Add Command and SQL steps (#1181)

* Add new steps

* update init file

* change command to shell

* bugfixes and changes

* lock update

* update callsql

* make username optional since some dbs take username in connect args instead

* fix test and filter url kwargs by none

* auto parse dict inputs

* bump version

* lint and update deps

patchwork/common/utils/input_parsing.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import json
 from collections.abc import Iterable, Mapping
 
 from typing_extensions import AnyStr, Union
@@ -69,3 +70,23 @@ def parse_to_list(
             continue
         rv.append(stripped_value)
     return rv
+
+
+def parse_to_dict(possible_dict, limit=-1):
+    if possible_dict is None and limit == 0:
+        return None
+
+    if isinstance(possible_dict, dict):
+        new_dict = dict()
+        for k, v in possible_dict.items():
+            new_dict[k] = parse_to_dict(v, limit - 1)
+        return new_dict
+    elif isinstance(possible_dict, str):
+        try:
+            new_dict = json.loads(possible_dict, strict=False)
+        except json.JSONDecodeError:
+            return possible_dict
+
+        return parse_to_dict(new_dict, limit - 1)
+    else:
+        return possible_dict

patchwork/common/utils/utils.py

@@ -2,10 +2,14 @@
 
 import atexit
 import dataclasses
+import random
 import signal
+import string
 import tempfile
+from collections.abc import Mapping
 from pathlib import Path
 
+import chevron
 import tiktoken
 from chardet.universaldetector import UniversalDetector
 from git import Head, Repo
@@ -19,6 +23,20 @@
 _NEWLINES = {"\n", "\r\n", "\r"}
 
 
+def mustache_render(template: str, data: Mapping) -> str:
+    if len(data.keys()) < 1:
+        return template
+
+    chevron.render.__globals__["_html_escape"] = lambda x: x
+    return chevron.render(
+        template=template,
+        data=data,
+        partials_path=None,
+        partials_ext="".join(random.choices(string.ascii_uppercase + string.digits, k=32)),
+        partials_dict=dict(),
+    )
+
+
 def detect_newline(path: str | Path) -> str | None:
     with open(path, "r", newline="") as f:
         lines = f.read().splitlines(keepends=True)

patchwork/step.py

@@ -11,7 +11,16 @@
 
 from enum import Enum
 
-from typing_extensions import Any, Dict, List, Optional, Union, is_typeddict
+from typing_extensions import (
+    Any,
+    Collection,
+    Dict,
+    List,
+    Optional,
+    Type,
+    Union,
+    is_typeddict,
+)
 
 from patchwork.logger import logger
 
@@ -45,10 +54,9 @@ def __init__(self, inputs: DataPoint):
         """
 
         # check if the inputs have the required keys
-        if self.__input_class is not None:
-            missing_keys = self.__input_class.__required_keys__.difference(inputs.keys())
-            if len(missing_keys) > 0:
-                raise ValueError(f"Missing required data: {list(missing_keys)}")
+        missing_keys = self.find_missing_inputs(inputs)
+        if len(missing_keys) > 0:
+            raise ValueError(f"Missing required data: {list(missing_keys)}")
 
         # store the inputs
         self.inputs = inputs
@@ -64,19 +72,25 @@ def __init__(self, inputs: DataPoint):
         self.original_run = self.run
         self.run = self.__managed_run
 
-    def __init_subclass__(cls, **kwargs):
-        input_class = kwargs.get("input_class", None) or getattr(cls, "input_class", None)
-        output_class = kwargs.get("output_class", None) or getattr(cls, "output_class", None)
+    def __init_subclass__(cls, input_class: Optional[Type] = None, output_class: Optional[Type] = None, **kwargs):
+        if cls.__name__ == "PreparePR":
+            print(1)
+        input_class = input_class or getattr(cls, "input_class", None)
+        if input_class is not None and not is_typeddict(input_class):
+            input_class = None
 
-        if input_class is not None and is_typeddict(input_class):
-            cls.__input_class = input_class
-        else:
-            cls.__input_class = None
+        output_class = output_class or getattr(cls, "output_class", None)
+        if output_class is not None and not is_typeddict(output_class):
+            output_class = None
 
-        if output_class is not None and is_typeddict(output_class):
-            cls.__output_class = output_class
-        else:
-            cls.__output_class = None
+        cls._input_class = input_class
+        cls._output_class = output_class
+
+    @classmethod
+    def find_missing_inputs(cls, inputs: DataPoint) -> Collection:
+        if getattr(cls, "_input_class", None) is None:
+            return []
+        return cls._input_class.__required_keys__.difference(inputs.keys())
 
     def __managed_run(self, *args, **kwargs) -> Any:
         self.debug(self.inputs)

patchwork/steps/CallSQL/CallSQL.py

@@ -0,0 +1,57 @@
+from __future__ import annotations
+
+from sqlalchemy import URL, create_engine, exc, text
+
+from patchwork.common.utils.input_parsing import parse_to_dict
+from patchwork.common.utils.utils import mustache_render
+from patchwork.logger import logger
+from patchwork.step import Step, StepStatus
+from patchwork.steps.CallSQL.typed import CallSQLInputs, CallSQLOutputs
+
+
+class CallSQL(Step, input_class=CallSQLInputs, output_class=CallSQLOutputs):
+    def __init__(self, inputs: dict):
+        super().__init__(inputs)
+        query_template_data = inputs.get("db_query_template_values", {})
+        self.query = mustache_render(inputs["db_query"], query_template_data)
+        self.__build_engine(inputs)
+
+    def __build_engine(self, inputs: dict):
+        dialect = inputs["db_dialect"]
+        driver = inputs.get("db_driver")
+        dialect_plus_driver = f"{dialect}+{driver}" if driver is not None else dialect
+        kwargs = dict(
+            username=inputs.get("db_username"),
+            host=inputs.get("db_host", "localhost"),
+            port=inputs.get("db_port", 5432),
+            password=inputs.get("db_password"),
+            database=inputs.get("db_database"),
+            query=parse_to_dict(inputs.get("db_params")),
+        )
+        connection_url = URL.create(
+            dialect_plus_driver,
+            **{k: v for k, v in kwargs.items() if v is not None},
+        )
+
+        connect_args = None
+        if inputs.get("db_driver_args") is not None:
+            connect_args = parse_to_dict(inputs.get("db_driver_args"))
+
+        self.engine = create_engine(connection_url, connect_args=connect_args)
+        with self.engine.connect() as conn:
+            conn.execute(text("SELECT 1"))
+        return self.engine
+
+    def run(self) -> dict:
+        try:
+            rv = []
+            with self.engine.begin() as conn:
+                cursor = conn.execute(text(self.query))
+                for row in cursor:
+                    result = row._asdict()
+                    rv.append(result)
+            logger.info(f"Retrieved {len(rv)} rows!")
+            return dict(results=rv)
+        except exc.InvalidRequestError as e:
+            self.set_status(StepStatus.FAILED, f"`{self.query}` failed with message:\n{e}")
+            return dict(results=[])

patchwork/steps/CallSQL/typed.py

@@ -0,0 +1,24 @@
+from __future__ import annotations
+
+from typing_extensions import Any, TypedDict
+
+
+class __RequiredCallSQLInputs(TypedDict):
+    db_dialect: str
+    db_query: str
+
+
+class CallSQLInputs(__RequiredCallSQLInputs, total=False):
+    db_driver: str
+    db_username: str
+    db_password: str
+    db_host: str
+    db_port: int
+    db_name: str
+    db_params: dict[str, Any]
+    db_driver_args: dict[str, Any]
+    db_query_template_values: dict[str, Any]
+
+
+class CallSQLOutputs(TypedDict):
+    results: list[dict[str, Any]]

patchwork/steps/CallShell/CallShell.py

@@ -0,0 +1,58 @@
+from __future__ import annotations
+
+import shlex
+import subprocess
+from pathlib import Path
+
+from patchwork.common.utils.utils import mustache_render
+from patchwork.logger import logger
+from patchwork.step import Step, StepStatus
+from patchwork.steps.CallShell.typed import CallShellInputs, CallShellOutputs
+
+
+class CallShell(Step, input_class=CallShellInputs, output_class=CallShellOutputs):
+    def __init__(self, inputs: dict):
+        super().__init__(inputs)
+        script_template_values = inputs.get("script_template_values", {})
+        self.script = mustache_render(inputs["script"], script_template_values)
+        self.working_dir = inputs.get("working_dir", Path.cwd())
+        self.env = self.__parse_env_text(inputs.get("env", ""))
+
+    @staticmethod
+    def __parse_env_text(env_text: str) -> dict[str, str]:
+        env_spliter = shlex.shlex(env_text, posix=True)
+        env_spliter.whitespace_split = True
+        env_spliter.whitespace += ";"
+
+        env: dict[str, str] = dict()
+        for env_assign in env_spliter:
+            env_assign_spliter = shlex.shlex(env_assign, posix=True)
+            env_assign_spliter.whitespace_split = True
+            env_assign_spliter.whitespace += "="
+            env_parts = list(env_assign_spliter)
+            if len(env_parts) < 1:
+                continue
+
+            env_assign_target = env_parts[0]
+            if len(env_parts) < 2:
+                logger.error(f"{env_assign_target} is not assigned anything, skipping...")
+                continue
+            if len(env_parts) > 2:
+                logger.error(f"{env_assign_target} has more than 1 assignment, skipping...")
+                continue
+            env[env_assign_target] = env_parts[1]
+
+        return env
+
+    def run(self) -> dict:
+        p = subprocess.run(self.script, shell=True, capture_output=True, text=True, cwd=self.working_dir, env=self.env)
+        try:
+            p.check_returncode()
+        except subprocess.CalledProcessError as e:
+            self.set_status(
+                StepStatus.FAILED,
+                f"Script failed.",
+            )
+        logger.info(f"stdout: \n{p.stdout}")
+        logger.info(f"stderr:\n{p.stderr}")
+        return dict(stdout_output=p.stdout, stderr_output=p.stderr)

patchwork/steps/CallShell/typed.py

@@ -0,0 +1,19 @@
+from __future__ import annotations
+
+from typing_extensions import Annotated, Any, TypedDict
+
+from patchwork.common.utils.step_typing import StepTypeConfig
+
+
+class __RequiredCallShellInputs(TypedDict):
+    script: str
+
+
+class CallShellInputs(__RequiredCallShellInputs, total=False):
+    working_dir: Annotated[str, StepTypeConfig(is_path=True)]
+    env: str
+    script_template_values: dict[str, Any]
+
+
+class CallShellOutputs(TypedDict):
+    stdout_output: str

patchwork/steps/FixIssue/FixIssue.py

@@ -1,10 +1,8 @@
-import difflib
 import re
 from pathlib import Path
 from typing import Any, Optional
 
-from git import Repo, InvalidGitRepositoryError
-from patchwork.logger import logger
+from git import InvalidGitRepositoryError, Repo
 from openai.types.chat import ChatCompletionMessageParam
 
 from patchwork.common.client.llm.aio import AioLlmClient
@@ -15,6 +13,7 @@
     AnalyzeImplementStrategy,
 )
 from patchwork.common.tools import CodeEditTool, Tool
+from patchwork.logger import logger
 from patchwork.step import Step
 from patchwork.steps.FixIssue.typed import FixIssueInputs, FixIssueOutputs
 
@@ -100,7 +99,7 @@ def is_stop(self, messages: list[ChatCompletionMessageParam]) -> bool:
 class FixIssue(Step, input_class=FixIssueInputs, output_class=FixIssueOutputs):
     def __init__(self, inputs):
         """Initialize the FixIssue step.
-        
+
         Args:
             inputs: Dictionary containing input parameters including:
                 - base_path: Optional path to the repository root
@@ -145,12 +144,12 @@ def __init__(self, inputs):
 
     def run(self):
         """Execute the FixIssue step.
-        
+
         This method:
         1. Executes the multi-turn LLM conversation to analyze and fix the issue
         2. Tracks file modifications made by the CodeEditTool
         3. Generates in-memory diffs for all modified files
-        
+
         Returns:
             dict: Dictionary containing list of modified files with their diffs
         """
@@ -162,8 +161,7 @@ def run(self):
             if not isinstance(tool, CodeEditTool):
                 continue
             tool_modified_files = [
-                dict(path=str(file_path.relative_to(cwd)), diff="")
-                for file_path in tool.tool_records["modified_files"]
+                dict(path=str(file_path.relative_to(cwd)), diff="") for file_path in tool.tool_records["modified_files"]
             ]
             modified_files.extend(tool_modified_files)
 
@@ -174,7 +172,7 @@ def run(self):
                 file = modified_file["path"]
                 try:
                     # Try to get the diff using git
-                    diff = self.repo.git.diff('HEAD', file)
+                    diff = self.repo.git.diff("HEAD", file)
                     modified_file["diff"] = diff or ""
                 except Exception as e:
                     # Git-specific errors (untracked files, etc) - keep empty diff

patchwork/steps/FixIssue/typed.py

@@ -1,4 +1,4 @@
-from typing_extensions import Annotated, Dict, List, TypedDict
+from typing_extensions import Annotated, List, TypedDict
 
 from patchwork.common.constants import TOKEN_URL
 from patchwork.common.utils.step_typing import StepTypeConfig
@@ -37,19 +37,21 @@ class FixIssueInputs(__FixIssueRequiredInputs, total=False):
 
 class ModifiedFile(TypedDict):
     """Represents a file that has been modified by the FixIssue step.
-    
+
     Attributes:
         path: The relative path to the modified file from the repository root
         diff: A unified diff string showing the changes made to the file.
               Generated using Python's difflib to compare the original and
               modified file contents in memory.
-    
+
     Note:
         The diff is generated by comparing file contents before and after
         modifications, without relying on version control systems.
     """
+
     path: str
     diff: str
 
+
 class FixIssueOutputs(TypedDict):
     modified_files: List[ModifiedFile]