gitrelevance 0.1.0__py3-none-any.whl

gitrelevance/providers/github.py

@@ -0,0 +1,419 @@
+"""GitHub provider implementation using PyGithub."""
+
+from __future__ import annotations
+
+import logging
+import os
+import re
+import tempfile
+import time
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from datetime import datetime
+from typing import Literal
+
+from diskcache import Cache
+from github import Auth, Github, GithubException, RateLimitExceededException
+
+from gitrelevance.config import load_github_token
+from gitrelevance.issues.models import Issue, PullRequest
+from gitrelevance.providers.base import Provider, RateLimitExceededError
+
+logger = logging.getLogger(__name__)
+
+# Regex pattern to extract closing issue numbers from PR body/title
+CLOSING_KEYWORD_PATTERN = re.compile(
+    r"(?:close|closes|closed|fix|fixes|fixed|resolve|resolves|resolved)\s+(?:#|gh-)(\d+)",
+    re.IGNORECASE,
+)
+
+
+class GitHubProvider:
+    """GitHub provider implementing the Provider protocol.
+
+    Provides access to GitHub issues, pull requests, and comments with disk caching
+    and rate limit error handling.
+    """
+
+    def __init__(
+        self,
+        owner: str,
+        repo: str,
+        token: str | None = None,
+        cache_dir: str | None = None,
+        ttl: int = 900,
+        timeout: int = 15,
+    ) -> None:
+        """Initialize the GitHub provider.
+
+        Args:
+            owner: Repository owner or organization name
+            repo: Repository name
+            token: Optional GitHub Personal Access Token. If None, loaded from GITHUB_TOKEN env var.
+            cache_dir: Directory for diskcache storage. If None, uses a temporary directory.
+            ttl: Cache TTL in seconds (default: 900 / 15 minutes).
+            timeout: HTTP request timeout in seconds (default: 15).
+        """
+        self.owner = owner
+        self.repo_name = repo
+        self._token = token or load_github_token()
+        self.ttl = ttl
+        self.timeout = timeout
+
+        if cache_dir is None:
+            cache_dir = os.path.join(tempfile.gettempdir(), "gitrelevance_cache")
+        self._cache = Cache(cache_dir)
+
+        # Explicit timeout and retry=0 to prevent indefinite socket hangs and silent backoff sleeping on 403
+        if self._token:
+            auth = Auth.Token(self._token)
+            self._client = Github(
+                auth=auth,
+                timeout=self.timeout,
+                retry=0,
+                per_page=100,
+            )
+        else:
+            self._client = Github(
+                timeout=self.timeout,
+                retry=0,
+                per_page=100,
+            )
+
+    @staticmethod
+    def parse_remote(remote_url: str) -> tuple[str, str] | None:
+        """Parse a remote URL into (owner, repo) pair for GitHub repositories.
+
+        Supported URL formats:
+        - git@github.com:owner/repo.git
+        - https://github.com/owner/repo.git
+        - https://github.com/owner/repo
+        - github.com/owner/repo
+
+        Args:
+            remote_url: Remote URL string
+
+        Returns:
+            Tuple of (owner, repo) if valid GitHub URL, None otherwise.
+        """
+        if not remote_url:
+            return None
+
+        url = remote_url.strip()
+
+        patterns = [
+            r"^git@github\.com:([a-zA-Z0-9_.-]+)/([a-zA-Z0-9_.-]+?)(?:\.git)?$",
+            r"^https?://github\.com/([a-zA-Z0-9_.-]+)/([a-zA-Z0-9_.-]+?)(?:\.git)?/?$",
+            r"^github\.com/([a-zA-Z0-9_.-]+)/([a-zA-Z0-9_.-]+?)(?:\.git)?/?$",
+        ]
+
+        for pattern in patterns:
+            match = re.match(pattern, url)
+            if match:
+                owner, repo = match.group(1), match.group(2)
+                return owner, repo
+
+        return None
+
+    def _get_github_repo(self) -> object:
+        """Get PyGithub Repository instance."""
+        return self._client.get_repo(f"{self.owner}/{self.repo_name}")
+
+    def get_issues(
+        self,
+        state: Literal["open", "closed", "all"] = "all",
+        force_refresh: bool = False,
+    ) -> list[Issue]:
+        """Fetch repository issues.
+
+        Args:
+            state: Issue state filter ("open", "closed", or "all")
+            force_refresh: If True, bypass cache and fetch fresh data.
+
+        Returns:
+            List of Issue dataclasses.
+
+        Raises:
+            RateLimitExceededError: If GitHub API rate limit is exceeded.
+        """
+        cache_key = (self.owner, self.repo_name, "issues", state)
+        if not force_refresh and self.ttl > 0:
+            cached_data = self._cache.get(cache_key)
+            if cached_data is not None:
+                logger.info("Loaded %d issues from disk cache for %s/%s", len(cached_data), self.owner, self.repo_name)
+                return cached_data
+
+        logger.info("Fetching issues from GitHub API for %s/%s (state=%s)...", self.owner, self.repo_name, state)
+        try:
+            gh_repo = self._get_github_repo()
+            gh_issues = gh_repo.get_issues(state=state)
+
+            issues: list[Issue] = []
+            for gh_issue in gh_issues:
+                # Filter out PRs (PyGithub get_issues returns issues + PRs unless filtered)
+                is_pr = False
+                raw_data = getattr(gh_issue, "_rawData", None)
+                if isinstance(raw_data, dict):
+                    is_pr = bool(raw_data.get("pull_request"))
+                else:
+                    try:
+                        is_pr = getattr(gh_issue, "pull_request", None) is not None
+                    except (RateLimitExceededException, KeyboardInterrupt):
+                        raise
+                    except Exception:
+                        is_pr = False
+
+                if is_pr:
+                    continue
+
+                linked_prs: list[int] = []
+                # Check timeline events if available
+                if hasattr(gh_issue, "get_timeline"):
+                    try:
+                        timeline = gh_issue.get_timeline()
+                        for event in timeline:
+                            event_name = getattr(event, "event", None)
+                            if event_name in ("cross-referenced", "connected", "referenced"):
+                                source = getattr(event, "source", None)
+                                if source and isinstance(source, dict):
+                                    issue_info = source.get("issue")
+                                    if issue_info and isinstance(issue_info, dict):
+                                        if "pull_request" in issue_info or issue_info.get("pull_request") is not None:
+                                            pr_num = issue_info.get("number")
+                                            if pr_num is not None and int(pr_num) not in linked_prs:
+                                                linked_prs.append(int(pr_num))
+                                elif source:
+                                    issue_info = getattr(source, "issue", None)
+                                    if issue_info:
+                                        if getattr(issue_info, "pull_request", None) is not None:
+                                            pr_num = getattr(issue_info, "number", None)
+                                            if pr_num is not None and int(pr_num) not in linked_prs:
+                                                linked_prs.append(int(pr_num))
+                    except (RateLimitExceededException, KeyboardInterrupt):
+                        raise
+                    except Exception as e:
+                        logger.debug("Could not fetch timeline for issue #%s: %s", gh_issue.number, e)
+
+                issue = Issue(
+                    number=gh_issue.number,
+                    title=gh_issue.title,
+                    body=gh_issue.body or "",
+                    state="closed" if gh_issue.state == "closed" else "open",
+                    created_at=gh_issue.created_at,
+                    closed_at=gh_issue.closed_at,
+                    labels=tuple(label.name for label in gh_issue.labels),
+                    linked_pr_numbers=tuple(linked_prs),
+                )
+                issues.append(issue)
+
+            logger.info("Successfully fetched %d issues for %s/%s", len(issues), self.owner, self.repo_name)
+            if self.ttl > 0:
+                self._cache.set(cache_key, issues, expire=self.ttl)
+            return issues
+
+        except RateLimitExceededException as e:
+            reset = getattr(e, "reset", None)
+            logger.warning("GitHub API rate limit exceeded for %s/%s", self.owner, self.repo_name)
+            raise RateLimitExceededError(reset_at=reset) from e
+        except GithubException as e:
+            if e.status in (403, 429) and "rate limit" in str(e).lower():
+                reset = getattr(e, "reset", None)
+                logger.warning("GitHub API rate limit exceeded (HTTP %d) for %s/%s", e.status, self.owner, self.repo_name)
+                raise RateLimitExceededError(reset_at=reset) from e
+            raise
+
+    def get_pull_requests(
+        self,
+        state: Literal["open", "closed", "all"] = "all",
+        force_refresh: bool = False,
+    ) -> list[PullRequest]:
+        """Fetch repository pull requests.
+
+        Args:
+            state: Pull request state filter ("open", "closed", or "all")
+            force_refresh: If True, bypass cache and fetch fresh data.
+
+        Returns:
+            List of PullRequest dataclasses.
+
+        Raises:
+            RateLimitExceededError: If GitHub API rate limit is exceeded.
+        """
+        cache_key = (self.owner, self.repo_name, "pull_requests", state)
+        if not force_refresh and self.ttl > 0:
+            cached_data = self._cache.get(cache_key)
+            if cached_data is not None:
+                logger.info("Loaded %d pull requests from disk cache for %s/%s", len(cached_data), self.owner, self.repo_name)
+                return cached_data
+
+        logger.info("Fetching pull requests from GitHub API for %s/%s (state=%s)...", self.owner, self.repo_name, state)
+        t0 = time.perf_counter()
+        try:
+            gh_repo = self._get_github_repo()
+
+            # Materialize the paginated PR list up-front so we can time it separately
+            # from the per-PR metadata extraction.  get_pulls() returns a lazy
+            # PaginatedList; each iteration may trigger a page-fetch API call.
+            logger.info("  Phase 0: fetching paginated PR list...")
+            t_p0 = time.perf_counter()
+            gh_prs_list: list[object] = list(gh_repo.get_pulls(state=state))
+            t_p1 = time.perf_counter()
+            logger.info("  Phase 0: got %d PR objects from API in %.2fs.", len(gh_prs_list), t_p1 - t_p0)
+
+            # ------------------------------------------------------------------
+            # Phase 1: Iterate PRs, parse metadata, decide which need file fetch
+            # ------------------------------------------------------------------
+            # IMPORTANT: Use _rawData dict to read PR properties instead of
+            # PyGithub object attributes.  PyGithub's lazy-loading triggers a
+            # separate API call for each attribute access (e.g. gh_pr.body)
+            # when the PaginatedList did not include it, causing ~1s per PR.
+            # Reading from _rawData avoids these per-PR API calls entirely.
+            pr_metadata: list[tuple[object, list[int], bool]] = []  # (gh_pr, closes_issues, should_fetch_files)
+            for gh_pr in gh_prs_list:
+                closes_issues: list[int] = []
+
+                # Read from _rawData to avoid per-PR lazy-load API calls.
+                # NOTE: The list endpoint does NOT include a "merged" boolean;
+                # it includes "merged_at" (timestamp or null).  We compute
+                # "merged" from that.
+                raw = getattr(gh_pr, "_rawData", None) or {}
+                pr_title = raw.get("title", "") or ""
+                pr_body = raw.get("body", "") or ""
+                text_to_search = f"{pr_title} {pr_body}"
+                for match in CLOSING_KEYWORD_PATTERN.findall(text_to_search):
+                    try:
+                        num = int(match)
+                        if num not in closes_issues:
+                            closes_issues.append(num)
+                    except ValueError:
+                        pass
+
+                is_merged = raw.get("merged_at") is not None
+                # Optimization: Skip get_files() for closed unmerged PRs that don't close any issues
+                # (avoids thousands of unnecessary API calls for spam/rejected PRs on public repos)
+                should_fetch_files = (
+                    is_merged
+                    or raw.get("state", "open") == "open"
+                    or len(closes_issues) > 0
+                )
+                pr_metadata.append((gh_pr, closes_issues, should_fetch_files))
+
+            t_metadata = time.perf_counter()
+            logger.info(
+                "Enumerated %d PRs in %.2fs (metadata phase).  %d need file changes fetched.",
+                len(pr_metadata),
+                t_metadata - t_p1,
+                sum(1 for _, _, sf in pr_metadata if sf),
+            )
+
+            # ------------------------------------------------------------------
+            # Phase 2: Parallel fetch of changed-file lists
+            # ------------------------------------------------------------------
+            def _fetch_pr_files(pr_obj: object) -> tuple[str, ...]:
+                """Fetch the file list for a single PR (I/O-bound, GIL-releasing)."""
+                try:
+                    return tuple(f.filename for f in pr_obj.get_files())
+                except (RateLimitExceededException, KeyboardInterrupt):
+                    raise
+                except Exception as exc:
+                    logger.debug("Could not fetch changed files for PR #%s: %s", pr_obj.number, exc)
+                    return ()
+
+            # Build file results map: PR number -> files_changed
+            files_map: dict[int, tuple[str, ...]] = {}
+            fetchable = [(gh_pr, idx) for idx, (gh_pr, _, sf) in enumerate(pr_metadata) if sf]
+
+            if fetchable:
+                max_workers = min(len(fetchable), 8)
+                with ThreadPoolExecutor(max_workers=max_workers) as executor:
+                    future_to_pr_num = {
+                        executor.submit(_fetch_pr_files, gh_pr): gh_pr.number
+                        for gh_pr, _ in fetchable
+                    }
+                    for future in as_completed(future_to_pr_num):
+                        pr_num = future_to_pr_num[future]
+                        files_map[pr_num] = future.result()
+
+            t_files = time.perf_counter()
+            logger.info(
+                "Fetched file lists for %d PRs in %.2fs (parallel, max_workers=%d).",
+                len(files_map),
+                t_files - t_metadata,
+                min(len(fetchable), 8) if fetchable else 0,
+            )
+
+            # ------------------------------------------------------------------
+            # Phase 3: Assemble PullRequest dataclass objects
+            # ------------------------------------------------------------------
+            prs: list[PullRequest] = []
+            for gh_pr, closes_issues, should_fetch in pr_metadata:
+                files_changed = files_map.get(gh_pr.number, ()) if should_fetch else ()
+                raw = getattr(gh_pr, "_rawData", None) or {}
+                pr = PullRequest(
+                    number=raw.get("number", gh_pr.number),
+                    title=raw.get("title", ""),
+                    merged=raw.get("merged_at") is not None,
+                    merge_commit_sha=raw.get("merge_commit_sha"),
+                    closes_issue_numbers=tuple(closes_issues),
+                    files_changed=files_changed,
+                )
+                prs.append(pr)
+
+            logger.info(
+                "Successfully fetched %d pull requests for %s/%s in %.2fs total.",
+                len(prs), self.owner, self.repo_name, time.perf_counter() - t0,
+            )
+            if self.ttl > 0:
+                self._cache.set(cache_key, prs, expire=self.ttl)
+            return prs
+
+        except RateLimitExceededException as e:
+            reset = getattr(e, "reset", None)
+            logger.warning("GitHub API rate limit exceeded when fetching PRs for %s/%s", self.owner, self.repo_name)
+            raise RateLimitExceededError(reset_at=reset) from e
+        except GithubException as e:
+            if e.status in (403, 429) and "rate limit" in str(e).lower():
+                reset = getattr(e, "reset", None)
+                logger.warning("GitHub API rate limit exceeded (HTTP %d) when fetching PRs for %s/%s", e.status, self.owner, self.repo_name)
+                raise RateLimitExceededError(reset_at=reset) from e
+            raise
+
+    def get_issue_comments(
+        self,
+        issue_number: int,
+        force_refresh: bool = False,
+    ) -> list[str]:
+        """Fetch comment bodies for a given issue.
+
+        Args:
+            issue_number: Issue number
+            force_refresh: If True, bypass cache and fetch fresh data.
+
+        Returns:
+            List of comment body strings.
+
+        Raises:
+            RateLimitExceededError: If GitHub API rate limit is exceeded.
+        """
+        cache_key = (self.owner, self.repo_name, "comments", issue_number)
+        if not force_refresh and self.ttl > 0:
+            cached_data = self._cache.get(cache_key)
+            if cached_data is not None:
+                return cached_data
+
+        try:
+            gh_repo = self._get_github_repo()
+            gh_issue = gh_repo.get_issue(issue_number)
+            comments = [comment.body for comment in gh_issue.get_comments() if comment.body]
+
+            if self.ttl > 0:
+                self._cache.set(cache_key, comments, expire=self.ttl)
+            return comments
+
+        except RateLimitExceededException as e:
+            reset = getattr(e, "reset", None)
+            raise RateLimitExceededError(reset_at=reset) from e
+        except GithubException as e:
+            if e.status in (403, 429) and "rate limit" in str(e).lower():
+                reset = getattr(e, "reset", None)
+                raise RateLimitExceededError(reset_at=reset) from e
+            raise

gitrelevance-0.1.0.dist-info/METADATA

@@ -0,0 +1,269 @@
+Metadata-Version: 2.5
+Name: gitrelevance
+Version: 0.1.0
+Summary: Analyze whether historical GitHub issues are still relevant to a codebase, using Git history as evidence
+Project-URL: Homepage, https://github.com/gitrelevance/gitrelevance
+Project-URL: Repository, https://github.com/gitrelevance/gitrelevance
+Project-URL: Issues, https://github.com/gitrelevance/gitrelevance/issues
+Author: GitRelevance Contributors
+License-Expression: MIT
+License-File: LICENSE
+Classifier: Development Status :: 4 - Beta
+Classifier: Environment :: Console
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Software Development :: Version Control :: Git
+Requires-Python: >=3.10
+Requires-Dist: diskcache>=5.6.0
+Requires-Dist: gitpython>=3.1.0
+Requires-Dist: pygithub>=2.0.0
+Requires-Dist: python-dotenv>=1.0.0
+Requires-Dist: rich>=13.0.0
+Requires-Dist: typer>=0.9.0
+Requires-Dist: typing-extensions; python_version < '3.11'
+Provides-Extra: dev
+Requires-Dist: build>=1.0.0; extra == 'dev'
+Requires-Dist: pytest>=7.0; extra == 'dev'
+Requires-Dist: responses>=0.23.0; extra == 'dev'
+Requires-Dist: twine>=5.0.0; extra == 'dev'
+Description-Content-Type: text/markdown
+
+# GitRelevance
+
+> Analyze whether historical GitHub issues are still relevant to a codebase, using Git history as evidence.
+
+[![CI](https://github.com/gitrelevance/gitrelevance/actions/workflows/ci.yml/badge.svg)](https://github.com/gitrelevance/gitrelevance/actions/workflows/ci.yml)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Python: >=3.10](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/downloads/)
+
+---
+
+## What is GitRelevance?
+
+In long-lived repositories, issue trackers often accumulate hundreds of historical bug reports and feature requests. Over time, code is refactored, features are deprecated, or fixes are merged without closing the original issue.
+
+**GitRelevance** audits your open and closed GitHub issues against your local Git history. It automatically correlates commit messages, pull request merges, file renames, deletions, and reverts to determine whether an issue is:
+- **`RESOLVED`** — A fix commit exists in `HEAD` history and was not reverted.
+- **`PROBABLY_RESOLVED`** — A fixing PR was merged or substantial fix evidence was detected.
+- **`STILL_RELEVANT`** — The issue is open and related code still actively exists in the tree.
+- **`OBSOLETE`** — The files or feature referenced by the issue were deleted without replacement.
+- **`UNKNOWN`** — No correlating commits or evidence exist in local Git history.
+
+Every classification is backed by an explicit, human-readable list of evidence items and a confidence score.
+
+---
+
+## Installation
+
+```bash
+pip install gitrelevance
+```
+
+For development:
+```bash
+git clone https://github.com/gitrelevance/gitrelevance.git
+cd gitrelevance
+pip install -e ".[dev]"
+```
+
+Requires Python ≥ 3.10.
+
+---
+
+## Quick Start (CLI)
+
+Run `gitrelevance analyze` from inside any local Git clone of a GitHub repository:
+
+```bash
+# Analyze all repository issues
+gitrelevance analyze
+
+# Filter by state (open, closed, all)
+gitrelevance analyze --state open
+
+# Filter by issue creation date
+gitrelevance analyze --since 2024-01-01
+
+# Output structured JSON for automation or CI pipelines
+gitrelevance analyze --json
+```
+
+### Example Terminal Output
+
+```
+GitRelevance
+
+Repository: github.com/owner/my-repo
+Branch: main
+HEAD: a81f23c
+
+Analyzing 24 issues...
+
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+RESOLVED
+
+#21 Login crashes after token expiration
+Confidence: 96%
+
+Evidence:
+  ✓ Fix commit is present in HEAD history (a81f23c)
+  ✓ Issue referenced in commit message (a81f23c)
+  ✓ All related files exist at HEAD
+  ✓ No revert of fix commit detected (a81f23c)
+
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+* Confidence is a heuristic evidence-strength score (0–100%), not a statistical probability.
+```
+
+---
+
+## How Classification Works
+
+GitRelevance evaluates evidence through a multi-stage deterministic pipeline:
+
+```
+┌─────────────────┐     ┌─────────────────────┐
+│ Local Git Repo  │     │ GitHub API Provider │
+└────────┬────────┘     └──────────┬──────────┘
+         │                         │
+         ▼                         ▼
+  ┌───────────────────────────────────────┐
+  │ Matcher (Correlates commits & PRs)    │
+  └──────────────────┬────────────────────┘
+                     │
+                     ▼
+  ┌───────────────────────────────────────┐
+  │ Current-State Analysis (HEAD status)  │
+  └──────────────────┬────────────────────┘
+                     │
+                     ▼
+  ┌───────────────────────────────────────┐
+  │ Evidence Collection (Signed weights)  │
+  └──────────────────┬────────────────────┘
+                     │
+                     ▼
+  ┌───────────────────────────────────────┐
+  │ Confidence & Classifier Decision Tree │
+  └──────────────────┬────────────────────┘
+                     │
+                     ▼
+  ┌───────────────────────────────────────┐
+  │ Output (Terminal with Rich or JSON)   │
+  └───────────────────────────────────────┘
+```
+
+### Evidence Rules & Signed Weights
+- **Fix commit in HEAD history** (`+3`): A commit referencing `#N` or `GH-N` is an ancestor of `HEAD`.
+- **Linked PR merged** (`+2`): A pull request linked to or closing issue `#N` was merged into the repository.
+- **Related files exist at HEAD** (`+2`): All files modified by the issue's commits/PRs still exist.
+- **No revert detected** (`+1`): The fix commit was not undone by a subsequent `git revert`.
+- **Related files deleted without replacement** (`-3`): Referenced files were deleted in git history and not renamed.
+- **Fix commit later reverted** (`-3`): Git's native revert trailer (`This reverts commit <sha>.`) indicates the fix was undone.
+
+### Confidence Score (Evidence Strength)
+
+Confidence is computed as an **evidence-strength heuristic** normalized and clamped to `[0.05, 0.98]`.
+
+```text
+Confidence = 0.50 + (sum(Weights) / (2 × MAX_ABS_WEIGHT))
+```
+> **Important Note:** Confidence is **NOT** a calibrated Bayesian probability. A confidence of 95% indicates an overwhelming accumulation of strong supporting evidence in the Git tree, not that the classification is statistically 95% likely to be correct.
+
+---
+
+## GitHub Authentication & Rate Limits
+
+GitRelevance works unauthenticated for public repositories out of the box.
+
+To analyze private repositories or avoid GitHub's unauthenticated API rate limits (60 requests/hour), supply a personal access token via the `GITHUB_TOKEN` environment variable or a local `.env` file:
+
+```bash
+export GITHUB_TOKEN="ghp_yourPersonalAccessTokenHere"
+gitrelevance analyze
+```
+
+---
+
+## Python API Usage
+
+GitRelevance can also be integrated directly into Python scripts:
+
+```python
+from gitrelevance.git.repository import GitRepository
+from gitrelevance.providers.github import GitHubProvider
+from gitrelevance.analysis.engine import AnalysisEngine
+from gitrelevance.output.terminal import TerminalRenderer
+from gitrelevance.output.json import to_json
+
+# 1. Initialize Git repository and GitHub provider
+repo = GitRepository(".")
+owner, repo_name = GitHubProvider.parse_remote(repo.remote_url("origin"))
+provider = GitHubProvider(owner=owner, repo=repo_name)
+
+# 2. Run analysis
+engine = AnalysisEngine(repo, provider)
+results = engine.analyze(state="open")
+
+# 3. Render results
+for result in results:
+    print(f"#{result.issue.number} {result.issue.title} -> {result.classification.value} ({result.confidence:.0%})")
+```
+
+---
+
+## Running Tests
+
+```bash
+# Run complete test suite
+pytest -v
+
+# Run individual test layers
+pytest tests/git/           # Git abstraction layer
+pytest tests/providers/     # Provider layer
+pytest tests/analysis/      # Evidence & classification unit tests
+pytest tests/output/        # Terminal & JSON rendering tests
+pytest tests/cli/           # Typer CLI tests
+pytest tests/scenarios/     # End-to-end repository scenarios
+```
+
+---
+
+## Adding a Scenario Test
+
+To add a new end-to-end scenario test:
+1. Create `tests/scenarios/test_scenario_<name>.py`.
+2. Import `FakeProvider`, `make_issue`, `make_pr`, and assertion helpers from `tests.scenarios.conftest`.
+3. Use `RepoBuilder` to construct a real temporary Git repository.
+4. Execute `AnalysisEngine(repo, provider).analyze()` and assert on the `AnalysisResult` fields.
+
+```python
+from gitrelevance.analysis.engine import AnalysisEngine
+from gitrelevance.git.repository import GitRepository
+from gitrelevance.models import Classification
+from tests.fixtures.repo_builder import RepoBuilder
+from tests.scenarios.conftest import FakeProvider, make_issue
+
+def test_my_custom_scenario():
+    builder = RepoBuilder().commit("Initial", files={"app.py": "code"})
+    path = builder.build()
+    try:
+        repo = GitRepository(path)
+        issue = make_issue(number=99, title="Custom issue", state="closed")
+        results = AnalysisEngine(repo, FakeProvider(issues=[issue])).analyze()
+        assert results[0].classification == Classification.UNKNOWN
+    finally:
+        builder.cleanup()
+```
+
+---
+
+## License
+
+This project is licensed under the [MIT License](LICENSE).