Get repo info from REST API if event file is unavailable (#576)

azeemshaikh38 · azeemsgoogle · web-flow · commit 2e062bc00045 · 2022-06-27T18:37:29.000Z
Co-authored-by: Azeem Shaikh &lt;azeems@google.com&gt;
diff --git a/github/github.go b/github/github.go
@@ -15,20 +15,18 @@
 package github
 
 import (
+	"bytes"
 	"context"
 	"encoding/json"
 	"fmt"
 	"io"
+	"io/ioutil"
+	"log"
 	"net/http"
 	"net/url"
-	"os"
 
 	"github.com/ossf/scorecard/v4/clients/githubrepo/roundtripper"
-	"github.com/ossf/scorecard/v4/log"
-)
-
-const (
-	baseRepoURL = "https://api.github.com/repos/"
+	sclog "github.com/ossf/scorecard/v4/log"
 )
 
 // RepoInfo is a struct for repository information.
@@ -43,9 +41,9 @@ type repo struct {
 
 		GITHUB_REPOSITORY_IS_FORK is true if the repository is a fork.
 	*/
-	DefaultBranch string `json:"default_branch"`
-	Fork          bool   `json:"fork"`
-	Private       bool   `json:"private"`
+	DefaultBranch *string `json:"default_branch"`
+	Fork          *bool   `json:"fork"`
+	Private       *bool   `json:"private"`
 }
 
 // Client holds a context and roundtripper for querying repo info from GitHub.
@@ -54,20 +52,6 @@ type Client struct {
 	rt  http.RoundTripper
 }
 
-// NewClient returns a new Client for querying repo info from GitHub.
-func NewClient(ctx context.Context) *Client {
-	c := &Client{}
-
-	defaultCtx := context.Background()
-	if ctx == nil {
-		ctx = defaultCtx
-	}
-
-	c.SetContext(ctx)
-	c.SetDefaultTransport()
-	return c
-}
-
 // SetContext sets a context for a GitHub client.
 func (c *Client) SetContext(ctx context.Context) {
 	c.ctx = ctx
@@ -80,82 +64,96 @@ func (c *Client) SetTransport(rt http.RoundTripper) {
 
 // SetDefaultTransport sets the scorecard roundtripper for a GitHub client.
 func (c *Client) SetDefaultTransport() {
-	logger := log.NewLogger(log.DefaultLevel)
+	logger := sclog.NewLogger(sclog.DefaultLevel)
 	rt := roundtripper.NewTransport(c.ctx, logger)
 	c.rt = rt
 }
 
-// WriteRepoInfo queries GitHub for repo info and writes it to a file.
-func WriteRepoInfo(ctx context.Context, repoName, path string) error {
-	c := NewClient(ctx)
-	repoInfo, err := c.RepoInfo(repoName)
-	if err != nil {
-		return fmt.Errorf("getting repo info: %w", err)
-	}
-
-	repoFile, err := os.Create(path)
-	if err != nil {
-		return fmt.Errorf("creating repo info file: %w", err)
-	}
-	defer repoFile.Close()
-
-	resp := repoInfo.respBytes
-	_, writeErr := repoFile.Write(resp)
-	if writeErr != nil {
-		return fmt.Errorf("writing repo info: %w", writeErr)
-	}
-
-	return nil
-}
-
-// RepoInfo is a function to get the repository information.
+// ParseFromURL is a function to get the repository information.
 // It is decided to not use the golang GitHub library because of the
 // dependency on the github.com/google/go-github/github library
 // which will in turn require other dependencies.
-func (c *Client) RepoInfo(repoName string) (RepoInfo, error) {
-	var r RepoInfo
-
+func (c *Client) ParseFromURL(baseRepoURL, repoName string) (RepoInfo, error) {
+	var ret RepoInfo
 	baseURL, err := url.Parse(baseRepoURL)
 	if err != nil {
-		return r, fmt.Errorf("parsing base repo URL: %w", err)
+		return ret, fmt.Errorf("parsing base repo URL: %w", err)
 	}
 
-	repoURL, err := baseURL.Parse(repoName)
+	repoURL, err := baseURL.Parse(fmt.Sprintf("repos/%s", repoName))
 	if err != nil {
-		return r, fmt.Errorf("parsing repo endpoint: %w", err)
+		return ret, fmt.Errorf("parsing repo endpoint: %w", err)
 	}
 
-	method := "GET"
+	log.Printf("getting repo info from URL: %s", repoURL.String())
 	req, err := http.NewRequestWithContext(
 		c.ctx,
-		method,
+		http.MethodGet,
 		repoURL.String(),
-		nil,
-	)
+		nil /*body*/)
 	if err != nil {
-		return r, fmt.Errorf("error creating request: %w", err)
+		return ret, fmt.Errorf("error creating request: %w", err)
 	}
 
 	resp, err := http.DefaultClient.Do(req)
 	if err != nil {
-		return r, fmt.Errorf("error creating request: %w", err)
+		return ret, fmt.Errorf("error creating request: %w", err)
 	}
 	defer resp.Body.Close()
 	if err != nil {
-		return r, fmt.Errorf("error reading response body: %w", err)
+		return ret, fmt.Errorf("error reading response body: %w", err)
 	}
 
 	respBytes, err := io.ReadAll(resp.Body)
 	if err != nil {
-		return r, fmt.Errorf("error reading response body: %w", err)
+		return ret, fmt.Errorf("error reading response body: %w", err)
 	}
 
-	r.respBytes = respBytes
+	prettyPrintJSON(respBytes)
+	ret.respBytes = respBytes
+	if err := json.Unmarshal(respBytes, &ret.Repo); err != nil {
+		return ret, fmt.Errorf("error decoding response body: %w", err)
+	}
+	return ret, nil
+}
+
+// ParseFromFile is a function to get the repository information
+// from GitHub event file.
+func (c *Client) ParseFromFile(filepath string) (RepoInfo, error) {
+	var ret RepoInfo
 
-	err = json.Unmarshal(respBytes, &r)
+	log.Printf("getting repo info from file: %s", filepath)
+	repoInfo, err := ioutil.ReadFile(filepath)
 	if err != nil {
-		return r, fmt.Errorf("error decoding response body: %w", err)
+		return ret, fmt.Errorf("reading GitHub event path: %w", err)
 	}
 
-	return r, nil
+	prettyPrintJSON(repoInfo)
+	if err := json.Unmarshal(repoInfo, &ret); err != nil {
+		return ret, fmt.Errorf("unmarshalling repo info: %w", err)
+	}
+
+	return ret, nil
+}
+
+// NewClient returns a new Client for querying repo info from GitHub.
+func NewClient(ctx context.Context) *Client {
+	c := &Client{
+		ctx: ctx,
+	}
+
+	if c.ctx == nil {
+		c.SetContext(context.Background())
+	}
+	c.SetDefaultTransport()
+	return c
+}
+
+func prettyPrintJSON(jsonBytes []byte) {
+	var buf bytes.Buffer
+	if err := json.Indent(&buf, jsonBytes, "", ""); err != nil {
+		log.Printf("%v", err)
+		return
+	}
+	log.Println(buf.String())
 }
diff --git a/options/env.go b/options/env.go
@@ -21,6 +21,7 @@ import (
 
 // Environment variables.
 // TODO(env): Remove once environment variables are not used for config.
+//
 //nolint:revive,nolintlint
 const (
 	EnvEnableSarif             = "ENABLE_SARIF"
diff --git a/options/options.go b/options/options.go
@@ -15,15 +15,14 @@
 package options
 
 import (
-	"encoding/json"
 	"errors"
 	"fmt"
-	"io/ioutil"
 	"os"
 	"strconv"
 	"strings"
 
 	"github.com/caarlos0/env/v6"
+	"golang.org/x/net/context"
 
 	"github.com/ossf/scorecard-action/github"
 	"github.com/ossf/scorecard/v4/checks"
@@ -44,6 +43,7 @@ var (
 	// Errors.
 	errGithubEventPathEmpty       = errors.New("GitHub event path is empty")
 	errResultsPathEmpty           = errors.New("results path is empty")
+	errGitHubRepoInfoUnavailable  = errors.New("GitHub repo info inaccessible")
 	errOnlyDefaultBranchSupported = errors.New("only default branch is supported")
 )
 
@@ -67,6 +67,7 @@ type Options struct {
 	GithubRef        string `env:"GITHUB_REF"`
 	GithubRepository string `env:"GITHUB_REPOSITORY"`
 	GithubWorkspace  string `env:"GITHUB_WORKSPACE"`
+	GithubAPIURL     string `env:"GITHUB_API_URL"`
 
 	DefaultBranch string `env:"SCORECARD_DEFAULT_BRANCH"`
 	// TODO(options): This may be better as a bool
@@ -87,6 +88,13 @@ func New() (*Options, error) {
 	if err := env.Parse(opts); err != nil {
 		return opts, fmt.Errorf("parsing entrypoint env vars: %w", err)
 	}
+	// GITHUB_AUTH_TOKEN
+	// Needs to be set *before* setRepoInfo() is invoked.
+	// setRepoInfo() uses the GITHUB_AUTH_TOKEN env for querying the REST API.
+	if _, tokenSet := os.LookupEnv(EnvGithubAuthToken); !tokenSet {
+		inputToken := os.Getenv(EnvInputRepoToken)
+		os.Setenv(EnvGithubAuthToken, inputToken)
+	}
 	if err := opts.setRepoInfo(); err != nil {
 		return opts, fmt.Errorf("parsing repo info: %w", err)
 	}
@@ -143,12 +151,6 @@ func (o *Options) Print() {
 
 func (o *Options) setScorecardOpts() {
 	o.ScorecardOpts = scopts.New()
-	// GITHUB_AUTH_TOKEN
-	_, tokenSet := os.LookupEnv(EnvGithubAuthToken)
-	if !tokenSet {
-		inputToken := os.Getenv(EnvInputRepoToken)
-		os.Setenv(EnvGithubAuthToken, inputToken)
-	}
 
 	// --repo= | --local
 	// This section restores functionality that was removed in
@@ -194,6 +196,8 @@ func (o *Options) setScorecardOpts() {
 // setPublishResults sets whether results should be published based on a
 // repository's visibility.
 func (o *Options) setPublishResults() {
+	inputVal := o.PublishResults
+	o.PublishResults = false
 	privateRepo, err := strconv.ParseBool(o.PrivateRepoStr)
 	if err != nil {
 		// TODO(options): Consider making this an error.
@@ -202,9 +206,10 @@ func (o *Options) setPublishResults() {
 			o.PrivateRepoStr,
 			err,
 		)
+		return
 	}
 
-	o.PublishResults = o.PublishResults && !privateRepo
+	o.PublishResults = inputVal && !privateRepo
 }
 
 // setRepoInfo gets the path to the GitHub event and sets the
@@ -217,21 +222,36 @@ func (o *Options) setRepoInfo() error {
 		return errGithubEventPathEmpty
 	}
 
-	repoInfo, err := ioutil.ReadFile(eventPath)
-	if err != nil {
-		return fmt.Errorf("reading GitHub event path: %w", err)
+	ghClient := github.NewClient(context.Background())
+	if repoInfo, err := ghClient.ParseFromFile(eventPath); err == nil &&
+		o.parseFromRepoInfo(repoInfo) {
+		return nil
 	}
 
-	var r github.RepoInfo
-	if err := json.Unmarshal(repoInfo, &r); err != nil {
-		return fmt.Errorf("unmarshalling repo info: %w", err)
+	if repoInfo, err := ghClient.ParseFromURL(o.GithubAPIURL, o.GithubRepository); err == nil &&
+		o.parseFromRepoInfo(repoInfo) {
+		return nil
 	}
 
-	o.PrivateRepoStr = strconv.FormatBool(r.Repo.Private)
-	o.IsForkStr = strconv.FormatBool(r.Repo.Fork)
-	o.DefaultBranch = r.Repo.DefaultBranch
+	return errGitHubRepoInfoUnavailable
+}
 
-	return nil
+func (o *Options) parseFromRepoInfo(repoInfo github.RepoInfo) bool {
+	if repoInfo.Repo.DefaultBranch == nil &&
+		repoInfo.Repo.Fork == nil &&
+		repoInfo.Repo.Private == nil {
+		return false
+	}
+	if repoInfo.Repo.Private != nil {
+		o.PrivateRepoStr = strconv.FormatBool(*repoInfo.Repo.Private)
+	}
+	if repoInfo.Repo.Fork != nil {
+		o.IsForkStr = strconv.FormatBool(*repoInfo.Repo.Fork)
+	}
+	if repoInfo.Repo.DefaultBranch != nil {
+		o.DefaultBranch = *repoInfo.Repo.DefaultBranch
+	}
+	return true
 }
 
 func (o *Options) isPullRequestEvent() bool {
