Initial upload: Local AI Commit Reviewer CLI with CI/CD workflow

src/core/review_engine.py

@@ -0,0 +1,421 @@
+import json
+import re
+from dataclasses import dataclass, field
+from enum import Enum
+from pathlib import Path
+
+from ..config import Config, StrictnessProfile
+from ..git import FileChange, GitRepo
+from ..llm import LLMProvider, OllamaProvider
+from ..llm.templates import ReviewPromptTemplates
+
+
+class IssueSeverity(str, Enum):
+    CRITICAL = "critical"
+    WARNING = "warning"
+    INFO = "info"
+
+
+class IssueCategory(str, Enum):
+    BUG = "bug"
+    SECURITY = "security"
+    STYLE = "style"
+    PERFORMANCE = "performance"
+    DOCUMENTATION = "documentation"
+
+
+@dataclass
+class Issue:
+    file: str
+    line: int
+    severity: IssueSeverity
+    category: IssueCategory
+    message: str
+    suggestion: str | None = None
+    raw_line: str | None = None
+
+    def to_dict(self) -> dict:
+        return {
+            "file": self.file,
+            "line": self.line,
+            "severity": self.severity.value,
+            "category": self.category.value,
+            "message": self.message,
+            "suggestion": self.suggestion,
+            "raw_line": self.raw_line
+        }
+
+    @classmethod
+    def from_dict(cls, data: dict) -> "Issue":
+        return cls(
+            file=data["file"],
+            line=data["line"],
+            severity=IssueSeverity(data["severity"]),
+            category=IssueCategory(data["category"]),
+            message=data["message"],
+            suggestion=data.get("suggestion"),
+            raw_line=data.get("raw_line")
+        )
+
+
+@dataclass
+class ReviewSummary:
+    critical_count: int = 0
+    warning_count: int = 0
+    info_count: int = 0
+    files_reviewed: int = 0
+    lines_changed: int = 0
+    overall_assessment: str = ""
+    issues_by_category: dict = field(default_factory=dict)
+    issues_by_file: dict = field(default_factory=dict)
+
+    def to_dict(self) -> dict:
+        return {
+            "critical_count": self.critical_count,
+            "warning_count": self.warning_count,
+            "info_count": self.info_count,
+            "files_reviewed": self.files_reviewed,
+            "lines_changed": self.lines_changed,
+            "overall_assessment": self.overall_assessment,
+            "issues_by_category": self.issues_by_category,
+            "issues_by_file": self.issues_by_file
+        }
+
+
+@dataclass
+class ReviewResult:
+    issues: list[Issue] = field(default_factory=list)
+    summary: ReviewSummary = field(default_factory=ReviewSummary)
+    model_used: str = ""
+    tokens_used: int = 0
+    review_mode: str = ""
+    error: str | None = None
+
+    def has_critical_issues(self) -> bool:
+        return any(issue.severity == IssueSeverity.CRITICAL for issue in self.issues)
+
+    def has_issues(self) -> bool:
+        return len(self.issues) > 0
+
+    def get_issues_by_severity(self, severity: IssueSeverity) -> list[Issue]:
+        return [issue for issue in self.issues if issue.severity == severity]
+
+    def get_issues_by_file(self, filename: str) -> list[Issue]:
+        return [issue for issue in self.issues if issue.file == filename]
+
+    def get_issues_by_category(self, category: IssueCategory) -> list[Issue]:
+        return [issue for issue in self.issues if issue.category == category]
+
+    def to_json(self) -> str:
+        return json.dumps({
+            "issues": [issue.to_dict() for issue in self.issues],
+            "summary": self.summary.to_dict(),
+            "model_used": self.model_used,
+            "tokens_used": self.tokens_used,
+            "review_mode": self.review_mode
+        }, indent=2)
+
+    def to_markdown(self) -> str:
+        lines = ["# AI Commit Review Results\n"]
+
+        lines.append("## Summary\n")
+        lines.append(f"- **Files Reviewed**: {self.summary.files_reviewed}")
+        lines.append(f"- **Lines Changed**: {self.summary.lines_changed}")
+        lines.append(f"- **Critical Issues**: {self.summary.critical_count}")
+        lines.append(f"- **Warnings**: {self.summary.warning_count}")
+        lines.append(f"- **Info**: {self.summary.info_count}")
+        lines.append(f"- **Assessment**: {self.summary.overall_assessment}\n")
+
+        if self.issues:
+            lines.append("## Issues Found\n")
+
+            for severity in [IssueSeverity.CRITICAL, IssueSeverity.WARNING, IssueSeverity.INFO]:
+                severity_issues = self.get_issues_by_severity(severity)
+                if severity_issues:
+                    lines.append(f"### {severity.value.upper()} ({len(severity_issues)})\n")
+                    for issue in severity_issues:
+                        lines.append(f"#### {issue.file}:{issue.line}")
+                        lines.append(f"- **Category**: {issue.category.value}")
+                        lines.append(f"- **Message**: {issue.message}")
+                        if issue.suggestion:
+                            lines.append(f"- **Suggestion**: {issue.suggestion}")
+                        lines.append("")
+
+        return "\n".join(lines)
+
+
+class ReviewEngine:
+    def __init__(
+        self,
+        config: Config | None = None,
+        llm_provider: LLMProvider | None = None
+    ):
+        self.config = config or Config()
+        self.llm_provider = llm_provider or OllamaProvider(
+            endpoint=self.config.llm.endpoint,
+            model=self.config.llm.model,
+            timeout=self.config.llm.timeout
+        )
+        self.repo: GitRepo | None = None
+
+    def set_repo(self, path: Path) -> None:
+        self.repo = GitRepo(path)
+
+    def _parse_llm_response(self, response_text: str, files: list[FileChange]) -> ReviewResult:
+        result = ReviewResult()
+
+        try:
+            json_match = re.search(r'\{[\s\S]*\}', response_text)
+            if json_match:
+                json_str = json_match.group()
+                data = json.loads(json_str)
+
+                issues_data = data.get("issues", [])
+                for issue_data in issues_data:
+                    try:
+                        issue = Issue.from_dict(issue_data)
+                        result.issues.append(issue)
+                    except Exception:
+                        pass
+
+                summary_data = data.get("summary", {})
+                result.summary.critical_count = summary_data.get("critical_count", 0)
+                result.summary.warning_count = summary_data.get("warning_count", 0)
+                result.summary.info_count = summary_data.get("info_count", 0)
+                result.summary.overall_assessment = summary_data.get("overall_assessment", "")
+            else:
+                text_issues = self._parse_text_response(response_text, files)
+                result.issues = text_issues
+
+        except json.JSONDecodeError:
+            result.issues = self._parse_text_response(response_text, files)
+
+        return result
+
+    def _parse_text_response(self, response_text: str, files: list[FileChange]) -> list[Issue]:
+        issues = []
+        lines = response_text.split("\n")
+
+        current_file = ""
+        for line in lines:
+            file_match = re.match(r'^\*\*(.+?)\*\*:\s*(\d+)', line)
+            if file_match:
+                current_file = file_match.group(1)
+                line_num = int(file_match.group(2))
+
+                severity = IssueSeverity.WARNING
+                if "critical" in line.lower():
+                    severity = IssueSeverity.CRITICAL
+                elif "security" in line.lower():
+                    severity = IssueSeverity.CRITICAL
+                    category = IssueCategory.SECURITY
+                else:
+                    category = IssueCategory.BUG
+
+                message = line
+                suggestion = None
+                if "->" in line:
+                    parts = line.split("->")
+                    message = parts[0].strip()
+                    suggestion = "->".join(parts[1:]).strip()
+
+                issues.append(Issue(
+                    file=current_file,
+                    line=line_num,
+                    severity=severity,
+                    category=category,
+                    message=message,
+                    suggestion=suggestion
+                ))
+
+        return issues
+
+    def _get_strictness_profile(self) -> StrictnessProfile:
+        return self.config.strictness_profiles.get_profile(
+            self.config.review.strictness
+        )
+
+    def _filter_issues_by_strictness(self, issues: list[Issue]) -> list[Issue]:
+        profile = self._get_strictness_profile()
+
+        severity_order = {
+            IssueSeverity.CRITICAL: 0,
+            IssueSeverity.WARNING: 1,
+            IssueSeverity.INFO: 2
+        }
+
+        min_severity = profile.min_severity.lower()
+        min_level = 2
+        if min_severity == "critical":
+            min_level = 0
+        elif min_severity == "warning":
+            min_level = 1
+
+        filtered = []
+        for issue in issues:
+            level = severity_order.get(issue.severity, 2)
+            if level <= min_level:
+                if issue.category == IssueCategory.SECURITY and not profile.check_security:
+                    continue
+                if issue.category == IssueCategory.BUG and not profile.check_bugs:
+                    continue
+                if issue.category == IssueCategory.STYLE and not profile.check_style:
+                    continue
+                if issue.category == IssueCategory.PERFORMANCE and not profile.check_performance:
+                    continue
+                if issue.category == IssueCategory.DOCUMENTATION and not profile.check_documentation:
+                    continue
+                filtered.append(issue)
+
+        return filtered
+
+    def _aggregate_summary(self, issues: list[Issue], files: list[FileChange]) -> ReviewSummary:
+        summary = ReviewSummary()
+        summary.files_reviewed = len(files)
+        summary.lines_changed = sum(
+            sum(1 for line in f.diff.split("\n") if line.startswith("+") and not line.startswith("+++"))
+            for f in files
+        )
+
+        for issue in issues:
+            if issue.severity == IssueSeverity.CRITICAL:
+                summary.critical_count += 1
+            elif issue.severity == IssueSeverity.WARNING:
+                summary.warning_count += 1
+            else:
+                summary.info_count += 1
+
+            if issue.category.value not in summary.issues_by_category:
+                summary.issues_by_category[issue.category.value] = []
+            summary.issues_by_category[issue.category.value].append(issue.file)
+
+            if issue.file not in summary.issues_by_file:
+                summary.issues_by_file[issue.file] = []
+            summary.issues_by_file[issue.file].append(issue.line)
+
+        if summary.critical_count > 0:
+            summary.overall_assessment = "Critical issues found. Review recommended before committing."
+        elif summary.warning_count > 0:
+            summary.overall_assessment = "Warnings found. Consider addressing before committing."
+        elif summary.info_count > 0:
+            summary.overall_assessment = "Minor issues found. Ready for commit with optional fixes."
+        else:
+            summary.overall_assessment = "No issues found. Code is ready for commit."
+
+        return summary
+
+    def review_staged_changes(
+        self,
+        files: list[FileChange] | None = None,
+        strictness: str | None = None,
+        language: str | None = None
+    ) -> ReviewResult:
+        if files is None:
+            if self.repo is None:
+                self.repo = GitRepo(Path.cwd())
+            files = self.repo.get_all_staged_changes()
+
+        if not files:
+            return ReviewResult(error="No staged changes found")
+
+        result = ReviewResult()
+        result.review_mode = strictness or self.config.review.strictness
+
+        if strictness is None:
+            strictness = self.config.review.strictness
+
+        all_issues = []
+
+        for file_change in files:
+            if not file_change.diff.strip():
+                continue
+
+            file_language = language or self.repo.get_file_language(file_change.filename)
+
+            prompt = ReviewPromptTemplates.get_prompt(
+                diff=file_change.diff,
+                strictness=strictness,
+                language=file_language
+            )
+
+            try:
+                if self.llm_provider.is_available():
+                    response = self.llm_provider.generate(
+                        prompt,
+                        max_tokens=self.config.llm.max_tokens,
+                        temperature=self.config.llm.temperature
+                    )
+                    result.model_used = response.model
+                    result.tokens_used += response.tokens_used
+
+                    file_result = self._parse_llm_response(response.text, [file_change])
+                    all_issues.extend(file_result.issues)
+                else:
+                    result.error = "LLM provider is not available"
+                    return result
+            except Exception as e:
+                result.error = f"Review failed: {e!s}"
+                return result
+
+        filtered_issues = self._filter_issues_by_strictness(all_issues)
+        max_issues = self.config.review.max_issues_per_file
+        limited_issues = filtered_issues[:max_issues * len(files)]
+        result.issues = limited_issues
+        result.summary = self._aggregate_summary(limited_issues, files)
+
+        return result
+
+    def review_commit(
+        self,
+        sha: str,
+        strictness: str | None = None
+    ) -> ReviewResult:
+        if self.repo is None:
+            self.repo = GitRepo(Path.cwd())
+
+        commit_info = self.repo.get_commit_info(sha)
+        if commit_info is None:
+            return ReviewResult(error=f"Commit {sha} not found")
+
+        result = ReviewResult()
+        result.review_mode = strictness or self.config.review.strictness
+
+        if strictness is None:
+            strictness = self.config.review.strictness
+
+        all_issues = []
+
+        for file_change in commit_info.changes:
+            if not file_change.diff.strip():
+                continue
+
+            prompt = ReviewPromptTemplates.get_commit_review_prompt(
+                diff=file_change.diff,
+                commit_message=commit_info.message,
+                strictness=strictness
+            )
+
+            try:
+                if self.llm_provider.is_available():
+                    response = self.llm_provider.generate(
+                        prompt,
+                        max_tokens=self.config.llm.max_tokens,
+                        temperature=self.config.llm.temperature
+                    )
+                    result.model_used = response.model
+                    result.tokens_used += response.tokens_used
+
+                    file_result = self._parse_llm_response(response.text, [file_change])
+                    all_issues.extend(file_result.issues)
+                else:
+                    result.error = "LLM provider is not available"
+                    return result
+            except Exception as e:
+                result.error = f"Review failed: {e!s}"
+                return result
+
+        filtered_issues = self._filter_issues_by_strictness(all_issues)
+        result.issues = filtered_issues
+        result.summary = self._aggregate_summary(filtered_issues, commit_info.changes)
+
+        return result