Use github graphql client as primary, fallback to rest client

2026-06-08 22:59:30 +00:00 · 2025-12-11 23:35:53 +00:00
parent 5115551543
commit 03d1ef430a
8 changed files with 929 additions and 100 deletions
@@ -41,6 +41,7 @@ func DefaultRetryConfig() RetryConfig {
 // Client wraps the GitHub API client with rate limiting and caching
 type Client struct {
 	gh       *github.Client
+	gql      *GraphQLClient // GraphQL client for batched queries
 	config   *config.Config
 	cache    cache.Cache
 	retry    RetryConfig
@@ -91,8 +92,15 @@ func NewClient(ctx context.Context, cfg *config.Config) (*Client, error) {
 		c = cache.NewNoopCache()
 	}

+	// Initialize GraphQL client if using token auth (GraphQL doesn't support GitHub App auth easily)
+	var gql *GraphQLClient
+	if cfg.HasGithubToken() && cfg.Options.UseGraphQL {
+		gql = NewGraphQLClient(cfg.Auth.GithubToken)
+	}
+
 	return &Client{
 		gh:       gh,
+		gql:      gql,
 		config:   cfg,
 		cache:    c,
 		retry:    DefaultRetryConfig(),
@@ -107,6 +115,73 @@ func (c *Client) SetProgressCallback(cb ProgressCallback) {
 	}
 }

+// HasGraphQL returns true if the GraphQL client is available
+func (c *Client) HasGraphQL() bool {
+	return c.gql != nil
+}
+
+// FetchPRsWithReviewsGraphQL fetches PRs and reviews using GraphQL (much fewer API calls)
+func (c *Client) FetchPRsWithReviewsGraphQL(ctx context.Context, owner, repo string, since, until *time.Time) ([]models.PullRequest, []models.Review, error) {
+	if c.gql == nil {
+		return nil, nil, fmt.Errorf("GraphQL client not initialized")
+	}
+
+	cacheKey := fmt.Sprintf("gql_prs_reviews:%s/%s:%v:%v", owner, repo, since, until)
+
+	// Check cache
+	type cachedData struct {
+		PRs     []models.PullRequest
+		Reviews []models.Review
+	}
+	if cached, ok := c.cache.Get(cacheKey); ok {
+		if data, ok := cached.(cachedData); ok {
+			c.progress("      Using cached PRs and reviews data (GraphQL)")
+			return data.PRs, data.Reviews, nil
+		}
+	}
+
+	prs, reviews, err := c.gql.FetchPRsWithReviews(ctx, owner, repo, since, until)
+	if err != nil {
+		return nil, nil, err
+	}
+
+	// Cache results
+	c.cache.Set(cacheKey, cachedData{PRs: prs, Reviews: reviews})
+
+	return prs, reviews, nil
+}
+
+// FetchIssuesWithCommentsGraphQL fetches issues and comments using GraphQL (much fewer API calls)
+func (c *Client) FetchIssuesWithCommentsGraphQL(ctx context.Context, owner, repo string, since, until *time.Time) ([]models.Issue, []models.IssueComment, error) {
+	if c.gql == nil {
+		return nil, nil, fmt.Errorf("GraphQL client not initialized")
+	}
+
+	cacheKey := fmt.Sprintf("gql_issues_comments:%s/%s:%v:%v", owner, repo, since, until)
+
+	// Check cache
+	type cachedData struct {
+		Issues   []models.Issue
+		Comments []models.IssueComment
+	}
+	if cached, ok := c.cache.Get(cacheKey); ok {
+		if data, ok := cached.(cachedData); ok {
+			c.progress("      Using cached issues and comments data (GraphQL)")
+			return data.Issues, data.Comments, nil
+		}
+	}
+
+	issues, comments, err := c.gql.FetchIssuesWithComments(ctx, owner, repo, since, until)
+	if err != nil {
+		return nil, nil, err
+	}
+
+	// Cache results
+	c.cache.Set(cacheKey, cachedData{Issues: issues, Comments: comments})
+
+	return issues, comments, nil
+}
+
 // SetRetryConfig sets the retry configuration
 func (c *Client) SetRetryConfig(rc RetryConfig) {
 	c.retry = rc
@@ -459,6 +534,7 @@ func (c *Client) fetchPRsForBranch(ctx context.Context, owner, repo, baseBranch
 		ResourceName:              "pull requests",
 		EarlyTermination:          true,
 		EarlyTerminationThreshold: 2,
+		Quiet:                     true, // Parent function handles progress
 	}

 	return FetchAllPages(ctx, c, "", config, fetcher) // Empty cache key - parent handles caching
@@ -489,6 +565,7 @@ func (c *Client) FetchReviews(ctx context.Context, owner, repo string, prNumber

 	config := DefaultFetchConfig("reviews")
 	config.EarlyTermination = false // Reviews don't need date-based early termination
+	config.Quiet = true             // Suppress per-page progress (called many times in parallel)

 	return FetchAllPages(ctx, c, cacheKey, config, fetcher)
 }
@@ -52,6 +52,8 @@ type FetchConfig struct {
 	EarlyTermination bool
 	// EarlyTerminationThreshold is the number of consecutive old pages before stopping
 	EarlyTerminationThreshold int
+	// Quiet suppresses per-page progress messages (useful for sub-fetches like reviews)
+	Quiet bool
 }

 // DefaultFetchConfig returns sensible defaults
@@ -91,8 +93,15 @@ func FetchAllPages[T any, R any](
 			return nil, fmt.Errorf("failed to fetch %s: %w", config.ResourceName, err)
 		}

-		c.progress(fmt.Sprintf("      Fetching %s page %d (%d %s so far)...",
-			config.ResourceName, page, len(allResults), config.ResourceName))
+		// Safety check for nil response
+		if resp == nil {
+			break
+		}
+
+		if !config.Quiet {
+			c.progress(fmt.Sprintf("      Fetching %s page %d (%d %s so far)...",
+				config.ResourceName, page, len(allResults), config.ResourceName))
+		}

 		oldInPage := 0
 		totalEligible := 0
@@ -121,8 +130,10 @@ func FetchAllPages[T any, R any](
 		if config.EarlyTermination && totalEligible > 0 && oldInPage == totalEligible {
 			consecutiveOldPages++
 			if consecutiveOldPages >= config.EarlyTerminationThreshold {
-				c.progress(fmt.Sprintf("      Reached %s older than date range, stopping early (page %d)",
-					config.ResourceName, page))
+				if !config.Quiet {
+					c.progress(fmt.Sprintf("      Reached %s older than date range, stopping early (page %d)",
+						config.ResourceName, page))
+				}
 				break
 			}
 		} else {
@@ -260,8 +271,15 @@ func FetchAllPagesWithEnrichment[T any, R any](
 			return nil, fmt.Errorf("failed to fetch %s: %w", config.ResourceName, err)
 		}

-		c.progress(fmt.Sprintf("      Fetching %s page %d (%d %s so far)...",
-			config.ResourceName, page, len(allResults), config.ResourceName))
+		// Safety check for nil response
+		if resp == nil {
+			break
+		}
+
+		if !config.Quiet {
+			c.progress(fmt.Sprintf("      Fetching %s page %d (%d %s so far)...",
+				config.ResourceName, page, len(allResults), config.ResourceName))
+		}

 		itemsInPage := 0
 		for i, item := range items {
@@ -0,0 +1,516 @@
+package github
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"os"
+	"time"
+
+	"github.com/charmbracelet/bubbles/progress"
+	"github.com/charmbracelet/lipgloss"
+	"github.com/lukaszraczylo/git-velocity/internal/domain/models"
+	"github.com/shurcooL/githubv4"
+	"golang.org/x/oauth2"
+)
+
+// progressBar handles terminal progress display
+type progressBar struct {
+	progress progress.Model
+	label    string
+	total    int
+	current  int
+	out      io.Writer
+}
+
+func newProgressBar(label string, total int) *progressBar {
+	p := progress.New(
+		progress.WithDefaultGradient(),
+		progress.WithWidth(40),
+	)
+	return &progressBar{
+		progress: p,
+		label:    label,
+		total:    total,
+		current:  0,
+		out:      os.Stderr,
+	}
+}
+
+func (p *progressBar) update(fetched int) {
+	p.current = fetched
+	percent := float64(p.current) / float64(p.total)
+	if percent > 1.0 {
+		percent = 1.0
+	}
+
+	labelStyle := lipgloss.NewStyle().Foreground(lipgloss.Color("205"))
+	countStyle := lipgloss.NewStyle().Foreground(lipgloss.Color("241"))
+
+	fmt.Fprintf(p.out, "\r%s %s %s",
+		labelStyle.Render(p.label),
+		p.progress.ViewAs(percent),
+		countStyle.Render(fmt.Sprintf("%d/%d", p.current, p.total)),
+	)
+}
+
+func (p *progressBar) done() {
+	p.update(p.total)
+	fmt.Fprintln(p.out)
+}
+
+// GraphQLClient wraps the githubv4 client for GitHub API
+type GraphQLClient struct {
+	client *githubv4.Client
+}
+
+// NewGraphQLClient creates a new GraphQL client for GitHub
+func NewGraphQLClient(token string) *GraphQLClient {
+	src := oauth2.StaticTokenSource(
+		&oauth2.Token{AccessToken: token},
+	)
+	httpClient := oauth2.NewClient(context.Background(), src)
+	client := githubv4.NewClient(httpClient)
+
+	return &GraphQLClient{
+		client: client,
+	}
+}
+
+// PageInfo contains pagination info from GraphQL responses
+type PageInfo struct {
+	HasNextPage bool
+	EndCursor   githubv4.String
+}
+
+// PageResult represents a page of results from GraphQL
+type PageResult[T any] struct {
+	TotalCount int
+	PageInfo   PageInfo
+	Nodes      []T
+}
+
+// GQLFetchConfig configures the generic paginated fetcher for GraphQL
+type GQLFetchConfig[Q any, T any, R any] struct {
+	Label         string
+	Query         *Q
+	GetPageResult func(q *Q) PageResult[T]
+	// ProcessNode returns items, whether this node is "old" (outside date range),
+	// and whether to hard stop immediately (past cutoff date)
+	ProcessNode func(node T, repo string) (items []R, isOld bool, hardStop bool)
+	// ConsecutiveOldPagesToStop controls early termination (default: 2)
+	ConsecutiveOldPagesToStop int
+}
+
+// fetchGQLPaginated is a generic paginated fetcher for GraphQL queries
+func fetchGQLPaginated[Q any, T any, R any](
+	ctx context.Context,
+	client *githubv4.Client,
+	owner, repo string,
+	config GQLFetchConfig[Q, T, R],
+) ([]R, error) {
+	var allResults []R
+
+	variables := map[string]interface{}{
+		"owner":  githubv4.String(owner),
+		"repo":   githubv4.String(repo),
+		"cursor": (*githubv4.String)(nil),
+	}
+
+	var pbar *progressBar
+	fetched := 0
+	repoFullName := fmt.Sprintf("%s/%s", owner, repo)
+	consecutiveOldPages := 0
+	pagesToStop := config.ConsecutiveOldPagesToStop
+	if pagesToStop == 0 {
+		pagesToStop = 2 // default
+	}
+
+	for {
+		if err := client.Query(ctx, config.Query, variables); err != nil {
+			return nil, fmt.Errorf("graphql query failed: %w", err)
+		}
+
+		page := config.GetPageResult(config.Query)
+
+		// Initialize progress bar on first query
+		if pbar == nil && page.TotalCount > 0 {
+			pbar = newProgressBar(config.Label, page.TotalCount)
+		}
+
+		oldInPage := 0
+		totalInPage := 0
+		shouldHardStop := false
+		for _, node := range page.Nodes {
+			fetched++
+			totalInPage++
+			items, isOld, hardStop := config.ProcessNode(node, repoFullName)
+			allResults = append(allResults, items...)
+			if isOld {
+				oldInPage++
+			}
+			if hardStop {
+				shouldHardStop = true
+				break
+			}
+		}
+
+		if pbar != nil {
+			pbar.update(fetched)
+		}
+
+		// Hard stop takes priority (past cutoff date)
+		if shouldHardStop {
+			if pbar != nil {
+				pbar.done()
+			}
+			break
+		}
+
+		// Track consecutive pages where all items are old
+		if totalInPage > 0 && oldInPage == totalInPage {
+			consecutiveOldPages++
+		} else {
+			consecutiveOldPages = 0
+		}
+
+		// Stop if we've seen enough consecutive old pages or no more pages
+		if consecutiveOldPages >= pagesToStop || !page.PageInfo.HasNextPage {
+			if pbar != nil {
+				pbar.done()
+			}
+			break
+		}
+		variables["cursor"] = githubv4.NewString(page.PageInfo.EndCursor)
+	}
+
+	return allResults, nil
+}
+
+// Query structs for PRs with reviews
+type gqlPRQuery struct {
+	Repository struct {
+		PullRequests struct {
+			TotalCount int
+			PageInfo   PageInfo
+			Nodes      []gqlPRNode
+		} `graphql:"pullRequests(first: 100, after: $cursor, states: [MERGED], orderBy: {field: UPDATED_AT, direction: DESC})"`
+	} `graphql:"repository(owner: $owner, name: $repo)"`
+}
+
+type gqlPRNode struct {
+	Number       int
+	Title        string
+	State        string
+	Merged       bool
+	Additions    int
+	Deletions    int
+	ChangedFiles int
+	CreatedAt    time.Time
+	UpdatedAt    time.Time
+	MergedAt     *time.Time
+	ClosedAt     *time.Time
+	BaseRefName  string
+	HeadRefName  string
+	URL          string
+	Commits      struct{ TotalCount int }
+	Author       gqlActor
+	Reviews      struct {
+		TotalCount int
+		Nodes      []gqlReviewNode
+		PageInfo   PageInfo
+	} `graphql:"reviews(first: 100)"`
+}
+
+type gqlActor struct {
+	Login     string
+	AvatarURL string `graphql:"avatarUrl"`
+}
+
+type gqlReviewNode struct {
+	ID          string `graphql:"id"`
+	Author      gqlActor
+	State       string
+	SubmittedAt *time.Time
+	Body        string
+	Comments    struct{ TotalCount int } `graphql:"comments"`
+}
+
+// Query struct for issues with comments
+type gqlIssueQuery struct {
+	Repository struct {
+		Issues struct {
+			TotalCount int
+			PageInfo   PageInfo
+			Nodes      []gqlIssueNode
+		} `graphql:"issues(first: 100, after: $cursor, orderBy: {field: CREATED_AT, direction: DESC})"`
+	} `graphql:"repository(owner: $owner, name: $repo)"`
+}
+
+type gqlIssueNode struct {
+	Number    int
+	Title     string
+	State     string
+	CreatedAt time.Time
+	UpdatedAt time.Time
+	ClosedAt  *time.Time
+	URL       string
+	Author    gqlActor
+	Labels    struct {
+		Nodes []struct{ Name string }
+	} `graphql:"labels(first: 10)"`
+	Comments struct {
+		TotalCount int
+		Nodes      []gqlCommentNode
+		PageInfo   PageInfo
+	} `graphql:"comments(first: 100)"`
+}
+
+type gqlCommentNode struct {
+	ID        string `graphql:"id"`
+	Author    gqlActor
+	Body      string
+	CreatedAt time.Time
+}
+
+// prWithReviews bundles a PR with its reviews for the generic fetcher
+type prWithReviews struct {
+	PR      models.PullRequest
+	Reviews []models.Review
+}
+
+// FetchPRsWithReviews fetches pull requests with their reviews using GraphQL
+func (g *GraphQLClient) FetchPRsWithReviews(ctx context.Context, owner, repo string, since, until *time.Time) ([]models.PullRequest, []models.Review, error) {
+	var query gqlPRQuery
+
+	// Hard cutoff: 1 week before start date - stop fetching entirely past this point
+	var hardCutoff *time.Time
+	if since != nil {
+		cutoff := since.AddDate(0, 0, -7)
+		hardCutoff = &cutoff
+	}
+
+	results, err := fetchGQLPaginated(ctx, g.client, owner, repo, GQLFetchConfig[gqlPRQuery, gqlPRNode, prWithReviews]{
+		Label:                     "      Fetching PRs:",
+		Query:                     &query,
+		ConsecutiveOldPagesToStop: 2,
+		GetPageResult: func(q *gqlPRQuery) PageResult[gqlPRNode] {
+			return PageResult[gqlPRNode]{
+				TotalCount: q.Repository.PullRequests.TotalCount,
+				PageInfo:   q.Repository.PullRequests.PageInfo,
+				Nodes:      q.Repository.PullRequests.Nodes,
+			}
+		},
+		ProcessNode: func(node gqlPRNode, repoName string) ([]prWithReviews, bool, bool) {
+			// Skip if not merged - not counted as "old"
+			if node.MergedAt == nil {
+				return nil, false, false
+			}
+
+			mergedAt := *node.MergedAt
+
+			// Hard cutoff check - stop entirely if past this date
+			if hardCutoff != nil && mergedAt.Before(*hardCutoff) {
+				return nil, true, true // Hard stop
+			}
+
+			// Check date range - skip if outside range
+			if until != nil && mergedAt.After(*until) {
+				return nil, false, false // Too new, not "old"
+			}
+			if since != nil && mergedAt.Before(*since) {
+				return nil, true, false // Too old - signal for early termination tracking
+			}
+
+			// Convert PR
+			pr := convertPRNode(node, repoName)
+
+			// Convert reviews
+			var reviews []models.Review
+			for _, r := range node.Reviews.Nodes {
+				reviews = append(reviews, convertReviewNode(r, repoName, node.Number))
+			}
+
+			return []prWithReviews{{PR: pr, Reviews: reviews}}, false, false
+		},
+	})
+	if err != nil {
+		return nil, nil, err
+	}
+
+	// Flatten results
+	var prs []models.PullRequest
+	var reviews []models.Review
+	for _, r := range results {
+		prs = append(prs, r.PR)
+		reviews = append(reviews, r.Reviews...)
+	}
+
+	return prs, reviews, nil
+}
+
+// issueWithComments bundles an issue with its comments for the generic fetcher
+type issueWithComments struct {
+	Issue    models.Issue
+	Comments []models.IssueComment
+}
+
+// FetchIssuesWithComments fetches issues with their comments using GraphQL
+func (g *GraphQLClient) FetchIssuesWithComments(ctx context.Context, owner, repo string, since, until *time.Time) ([]models.Issue, []models.IssueComment, error) {
+	var query gqlIssueQuery
+
+	// Hard cutoff: 1 week before start date - stop fetching entirely past this point
+	var hardCutoff *time.Time
+	if since != nil {
+		cutoff := since.AddDate(0, 0, -7)
+		hardCutoff = &cutoff
+	}
+
+	results, err := fetchGQLPaginated(ctx, g.client, owner, repo, GQLFetchConfig[gqlIssueQuery, gqlIssueNode, issueWithComments]{
+		Label:                     "      Fetching issues:",
+		Query:                     &query,
+		ConsecutiveOldPagesToStop: 2,
+		GetPageResult: func(q *gqlIssueQuery) PageResult[gqlIssueNode] {
+			return PageResult[gqlIssueNode]{
+				TotalCount: q.Repository.Issues.TotalCount,
+				PageInfo:   q.Repository.Issues.PageInfo,
+				Nodes:      q.Repository.Issues.Nodes,
+			}
+		},
+		ProcessNode: func(node gqlIssueNode, repoName string) ([]issueWithComments, bool, bool) {
+			// Hard cutoff check - stop entirely if past this date
+			if hardCutoff != nil && node.CreatedAt.Before(*hardCutoff) {
+				return nil, true, true // Hard stop
+			}
+
+			// Check date range
+			if until != nil && node.CreatedAt.After(*until) {
+				return nil, false, false // Too new, not "old"
+			}
+			if since != nil && node.CreatedAt.Before(*since) {
+				return nil, true, false // Too old - signal for early termination tracking
+			}
+
+			// Convert issue
+			issue := convertIssueNode(node, repoName)
+
+			// Convert comments within date range
+			var comments []models.IssueComment
+			for _, c := range node.Comments.Nodes {
+				if until != nil && c.CreatedAt.After(*until) {
+					continue
+				}
+				if since != nil && c.CreatedAt.Before(*since) {
+					continue
+				}
+				comments = append(comments, convertCommentNode(c, repoName, node.Number))
+			}
+
+			return []issueWithComments{{Issue: issue, Comments: comments}}, false, false
+		},
+	})
+	if err != nil {
+		return nil, nil, err
+	}
+
+	// Flatten results
+	var issues []models.Issue
+	var comments []models.IssueComment
+	for _, r := range results {
+		issues = append(issues, r.Issue)
+		comments = append(comments, r.Comments...)
+	}
+
+	return issues, comments, nil
+}
+
+// Conversion helpers
+
+func convertActor(a gqlActor) models.Author {
+	return models.Author{
+		Login:     a.Login,
+		AvatarURL: a.AvatarURL,
+	}
+}
+
+func convertPRNode(node gqlPRNode, repoName string) models.PullRequest {
+	state := models.PRStateOpen
+	if node.Merged {
+		state = models.PRStateMerged
+	} else if node.State == "CLOSED" {
+		state = models.PRStateClosed
+	}
+
+	return models.PullRequest{
+		Number:       node.Number,
+		Title:        node.Title,
+		State:        state,
+		Author:       convertActor(node.Author),
+		Repository:   repoName,
+		BaseBranch:   node.BaseRefName,
+		HeadBranch:   node.HeadRefName,
+		CreatedAt:    node.CreatedAt,
+		UpdatedAt:    node.UpdatedAt,
+		MergedAt:     node.MergedAt,
+		ClosedAt:     node.ClosedAt,
+		Additions:    node.Additions,
+		Deletions:    node.Deletions,
+		FilesChanged: node.ChangedFiles,
+		CommitCount:  node.Commits.TotalCount,
+		Comments:     node.Reviews.TotalCount,
+		URL:          node.URL,
+	}
+}
+
+func convertReviewNode(node gqlReviewNode, repoName string, prNumber int) models.Review {
+	var submittedAt time.Time
+	if node.SubmittedAt != nil {
+		submittedAt = *node.SubmittedAt
+	}
+
+	return models.Review{
+		PullRequest:   prNumber,
+		Repository:    repoName,
+		Author:        convertActor(node.Author),
+		State:         models.ReviewState(node.State),
+		SubmittedAt:   submittedAt,
+		Body:          node.Body,
+		CommentsCount: node.Comments.TotalCount,
+	}
+}
+
+func convertIssueNode(node gqlIssueNode, repoName string) models.Issue {
+	state := models.IssueStateOpen
+	if node.State == "CLOSED" {
+		state = models.IssueStateClosed
+	}
+
+	var labels []string
+	for _, l := range node.Labels.Nodes {
+		labels = append(labels, l.Name)
+	}
+
+	return models.Issue{
+		Number:     node.Number,
+		Title:      node.Title,
+		State:      state,
+		Author:     convertActor(node.Author),
+		Repository: repoName,
+		CreatedAt:  node.CreatedAt,
+		UpdatedAt:  node.UpdatedAt,
+		ClosedAt:   node.ClosedAt,
+		Comments:   node.Comments.TotalCount,
+		Labels:     labels,
+		URL:        node.URL,
+	}
+}
+
+func convertCommentNode(node gqlCommentNode, repoName string, issueNumber int) models.IssueComment {
+	return models.IssueComment{
+		Issue:      issueNumber,
+		Repository: repoName,
+		Author:     convertActor(node.Author),
+		Body:       node.Body,
+		CreatedAt:  node.CreatedAt,
+	}
+}