ai-cr 0.4.8__tar.gz → 0.6.0__tar.gz

{ai_cr-0.4.8 → ai_cr-0.6.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: ai-cr
-Version: 0.4.8
+Version: 0.6.0
 Summary: LLM-agnostic GitHub AI Code Review Tool with integration to GitHub actions
 License: MIT
 Keywords: static code analysis,code review,code quality,ai,coding,assistant,llm,github,automation,devops,developer tools,github actions,workflows,git
@@ -15,13 +15,12 @@ Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Classifier: Topic :: Software Development
-Requires-Dist: GitPython (==3.1.44)
-Requires-Dist: ai-microcore (==4.0.0.dev16)
-Requires-Dist: anthropic (==0.49.0)
-Requires-Dist: async-typer (==0.1.8)
-Requires-Dist: google-generativeai (==0.8.5)
-Requires-Dist: typer (==0.9.4)
-Requires-Dist: unidiff (==0.7.5)
+Requires-Dist: GitPython (>=3.1.44,<4.0.0)
+Requires-Dist: ai-microcore (==4.0.0.dev19)
+Requires-Dist: anthropic (>=0.52.2,<0.53.0)
+Requires-Dist: google-generativeai (>=0.8.5,<0.9.0)
+Requires-Dist: typer (>=0.16.0,<0.17.0)
+Requires-Dist: unidiff (>=0.7.5,<0.8.0)
 Project-URL: Homepage, https://github.com/Nayjest/github-ai-code-review
 Project-URL: Repository, https://github.com/Nayjest/github-ai-code-review
 Description-Content-Type: text/markdown
@@ -30,6 +29,7 @@ Description-Content-Type: text/markdown
 <a href="https://pypi.org/project/ai-code-review/" target="_blank"><img src="https://badge.fury.io/py/ai-code-review.svg" alt="PYPI Release"></a>
 <a href="https://github.com/Nayjest/ai-code-review/actions/workflows/code-style.yml" target="_blank"><img src="https://github.com/Nayjest/ai-code-review/actions/workflows/code-style.yml/badge.svg" alt="Pylint"></a>
 <a href="https://github.com/Nayjest/ai-code-review/actions/workflows/tests.yml" target="_blank"><img src="https://github.com/Nayjest/ai-code-review/actions/workflows/tests.yml/badge.svg" alt="Tests"></a>
+<img src="https://github.com/Nayjest/ai-code-review/blob/main/coverage.svg" alt="Code Coverage">
 <a href="https://github.com/Nayjest/ai-code-review/blob/main/LICENSE" target="_blank"><img src="https://img.shields.io/static/v1?label=license&message=MIT&color=d08aff" alt="License"></a>
 </p>
 
@@ -70,7 +70,7 @@ jobs:
       uses: actions/setup-python@v5
       with: { python-version: "3.13" }
     - name: Install AI Code Review tool
-      run: pip install ai-code-review==0.4.7
+      run: pip install ai-code-review==0.5.0
     - name: Run AI code analysis
       env:
         LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
@@ -91,7 +91,8 @@ jobs:
 > ⚠️ Make sure to add `LLM_API_KEY` to your repository’s GitHub secrets.
 
 💪 Done!  
-PRs to your repository will now receive AI code reviews automatically. ✨
+PRs to your repository will now receive AI code reviews automatically. ✨  
+See [GitHub Setup Guide](https://github.com/Nayjest/ai-code-review/blob/main/documentation/github_setup.md) for more details.
 
 ### 2. Run Code Analysis Locally
 
@@ -151,7 +152,7 @@ pytest
 
 ## 🤝 Contributing
 
-**Looking for a specific feature or having trouble?**
+**Looking for a specific feature or having trouble?**  
 Contributions are welcome! ❤️  
 See [CONTRIBUTING.md](https://github.com/Nayjest/ai-code-review/blob/main/CONTRIBUTING.md) for details.
 

{ai_cr-0.4.8 → ai_cr-0.6.0}/README.md

@@ -2,6 +2,7 @@
 <a href="https://pypi.org/project/ai-code-review/" target="_blank"><img src="https://badge.fury.io/py/ai-code-review.svg" alt="PYPI Release"></a>
 <a href="https://github.com/Nayjest/ai-code-review/actions/workflows/code-style.yml" target="_blank"><img src="https://github.com/Nayjest/ai-code-review/actions/workflows/code-style.yml/badge.svg" alt="Pylint"></a>
 <a href="https://github.com/Nayjest/ai-code-review/actions/workflows/tests.yml" target="_blank"><img src="https://github.com/Nayjest/ai-code-review/actions/workflows/tests.yml/badge.svg" alt="Tests"></a>
+<img src="https://github.com/Nayjest/ai-code-review/blob/main/coverage.svg" alt="Code Coverage">
 <a href="https://github.com/Nayjest/ai-code-review/blob/main/LICENSE" target="_blank"><img src="https://img.shields.io/static/v1?label=license&message=MIT&color=d08aff" alt="License"></a>
 </p>
 
@@ -42,7 +43,7 @@ jobs:
       uses: actions/setup-python@v5
       with: { python-version: "3.13" }
     - name: Install AI Code Review tool
-      run: pip install ai-code-review==0.4.7
+      run: pip install ai-code-review==0.5.0
     - name: Run AI code analysis
       env:
         LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
@@ -63,7 +64,8 @@ jobs:
 > ⚠️ Make sure to add `LLM_API_KEY` to your repository’s GitHub secrets.
 
 💪 Done!  
-PRs to your repository will now receive AI code reviews automatically. ✨
+PRs to your repository will now receive AI code reviews automatically. ✨  
+See [GitHub Setup Guide](https://github.com/Nayjest/ai-code-review/blob/main/documentation/github_setup.md) for more details.
 
 ### 2. Run Code Analysis Locally
 
@@ -123,7 +125,7 @@ pytest
 
 ## 🤝 Contributing
 
-**Looking for a specific feature or having trouble?**
+**Looking for a specific feature or having trouble?**  
 Contributions are welcome! ❤️  
 See [CONTRIBUTING.md](https://github.com/Nayjest/ai-code-review/blob/main/CONTRIBUTING.md) for details.
 

ai_cr-0.6.0/ai_code_review/__main__.py

@@ -0,0 +1,4 @@
+from .cli import main
+
+if __name__ == "__main__":
+    main()

{ai_cr-0.4.8 → ai_cr-0.6.0}/ai_code_review/bootstrap.py

@@ -41,7 +41,10 @@ def bootstrap():
         if is_running_in_github_action():
             ref = os.getenv("GITHUB_WORKFLOW_REF", "")
             if ref:
-                ref = f" ({ref})"
+                # example value: 'owner/repo/.github/workflows/ai-code-review.yml@refs/pull/1/merge'
+                ref = ref.split("@")[0]
+                ref = ref.split(".github/workflows/")[-1]
+                ref = f" (.github/workflows/{ref})"
             msg += (
                 f"\nPlease check your GitHub Action Secrets "
                 f"and `env` configuration section of the corresponding workflow step{ref}."
@@ -56,4 +59,4 @@ def bootstrap():
     except Exception as e:
         logging.error(f"Unexpected configuration error: {e}")
         raise SystemExit(3)
-    mc.logging.LoggingConfig.STRIP_REQUEST_LINES = [100, 15]
+    mc.logging.LoggingConfig.STRIP_REQUEST_LINES = [300, 15]

ai_cr-0.6.0/ai_code_review/cli.py

@@ -0,0 +1,236 @@
+import asyncio
+import logging
+import sys
+import os
+import textwrap
+import tempfile
+import requests
+
+import microcore as mc
+import typer
+from git import Repo
+
+from .core import review, get_diff, filter_diff
+from .report_struct import Report
+from .constants import ENV_CONFIG_FILE
+from .bootstrap import bootstrap
+from .project_config import ProjectConfig
+from .utils import no_subcommand, parse_refs_pair
+
+app = typer.Typer(pretty_exceptions_show_locals=False)
+app_no_subcommand = typer.Typer(pretty_exceptions_show_locals=False)
+
+
+def main():
+    if sys.platform == "win32":
+        asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy())
+    if no_subcommand(app):
+        bootstrap()
+        app_no_subcommand()
+    else:
+        app()
+
+
+@app.callback(invoke_without_command=True)
+def cli(ctx: typer.Context):
+    if ctx.invoked_subcommand != "setup":
+        bootstrap()
+
+
+def args_to_target(refs, what, against) -> tuple[str | None, str | None]:
+    _what, _against = parse_refs_pair(refs)
+    if _what:
+        if what:
+            raise typer.BadParameter(
+                "You cannot specify both 'refs' <WHAT>..<AGAINST> and '--what'. Use one of them."
+            )
+    else:
+        _what = what
+    if _against:
+        if against:
+            raise typer.BadParameter(
+                "You cannot specify both 'refs' <WHAT>..<AGAINST> and '--against'. Use one of them."
+            )
+    else:
+        _against = against
+    return _what, _against
+
+
+def arg_refs() -> typer.Argument:
+    return typer.Argument(
+        default=None,
+        help="Git refs to review, [what]..[against] e.g. 'HEAD..HEAD~1'"
+    )
+
+
+def arg_what() -> typer.Option:
+    return typer.Option(None, "--what", "-w", help="Git ref to review")
+
+
+def arg_filters() -> typer.Option:
+    return typer.Option(
+        "", "--filter", "-f", "--filters",
+        help="""
+            filter reviewed files by glob / fnmatch pattern(s),
+            e.g. 'src/**/*.py', may be comma-separated
+            """,
+    )
+
+
+def arg_out() -> typer.Option:
+    return typer.Option(
+        None,
+        "--out", "-o", "--output",
+        help="Output folder for the code review report"
+    )
+
+
+def arg_against() -> typer.Option:
+    return typer.Option(
+        None,
+        "--against", "-vs", "--vs",
+        help="Git ref to compare against"
+    )
+
+
+@app_no_subcommand.command(name="review", help="Perform code review")
+@app.command(name="review", help="Perform code review")
+def cmd_review(
+    refs: str = arg_refs(),
+    what: str = arg_what(),
+    against: str = arg_against(),
+    filters: str = arg_filters(),
+    merge_base: bool = typer.Option(default=True, help="Use merge base for comparison"),
+    out: str = arg_out()
+):
+    _what, _against = args_to_target(refs, what, against)
+    asyncio.run(review(
+        what=_what,
+        against=_against,
+        filters=filters,
+        use_merge_base=merge_base,
+        out_folder=out,
+    ))
+
+
+@app.command(help="Configure LLM for local usage interactively")
+def setup():
+    mc.interactive_setup(ENV_CONFIG_FILE)
+
+
+@app.command()
+def render(format: str = Report.Format.MARKDOWN):
+    print(Report.load().render(format=format))
+
+
+@app.command(help="Review remote code")
+def remote(
+    url: str = typer.Argument(..., help="Git repository URL"),
+    refs: str = arg_refs(),
+    what: str = arg_what(),
+    against: str = arg_against(),
+    filters: str = arg_filters(),
+    merge_base: bool = typer.Option(default=True, help="Use merge base for comparison"),
+    out: str = arg_out()
+):
+    _what, _against = args_to_target(refs, what, against)
+    with tempfile.TemporaryDirectory() as temp_dir:
+        logging.info(f"Cloning [{mc.ui.green(url)}] to {mc.utils.file_link(temp_dir)} ...")
+        repo = Repo.clone_from(url, branch=_what, to_path=temp_dir)
+        asyncio.run(review(
+            repo=repo,
+            what=_what,
+            against=_against,
+            filters=filters,
+            use_merge_base=merge_base,
+            out_folder=out or '.',
+        ))
+        repo.close()
+
+
+@app.command(help="Leave a GitHub PR comment with the review.")
+def github_comment(
+    token: str = typer.Option(
+        os.environ.get("GITHUB_TOKEN", ""), help="GitHub token (or set GITHUB_TOKEN env var)"
+    ),
+):
+    """
+    Leaves a comment with the review on the current GitHub pull request.
+    """
+    file = "code-review-report.md"
+    if not os.path.exists(file):
+        print(f"Review file not found: {file}")
+        raise typer.Exit(4)
+
+    with open(file, "r", encoding="utf-8") as f:
+        body = f.read()
+
+    if not token:
+        print("GitHub token is required (--token or GITHUB_TOKEN env var).")
+        raise typer.Exit(1)
+
+    github_env = ProjectConfig.load().prompt_vars["github_env"]
+    repo = github_env.get("github_repo", "")
+    pr_env_val = github_env.get("github_pr_number", "")
+    logging.info(f"github_pr_number = {pr_env_val}")
+
+    # e.g. could be "refs/pull/123/merge" or a direct number
+    if "/" in pr_env_val and "pull" in pr_env_val:
+        # refs/pull/123/merge
+        try:
+            pr_num_candidate = pr_env_val.strip("/").split("/")
+            idx = pr_num_candidate.index("pull")
+            pr = int(pr_num_candidate[idx + 1])
+        except Exception:
+            pr = 0
+    else:
+        try:
+            pr = int(pr_env_val)
+        except Exception:
+            pr = 0
+
+    api_url = f"https://api.github.com/repos/{repo}/issues/{pr}/comments"
+    headers = {
+        "Authorization": f"token {token}",
+        "Accept": "application/vnd.github+json",
+    }
+    data = {"body": body}
+
+    resp = requests.post(api_url, headers=headers, json=data)
+    if 200 <= resp.status_code < 300:
+        logging.info(f"Posted review comment to PR #{pr} in {repo}")
+    else:
+        logging.error(f"Failed to post comment: {resp.status_code} {resp.reason}\n{resp.text}")
+        raise typer.Exit(5)
+
+
+@app.command(help="List files in the diff. Might be useful to check what will be reviewed.")
+def files(
+    refs: str = arg_refs(),
+    what: str = arg_what(),
+    against: str = arg_against(),
+    filters: str = arg_filters(),
+    merge_base: bool = typer.Option(default=True, help="Use merge base for comparison"),
+    diff: bool = typer.Option(default=False, help="Show diff content")
+):
+    _what, _against = args_to_target(refs, what, against)
+    repo = Repo(".")
+    patch_set = get_diff(repo=repo, what=_what, against=_against, use_merge_base=merge_base)
+    patch_set = filter_diff(patch_set, filters)
+    print(
+        f"Changed files: "
+        f"{mc.ui.green(_what or 'INDEX')} vs "
+        f"{mc.ui.yellow(_against or repo.remotes.origin.refs.HEAD.reference.name)}"
+        f"{' filtered by '+mc.ui.cyan(filters) if filters else ''}"
+    )
+
+    for patch in patch_set:
+        if patch.is_added_file:
+            color = mc.ui.green
+        elif patch.is_removed_file:
+            color = mc.ui.red
+        else:
+            color = mc.ui.blue
+        print(f"- {color(patch.path)}")
+        if diff:
+            print(mc.ui.gray(textwrap.indent(str(patch), "  ")))

{ai_cr-0.4.8 → ai_cr-0.6.0}/ai_code_review/constants.py

@@ -3,5 +3,5 @@ from pathlib import Path
 
 PROJECT_CONFIG_FILE = Path(".ai-code-review.toml")
 PROJECT_CONFIG_DEFAULTS_FILE = Path(__file__).resolve().parent / PROJECT_CONFIG_FILE
-ENV_CONFIG_FILE = ""#Path("~/.env.ai-code-review").expanduser()
+ENV_CONFIG_FILE = Path("~/.env.ai-code-review").expanduser()
 JSON_REPORT_FILE_NAME = "code-review-report.json"

ai_cr-0.6.0/ai_code_review/core.py

@@ -0,0 +1,191 @@
+import fnmatch
+import logging
+from os import PathLike
+from typing import Iterable
+from pathlib import Path
+
+import microcore as mc
+from git import Repo
+from unidiff import PatchSet, PatchedFile
+from unidiff.constants import DEV_NULL
+
+from .project_config import ProjectConfig
+from .report_struct import Report
+from .constants import JSON_REPORT_FILE_NAME
+
+
+def is_binary_file(repo: Repo, file_path: str) -> bool:
+    """
+    Check if a file is binary by attempting to read it as text.
+    Returns True if the file is binary, False otherwise.
+    """
+    try:
+        # Attempt to read the file content from the repository tree
+        content = repo.tree()[file_path].data_stream.read()
+        # Try decoding as UTF-8; if it fails, it's likely binary
+        content.decode("utf-8")
+        return False
+    except (UnicodeDecodeError, KeyError):
+        return True
+    except Exception as e:
+        logging.warning(f"Error checking if file {file_path} is binary: {e}")
+        return True  # Conservatively treat errors as binary to avoid issues
+
+
+def get_diff(
+    repo: Repo = None,
+    what: str = None,
+    against: str = None,
+    use_merge_base: bool = True,
+) -> PatchSet | list[PatchedFile]:
+    repo = repo or Repo(".")
+    if not against:
+        against = repo.remotes.origin.refs.HEAD.reference.name  # origin/main
+    if not what:
+        what = None  # working copy
+    if use_merge_base:
+        if what is None:
+            try:
+                current_ref = repo.active_branch.name
+            except TypeError:
+                # In detached HEAD state, use HEAD directly
+                current_ref = "HEAD"
+                logging.info(
+                    "Detected detached HEAD state, using HEAD as current reference"
+                )
+        else:
+            current_ref = what
+        merge_base = repo.merge_base(current_ref or repo.active_branch.name, against)[0]
+        against = merge_base.hexsha
+        logging.info(
+            f"Using merge base: {mc.ui.cyan(merge_base.hexsha[:8])} ({merge_base.summary})"
+        )
+    logging.info(
+        f"Making diff: {mc.ui.green(what or 'INDEX')} vs {mc.ui.yellow(against)}"
+    )
+    diff_content = repo.git.diff(against, what)
+    diff = PatchSet.from_string(diff_content)
+    diff = PatchSet.from_string(diff_content)
+
+    # Filter out binary files
+    non_binary_diff = PatchSet([])
+    for patched_file in diff:
+        # Check if the file is binary using the source or target file path
+        file_path = (
+            patched_file.target_file
+            if patched_file.target_file != DEV_NULL
+            else patched_file.source_file
+        )
+        if file_path == DEV_NULL or is_binary_file(repo, file_path.lstrip("b/")):
+            logging.info(f"Skipping binary file: {patched_file.path}")
+            continue
+        non_binary_diff.append(patched_file)
+    return non_binary_diff
+
+
+def filter_diff(
+    patch_set: PatchSet | Iterable[PatchedFile], filters: str | list[str]
+) -> PatchSet | Iterable[PatchedFile]:
+    """
+    Filter the diff files by the given fnmatch filters.
+    """
+    assert isinstance(filters, (list, str))
+    if not isinstance(filters, list):
+        filters = [f.strip() for f in filters.split(",") if f.strip()]
+    if not filters:
+        return patch_set
+    files = [
+        file
+        for file in patch_set
+        if any(fnmatch.fnmatch(file.path, pattern) for pattern in filters)
+    ]
+    return files
+
+
+def file_lines(repo: Repo, file: str, max_tokens: int = None) -> str:
+    text = repo.tree()[file].data_stream.read().decode()
+    lines = [f"{i + 1}: {line}\n" for i, line in enumerate(text.splitlines())]
+    if max_tokens:
+        lines, removed_qty = mc.tokenizing.fit_to_token_size(lines, max_tokens)
+        if removed_qty:
+            lines.append(
+                f"(!) DISPLAYING ONLY FIRST {len(lines)} LINES DUE TO LARGE FILE SIZE\n"
+            )
+    return "".join(lines)
+
+
+def make_cr_summary(cfg: ProjectConfig, report: Report, diff):
+    return (
+        mc.prompt(
+            cfg.summary_prompt,
+            diff=mc.tokenizing.fit_to_token_size(diff, cfg.max_code_tokens)[0],
+            issues=report.issues,
+            **cfg.prompt_vars,
+        ).to_llm()
+        if cfg.summary_prompt
+        else ""
+    )
+
+
+async def review(
+    repo: Repo = None,
+    what: str = None,
+    against: str = None,
+    filters: str | list[str] = "",
+    use_merge_base: bool = True,
+    out_folder: str | PathLike | None = None,
+):
+    cfg = ProjectConfig.load()
+    repo = repo or Repo(".")
+    out_folder = Path(out_folder or repo.working_tree_dir)
+    diff = get_diff(
+        repo=repo, what=what, against=against, use_merge_base=use_merge_base
+    )
+    diff = filter_diff(diff, filters)
+    if not diff:
+        logging.error("Nothing to review")
+        return
+    lines = {
+        file_diff.path: (
+            file_lines(
+                repo,
+                file_diff.path,
+                cfg.max_code_tokens
+                - mc.tokenizing.num_tokens_from_string(str(file_diff)),
+            )
+            if file_diff.target_file != DEV_NULL and not file_diff.is_added_file
+            else ""
+        )
+        for file_diff in diff
+    }
+    responses = await mc.llm_parallel(
+        [
+            mc.prompt(
+                cfg.prompt,
+                input=file_diff,
+                file_lines=lines[file_diff.path],
+                **cfg.prompt_vars,
+            )
+            for file_diff in diff
+        ],
+        retries=cfg.retries,
+        parse_json=True,
+    )
+    issues = {file.path: issues for file, issues in zip(diff, responses) if issues}
+    for file, file_issues in issues.items():
+        for issue in file_issues:
+            for i in issue.get("affected_lines", []):
+                if lines[file]:
+                    f_lines = [""] + lines[file].splitlines()
+                    i["affected_code"] = "\n".join(
+                        f_lines[i["start_line"]: i["end_line"] + 1]
+                    )
+    exec(cfg.post_process, {"mc": mc, **locals()})
+    out_folder.mkdir(parents=True, exist_ok=True)
+    report = Report(issues=issues, number_of_processed_files=len(diff))
+    report.summary = make_cr_summary(cfg, report, diff)
+    report.save(file_name=out_folder / JSON_REPORT_FILE_NAME)
+    report_text = report.render(cfg, Report.Format.MARKDOWN)
+    print(mc.ui.yellow(report_text))
+    text_report_path = out_folder / "code-review-report.md"
+    text_report_path.write_text(report_text, encoding="utf-8")

{ai_cr-0.4.8 → ai_cr-0.6.0}/ai_code_review/utils.py

@@ -1,5 +1,8 @@
+import sys
 import os
 from pathlib import Path
+import typer
+
 
 _EXT_TO_HINT: dict[str, str] = {
     # scripting & languages
@@ -87,3 +90,27 @@ def syntax_hint(file_path: str | Path) -> str:
 
 def is_running_in_github_action():
     return os.getenv("GITHUB_ACTIONS") == "true"
+
+
+def no_subcommand(app: typer.Typer) -> bool:
+    """
+    Checks if the current script is being invoked as a command in a target Typer application.
+    """
+    return not (
+        (first_arg := next((a for a in sys.argv[1:] if not a.startswith('-')), None))
+        and first_arg in (
+            cmd.name or cmd.callback.__name__.replace('_', '-')
+            for cmd in app.registered_commands
+        )
+        or '--help' in sys.argv
+    )
+
+
+def parse_refs_pair(refs: str) -> tuple[str | None, str | None]:
+    SEPARATOR = '..'
+    if not refs:
+        return None, None
+    if SEPARATOR not in refs:
+        return refs, None
+    what, against = refs.split(SEPARATOR)
+    return what or None, against or None

{ai_cr-0.4.8 → ai_cr-0.6.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ai-cr"
-version = "0.4.8"
+version = "0.6.0"
 description = "LLM-agnostic GitHub AI Code Review Tool with integration to GitHub actions"
 authors = ["Nayjest <mail@vitaliy.in>"]
 readme = "README.md"
@@ -21,13 +21,12 @@ packages = [
 
 [tool.poetry.dependencies]
 python = "^3.11"
-ai-microcore = "4.0.0.dev16"
-GitPython = "3.1.44"
-unidiff = "0.7.5"
-google-generativeai = "0.8.5"
-anthropic = "0.49.0"
-typer = "0.9.4"
-async-typer = "0.1.8"
+ai-microcore = "4.0.0.dev19"
+GitPython = "^3.1.44"
+unidiff = "^0.7.5"
+google-generativeai = "^0.8.5"
+anthropic = "^0.52.2"
+typer = "^0.16.0"
 
 [tool.poetry.group.dev.dependencies]
 flake8 = "*"
@@ -49,7 +48,7 @@ requires = ["poetry-core"]
 build-backend = "poetry.core.masonry.api"
 
 [tool.poetry.scripts]
-ai-cr = "ai_code_review.cli:app"
+ai-cr = "ai_code_review.cli:main"
 
 [tool.pytest.ini_options]
 minversion = "6.0"

ai_cr-0.4.8/ai_code_review/cli.py

@@ -1,112 +0,0 @@
-import asyncio
-import logging
-import sys
-import os
-import shutil
-
-import microcore as mc
-import async_typer
-import typer
-from .core import review
-from .report_struct import Report
-from git import Repo
-import requests
-
-from .constants import ENV_CONFIG_FILE
-from .bootstrap import bootstrap
-from .project_config import ProjectConfig
-
-app = async_typer.AsyncTyper(
-    pretty_exceptions_show_locals=False,
-)
-
-
-if sys.platform == "win32":
-    asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy())
-
-
-@app.callback(invoke_without_command=True)
-def cli(ctx: typer.Context, filters=typer.Option("", "--filter", "-f", "--filters")):
-    if ctx.invoked_subcommand != "setup":
-        bootstrap()
-    if not ctx.invoked_subcommand:
-        asyncio.run(review(filters=filters))
-
-
-@app.async_command(help="Configure LLM for local usage interactively")
-async def setup():
-    mc.interactive_setup(ENV_CONFIG_FILE)
-
-
-@app.async_command()
-async def render(format: str = Report.Format.MARKDOWN):
-    print(Report.load().render(format=format))
-
-
-@app.async_command(help="Review remote code")
-async def remote(url=typer.Option(), branch=typer.Option()):
-    if os.path.exists("reviewed-repo"):
-        shutil.rmtree("reviewed-repo")
-    Repo.clone_from(url, branch=branch, to_path="reviewed-repo")
-    prev_dir = os.getcwd()
-    try:
-        os.chdir("reviewed-repo")
-        await review()
-    finally:
-        os.chdir(prev_dir)
-
-
-@app.async_command(help="Leave a GitHub PR comment with the review.")
-async def github_comment(
-    token: str = typer.Option(
-        os.environ.get("GITHUB_TOKEN", ""), help="GitHub token (or set GITHUB_TOKEN env var)"
-    ),
-):
-    """
-    Leaves a comment with the review on the current GitHub pull request.
-    """
-    file = "code-review-report.txt"
-    if not os.path.exists(file):
-        print(f"Review file not found: {file}")
-        raise typer.Exit(4)
-
-    with open(file, "r", encoding="utf-8") as f:
-        body = f.read()
-
-    if not token:
-        print("GitHub token is required (--token or GITHUB_TOKEN env var).")
-        raise typer.Exit(1)
-
-    github_env = ProjectConfig.load().prompt_vars["github_env"]
-    repo = github_env.get("github_repo", "")
-    pr_env_val = github_env.get("github_pr_number", "")
-    logging.info(f"github_pr_number = {pr_env_val}")
-
-    # e.g. could be "refs/pull/123/merge" or a direct number
-    if "/" in pr_env_val and "pull" in pr_env_val:
-        # refs/pull/123/merge
-        try:
-            pr_num_candidate = pr_env_val.strip("/").split("/")
-            idx = pr_num_candidate.index("pull")
-            pr = int(pr_num_candidate[idx + 1])
-        except Exception:
-            pr = 0
-    else:
-        try:
-            pr = int(pr_env_val)
-        except Exception:
-            pr = 0
-
-    api_url = f"https://api.github.com/repos/{repo}/issues/{pr}/comments"
-    headers = {
-        "Authorization": f"token {token}",
-        "Accept": "application/vnd.github+json",
-    }
-    data = {"body": body}
-
-    resp = requests.post(api_url, headers=headers, json=data)
-    if 200 <= resp.status_code < 300:
-        logging.info(f"Posted review comment to PR #{pr} in {repo}")
-    else:
-        logging.error(f"Failed to post comment: {resp.status_code} {resp.reason}\n{resp.text}")
-        raise typer.Exit(5)

ai_cr-0.4.8/ai_code_review/core.py

@@ -1,114 +0,0 @@
-import fnmatch
-import logging
-from typing import Iterable
-
-import microcore as mc
-from git import Repo
-from unidiff import PatchSet, PatchedFile
-from unidiff.constants import DEV_NULL
-
-from .project_config import ProjectConfig
-from .report_struct import Report
-
-
-def get_diff(repo: Repo = None, against: str = "HEAD") -> PatchSet | list[PatchedFile]:
-    repo = repo or Repo(".")
-    base = repo.remotes.origin.refs.HEAD.reference.name
-    logging.info(f"{base}...{against}")
-    diff_content = repo.git.diff(base, against)
-    diff = PatchSet.from_string(diff_content)
-    return diff
-
-
-def filter_diff(
-    patch_set: PatchSet | Iterable[PatchedFile], filters: str | list[str]
-) -> PatchSet | Iterable[PatchedFile]:
-    """
-    Filter the diff files by the given fnmatch filters.
-    """
-    print([f.path for f in patch_set])
-    assert isinstance(filters, (list, str))
-    if not isinstance(filters, list):
-        filters = [f.strip() for f in filters.split(",") if f.strip()]
-    if not filters:
-        return patch_set
-    files = [
-        file
-        for file in patch_set
-        if any(fnmatch.fnmatch(file.path, pattern) for pattern in filters)
-    ]
-    print([f.path for f in files])
-    return files
-
-
-def file_lines(repo: Repo, file: str, max_tokens: int = None) -> str:
-    text = repo.tree()[file].data_stream.read().decode()
-    lines = [f"{i + 1}: {line}\n" for i, line in enumerate(text.splitlines())]
-    if max_tokens:
-        lines, removed_qty = mc.tokenizing.fit_to_token_size(lines, max_tokens)
-        if removed_qty:
-            lines.append(
-                f"(!) DISPLAYING ONLY FIRST {len(lines)} LINES DUE TO LARGE FILE SIZE\n"
-            )
-    return "".join(lines)
-
-
-def make_cr_summary(cfg: ProjectConfig, report: Report, diff):
-    return mc.prompt(
-        cfg.summary_prompt,
-        diff=mc.tokenizing.fit_to_token_size(diff, cfg.max_code_tokens)[0],
-        issues=report.issues,
-        **cfg.prompt_vars,
-    ).to_llm() if cfg.summary_prompt else ""
-
-
-async def review(filters: str | list[str] = ""):
-    cfg = ProjectConfig.load()
-    repo = Repo(".")
-    diff = get_diff(repo=repo, against="HEAD")
-    diff = filter_diff(diff, filters)
-    if not diff:
-        logging.error("Nothing to review")
-        return
-    lines = {
-        file_diff.path: (
-            file_lines(
-                repo,
-                file_diff.path,
-                cfg.max_code_tokens
-                - mc.tokenizing.num_tokens_from_string(str(file_diff)),
-            )
-            if file_diff.target_file != DEV_NULL
-            else ""
-        )
-        for file_diff in diff
-    }
-    responses = await mc.llm_parallel(
-        [
-            mc.prompt(
-                cfg.prompt,
-                input=file_diff,
-                file_lines=lines[file_diff.path],
-                **cfg.prompt_vars,
-            )
-            for file_diff in diff
-        ],
-        retries=cfg.retries,
-        parse_json=True,
-    )
-    issues = {file.path: issues for file, issues in zip(diff, responses) if issues}
-    for file, file_issues in issues.items():
-        for issue in file_issues:
-            for i in issue.get("affected_lines", []):
-                if lines[file]:
-                    f_lines = [""] + lines[file].splitlines()
-                    i["affected_code"] = "\n".join(
-                        f_lines[i["start_line"]: i["end_line"]+1]
-                    )
-    exec(cfg.post_process, {"mc": mc, **locals()})
-    report = Report(issues=issues, number_of_processed_files=len(diff))
-    report.summary = make_cr_summary(cfg, report, diff)
-    report.save()
-    report_text = report.render(cfg, Report.Format.MARKDOWN)
-    print(mc.ui.yellow(report_text))
-    open("code-review-report.txt", "w", encoding="utf-8").write(report_text)