os.makedirs(memory_dir, exist_ok=True)
+
+    task_id = task_memory.id
+    if not task_id:
+        raise Exception("Task id can not be empty")
+    memory_data = format_memory(task_memory, relevant_files)
+
+    memory_file = os.path.join(memory_dir, f"{task_id}.json")
+    memory_file_full = os.path.join(memory_dir, f"{task_id}.txt")
+
+    with open(memory_file_full, "w") as f:
+        f.write(memory_data)
+
+    with open(memory_file, "w") as f:
+        f.write(task_memory.model_dump_json())
+
+    return memory_file_full
+
+
+def load_memory(task_id: str) -> KnowledgeTransfer:
+    app_dir = get_app_dir_xdg()
+    memory_dir = os.path.join(app_dir, "memory")
+    memory_file = os.path.join(memory_dir, f"{task_id}.json")
+
+    with open(memory_file, "r") as f:
+        task_save = KnowledgeTransfer.model_validate_json(f.read())
+    return task_save

wcgw/client/openai_client.py

@@ -27,17 +27,20 @@ from ..types_ import (
     BashCommand,
     BashInteraction,
     FileEdit,
+    KnowledgeTransfer,
     ReadFiles,
     ReadImage,
     ResetShell,
     WriteIfEmpty,
 )
 from .common import CostData, History, Models, discard_input
+from .memory import load_memory
 from .openai_utils import get_input_cost, get_output_cost
 from .tools import (
     DoneFlag,
     ImageData,
     get_tool_output,
+    initialize,
     which_tool,
 )
 
@@ -117,19 +120,24 @@ def loop(
 
     history: History = []
     waiting_for_assistant = False
+
+    memory = None
     if resume:
-        if resume == "latest":
-            resume_path = sorted(Path(".wcgw").iterdir(), key=os.path.getmtime)[-1]
-        else:
-            resume_path = Path(resume)
-        if not resume_path.exists():
-            raise FileNotFoundError(f"File {resume} not found")
-        with resume_path.open() as f:
-            history = json.load(f)
-        if len(history) <= 2:
-            raise ValueError("Invalid history file")
-        first_message = ""
-        waiting_for_assistant = history[-1]["role"] != "assistant"
+        try:
+            memory = load_memory(resume)
+        except OSError:
+            if resume == "latest":
+                resume_path = sorted(Path(".wcgw").iterdir(), key=os.path.getmtime)[-1]
+            else:
+                resume_path = Path(resume)
+            if not resume_path.exists():
+                raise FileNotFoundError(f"File {resume} not found")
+            with resume_path.open() as f:
+                history = json.load(f)
+            if len(history) <= 2:
+                raise ValueError("Invalid history file")
+            first_message = ""
+            waiting_for_assistant = history[-1]["role"] != "assistant"
 
     my_dir = os.path.dirname(__file__)
 
@@ -202,10 +210,29 @@ def loop(
             ResetShell,
             description="Resets the shell. Use only if all interrupts and prompt reset attempts have failed repeatedly.",
         ),
+        openai.pydantic_function_tool(
+            KnowledgeTransfer,
+            description="""
+Write detailed description in order to do a KT, if the user asks for it.
+Save all information necessary for a person to understand the task and the problems.
+
+- `all_user_instructions` should contain all instructions user shared in the conversation.
+- `current_status_of_the_task` should contain only what is already achieved, not what's remaining.
+- `all_issues_snippets` should only contain snippets of error, traceback, file snippets, commands, etc., no comments or solutions (important!).
+- Be very verbose in `all_issues_snippets` providing as much error context as possible.
+- Provide an id if the user hasn't provided one. 
+- This tool will return a text file path where the information is saved.
+- After the tool completes succesfully, tell the user the task id and the generate file path. (important!)
+- Leave arguments as empty string if they aren't relevant.
+- This tool marks end of your conversation, do not run any further tools after calling this.
+- Provide absolute file paths only in `relevant_file_paths` containing all relevant files.
+""",
+        ),
     ]
-    uname_sysname = os.uname().sysname
-    uname_machine = os.uname().machine
 
+    initial_info = initialize(
+        os.getcwd(), [], resume if (memory and resume) else "", 8000
+    )
     system = f"""
 You're an expert software engineer with shell and code knowledge.
 
@@ -217,10 +244,7 @@ Instructions:
     - Do not provide code snippets unless asked by the user, instead directly add/edit the code.
     - Do not install new tools/packages before ensuring no such tools/package or an alternative already exists.
 
-System information:
-    - System: {uname_sysname}
-    - Machine: {uname_machine}
-    - Current directory: {os.getcwd()}
+{initial_info}
 
 """