feat(#22): add ability to review local changes without a PR

README.md

@@ -18,6 +18,7 @@ lgtm-ai is your AI-powered code review companion. It generates code reviews usin
 **Table of Contents**
 - [Quick Usage](#quick-usage)
   - [Review](#review)
+    - [Local Changes](#local-changes)
   - [Reviewer Guide](#reviewer-guide)
 - [Installation](#installation)
 - [How it works](#how-it-works)
@@ -64,6 +65,17 @@ This will generate a **review** like this one:
 <img src="./assets/review-comment.png" alt="lgtm-review-comment" height="250"/>
 
 
+#### Local Changes
+
+You can also review local changes without a pull request:
+
+```sh
+lgtm review --ai-api-key $OPENAI_API_KEY \
+            --model gpt-4.1 \
+            --compare main \
+            path/to/git/repo
+```
+
 ### Reviewer Guide
 
 ```sh
@@ -384,11 +396,12 @@ When it comes to preference for selecting options, lgtm follows this preference
 | silent               | Main (review + guide)  | 🟢 Optional                   | Suppress terminal output. Default: false.                                        |
 | ai_retries           | Main (review + guide)  | 🟢 Optional                   | Number of retries for AI agent queries. Default: 1.                              |
 | ai_input_tokens_limit| Main (review + guide)  | 🟢 Optional                   | Max input tokens for LLM. Default: 500,000. Use `"no-limit"` to disable.        |
-| git_api_key          | Main (review + guide)  | 🔴 Required*                  | API key for git service (GitHub/GitLab). Can't be given through config file. Also available through env variable `LGTM_GIT_API_KEY`.     |
+| git_api_key          | Main (review + guide)  | 🟡 Conditionally required     | API key for git service (GitHub/GitLab). Can't be given through config file. Also available through env variable `LGTM_GIT_API_KEY`. Required if reviewing a PR URL from a remote repository service (GitHub, GitLab, etc.).     |
 | ai_api_key           | Main (review + guide)  | 🔴 Required*                  | API key for AI model. Can't be given through config file. Also available through env variable `LGTM_AI_API_KEY`.                        |
 | technologies         | Review Only          | 🟢 Optional                   | List of technologies for reviewer expertise.                                     |
 | categories           | Review Only          | 🟢 Optional                   | Review categories. Defaults to all (`Quality`, `Correctness`, `Testing`, `Security`). |
 | additional_context   | Review Only          | 🟢 Optional                   | Extra context for the LLM (array of prompts/paths/URLs). Can't be given through the CLI |
+| compare              | Review Only          | 🟢 Optional                   | If reviewing local changes, what to compare against (branch, commit, range, etc.). CLI only. |
 | issues_url           | Issues Integration   | 🟢 Optional                   | Enables issue context. If set, `issues_platform` becomes required.                 |
 | issues_platform        | Issues Integration   | 🟡 Conditionally required     | Required if `issues_url` is set.                                                 |
 | issues_regex         | Issues Integration   | 🟢 Optional                   | Regex for issue ID extraction. Defaults to conventional commit compatible regex. |
@@ -409,7 +422,7 @@ These options apply to both reviews and guides generated by lgtm.
 - **silent**: Do not print the review in the terminal. Default is `false`.
 - **ai_retries**: How many times to retry calls to the LLM when they do not succeed. By default, this is set to 1 (no retries at all).
 - **ai_input_tokens_limit**: Set a limit on the input tokens sent to the LLM in total. Default is 500,000. To disable the limit, you can pass the string `"no-limit"`.
-- **git_api_key**: API key to post the review in the source system of the PR. Can be given as a CLI argument, or as an environment variable (`LGTM_GIT_API_KEY`).
+- **git_api_key**: API key to post the review in the source system of the PR. Can be given as a CLI argument, or as an environment variable (`LGTM_GIT_API_KEY`). You can omit this option if reviewing local changes.
 - **ai_api_key**: API key to call the selected AI model. Can be given as a CLI argument, or as an environment variable (`LGTM_AI_API_KEY`).
 
 #### Review options
@@ -419,6 +432,7 @@ These options are only used when performing reviews through the command `lgtm re
 - **technologies**: Specify, as a list of free strings, which technologies lgtm specializes in. This can help direct the reviewer towards specific technologies. By default, lgtm won't assume any technology and will just review the PR considering itself an "expert" in it.
 - **categories**: lgtm will, by default, evaluate several areas of the given PR (`Quality`, `Correctness`, `Testing`, and `Security`). You can choose any subset of these (e.g., if you are only interested in `Correctness`, you can configure `categories` so that lgtm does not evaluate the other missing areas).
 - **additional_context**: TOML array of extra context to send to the LLM. It supports setting the context directly in the `context` field, passing a relative file path so that lgtm downloads it from the repository, or passing any URL from which to download the context. Each element of the array must contain `prompt`, and either `context` (directly injecting context) or `file_url` (for directing lgtm to download it from there).
+- **compare**: When reviewing local changes (the positional argument to `lgtm` is a valid `git` path), you can choose what to compare against to generate a git diff. You can pass branch names, commits, etc. Default is `HEAD`. Only available as a CLI option.
 
 #### Issues Integration options
 

pyproject.toml

@@ -39,6 +39,7 @@ dependencies = [
     "pygithub>=2.6.1,<3.0.0",
     "httpx (>=0.28.1,<0.29.0)",
     "jinja2 (>=3.1.6,<4.0.0)",
+    "gitpython (>=3.1.45,<4.0.0)",
 ]
 
 [tool.poetry.group.dev.dependencies]

scripts/evaluate_review_quality.py

@@ -15,7 +15,7 @@
 from lgtm_ai.git_client.gitlab import GitlabClient
 from lgtm_ai.review import CodeReviewer
 from lgtm_ai.review.context import ContextRetriever
-from lgtm_ai.validators import parse_pr_url
+from lgtm_ai.validators import parse_target
 from rich.logging import RichHandler
 
 PRS_FOR_EVALUATION = {
@@ -94,7 +94,7 @@ def get_git_branch() -> str:
 def perform_review(
     output_dir: str, pr_url: str, pr_name: str, sample: int, model: SupportedAIModels, git_api_key: str, ai_api_key: str
 ) -> None:
-    url = parse_pr_url(mock.Mock(), "pr_url", pr_url)
+    url = parse_target(mock.Mock(), "pr_url", pr_url)
     git_client = GitlabClient(client=gitlab.Gitlab(private_token=git_api_key), formatter=MarkDownFormatter())
     code_reviewer = CodeReviewer(
         reviewer_agent=get_reviewer_agent_with_settings(),
@@ -106,7 +106,7 @@ def perform_review(
         ),
         config=ResolvedConfig(model=model, technologies=("Python", "Django", "FastAPI")),
     )
-    review = code_reviewer.review_pull_request(pr_url=url)
+    review = code_reviewer.review_pull_request(target=url)
     write_review_to_dir(model, output_dir, pr_name, sample, review)
 
 

src/lgtm_ai/__main__.py

@@ -14,7 +14,7 @@
 )
 from lgtm_ai.ai.schemas import AgentSettings, CommentCategory, SupportedAIModels, SupportedAIModelsList
 from lgtm_ai.base.constants import DEFAULT_HTTPX_TIMEOUT
-from lgtm_ai.base.schemas import IntOrNoLimit, IssuesPlatform, OutputFormat, PRUrl
+from lgtm_ai.base.schemas import IntOrNoLimit, IssuesPlatform, LocalRepository, OutputFormat, PRUrl
 from lgtm_ai.base.utils import git_source_supports_multiline_suggestions
 from lgtm_ai.config.constants import DEFAULT_INPUT_TOKEN_LIMIT
 from lgtm_ai.config.handler import ConfigHandler, PartialConfig, ResolvedConfig
@@ -31,7 +31,7 @@
 from lgtm_ai.validators import (
     IntOrNoLimitType,
     ModelChoice,
-    parse_pr_url,
+    parse_target,
     validate_model_url,
 )
 from rich.console import Console
@@ -56,7 +56,7 @@ def entry_point() -> None:
 def _common_options[**P, T](func: Callable[P, T]) -> Callable[P, T]:
     """Wrap a click command and adds common options for lgtm commands."""
 
-    @click.argument("pr-url", required=True, callback=parse_pr_url)
+    @click.argument("target", required=True, callback=parse_target)
     @click.option(
         "--model",
         type=ModelChoice(SupportedAIModelsList),
@@ -69,7 +69,10 @@ def _common_options[**P, T](func: Callable[P, T]) -> Callable[P, T]:
         default=None,
         callback=validate_model_url,
     )
-    @click.option("--git-api-key", help="The API key to the git service (GitLab, GitHub, etc.)")
+    @click.option(
+        "--git-api-key",
+        help="The API key to the git service (GitLab, GitHub, etc.). Required if the target is a PR URL.",
+    )
     @click.option("--ai-api-key", help="The API key to the AI model service (OpenAI, etc.)")
     @click.option("--config", type=click.STRING, help="Path to the configuration file.")
     @click.option(
@@ -134,8 +137,13 @@ def wrapper(*args: P.args, **kwargs: P.kwargs) -> T:
     type=click.Choice(get_args(CommentCategory)),
     help="List of categories the reviewer should focus on. If not provided, the reviewer will focus on all categories.",
 )
+@click.option(
+    "--compare",
+    default=None,
+    help="If reviewing a local repository, what to compare against (branch, commit, or HEAD for working dir). Default: HEAD",
+)
 def review(
-    pr_url: PRUrl,
+    target: PRUrl | LocalRepository,
     model: SupportedAIModels | None,
     model_url: str | None,
     git_api_key: str | None,
@@ -155,16 +163,26 @@ def review(
     issues_platform: IssuesPlatform | None,
     issues_api_key: str | None,
     issues_user: str | None,
+    compare: str | None,
 ) -> None:
-    """Review a Pull Request using AI.
+    """Review a Pull Request or local repository using AI.
+
+    TARGET can be either:
 
-    PR_URL is the URL of the pull request to review.
+        - A pull request URL (GitHub, GitLab, etc.).
+
+        - A local directory path (use --compare to specify what to compare against).
     """
     _set_logging_level(logger, verbose)
 
+    if compare and not isinstance(target, LocalRepository):
+        logger.warning(
+            "`--compare` option is only used when reviewing a local repository. Ignoring the provided value."
+        )
+
     logger.info("lgtm-ai version: %s", __version__)
-    logger.debug("Parsed PR URL: %s", pr_url)
-    logger.info("Starting review of %s", pr_url.full_url)
+    logger.debug("Parsed PR URL: %s", target)
+    logger.info("Starting review of %s", target.full_url)
     resolved_config = ConfigHandler(
         cli_args=PartialConfig(
             technologies=technologies,
@@ -184,14 +202,16 @@ def review(
             issues_platform=issues_platform,
             issues_api_key=issues_api_key,
             issues_user=issues_user,
+            compare=compare,
         ),
         config_file=config,
-    ).resolve_config()
+    ).resolve_config(target)
+
     agent_extra_settings = AgentSettings(retries=resolved_config.ai_retries)
     formatter: Formatter[Any] = MarkDownFormatter(
-        add_ranges_to_suggestions=git_source_supports_multiline_suggestions(pr_url.source)
+        add_ranges_to_suggestions=git_source_supports_multiline_suggestions(target.source)
     )
-    git_client = get_git_client(source=pr_url.source, token=resolved_config.git_api_key, formatter=formatter)
+    git_client = get_git_client(source=target.source, token=resolved_config.git_api_key, formatter=formatter)
     issues_client = _get_issues_client(resolved_config, git_client, formatter)
 
     code_reviewer = CodeReviewer(
@@ -206,7 +226,7 @@ def review(
         git_client=git_client,
         config=resolved_config,
     )
-    review = code_reviewer.review_pull_request(pr_url=pr_url)
+    review = code_reviewer.review_pull_request(target=target)
     logger.info("Review completed, total comments: %d", len(review.review_response.comments))
 
     if not resolved_config.silent:
@@ -216,16 +236,16 @@ def review(
         if review.review_response.comments:
             printer(formatter.format_review_comments_section(review.review_response.comments))
 
-    if resolved_config.publish:
+    if resolved_config.publish and isinstance(target, PRUrl) and git_client:
         logger.info("Publishing review to git service")
-        git_client.publish_review(pr_url=pr_url, review=review)
+        git_client.publish_review(pr_url=target, review=review)
         logger.info("Review published successfully")
 
 
 @entry_point.command()
 @_common_options
 def guide(
-    pr_url: PRUrl,
+    target: PRUrl,
     model: SupportedAIModels | None,
     model_url: str | None,
     git_api_key: str | None,
@@ -241,13 +261,13 @@ def guide(
 ) -> None:
     """Generate a review guide for a Pull Request using AI.
 
-    PR_URL is the URL of the pull request to generate a guide for.
+    TARGET is the URL of the pull request to generate a guide for.
     """
     _set_logging_level(logger, verbose)
 
     logger.info("lgtm-ai version: %s", __version__)
-    logger.debug("Parsed PR URL: %s", pr_url)
-    logger.info("Starting generating guide of %s", pr_url.full_url)
+    logger.debug("Parsed PR URL: %s", target)
+    logger.info("Starting generating guide of %s", target.full_url)
     resolved_config = ConfigHandler(
         cli_args=PartialConfig(
             exclude=exclude,
@@ -262,9 +282,9 @@ def guide(
             ai_input_tokens_limit=ai_input_tokens_limit,
         ),
         config_file=config,
-    ).resolve_config()
+    ).resolve_config(target)
     agent_extra_settings = AgentSettings(retries=resolved_config.ai_retries)
-    git_client = get_git_client(source=pr_url.source, token=resolved_config.git_api_key, formatter=MarkDownFormatter())
+    git_client = get_git_client(source=target.source, token=resolved_config.git_api_key, formatter=MarkDownFormatter())
     review_guide = ReviewGuideGenerator(
         guide_agent=get_guide_agent_with_settings(agent_extra_settings),
         model=get_ai_model(
@@ -273,16 +293,16 @@ def guide(
         git_client=git_client,
         config=resolved_config,
     )
-    guide = review_guide.generate_review_guide(pr_url=pr_url)
+    guide = review_guide.generate_review_guide(pr_url=target)
 
     if not resolved_config.silent:
         logger.info("Printing review to console")
         formatter, printer = _get_formatter_and_printer(resolved_config.output_format)
         printer(formatter.format_guide(guide))
 
-    if resolved_config.publish:
+    if resolved_config.publish and git_client:
         logger.info("Publishing review guide to git service")
-        git_client.publish_guide(pr_url=pr_url, guide=guide)
+        git_client.publish_guide(pr_url=target, guide=guide)
         logger.info("Review Guide published successfully")
 
 
@@ -310,16 +330,16 @@ def _get_formatter_and_printer(output_format: OutputFormat) -> tuple[Formatter[A
 
 
 def _get_issues_client(
-    resolved_config: ResolvedConfig, git_client: GitClient, formatter: Formatter[Any]
-) -> IssuesClient:
+    resolved_config: ResolvedConfig, git_client: GitClient | None, formatter: Formatter[Any]
+) -> IssuesClient | None:
     """Select a different issues client for retrieving issues.
 
     Will only return a different client if all of the following are true:
         1) Be used at all
         2) Be retrieved from a git platform and not elsewhere (e.g., Jira, Asana, etc.)
         3) Have a specific API key configured
     """
-    issues_client: IssuesClient = git_client
+    issues_client: IssuesClient | None = git_client
     if not resolved_config.issues_url or not resolved_config.issues_platform or not resolved_config.issues_regex:
         return issues_client
     if resolved_config.issues_platform.is_git_platform:

src/lgtm_ai/base/schemas.py

@@ -1,3 +1,4 @@
+import pathlib
 from dataclasses import dataclass
 from enum import StrEnum
 from typing import Literal
@@ -9,6 +10,7 @@
 class PRSource(StrEnum):
     github = "github"
     gitlab = "gitlab"
+    local = "local"
 
 
 class IssuesPlatform(StrEnum):
@@ -29,6 +31,16 @@ class PRUrl:
     source: PRSource
 
 
+@dataclass(frozen=True, slots=True)
+class LocalRepository:
+    repo_path: pathlib.Path
+    source: PRSource = PRSource.local
+
+    @property
+    def full_url(self) -> str:
+        return self.repo_path.as_posix()
+
+
 class OutputFormat(StrEnum):
     pretty = "pretty"
     json = "json"