gac 3.10.3__py3-none-any.whl

gac/git_state_validator.py

@@ -0,0 +1,184 @@
+#!/usr/bin/env python3
+"""Git state validation and management for gac."""
+
+import logging
+import subprocess
+import sys
+from typing import Any, NamedTuple
+
+from rich.console import Console
+
+from gac.config import GACConfig
+from gac.errors import ConfigError, GitError, handle_error
+from gac.git import get_staged_files, get_staged_status, run_git_command
+from gac.preprocess import preprocess_diff
+from gac.security import get_affected_files, scan_staged_diff
+
+logger = logging.getLogger(__name__)
+console = Console()
+
+
+class GitState(NamedTuple):
+    """Structured representation of git repository state."""
+
+    repo_root: str
+    staged_files: list[str]
+    status: str
+    diff: str
+    diff_stat: str
+    processed_diff: str
+    has_secrets: bool
+    secrets: list[Any]
+
+
+class GitStateValidator:
+    """Validates and manages git repository state."""
+
+    def __init__(self, config: GACConfig):
+        self.config = config
+
+    def validate_repository(self) -> str:
+        """Validate that we're in a git repository and return the repo root."""
+        try:
+            git_dir = run_git_command(["rev-parse", "--show-toplevel"])
+            if not git_dir:
+                raise GitError("Not in a git repository")
+            return git_dir
+        except (subprocess.SubprocessError, GitError, OSError) as e:
+            logger.error(f"Error checking git repository: {e}")
+            handle_error(GitError("Not in a git repository"), exit_program=True)
+            return ""  # Never reached, but required for type safety
+
+    def stage_all_if_requested(self, stage_all: bool, dry_run: bool) -> None:
+        """Stage all changes if requested and not in dry run mode."""
+        if stage_all and (not dry_run):
+            logger.info("Staging all changes")
+            run_git_command(["add", "--all"])
+
+    def get_git_state(
+        self,
+        stage_all: bool = False,
+        dry_run: bool = False,
+        skip_secret_scan: bool = False,
+        quiet: bool = False,
+        model: str | None = None,
+        hint: str = "",
+        one_liner: bool = False,
+        infer_scope: bool = False,
+        verbose: bool = False,
+        language: str | None = None,
+    ) -> GitState:
+        """Get complete git state including validation and processing."""
+        from gac.constants import Utility
+
+        # Validate repository
+        repo_root = self.validate_repository()
+
+        # Stage files if requested
+        self.stage_all_if_requested(stage_all, dry_run)
+
+        # Get staged files
+        staged_files = get_staged_files(existing_only=False)
+
+        if not staged_files:
+            console.print(
+                "[yellow]No staged changes found. Stage your changes with git add first or use --add-all.[/yellow]"
+            )
+            sys.exit(0)
+
+        # Get git status and diffs
+        status = get_staged_status()
+        diff = run_git_command(["diff", "--staged"])
+        diff_stat = " " + run_git_command(["diff", "--stat", "--cached"])
+
+        # Scan for secrets
+        has_secrets = False
+        secrets = []
+        if not skip_secret_scan:
+            logger.info("Scanning staged changes for potential secrets...")
+            secrets = scan_staged_diff(diff)
+            has_secrets = bool(secrets)
+
+        # Process diff for AI consumption
+        logger.debug(f"Preprocessing diff ({len(diff)} characters)")
+        if model is None:
+            raise ConfigError("Model must be specified via GAC_MODEL environment variable or --model flag")
+        processed_diff = preprocess_diff(diff, token_limit=Utility.DEFAULT_DIFF_TOKEN_LIMIT, model=model)
+        logger.debug(f"Processed diff ({len(processed_diff)} characters)")
+
+        return GitState(
+            repo_root=repo_root,
+            staged_files=staged_files,
+            status=status,
+            diff=diff,
+            diff_stat=diff_stat,
+            processed_diff=processed_diff,
+            has_secrets=has_secrets,
+            secrets=secrets,
+        )
+
+    def handle_secret_detection(self, secrets: list[Any], quiet: bool = False) -> bool:
+        """Handle secret detection and user interaction. Returns True if commit should continue."""
+        if not secrets:
+            return True
+
+        if not quiet:
+            console.print("\n[bold red]⚠️  SECURITY WARNING: Potential secrets detected![/bold red]")
+            console.print("[red]The following sensitive information was found in your staged changes:[/red]\n")
+
+        for secret in secrets:
+            location = f"{secret.file_path}:{secret.line_number}" if secret.line_number else secret.file_path
+            if not quiet:
+                console.print(f"  • [yellow]{secret.secret_type}[/yellow] in [cyan]{location}[/cyan]")
+                console.print(f"    Match: [dim]{secret.matched_text}[/dim]\n")
+
+        if not quiet:
+            console.print("\n[bold]Options:[/bold]")
+            console.print("  \\[a] Abort commit (recommended)")
+            console.print("  \\[c] [yellow]Continue anyway[/yellow] (not recommended)")
+            console.print("  \\[r] Remove affected file(s) and continue")
+
+        try:
+            import click
+
+            choice = (
+                click.prompt(
+                    "\nChoose an option",
+                    type=click.Choice(["a", "c", "r"], case_sensitive=False),
+                    default="a",
+                    show_choices=True,
+                    show_default=True,
+                )
+                .strip()
+                .lower()
+            )
+        except (EOFError, KeyboardInterrupt):
+            console.print("\n[red]Aborted by user.[/red]")
+            sys.exit(0)
+
+        if choice == "a":
+            console.print("[yellow]Commit aborted.[/yellow]")
+            sys.exit(0)
+        elif choice == "c":
+            console.print("[bold yellow]⚠️  Continuing with potential secrets in commit...[/bold yellow]")
+            logger.warning("User chose to continue despite detected secrets")
+            return True
+        elif choice == "r":
+            affected_files = get_affected_files(secrets)
+            for file_path in affected_files:
+                try:
+                    run_git_command(["reset", "HEAD", file_path])
+                    console.print(f"[green]Unstaged: {file_path}[/green]")
+                except GitError as e:
+                    console.print(f"[red]Failed to unstage {file_path}: {e}[/red]")
+
+            # Check if there are still staged files
+            remaining_staged = get_staged_files(existing_only=False)
+            if not remaining_staged:
+                console.print("[yellow]No files remain staged. Commit aborted.[/yellow]")
+                sys.exit(0)
+
+            console.print(f"[green]Continuing with {len(remaining_staged)} staged file(s)...[/green]")
+            return False
+
+        return True

gac/grouped_commit_workflow.py

@@ -0,0 +1,423 @@
+#!/usr/bin/env python3
+"""Grouped commit workflow handling for gac."""
+
+import json
+import logging
+import subprocess
+import sys
+from collections import Counter
+from typing import Any, NamedTuple
+
+import click
+from rich.console import Console
+from rich.panel import Panel
+
+from gac.ai import generate_grouped_commits
+from gac.ai_utils import count_tokens
+from gac.config import GACConfig
+from gac.constants import EnvDefaults
+from gac.errors import AIError, ConfigError, GitError
+from gac.git import detect_rename_mappings, get_staged_files, run_git_command
+from gac.git_state_validator import GitState
+from gac.workflow_utils import check_token_warning, execute_commit, restore_staging
+
+logger = logging.getLogger(__name__)
+console = Console()
+
+
+class GroupedCommitResult(NamedTuple):
+    """Result of grouped commit generation."""
+
+    commits: list[dict[str, Any]]
+    raw_response: str
+
+
+class GroupedCommitWorkflow:
+    """Handles multi-file grouping logic and per-group AI calls."""
+
+    def __init__(self, config: GACConfig):
+        self.config = config
+
+    def validate_grouped_files_or_feedback(
+        self, staged: set[str], grouped_result: dict[str, Any]
+    ) -> tuple[bool, str, str]:
+        """Validate that grouped commits cover all staged files correctly."""
+        commits = grouped_result.get("commits", []) if isinstance(grouped_result, dict) else []
+        all_files: list[str] = []
+        for commit in commits:
+            files = commit.get("files", []) if isinstance(commit, dict) else []
+            all_files.extend([str(p) for p in files])
+
+        counts = Counter(all_files)
+        union_set = set(all_files)
+
+        duplicates = sorted([f for f, c in counts.items() if c > 1])
+        missing = sorted(staged - union_set)
+        unexpected = sorted(union_set - staged)
+
+        if not duplicates and not missing and not unexpected:
+            return True, "", ""
+
+        problems: list[str] = []
+        if missing:
+            problems.append(f"Missing: {', '.join(missing)}")
+        if unexpected:
+            problems.append(f"Not staged: {', '.join(unexpected)}")
+        if duplicates:
+            problems.append(f"Duplicates: {', '.join(duplicates)}")
+
+        feedback = f"{'; '.join(problems)}. Required files: {', '.join(sorted(staged))}. Respond with ONLY valid JSON."
+        return False, feedback, "; ".join(problems)
+
+    def handle_validation_retry(
+        self,
+        attempts: int,
+        content_retry_budget: int,
+        raw_response: str,
+        feedback_message: str,
+        error_message: str,
+        conversation_messages: list[dict[str, str]],
+        quiet: bool,
+        retry_context: str,
+    ) -> bool:
+        """Handle validation retry logic. Returns True if should exit, False if should retry."""
+        conversation_messages.append({"role": "assistant", "content": raw_response})
+        conversation_messages.append({"role": "user", "content": feedback_message})
+        if attempts >= content_retry_budget:
+            logger.error(error_message)
+            console.print(f"\n[red]{error_message}[/red]")
+            console.print("\n[yellow]Raw model output:[/yellow]")
+            console.print(Panel(raw_response, title="Model Output", border_style="yellow"))
+            return True
+        if not quiet:
+            console.print(f"[yellow]Retry {attempts} of {content_retry_budget - 1}: {retry_context}[/yellow]")
+        return False
+
+    def parse_and_validate_json_response(self, raw_response: str) -> dict[str, Any] | None:
+        """Parse and validate JSON response from AI."""
+        parsed: dict[str, Any] | None = None
+        extract = raw_response
+        first_brace = raw_response.find("{")
+        last_brace = raw_response.rfind("}")
+        if first_brace != -1 and last_brace != -1 and first_brace < last_brace:
+            extract = raw_response[first_brace : last_brace + 1]
+
+        try:
+            parsed = json.loads(extract)
+        except json.JSONDecodeError as e:
+            parsed = None
+            logger.debug(
+                f"JSON parsing failed: {e}. Extract length: {len(extract)}, Response length: {len(raw_response)}"
+            )
+
+        if parsed is None:
+            raise ValueError("Invalid JSON response")
+
+        # Validate structure
+        if "commits" not in parsed or not isinstance(parsed["commits"], list):
+            raise ValueError("Response missing 'commits' array")
+        if len(parsed["commits"]) == 0:
+            raise ValueError("No commits in response")
+        for idx, commit in enumerate(parsed["commits"]):
+            if "files" not in commit or not isinstance(commit["files"], list):
+                raise ValueError(f"Commit {idx + 1} missing 'files' array")
+            if "message" not in commit or not isinstance(commit["message"], str):
+                raise ValueError(f"Commit {idx + 1} missing 'message' string")
+            if len(commit["files"]) == 0:
+                raise ValueError(f"Commit {idx + 1} has empty files list")
+            if not commit["message"].strip():
+                raise ValueError(f"Commit {idx + 1} has empty message")
+
+        return parsed
+
+    def generate_grouped_commits_with_retry(
+        self,
+        model: str,
+        conversation_messages: list[dict[str, str]],
+        temperature: float,
+        max_output_tokens: int,
+        max_retries: int,
+        quiet: bool,
+        staged_files_set: set[str],
+        require_confirmation: bool = True,
+    ) -> GroupedCommitResult:
+        """Generate grouped commits with validation and retry logic."""
+        first_iteration = True
+        content_retry_budget = max(3, int(max_retries))
+        attempts = 0
+
+        warning_limit_val = self.config.get("warning_limit_tokens", EnvDefaults.WARNING_LIMIT_TOKENS)
+        if warning_limit_val is None:
+            raise ConfigError("warning_limit_tokens configuration missing")
+        warning_limit = int(warning_limit_val)
+
+        while True:
+            prompt_tokens = count_tokens(conversation_messages, model)
+
+            if first_iteration:
+                if not check_token_warning(prompt_tokens, warning_limit, require_confirmation):
+                    sys.exit(0)
+            first_iteration = False
+
+            raw_response = generate_grouped_commits(
+                model=model,
+                prompt=conversation_messages,
+                temperature=temperature,
+                max_tokens=max_output_tokens,
+                max_retries=max_retries,
+                quiet=quiet,
+                skip_success_message=True,
+            )
+
+            try:
+                parsed = self.parse_and_validate_json_response(raw_response)
+            except ValueError as e:
+                attempts += 1
+                feedback = f"Invalid response structure: {e}. Please return ONLY valid JSON following the schema with a non-empty 'commits' array of objects containing 'files' and 'message'."
+                error_msg = f"Invalid grouped commits structure after {attempts} retries: {e}"
+                if self.handle_validation_retry(
+                    attempts,
+                    content_retry_budget,
+                    raw_response,
+                    feedback,
+                    error_msg,
+                    conversation_messages,
+                    quiet,
+                    "Structure validation failed, asking model to fix...",
+                ):
+                    sys.exit(1)
+                continue
+
+            # Assert parsed is not None for mypy - ValueError would have been raised earlier
+            assert parsed is not None
+            ok, feedback, detail_msg = self.validate_grouped_files_or_feedback(staged_files_set, parsed)
+            if not ok:
+                attempts += 1
+                error_msg = f"Grouped commits file set mismatch after {attempts} retries{': ' + detail_msg if detail_msg else ''}"
+                if self.handle_validation_retry(
+                    attempts,
+                    content_retry_budget,
+                    raw_response,
+                    feedback,
+                    error_msg,
+                    conversation_messages,
+                    quiet,
+                    "File coverage mismatch, asking model to fix...",
+                ):
+                    sys.exit(1)
+                continue
+
+            conversation_messages.append({"role": "assistant", "content": raw_response})
+            # Assert parsed is not None for mypy - ValueError would have been raised earlier
+            assert parsed is not None
+            return GroupedCommitResult(commits=parsed["commits"], raw_response=raw_response)
+
+    def display_grouped_commits(self, result: GroupedCommitResult, model: str, prompt_tokens: int, quiet: bool) -> None:
+        """Display the generated grouped commits to the user."""
+        provider, model_name = model.split(":", 1)
+
+        if not quiet:
+            console.print(f"[green]✔ Generated commit messages with {provider} {model_name}[/green]")
+            num_commits = len(result.commits)
+            console.print(f"[bold green]Proposed Commits ({num_commits}):[/bold green]\n")
+            for idx, commit in enumerate(result.commits, 1):
+                files = commit["files"]
+                files_display = ", ".join(files)
+                console.print(f"[dim]{files_display}[/dim]")
+                commit_msg = commit["message"].strip()
+                console.print(Panel(commit_msg, title=f"Commit Message {idx}/{num_commits}", border_style="cyan"))
+                console.print()
+
+            completion_tokens = count_tokens(result.raw_response, model)
+            total_tokens = prompt_tokens + completion_tokens
+            console.print(
+                f"[dim]Token usage: {prompt_tokens} prompt + {completion_tokens} completion = {total_tokens} total[/dim]"
+            )
+
+    def handle_grouped_commit_confirmation(self, result: GroupedCommitResult) -> bool:
+        """Handle user confirmation for grouped commits. Returns True if accepted."""
+        num_commits = len(result.commits)
+        while True:
+            response = click.prompt(
+                f"Proceed with {num_commits} commits above? [y/n/r/<feedback>]",
+                type=str,
+                show_default=False,
+            ).strip()
+            response_lower = response.lower()
+
+            if response_lower in ["y", "yes"]:
+                return True
+            if response_lower in ["n", "no"]:
+                console.print("[yellow]Commits not accepted. Exiting...[/yellow]")
+                sys.exit(0)
+            if response == "":
+                continue
+            if response_lower in ["r", "reroll"]:
+                console.print("[cyan]Regenerating commit groups...[/cyan]")
+                return False  # Signal to regenerate
+
+    def execute_grouped_commits(
+        self,
+        result: GroupedCommitResult,
+        dry_run: bool,
+        push: bool,
+        no_verify: bool,
+        hook_timeout: int,
+    ) -> None:
+        """Execute the grouped commits by creating multiple individual commits."""
+        num_commits = len(result.commits)
+
+        if dry_run:
+            console.print(f"[yellow]Dry run: Would create {num_commits} commits[/yellow]")
+            for idx, commit in enumerate(result.commits, 1):
+                console.print(f"\n[cyan]Commit {idx}/{num_commits}:[/cyan]")
+                console.print(f"  Files: {', '.join(commit['files'])}")
+                console.print(f"  Message: {commit['message'].strip()[:50]}...")
+        else:
+            original_staged_files = get_staged_files(existing_only=False)
+            original_staged_diff = run_git_command(["diff", "--cached", "--binary"], silent=True)
+            run_git_command(["reset", "HEAD"])
+
+            try:
+                # Detect file renames to handle them properly
+                rename_mappings = detect_rename_mappings(original_staged_diff)
+
+                for idx, commit in enumerate(result.commits, 1):
+                    try:
+                        for file_path in commit["files"]:
+                            # Check if this file is the destination of a rename
+                            if file_path in rename_mappings:
+                                old_file = rename_mappings[file_path]
+                                # For renames, stage both the old file (for deletion) and new file
+                                # This ensures the complete rename operation is preserved
+                                run_git_command(["add", "-A", old_file])
+                                run_git_command(["add", "-A", file_path])
+                            else:
+                                run_git_command(["add", "-A", file_path])
+                        execute_commit(commit["message"].strip(), no_verify, hook_timeout)
+                        console.print(f"[green]✓ Commit {idx}/{num_commits} created[/green]")
+                    except (AIError, ConfigError, GitError, subprocess.SubprocessError, OSError) as e:
+                        console.print(f"[red]✗ Failed at commit {idx}/{num_commits}: {e}[/red]")
+                        console.print(f"[yellow]Completed {idx - 1}/{num_commits} commits.[/yellow]")
+                        if idx == 1:
+                            console.print("[yellow]Restoring original staging area...[/yellow]")
+                            restore_staging(original_staged_files, original_staged_diff)
+                            console.print("[green]Original staging area restored.[/green]")
+                        sys.exit(1)
+            except KeyboardInterrupt:
+                console.print("\n[yellow]Interrupted by user. Restoring original staging area...[/yellow]")
+                restore_staging(original_staged_files, original_staged_diff)
+                console.print("[green]Original staging area restored.[/green]")
+                sys.exit(1)
+
+        if push:
+            try:
+                if dry_run:
+                    console.print("[yellow]Dry run: Would push changes[/yellow]")
+                    sys.exit(0)
+                from gac.git import push_changes
+
+                if push_changes():
+                    logger.info("Changes pushed successfully")
+                    console.print("[green]Changes pushed successfully[/green]")
+                else:
+                    console.print(
+                        "[red]Failed to push changes. Check your remote configuration and network connection.[/red]"
+                    )
+                    sys.exit(1)
+            except (GitError, OSError) as e:
+                console.print(f"[red]Error pushing changes: {e}[/red]")
+                sys.exit(1)
+
+        sys.exit(0)
+
+    def execute_workflow(
+        self,
+        system_prompt: str,
+        user_prompt: str,
+        model: str,
+        temperature: float,
+        max_output_tokens: int,
+        max_retries: int,
+        require_confirmation: bool,
+        quiet: bool,
+        no_verify: bool,
+        dry_run: bool,
+        push: bool,
+        show_prompt: bool,
+        interactive: bool,
+        message_only: bool,
+        git_state: GitState,
+        hint: str,
+        hook_timeout: int = 120,
+    ) -> None:
+        """Execute the complete grouped commit workflow."""
+        if show_prompt:
+            full_prompt = f"SYSTEM PROMPT:\n{system_prompt}\n\nUSER PROMPT:\n{user_prompt}"
+            console.print(Panel(full_prompt, title="Prompt for LLM", border_style="bright_blue"))
+
+        conversation_messages: list[dict[str, str]] = []
+        if system_prompt:
+            conversation_messages.append({"role": "system", "content": system_prompt})
+        conversation_messages.append({"role": "user", "content": user_prompt})
+
+        # Get staged files for validation
+        staged_files_set = set(get_staged_files(existing_only=False))
+
+        # Handle interactive questions if enabled
+        if interactive and not message_only:
+            from gac.interactive_mode import InteractiveMode
+
+            interactive_mode = InteractiveMode(self.config)
+            interactive_mode.handle_interactive_flow(
+                model=model,
+                user_prompt=user_prompt,
+                git_state=git_state,
+                hint=hint,
+                conversation_messages=conversation_messages,
+                temperature=temperature,
+                max_tokens=max_output_tokens,
+                max_retries=max_retries,
+                quiet=quiet,
+            )
+
+        while True:
+            # Generate grouped commits
+            result = self.generate_grouped_commits_with_retry(
+                model=model,
+                conversation_messages=conversation_messages,
+                temperature=temperature,
+                max_output_tokens=max_output_tokens,
+                max_retries=max_retries,
+                quiet=quiet,
+                staged_files_set=staged_files_set,
+                require_confirmation=require_confirmation,
+            )
+
+            # Display results
+            prompt_tokens = count_tokens(conversation_messages, model)
+            self.display_grouped_commits(result, model, prompt_tokens, quiet)
+
+            # Handle confirmation
+            if require_confirmation:
+                if self.handle_grouped_commit_confirmation(result):
+                    # User accepted, execute commits
+                    self.execute_grouped_commits(
+                        result=result,
+                        dry_run=dry_run,
+                        push=push,
+                        no_verify=no_verify,
+                        hook_timeout=hook_timeout,
+                    )
+                else:
+                    # User wants to regenerate, continue loop
+                    continue
+            else:
+                # No confirmation required, execute directly
+                self.execute_grouped_commits(
+                    result=result,
+                    dry_run=dry_run,
+                    push=push,
+                    no_verify=no_verify,
+                    hook_timeout=hook_timeout,
+                )

gac/init_cli.py

@@ -0,0 +1,70 @@
+"""CLI for initializing gac configuration interactively."""
+
+from pathlib import Path
+from typing import cast
+
+import click
+import questionary
+from dotenv import dotenv_values
+
+from gac.language_cli import configure_language_init_workflow
+from gac.model_cli import _configure_model
+
+GAC_ENV_PATH = Path.home() / ".gac.env"
+
+
+def _prompt_required_text(prompt: str) -> str | None:
+    """Prompt until a non-empty string is provided or the user cancels."""
+    while True:
+        response = questionary.text(prompt).ask()
+        if response is None:
+            return None
+        value = response.strip()
+        if value:
+            return cast(str, value)
+        click.echo("A value is required. Please try again.")
+
+
+def _load_existing_env() -> dict[str, str]:
+    """Ensure the env file exists and return its current values."""
+    existing_env: dict[str, str] = {}
+    if GAC_ENV_PATH.exists():
+        click.echo(f"$HOME/.gac.env already exists at {GAC_ENV_PATH}. Values will be updated.")
+        existing_env = {k: v for k, v in dotenv_values(str(GAC_ENV_PATH)).items() if v is not None}
+    else:
+        GAC_ENV_PATH.touch()
+        click.echo(f"Created $HOME/.gac.env at {GAC_ENV_PATH}.")
+    return existing_env
+
+
+def _configure_language(existing_env: dict[str, str]) -> None:
+    """Run the language configuration flow using consolidated logic."""
+    click.echo("\n")
+
+    # Use the consolidated language configuration from language_cli
+    success = configure_language_init_workflow(GAC_ENV_PATH)
+
+    if not success:
+        click.echo("Language configuration cancelled or failed.")
+    else:
+        click.echo("Language configuration completed.")
+
+
+@click.command()
+def init() -> None:
+    """Interactively set up $HOME/.gac.env for gac."""
+    click.echo("Welcome to gac initialization!\n")
+
+    existing_env = _load_existing_env()
+
+    if not _configure_model(existing_env):
+        click.echo("Model configuration cancelled. Exiting.")
+        return
+
+    _configure_language(existing_env)
+
+    click.echo("\ngac environment setup complete 🎉")
+    click.echo("Configuration saved to:")
+    click.echo(f"  {GAC_ENV_PATH}")
+    click.echo("\nYou can now run 'gac' in any Git repository to generate commit messages.")
+    click.echo("Run 'gac --help' to see available options.")