cmd/generate: Factor forge_*.go files out of src_contribs.go

1 files changed, 28 insertions, 369 deletions

diff --git a/cmd/generate/src_contribs.go b/cmd/generate/src_contribs.go
index 2ace0fd..b5345e3 100644
--- a/cmd/generate/src_contribs.go
+++ b/cmd/generate/src_contribs.go
@@ -2,9 +2,7 @@ package main
 
 import (
 	"fmt"
-	"net/url"
 	"os"
-	"regexp"
 	"strings"
 	"time"
 
@@ -96,163 +94,48 @@ func classifyStatus(status string) (string, error) {
 	}
 }
 
-var (
-	reGoLangGerritCL = regexp.MustCompile(`https://go-review\.googlesource\.com/c/([^/?#]+)/\+/([0-9]+)(?:\?[^#]*)?(?:#.*)?$`)
-	reGitHubPR       = regexp.MustCompile(`^https://github\.com/([^/?#]+)/([^/?#]+)/pull/([0-9]+)(?:\?[^#]*)?(?:#.*)?$`)
-	reGitHubCommit   = regexp.MustCompile(`^https://github\.com/([^/?#]+)/([^/?#]+)/commit/([0-9a-f]+)(?:\?[^#]*)?(?:#.*)?$`)
-	reGitLabMR       = regexp.MustCompile(`^https://([^/]+)/([^?#]+)/-/merge_requests/([0-9]+)(?:\?[^#]*)?(?:#.*)?$`)
-	rePiperMailDate  = regexp.MustCompile(`^\s*<I>([^<]+)</I>\s*$`)
-)
-
 const (
 	statusOpen        = "open"
 	statusMerged      = "merged, not yet in a release"
 	statusReleasedFmt = "merged, released in %s"
 )
 
-func (c Contribution) fetchStatus() (string, error) {
-	if m := reGitHubPR.FindStringSubmatch(c.URLs[0]); m != nil {
-		user := m[1]
-		repo := m[2]
-		prnum := m[3]
-
-		urlStr := "https://api.github.com/repos/" + user + "/" + repo + "/pulls/" + prnum
-
-		var obj struct {
-			// State values are "open" and "closed".
-			State          string `json:"state"`
-			Merged         bool   `json:"merged"`
-			MergeCommitSha string `json:"merge_commit_sha"`
-		}
-		if err := httpGetJSON(urlStr, nil, &obj); err != nil {
-			return "", err
-		}
-		ret := obj.State
-		if obj.Merged {
-			ret = statusMerged
-			tag, err := getGitTagThatContainsAll("https://github.com/"+user+"/"+repo, obj.MergeCommitSha)
-			if err != nil {
-				return "", err
-			}
-			if tag != "" {
-				ret = fmt.Sprintf(statusReleasedFmt, tag)
-			}
-		}
-
-		return ret, nil
-	}
-	if m := reGitLabMR.FindStringSubmatch(c.URLs[0]); m != nil {
-		authority := m[1]
-		projectID := m[2]
-		mrnum := m[3]
+type Forge interface {
+	FetchStatus(urls []string) (string, error)
+	FetchSubmittedAt(urls []string) (time.Time, error)
+	FetchLastUpdated(urls []string) (time.Time, User, error)
+}
 
-		urlStr := "https://" + authority + "/api/v4/projects/" + url.QueryEscape(projectID) + "/merge_requests/" + mrnum
+var forges = []Forge{
+	// highest precedence
+	Gerrit{}, // must be higher than GitHub because of golang
+	GitHub{},
+	GitLab{},
+	PiperMail{},
+	// lowest precedence
+}
 
-		var obj struct {
-			// State values are "opened", "closed", "locked", and "merged".
-			State           string `json:"state"`
-			MergeCommitSha  string `json:"merge_commit_sha"`
-			SquashCommitSha string `json:"squash_commit_sha"`
-		}
-		if err := httpGetJSON(urlStr, nil, &obj); err != nil {
+func (c Contribution) fetchStatus() (string, error) {
+	for _, forge := range forges {
+		status, err := forge.FetchStatus(c.URLs)
+		if err != nil {
 			return "", err
 		}
-
-		ret := obj.State
-		if ret == "opened" {
-			ret = statusOpen
-		}
-
-		if ret == "merged" {
-			ret = statusMerged
-			var mergeCommit string
-			if obj.MergeCommitSha != "" {
-				mergeCommit = obj.MergeCommitSha
-			}
-			if obj.SquashCommitSha != "" {
-				mergeCommit = obj.SquashCommitSha
-			}
-			if mergeCommit != "" {
-				tag, err := getGitTagThatContainsAll("https://"+authority+"/"+projectID+".git", mergeCommit)
-				if err != nil {
-					return "", err
-				}
-				if tag != "" {
-					ret = fmt.Sprintf(statusReleasedFmt, tag)
-				}
-			}
-		}
-
-		return ret, nil
-	}
-	if len(c.URLs) > 1 {
-		var gitURL string
-		var gitCommits []string
-		for _, u := range c.URLs[1:] {
-			if m := reGitHubCommit.FindStringSubmatch(u); m != nil {
-				user := m[1]
-				repo := m[2]
-				hash := m[3]
-
-				gitURL = "https://github.com/" + user + "/" + repo
-				gitCommits = append(gitCommits, hash)
-			}
-		}
-		if len(gitCommits) > 0 {
-			ret := statusMerged
-			tag, err := getGitTagThatContainsAll(gitURL, gitCommits...)
-			if err != nil {
-				return "", err
-			}
-			if tag != "" {
-				ret = fmt.Sprintf(statusReleasedFmt, tag)
-			}
-			return ret, nil
+		if status != "" {
+			return status, nil
 		}
 	}
 	return "", fmt.Errorf("idk how to get status for %q", c.URLs[0])
 }
 
 func (c Contribution) fetchSubmittedAt() (time.Time, error) {
-	if m := reGitHubPR.FindStringSubmatch(c.URLs[0]); m != nil {
-		user := m[1]
-		repo := m[2]
-		prnum := m[3]
-
-		urlStr := "https://api.github.com/repos/" + user + "/" + repo + "/pulls/" + prnum
-
-		var obj struct {
-			CreatedAt time.Time `json:"created_at"`
-		}
-		if err := httpGetJSON(urlStr, nil, &obj); err != nil {
-			return time.Time{}, err
-		}
-		return obj.CreatedAt, nil
-	}
-	if m := reGitLabMR.FindStringSubmatch(c.URLs[0]); m != nil {
-		authority := m[1]
-		projectID := m[2]
-		mrnum := m[3]
-
-		urlStr := "https://" + authority + "/api/v4/projects/" + url.QueryEscape(projectID) + "/merge_requests/" + mrnum
-
-		var obj struct {
-			CreatedAt time.Time `json:"created_at"`
-		}
-		if err := httpGetJSON(urlStr, nil, &obj); err != nil {
-			return time.Time{}, err
-		}
-		return obj.CreatedAt, nil
-	}
-	if strings.Contains(c.URLs[0], "/pipermail/") {
-		htmlStr, err := httpGet(c.URLs[0], nil)
+	for _, forge := range forges {
+		submittedAt, err := forge.FetchSubmittedAt(c.URLs)
 		if err != nil {
 			return time.Time{}, err
 		}
-		for _, line := range strings.Split(htmlStr, "\n") {
-			if m := rePiperMailDate.FindStringSubmatch(line); m != nil {
-				return time.Parse(time.UnixDate, m[1])
-			}
+		if !submittedAt.IsZero() {
+			return submittedAt, nil
 		}
 	}
 	return time.Time{}, fmt.Errorf("idk how to get created timestamp for %q", c.URLs[0])
@@ -267,238 +150,14 @@ func withinOneSecond(a, b time.Time) bool {
 }
 
 func (c Contribution) fetchLastUpdated() (time.Time, User, error) {
-	for _, u := range c.URLs {
-		if m := reGoLangGerritCL.FindStringSubmatch(u); m != nil {
-			projectID := m[1]
-			changeID := m[2]
-
-			urlStr := "https://go-review.googlesource.com/changes/" + projectID + "~" + changeID + "?o=MESSAGES&o=DETAILED_ACCOUNTS"
-
-			var obj struct {
-				Updated  GerritTime `json:"updated"`
-				Messages []struct {
-					Author struct {
-						AccountID   int    `json:"_account_id"`
-						Name        string `json:"name"`
-						DisplayName string `json:"display_name"`
-					} `json:"author"`
-					Date GerritTime `json:"date"`
-				} `json:"messages"`
-			}
-			if err := httpGetGerritJSON(urlStr, nil, &obj); err != nil {
-				return time.Time{}, User{}, err
-			}
-			retUpdatedAt := obj.Updated.Val
-			var retUser User
-			for _, message := range obj.Messages {
-				if withinOneSecond(message.Date.Val, retUpdatedAt) {
-					if message.Author.DisplayName != "" {
-						retUser.Name = message.Author.DisplayName
-					} else {
-						retUser.Name = message.Author.Name
-					}
-					retUser.URL = fmt.Sprintf("https://go-review.googlesource.com/dashboard/%d", message.Author.AccountID)
-					break
-				}
-			}
-			return retUpdatedAt, retUser, nil
-		}
-	}
-	if m := reGitHubPR.FindStringSubmatch(c.URLs[0]); m != nil {
-		user := m[1]
-		repo := m[2]
-		prnum := m[3]
-
-		var obj struct {
-			UpdatedAt time.Time `json:"updated_at"`
-
-			CreatedAt time.Time `json:"created_at"`
-			CreatedBy struct {
-				Login   string `json:"login"`
-				HTMLURL string `json:"html_url"`
-			} `json:"user"`
-
-			MergedAt time.Time `json:"merged_at"`
-			MergedBy struct {
-				Login   string `json:"login"`
-				HTMLURL string `json:"html_url"`
-			} `json:"merged_by"`
-		}
-		if err := httpGetJSON("https://api.github.com/repos/"+user+"/"+repo+"/pulls/"+prnum, nil, &obj); err != nil {
-			return time.Time{}, User{}, err
-		}
-
-		retUpdatedAt := obj.UpdatedAt
-		var retUser User
-
-		if retUser == (User{}) && withinOneSecond(obj.CreatedAt, retUpdatedAt) {
-			retUser.Name = obj.CreatedBy.Login
-			retUser.URL = obj.CreatedBy.HTMLURL
-		}
-		if retUser == (User{}) && withinOneSecond(obj.MergedAt, retUpdatedAt) {
-			retUser.Name = obj.MergedBy.Login
-			retUser.URL = obj.MergedBy.HTMLURL
-		}
-		if retUser == (User{}) {
-			// "normal" comments
-			var comments []struct {
-				UpdatedAt time.Time `json:"updated_at"`
-				User      struct {
-					Login   string `json:"login"`
-					HTMLURL string `json:"html_url"`
-				} `json:"user"`
-			}
-			if err := httpGetPaginatedJSON("https://api.github.com/repos/"+user+"/"+repo+"/issues/"+prnum+"/comments", nil, &comments, githubPagination); err != nil {
-				return time.Time{}, User{}, err
-			}
-			for _, comment := range comments {
-				if withinOneSecond(comment.UpdatedAt, retUpdatedAt) {
-					retUser.Name = comment.User.Login
-					retUser.URL = comment.User.HTMLURL
-					break
-				}
-			}
-		}
-		if retUser == (User{}) {
-			// comments on a specific part of the diff
-			var reviewComments []struct {
-				UpdatedAt time.Time `json:"updated_at"`
-				User      struct {
-					Login   string `json:"login"`
-					HTMLURL string `json:"html_url"`
-				} `json:"user"`
-			}
-			if err := httpGetPaginatedJSON("https://api.github.com/repos/"+user+"/"+repo+"/pulls/"+prnum+"/comments", nil, &reviewComments, githubPagination); err != nil {
-				return time.Time{}, User{}, err
-			}
-			for _, comment := range reviewComments {
-				if withinOneSecond(comment.UpdatedAt, retUpdatedAt) {
-					retUser.Name = comment.User.Login
-					retUser.URL = comment.User.HTMLURL
-					break
-				}
-			}
-		}
-		if retUser == (User{}) {
-			var events []struct {
-				CreatedAt time.Time `json:"created_at"`
-				Actor     struct {
-					Login   string `json:"login"`
-					HTMLURL string `json:"html_url"`
-				} `json:"actor"`
-			}
-			if err := httpGetJSON("https://api.github.com/repos/"+user+"/"+repo+"/issues/"+prnum+"/events", nil, &events); err != nil {
-				return time.Time{}, User{}, err
-			}
-			for _, event := range events {
-				if withinOneSecond(event.CreatedAt, retUpdatedAt) {
-					retUser.Name = event.Actor.Login
-					retUser.URL = event.Actor.HTMLURL
-					break
-				}
-			}
-		}
-
-		return retUpdatedAt, retUser, nil
-	}
-	if m := reGitLabMR.FindStringSubmatch(c.URLs[0]); m != nil {
-		authority := m[1]
-		projectID := m[2]
-		mrnum := m[3]
-
-		urlStr := "https://" + authority + "/api/v4/projects/" + url.QueryEscape(projectID) + "/merge_requests/" + mrnum
-
-		var obj struct {
-			ID int `json:"id"`
-
-			UpdatedAt time.Time `json:"updated_at"`
-
-			CreatedAt time.Time `json:"created_at"`
-			CreatedBy struct {
-				Username string `json:"username"`
-				WebURL   string `json:"web_url"`
-			} `json:"author"`
-
-			MergedAt time.Time `json:"merged_at"`
-			MergedBy struct {
-				Username string `json:"username"`
-				WebURL   string `json:"web_url"`
-			} `json:"merged_by"`
-		}
-		if err := httpGetJSON(urlStr, nil, &obj); err != nil {
+	for _, forge := range forges {
+		updatedAt, updatedBy, err := forge.FetchLastUpdated(c.URLs)
+		if err != nil {
 			return time.Time{}, User{}, err
 		}
-
-		retUpdatedAt := obj.UpdatedAt
-		var retUser User
-
-		if retUser == (User{}) && withinOneSecond(obj.CreatedAt, retUpdatedAt) {
-			retUser.Name = obj.CreatedBy.Username
-			retUser.URL = obj.CreatedBy.WebURL
-		}
-		if retUser == (User{}) && withinOneSecond(obj.MergedAt, retUpdatedAt) {
-			retUser.Name = obj.MergedBy.Username
-			retUser.URL = obj.MergedBy.WebURL
-		}
-		if retUser == (User{}) {
-			var notes struct {
-				Notes []struct {
-					UpdatedAt time.Time `json:"updated_at"`
-					Author    struct {
-						Username string `json:"username"`
-						WebURL   string `json:"web_url"`
-					} `json:"author"`
-				} `json:"notes"`
-			}
-			if err := httpGetJSON(fmt.Sprintf("https://%s/%s/noteable/merge_request/%d/notes", authority, projectID, obj.ID), map[string]string{"X-Last-Fetched-At": "0"}, &notes); err != nil {
-				return time.Time{}, User{}, err
-			}
-			for _, note := range notes.Notes {
-				if withinOneSecond(note.UpdatedAt, retUpdatedAt) {
-					retUser.Name = note.Author.Username
-					retUser.URL = note.Author.WebURL
-					break
-				}
-			}
+		if !updatedAt.IsZero() {
+			return updatedAt, updatedBy, nil
 		}
-
-		return retUpdatedAt, retUser, nil
 	}
-
-	var ret time.Time
-	if len(c.URLs) > 1 {
-		for _, u := range c.URLs[1:] {
-			if m := reGitHubCommit.FindStringSubmatch(u); m != nil {
-				user := m[1]
-				repo := m[2]
-				hash := m[3]
-
-				urlStr := "https://api.github.com/repos/" + user + "/" + repo + "/commits/" + hash
-				var obj struct {
-					Commit struct {
-						Author struct {
-							Date time.Time `json:"date"`
-						} `json:"author"`
-						Committer struct {
-							Date time.Time `json:"date"`
-						} `json:"committer"`
-					} `json:"commit"`
-				}
-				if err := httpGetJSON(urlStr, nil, &obj); err != nil {
-					return time.Time{}, User{}, err
-				}
-				if obj.Commit.Author.Date.After(ret) {
-					ret = obj.Commit.Author.Date
-				}
-				if obj.Commit.Committer.Date.After(ret) {
-					ret = obj.Commit.Committer.Date
-				}
-			}
-		}
-	}
-	if !ret.IsZero() {
-		return ret, User{}, nil
-	}
-
 	return time.Time{}, User{}, nil //fmt.Errorf("idk how to get updated timestamp for %q", c.URLs[0])
 }