review-bot/github/client.go

// Package github provides a client for the GitHub API.
// It supports pull request operations, file content retrieval,
// and review submission for both github.com and GitHub Enterprise.
package github

import (
	"bytes"
	"context"
	"encoding/base64"
	"encoding/json"
	"errors"
	"fmt"
	"io"
	"log/slog"
	"net/http"
	"net/url"
	"os"
	"strconv"
	"strings"
	"time"
)

const (
	defaultBaseURL = "https://api.github.com"

	// maxRetryAttempts is the number of times doRequest will attempt a request.
	maxRetryAttempts = 3

	// maxRetryAfter caps the maximum delay from a Retry-After header to prevent
	// a server from stalling the client indefinitely.
	maxRetryAfter = 60 * time.Second

	// maxErrorBodyBytes limits how much of an error response body we read
	// to protect against malicious servers sending unbounded data.
	maxErrorBodyBytes = 64 * 1024 // 64 KB

	// maxResponseBodyBytes limits how much of a successful response body we read
	// for defense-in-depth against servers returning excessively large payloads.
	maxResponseBodyBytes = 10 * 1024 * 1024 // 10 MB
)

// APIError represents an HTTP error response from the GitHub API.
// It carries the status code so callers can distinguish between
// different failure modes (e.g. 404 vs 500).
//
// The Body field stores up to 64 KiB of the raw response for programmatic
// inspection. Error() truncates to 200 bytes for safe logging, but callers
// should avoid logging or propagating Body directly in production since it may
// contain sensitive details from the upstream server.
type APIError struct {
	StatusCode int
	Body       string
}

func (e *APIError) Error() string {
	body := e.Body
	if len(body) > 200 {
		body = body[:200] + "...(truncated)"
	}
	// Sanitize newlines to prevent log injection from upstream response bodies.
	body = strings.ReplaceAll(body, "\n", " ")
	body = strings.ReplaceAll(body, "\r", " ")
	return fmt.Sprintf("HTTP %d: %s", e.StatusCode, body)
}

// IsNotFound reports whether an error is an API 404 response.
func IsNotFound(err error) bool {
	if apiErr, ok := asAPIError(err); ok {
		return apiErr.StatusCode == http.StatusNotFound
	}
	return false
}

// IsUnauthorized reports whether an error is an API 401 response.
func IsUnauthorized(err error) bool {
	if apiErr, ok := asAPIError(err); ok {
		return apiErr.StatusCode == http.StatusUnauthorized
	}
	return false
}

func asAPIError(err error) (*APIError, bool) {
	if err == nil {
		return nil, false
	}
	var target *APIError
	if errors.As(err, &target) {
		return target, true
	}
	return nil, false
}

// Client interacts with the GitHub API.
// A Client is safe for concurrent use by multiple goroutines.
// SetHTTPClient and SetRetryBackoff are intended for test setup only and must
// be called before any goroutines issue requests; they have no synchronization.
type Client struct {
	baseURL    string
	token      string
	httpClient *http.Client

	// allowInsecureHTTP permits requests to HTTP (non-TLS) endpoints.
	// When false, doRequest rejects URLs with an http:// scheme.
	allowInsecureHTTP bool

	// retryBackoff defines the delays between retry attempts for 429 responses.
	// retryBackoff[i] is the delay before attempt i+1 (after attempt i fails).
	// If nil, defaults to {1s, 2s}.
	retryBackoff []time.Duration

	// now returns the current time. Defaults to time.Now.
	// Override in tests to control HTTP-date Retry-After calculations.
	now func() time.Time
}

// defaultCheckRedirect is the redirect policy used by NewClient.
// NOTE: This function is intentionally duplicated in gitea/client.go (and vice versa)
// because the packages are separate. Changes here must be mirrored there.
// It rejects HTTPS->HTTP protocol downgrades (to prevent plaintext leakage)
// and cross-host redirects (to prevent following responses from untrusted
// endpoints). Same-host, same-or-upgraded-scheme redirects are allowed.
func defaultCheckRedirect(req *http.Request, via []*http.Request) error {
	if len(via) >= 10 {
		return fmt.Errorf("stopped after 10 redirects")
	}
	// Guard for direct invocation in tests and any future callers;
	// net/http guarantees len(via) >= 1 during actual redirects.
	if len(via) == 0 {
		return nil
	}
	prev := via[len(via)-1]
	// Reject protocol downgrade: HTTPS->HTTP leaks request metadata over plaintext.
	if prev.URL.Scheme == "https" && req.URL.Scheme == "http" {
		return fmt.Errorf("refusing redirect: HTTPS to HTTP downgrade (%s -> %s)", prev.URL.Host, req.URL.Host)
	}
	// Reject cross-host redirect entirely to avoid consuming responses
	// from untrusted endpoints.
	if req.URL.Host != prev.URL.Host {
		return fmt.Errorf("refusing redirect: cross-host (%s -> %s)", prev.URL.Host, req.URL.Host)
	}
	return nil
}

// ClientOption configures optional behavior of a Client.
type ClientOption func(*clientConfig)

type clientConfig struct {
	allowInsecureHTTP    bool
	insecureIsTestBypass bool
}

// AllowInsecureHTTP permits sending credentials over plaintext HTTP connections.
// In production, this option is gated by the REVIEW_BOT_ALLOW_INSECURE=1
// environment variable. Without the env var set, the option is ignored
// and a warning is logged.
//
// For tests, use AllowInsecureHTTPForTest (defined in a _test.go file in the same package) which bypasses the env gate.
func AllowInsecureHTTP() ClientOption {
	return func(cfg *clientConfig) {
		cfg.allowInsecureHTTP = true
	}
}

// NewClient creates a new GitHub API client.
// If baseURL is empty, it defaults to https://api.github.com.
// For GitHub Enterprise, pass the API base URL (e.g. https://github.concur.com/api/v3).
func NewClient(token, baseURL string, opts ...ClientOption) *Client {
	if baseURL == "" {
		baseURL = defaultBaseURL
	}

	var cfg clientConfig
	for _, opt := range opts {
		opt(&cfg)
	}

	if cfg.allowInsecureHTTP && !cfg.insecureIsTestBypass {
		if os.Getenv("REVIEW_BOT_ALLOW_INSECURE") != "1" {
			slog.Warn("AllowInsecureHTTP ignored: set REVIEW_BOT_ALLOW_INSECURE=1 to enable")
			cfg.allowInsecureHTTP = false
		} else {
			slog.Warn("AllowInsecureHTTP enabled — credentials may be sent over plaintext",
				"env", "REVIEW_BOT_ALLOW_INSECURE=1")
		}
	}

	return &Client{
		baseURL:           strings.TrimRight(baseURL, "/"),
		token:             token,
		allowInsecureHTTP: cfg.allowInsecureHTTP,
		httpClient: &http.Client{
			Timeout:       30 * time.Second,
			CheckRedirect: defaultCheckRedirect,
		},
		now: time.Now,
	}
}

// SetHTTPClient sets the underlying HTTP client used for requests.
// This is intended for test setup only to inject mock transports; it must be
// called before any goroutines issue requests.
//
// Passing nil restores the default client (30s timeout + redirect-rejecting
// CheckRedirect policy matching NewClient).
//
// Callers providing a non-nil client are responsible for configuring a safe
// CheckRedirect policy. Without one, the default net/http behavior will follow
// redirects and may forward the Authorization header to untrusted hosts.
func (c *Client) SetHTTPClient(hc *http.Client) {
	if hc == nil {
		hc = &http.Client{
			Timeout:       30 * time.Second,
			CheckRedirect: defaultCheckRedirect,
		}
	}
	c.httpClient = hc
}

// SetRetryBackoff sets the delays between retry attempts.
// This is intended for testing to speed up retry tests.
//
// Note: if an empty non-nil slice is provided, Retry-After delays parsed from
// server responses will be computed and capped but not applied (because
// attempt < len(backoff) is always false). This is acceptable for the
// test-only use case but callers should be aware of this edge case.
func (c *Client) SetRetryBackoff(backoff []time.Duration) {
	c.retryBackoff = backoff
}

// parseRetryAfter parses a Retry-After header value, supporting both integer
// seconds (e.g. "120") and HTTP-date format (e.g. "Thu, 01 Dec 2025 16:00:00 GMT")
// as specified in RFC 7231 §7.1.3.
//
// For integer values, it returns the duration directly.
// For HTTP-date values, it computes the delay as the difference between the
// parsed time and now. If the date is in the past, it returns 0.
//
// Returns (0, false) if the value cannot be parsed as either format.
func (c *Client) parseRetryAfter(value string) (time.Duration, bool) {
	value = strings.TrimSpace(value)

	// Try integer seconds first (most common from GitHub).
	// RFC 7231 allows delta-seconds of 0 to indicate immediate retry.
	if seconds, err := strconv.Atoi(value); err == nil && seconds >= 0 {
		return time.Duration(seconds) * time.Second, true
	}

	// Try HTTP-date format (RFC 7231 §7.1.3).
	// http.ParseTime handles RFC 1123, RFC 850, and ASCTIME formats.
	if retryAt, err := http.ParseTime(value); err == nil {
		delay := retryAt.Sub(c.now())
		if delay < 0 {
			delay = 0
		}
		return delay, true
	}

	return 0, false
}

// redactURL redacts sensitive components from a URL for safe inclusion in error
// messages and log output. It removes userinfo (e.g., user:pass@) and replaces
// query parameters with a placeholder.
func redactURL(rawURL string) string {
	u, err := url.Parse(rawURL)
	if err != nil {
		return "<unparseable URL>"
	}
	u.User = nil

	if u.RawQuery != "" {
		u.RawQuery = "<redacted>"
	}
	return u.String()
}

// doRequest performs an HTTP request with retry on 429 rate limit responses.
// It respects the Retry-After header when present, supporting both integer
// seconds and HTTP-date formats (capped at maxRetryAfter).
func (c *Client) doRequest(ctx context.Context, method, reqURL string, accept string) ([]byte, error) {
	// NOTE: This parses reqURL a second time (http.NewRequestWithContext parses it
	// again internally). Acceptable cost: URL parsing is cheap and threading the
	// parsed *url.URL through would complicate the interface for negligible gain.
	if !c.allowInsecureHTTP {
		parsed, err := url.Parse(reqURL)
		if err != nil {
			return nil, fmt.Errorf("parse request URL: %w", err)
		}
		if strings.EqualFold(parsed.Scheme, "http") {
			return nil, fmt.Errorf("refusing HTTP request to %s: use HTTPS or set AllowInsecureHTTP option", redactURL(reqURL))
		}
	}

	var backoff []time.Duration
	if c.retryBackoff != nil {
		backoff = append([]time.Duration(nil), c.retryBackoff...)
	} else {
		backoff = []time.Duration{1 * time.Second, 2 * time.Second}
	}

	var lastErr error
	for attempt := 0; attempt < maxRetryAttempts; attempt++ {
		if attempt > 0 {
			var delay time.Duration
			if attempt-1 < len(backoff) {
				delay = backoff[attempt-1]
			}
			if delay > 0 {
				timer := time.NewTimer(delay)
				select {
				case <-timer.C:
					timer.Stop() // no-op after fire; kept for symmetry with the ctx.Done case
				case <-ctx.Done():
					timer.Stop()
					return nil, ctx.Err()
				}
			}
		}

		req, err := http.NewRequestWithContext(ctx, method, reqURL, nil)
		if err != nil {
			return nil, fmt.Errorf("create request: %w", err)
		}
		req.Header.Set("Authorization", "Bearer "+c.token)
		if accept != "" {
			req.Header.Set("Accept", accept)
		} else {
			req.Header.Set("Accept", "application/vnd.github+json")
		}

		resp, err := c.httpClient.Do(req)
		if err != nil {
			return nil, fmt.Errorf("do request: %w", err)
		}

		if resp.StatusCode >= 200 && resp.StatusCode < 300 {
			body, err := io.ReadAll(io.LimitReader(resp.Body, maxResponseBodyBytes))
			resp.Body.Close()
			if err != nil {
				return nil, fmt.Errorf("read response body: %w", err)
			}
			return body, nil
		}

		errBody, _ := io.ReadAll(io.LimitReader(resp.Body, maxErrorBodyBytes))
		resp.Body.Close()

		lastErr = &APIError{StatusCode: resp.StatusCode, Body: string(errBody)}

		// Retry on 429 rate limit
		if resp.StatusCode == http.StatusTooManyRequests && attempt < maxRetryAttempts-1 {
			// Check for Retry-After header and override backoff if present.
			// Supports both integer seconds (common) and HTTP-date format (RFC 7231).
			if ra := resp.Header.Get("Retry-After"); ra != "" {
				if delay, ok := c.parseRetryAfter(ra); ok {
					if delay > maxRetryAfter {
						delay = maxRetryAfter
					}
					if attempt < len(backoff) {
						backoff[attempt] = delay
					}
				}
			}
			continue
		}

		// Don't retry other errors
		return nil, lastErr
	}

	return nil, lastErr
}

// doGet is a convenience wrapper for GET requests with the default Accept header.
func (c *Client) doGet(ctx context.Context, url string) ([]byte, error) {
	return c.doRequest(ctx, http.MethodGet, url, "")
}

// doRequestWithBody performs an HTTP request with an optional body, applying the
// same HTTPS enforcement as doRequest. It is used by write methods (POST, PUT,
// DELETE) that bypass the retry loop in doRequest because write operations are
// not idempotent.
//
// body may be nil for requests that carry no payload (e.g. DELETE).
// When body is non-nil, Content-Type is set to application/json.
func (c *Client) doRequestWithBody(ctx context.Context, method, reqURL string, body []byte) ([]byte, error) {
	if !c.allowInsecureHTTP {
		parsed, err := url.Parse(reqURL)
		if err != nil {
			return nil, fmt.Errorf("parse request URL: %w", err)
		}
		if strings.EqualFold(parsed.Scheme, "http") {
			return nil, fmt.Errorf("refusing HTTP request to %s: use HTTPS or set AllowInsecureHTTP option", redactURL(reqURL))
		}
	}

	var reqBody io.Reader
	if body != nil {
		reqBody = bytes.NewReader(body)
	}

	req, err := http.NewRequestWithContext(ctx, method, reqURL, reqBody)
	if err != nil {
		return nil, fmt.Errorf("create request: %w", err)
	}
	req.Header.Set("Authorization", "Bearer "+c.token)
	req.Header.Set("Accept", "application/vnd.github+json")
	if body != nil {
		req.Header.Set("Content-Type", "application/json")
	}

	resp, err := c.httpClient.Do(req)
	if err != nil {
		return nil, fmt.Errorf("do request: %w", err)
	}
	defer resp.Body.Close()

	if resp.StatusCode >= 200 && resp.StatusCode < 300 {
		respBody, err := io.ReadAll(io.LimitReader(resp.Body, maxResponseBodyBytes))
		if err != nil {
			return nil, fmt.Errorf("read response body: %w", err)
		}
		return respBody, nil
	}

	errBody, _ := io.ReadAll(io.LimitReader(resp.Body, maxErrorBodyBytes))
	return nil, &APIError{StatusCode: resp.StatusCode, Body: string(errBody)}
}

// --- API types ---

// PullRequest holds relevant PR metadata.
type PullRequest struct {
	Title string `json:"title"`
	Body  string `json:"body"`
	Head  struct {
		Sha string `json:"sha"`
		Ref string `json:"ref"`
	} `json:"head"`
	Draft bool `json:"draft"`
}

// CommitStatus represents a single CI status entry.
// GitHub returns "state" not "status"; this type uses Status for consistency
// with the gitea package (both are normalized before use).
type CommitStatus struct {
	Status      string `json:"state"` // GitHub field is "state"
	Context     string `json:"context"`
	Description string `json:"description"`
	TargetURL   string `json:"target_url"`
}

// ChangedFile represents a file modified in a PR.
type ChangedFile struct {
	Filename string `json:"filename"`
	Status   string `json:"status"`
}

// ReviewComment represents an inline comment to attach to a review.
// GitHub uses "position" (diff hunk position), whereas Gitea uses "new_position" (line number).
// When posting inline comments on GitHub, position is required; line numbers
// from the diff cannot be used directly.
type ReviewComment struct {
	ID       int64  `json:"id,omitempty"`
	Path     string `json:"path"`
	Position int64  `json:"position,omitempty"` // GitHub diff hunk position
	Line     int64  `json:"line,omitempty"`     // GitHub absolute line number (alternative to position)
	Side     string `json:"side,omitempty"`     // "RIGHT" or "LEFT"
	Body     string `json:"body"`
}

// Review represents a pull request review from the GitHub API.
type Review struct {
	ID   int64  `json:"id"`
	Body string `json:"body"`
	User struct {
		Login string `json:"login"`
	} `json:"user"`
	State string `json:"state"`
}

// contentResponse is the GitHub contents API response for a single file.
type contentResponse struct {
	Name     string `json:"name"`
	Path     string `json:"path"`
	Type     string `json:"type"`     // "file" or "dir" or "symlink" or "submodule"
	Content  string `json:"content"`  // Base64-encoded file content (with embedded newlines)
	Encoding string `json:"encoding"` // "base64" or ""
}

// ContentEntry represents a file or directory entry from the contents API.
type ContentEntry struct {
	Name string `json:"name"`
	Path string `json:"path"`
	Type string `json:"type"` // "file" or "dir"
}

// --- PR methods ---

// GetPullRequest fetches PR metadata.
func (c *Client) GetPullRequest(ctx context.Context, owner, repo string, number int) (*PullRequest, error) {
	reqURL := fmt.Sprintf("%s/repos/%s/%s/pulls/%d",
		c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number)
	body, err := c.doGet(ctx, reqURL)
	if err != nil {
		return nil, fmt.Errorf("fetch PR: %w", err)
	}
	var pr PullRequest
	if err := json.Unmarshal(body, &pr); err != nil {
		return nil, fmt.Errorf("parse PR JSON: %w", err)
	}
	return &pr, nil
}

// GetPullRequestDiff fetches the unified diff for a PR.
func (c *Client) GetPullRequestDiff(ctx context.Context, owner, repo string, number int) (string, error) {
	reqURL := fmt.Sprintf("%s/repos/%s/%s/pulls/%d",
		c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number)
	body, err := c.doRequest(ctx, http.MethodGet, reqURL, "application/vnd.github.diff")
	if err != nil {
		return "", fmt.Errorf("fetch diff: %w", err)
	}
	return string(body), nil
}

// GetPullRequestFiles fetches the list of files changed in a PR.
// GitHub paginates this endpoint (100 per page max).
func (c *Client) GetPullRequestFiles(ctx context.Context, owner, repo string, number int) ([]ChangedFile, error) {
	const perPage = 100
	var all []ChangedFile
	for page := 1; ; page++ {
		reqURL := fmt.Sprintf("%s/repos/%s/%s/pulls/%d/files?per_page=%d&page=%d",
			c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number, perPage, page)
		body, err := c.doGet(ctx, reqURL)
		if err != nil {
			return nil, fmt.Errorf("fetch PR files (page %d): %w", page, err)
		}
		var batch []ChangedFile
		if err := json.Unmarshal(body, &batch); err != nil {
			return nil, fmt.Errorf("parse PR files JSON (page %d): %w", page, err)
		}
		all = append(all, batch...)
		if len(batch) < perPage {
			break
		}
	}
	return all, nil
}

// GetCommitStatuses fetches CI statuses for a commit SHA.
// GitHub has two status systems: legacy "commit statuses" and newer "check runs".
// This method returns commit statuses only; check runs are a separate API.
// Note: GitHub returns "state" in the JSON; CommitStatus.Status is tagged accordingly.
func (c *Client) GetCommitStatuses(ctx context.Context, owner, repo, sha string) ([]CommitStatus, error) {
	const perPage = 100
	var all []CommitStatus
	for page := 1; ; page++ {
		reqURL := fmt.Sprintf("%s/repos/%s/%s/commits/%s/statuses?per_page=%d&page=%d",
			c.baseURL, url.PathEscape(owner), url.PathEscape(repo), url.PathEscape(sha), perPage, page)
		body, err := c.doGet(ctx, reqURL)
		if err != nil {
			return nil, fmt.Errorf("fetch commit statuses (page %d): %w", page, err)
		}
		var batch []CommitStatus
		if err := json.Unmarshal(body, &batch); err != nil {
			return nil, fmt.Errorf("parse statuses JSON (page %d): %w", page, err)
		}
		all = append(all, batch...)
		if len(batch) < perPage {
			break
		}
	}
	return all, nil
}

// --- File content methods ---

// GetFileContent fetches a file from the default branch of a repo.
// GitHub returns base64-encoded content; this method decodes it.
func (c *Client) GetFileContent(ctx context.Context, owner, repo, filepath string) (string, error) {
	return c.getFileContentAtRef(ctx, owner, repo, filepath, "")
}

// GetFileContentRef fetches a file from a specific ref (branch/tag/sha).
func (c *Client) GetFileContentRef(ctx context.Context, owner, repo, filepath, ref string) (string, error) {
	return c.getFileContentAtRef(ctx, owner, repo, filepath, ref)
}

// getFileContentAtRef fetches a file at the given ref (empty = default branch).
// GitHub's contents API returns base64-encoded file content.
func (c *Client) getFileContentAtRef(ctx context.Context, owner, repo, filepath, ref string) (string, error) {
	reqURL := fmt.Sprintf("%s/repos/%s/%s/contents/%s",
		c.baseURL, url.PathEscape(owner), url.PathEscape(repo), escapePath(filepath))
	if ref != "" {
		reqURL += "?ref=" + url.QueryEscape(ref)
	}
	body, err := c.doGet(ctx, reqURL)
	if err != nil {
		return "", fmt.Errorf("fetch file %s: %w", filepath, err)
	}
	var resp contentResponse
	if err := json.Unmarshal(body, &resp); err != nil {
		return "", fmt.Errorf("parse file content JSON for %s: %w", filepath, err)
	}
	if resp.Type != "file" {
		return "", fmt.Errorf("path %s is a %s, not a file", filepath, resp.Type)
	}
	if resp.Encoding == "base64" {
		// GitHub embeds newlines in the base64 content for readability.
		// Strip them before decoding.
		cleaned := strings.ReplaceAll(resp.Content, "\n", "")
		decoded, err := base64.StdEncoding.DecodeString(cleaned)
		if err != nil {
			return "", fmt.Errorf("decode base64 content for %s: %w", filepath, err)
		}
		return string(decoded), nil
	}
	// Non-base64 encoding (shouldn't happen normally, but handle gracefully).
	return resp.Content, nil
}

// ListContents lists files and directories at a given path.
// Pass an empty path to list the repository root.
// GitHub returns a single object (not array) when path is a file — this
// method normalizes both cases to a slice, matching Gitea's behavior.
func (c *Client) ListContents(ctx context.Context, owner, repo, path string) ([]ContentEntry, error) {
	var reqURL string
	if path == "" || path == "." {
		reqURL = fmt.Sprintf("%s/repos/%s/%s/contents",
			c.baseURL, url.PathEscape(owner), url.PathEscape(repo))
	} else {
		reqURL = fmt.Sprintf("%s/repos/%s/%s/contents/%s",
			c.baseURL, url.PathEscape(owner), url.PathEscape(repo), escapePath(path))
	}
	body, err := c.doGet(ctx, reqURL)
	if err != nil {
		return nil, fmt.Errorf("list contents %s: %w", path, err)
	}

	var entries []ContentEntry
	if err := json.Unmarshal(body, &entries); err != nil {
		// GitHub returns a single object when path is a file (not an array).
		var single contentResponse
		if err2 := json.Unmarshal(body, &single); err2 != nil {
			return nil, fmt.Errorf("parse contents JSON: %w", err)
		}
		if single.Name == "" && single.Path == "" {
			return nil, fmt.Errorf("parse contents JSON: empty response for path %q", path)
		}
		entries = []ContentEntry{{
			Name: single.Name,
			Path: single.Path,
			Type: single.Type,
		}}
	}
	return entries, nil
}

// GetAllFilesInPath recursively fetches all file contents under a path.
// If the path is a file, returns just that file's content.
// If the path is a directory, recursively fetches all files within it.
func (c *Client) GetAllFilesInPath(ctx context.Context, owner, repo, path string) (map[string]string, error) {
	results := make(map[string]string)

	entries, err := c.ListContents(ctx, owner, repo, path)
	if err != nil {
		if !IsNotFound(err) {
			return nil, fmt.Errorf("list contents %q: %w", path, err)
		}
		// 404 means path may be a file — try fetching directly.
		content, fileErr := c.GetFileContent(ctx, owner, repo, path)
		if fileErr != nil {
			return nil, fmt.Errorf("path %q is neither a file nor directory: %w", path, fileErr)
		}
		results[path] = content
		return results, nil
	}

	for _, entry := range entries {
		switch entry.Type {
		case "file":
			content, err := c.GetFileContent(ctx, owner, repo, entry.Path)
			if err != nil {
				slog.Warn("could not fetch file from patterns repo", "file", entry.Path, "error", err)
				continue
			}
			results[entry.Path] = content
		case "dir":
			subResults, err := c.GetAllFilesInPath(ctx, owner, repo, entry.Path)
			if err != nil {
				slog.Warn("could not recurse into directory", "dir", entry.Path, "error", err)
				continue
			}
			for k, v := range subResults {
				results[k] = v
			}
		}
	}
	return results, nil
}

// --- Review methods ---

// PostReview submits a review to a PR.
// event should be one of "APPROVE", "REQUEST_CHANGES", or "COMMENT".
// commitID anchors the review to a specific commit SHA. If empty, defaults to current HEAD.
// comments are optional inline comments; GitHub uses diff hunk position (not line numbers).
// Note: unlike Gitea, GitHub does not support deleting submitted reviews.
// Use COMMENT event to supersede old reviews.
func (c *Client) PostReview(ctx context.Context, owner, repo string, number int, event, body, commitID string, comments []ReviewComment) (*Review, error) {
	reqURL := fmt.Sprintf("%s/repos/%s/%s/pulls/%d/reviews",
		c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number)

	payload := struct {
		Body     string          `json:"body"`
		Event    string          `json:"event"`
		CommitID string          `json:"commit_id,omitempty"`
		Comments []ReviewComment `json:"comments,omitempty"`
	}{
		Body:     body,
		Event:    event,
		CommitID: commitID,
		Comments: comments,
	}

	data, err := json.Marshal(payload)
	if err != nil {
		return nil, fmt.Errorf("marshal review payload: %w", err)
	}

	respBody, err := c.doRequestWithBody(ctx, http.MethodPost, reqURL, data)
	if err != nil {
		return nil, fmt.Errorf("post review: %w", err)
	}

	var review Review
	if err := json.Unmarshal(respBody, &review); err != nil {
		return nil, fmt.Errorf("parse review response: %w", err)
	}
	return &review, nil
}

// ListReviews returns all reviews on a pull request.
// GitHub paginates via Link header; this method uses per_page=100.
func (c *Client) ListReviews(ctx context.Context, owner, repo string, number int) ([]Review, error) {
	const perPage = 100
	var all []Review
	for page := 1; ; page++ {
		reqURL := fmt.Sprintf("%s/repos/%s/%s/pulls/%d/reviews?per_page=%d&page=%d",
			c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number, perPage, page)
		body, err := c.doGet(ctx, reqURL)
		if err != nil {
			return nil, fmt.Errorf("list reviews (page %d): %w", page, err)
		}
		var batch []Review
		if err := json.Unmarshal(body, &batch); err != nil {
			return nil, fmt.Errorf("parse reviews (page %d): %w", page, err)
		}
		all = append(all, batch...)
		if len(batch) < perPage {
			break
		}
	}
	return all, nil
}

// DeleteReview attempts to delete a pull request review.
// GitHub only allows deleting PENDING (draft) reviews. Submitted reviews cannot
// be deleted via the API; this method returns a descriptive error in that case.
// review-bot callers should handle this error gracefully (e.g., by not attempting
// supersede and instead posting a new review alongside the old one).
func (c *Client) DeleteReview(ctx context.Context, owner, repo string, number int, reviewID int64) error {
	reqURL := fmt.Sprintf("%s/repos/%s/%s/pulls/%d/reviews/%d",
		c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number, reviewID)

	// nil body: the GitHub DELETE endpoint for reviews requires no request body.
	_, err := c.doRequestWithBody(ctx, http.MethodDelete, reqURL, nil)
	if err != nil {
		return fmt.Errorf("delete review: %w", err)
	}
	return nil
}

// GetAuthenticatedUser returns the login of the authenticated user.
func (c *Client) GetAuthenticatedUser(ctx context.Context) (string, error) {
	reqURL := c.baseURL + "/user"
	body, err := c.doGet(ctx, reqURL)
	if err != nil {
		return "", fmt.Errorf("get authenticated user: %w", err)
	}
	var result struct {
		Login string `json:"login"`
	}
	if err := json.Unmarshal(body, &result); err != nil {
		return "", fmt.Errorf("parse user response: %w", err)
	}
	return result.Login, nil
}

// RequestReviewer adds a user as a requested reviewer on a pull request.
// This is idempotent — requesting an already-requested reviewer is a no-op.
func (c *Client) RequestReviewer(ctx context.Context, owner, repo string, number int, reviewer string) error {
	reqURL := fmt.Sprintf("%s/repos/%s/%s/pulls/%d/requested_reviewers",
		c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number)

	payload := struct {
		Reviewers []string `json:"reviewers"`
	}{Reviewers: []string{reviewer}}
	data, err := json.Marshal(payload)
	if err != nil {
		return fmt.Errorf("marshal reviewer request: %w", err)
	}

	_, err = c.doRequestWithBody(ctx, http.MethodPost, reqURL, data)
	if err != nil {
		return fmt.Errorf("request reviewer: %w", err)
	}
	return nil
}

// --- helpers ---

// escapePath escapes each segment of a relative file path for use in URLs.
// Slashes are preserved as path separators; other special characters are escaped.
func escapePath(p string) string {
	parts := strings.Split(p, "/")
	for i, part := range parts {
		parts[i] = url.PathEscape(part)
	}
	return strings.Join(parts, "/")
}