@@ -23,6 +23,15 @@ const (
// githubImporter implement the Importer interface
type githubImporter struct {
conf core.Configuration
+
+ // iterator
+ iterator *iterator
+
+ // number of imported issues
+ importedIssues int
+
+ // number of imported identities
+ importedIdentities int
}
func (gi *githubImporter) Init(conf core.Configuration) error {
@@ -30,54 +39,26 @@ func (gi *githubImporter) Init(conf core.Configuration) error {
return nil
}
-// ImportAll .
+// ImportAll iterate over all the configured repository issues and ensure the creation of the
+// missing issues / timeline items / edits / label events ...
func (gi *githubImporter) ImportAll(repo *cache.RepoCache, since time.Time) error {
- iterator := NewIterator(gi.conf[keyUser], gi.conf[keyProject], gi.conf[keyToken], since)
+ gi.iterator = NewIterator(gi.conf[keyUser], gi.conf[keyProject], gi.conf[keyToken], since)
// Loop over all matching issues
- for iterator.NextIssue() {
- issue := iterator.IssueValue()
-
- fmt.Printf("importing issue: %v %v\n", iterator.importedIssues, issue.Title)
- // get issue edits
- issueEdits := []userContentEdit{}
- for iterator.NextIssueEdit() {
- // issueEdit.Diff == nil happen if the event is older than early 2018, Github doesn't have the data before that.
- // Best we can do is to ignore the event.
- if issueEdit := iterator.IssueEditValue(); issueEdit.Diff != nil && string(*issueEdit.Diff) != "" {
- issueEdits = append(issueEdits, issueEdit)
- }
- }
+ for gi.iterator.NextIssue() {
+ issue := gi.iterator.IssueValue()
+ fmt.Printf("importing issue: %v\n", issue.Title)
// create issue
- b, err := gi.ensureIssue(repo, issue, issueEdits)
+ b, err := gi.ensureIssue(repo, issue)
if err != nil {
return fmt.Errorf("issue creation: %v", err)
}
// loop over timeline items
- for iterator.NextTimeline() {
- item := iterator.TimelineValue()
-
- // if item is comment
- if item.Typename == "IssueComment" {
- // collect all edits
- commentEdits := []userContentEdit{}
- for iterator.NextCommentEdit() {
- if commentEdit := iterator.CommentEditValue(); commentEdit.Diff != nil && string(*commentEdit.Diff) != "" {
- commentEdits = append(commentEdits, commentEdit)
- }
- }
-
- err := gi.ensureTimelineComment(repo, b, item.IssueComment, commentEdits)
- if err != nil {
- return fmt.Errorf("timeline comment creation: %v", err)
- }
-
- } else {
- if err := gi.ensureTimelineItem(repo, b, item); err != nil {
- return fmt.Errorf("timeline event creation: %v", err)
- }
+ for gi.iterator.NextTimelineItem() {
+ if err := gi.ensureTimelineItem(repo, b, gi.iterator.TimelineItemValue()); err != nil {
+ return fmt.Errorf("timeline item creation: %v", err)
}
}
@@ -87,16 +68,16 @@ func (gi *githubImporter) ImportAll(repo *cache.RepoCache, since time.Time) erro
}
}
- if err := iterator.Error(); err != nil {
+ if err := gi.iterator.Error(); err != nil {
fmt.Printf("import error: %v\n", err)
return err
}
- fmt.Printf("Successfully imported %v issues from Github\n", iterator.ImportedIssues())
+ fmt.Printf("Successfully imported %d issues and %d identities from Github\n", gi.importedIssues, gi.importedIdentities)
return nil
}
-func (gi *githubImporter) ensureIssue(repo *cache.RepoCache, issue issueTimeline, issueEdits []userContentEdit) (*cache.BugCache, error) {
+func (gi *githubImporter) ensureIssue(repo *cache.RepoCache, issue issueTimeline) (*cache.BugCache, error) {
// ensure issue author
author, err := gi.ensurePerson(repo, issue.Author)
if err != nil {
@@ -109,6 +90,12 @@ func (gi *githubImporter) ensureIssue(repo *cache.RepoCache, issue issueTimeline
return nil, err
}
+ // get issue edits
+ issueEdits := []userContentEdit{}
+ for gi.iterator.NextIssueEdit() {
+ issueEdits = append(issueEdits, gi.iterator.IssueEditValue())
+ }
+
// if issueEdits is empty
if len(issueEdits) == 0 {
if err == bug.ErrBugNotExist {
@@ -131,6 +118,9 @@ func (gi *githubImporter) ensureIssue(repo *cache.RepoCache, issue issueTimeline
if err != nil {
return nil, err
}
+
+ // importing a new bug
+ gi.importedIssues++
}
} else {
@@ -165,6 +155,9 @@ func (gi *githubImporter) ensureIssue(repo *cache.RepoCache, issue issueTimeline
return nil, err
}
+ // importing a new bug
+ gi.importedIssues++
+
continue
}
@@ -189,6 +182,16 @@ func (gi *githubImporter) ensureTimelineItem(repo *cache.RepoCache, b *cache.Bug
switch item.Typename {
case "IssueComment":
+ // collect all comment edits
+ commentEdits := []userContentEdit{}
+ for gi.iterator.NextCommentEdit() {
+ commentEdits = append(commentEdits, gi.iterator.CommentEditValue())
+ }
+
+ err := gi.ensureTimelineComment(repo, b, item.IssueComment, commentEdits)
+ if err != nil {
+ return fmt.Errorf("timeline comment creation: %v", err)
+ }
case "LabeledEvent":
id := parseId(item.LabeledEvent.Id)
@@ -455,6 +458,9 @@ func (gi *githubImporter) ensurePerson(repo *cache.RepoCache, actor *actor) (*ca
return nil, err
}
+ // importing a new identity
+ gi.importedIdentities++
+
var name string
var email string
@@ -528,10 +534,3 @@ func (gi *githubImporter) getGhost(repo *cache.RepoCache) (*cache.IdentityCache,
func parseId(id githubv4.ID) string {
return fmt.Sprintf("%v", id)
}
-
-func reverseEdits(edits []userContentEdit) []userContentEdit {
- for i, j := 0, len(edits)-1; i < j; i, j = i+1, j-1 {
- edits[i], edits[j] = edits[j], edits[i]
- }
- return edits
-}
@@ -49,9 +49,6 @@ type iterator struct {
// sticky error
err error
- // number of imported issues
- importedIssues int
-
// timeline iterator
timeline timelineIterator
@@ -62,8 +59,9 @@ type iterator struct {
commentEdit commentEditIterator
}
+// NewIterator create and initalize a new iterator
func NewIterator(user, project, token string, since time.Time) *iterator {
- return &iterator{
+ i := &iterator{
gc: buildClient(token),
since: since,
capacity: 10,
@@ -91,6 +89,9 @@ func NewIterator(user, project, token string, since time.Time) *iterator {
},
},
}
+
+ i.initTimelineQueryVariables()
+ return i
}
// init issue timeline variables
@@ -145,11 +146,6 @@ func (i *iterator) Error() error {
return i.err
}
-// ImportedIssues return the number of issues we iterated over
-func (i *iterator) ImportedIssues() int {
- return i.importedIssues
-}
-
func (i *iterator) queryIssue() bool {
if err := i.gc.Query(context.TODO(), &i.timeline.query, i.timeline.variables); err != nil {
i.err = err
@@ -161,18 +157,18 @@ func (i *iterator) queryIssue() bool {
}
i.reverseTimelineEditNodes()
- i.importedIssues++
return true
}
-// Next issue
+// NextIssue try to query the next issue and return true. Only one issue is
+// queried at each call.
func (i *iterator) NextIssue() bool {
- // we make the first move
- if i.importedIssues == 0 {
-
- // init variables and goto queryIssue block
- i.initTimelineQueryVariables()
- return i.queryIssue()
+ // if $issueAfter variable is nil we can directly make the first query
+ if i.timeline.variables["issueAfter"] == (*githubv4.String)(nil) {
+ nextIssue := i.queryIssue()
+ // prevent from infinite loop by setting a non nil cursor
+ i.timeline.variables["issueAfter"] = i.timeline.query.Repository.Issues.PageInfo.EndCursor
+ return nextIssue
}
if i.err != nil {
@@ -195,11 +191,14 @@ func (i *iterator) NextIssue() bool {
return i.queryIssue()
}
+// IssueValue return the actual issue value
func (i *iterator) IssueValue() issueTimeline {
return i.timeline.query.Repository.Issues.Nodes[0]
}
-func (i *iterator) NextTimeline() bool {
+// NextTimelineItem return true if there is a next timeline item and increments the index by one.
+// It is used iterates over all the timeline items. Extra queries are made if it is necessary.
+func (i *iterator) NextTimelineItem() bool {
if i.err != nil {
return false
}
@@ -231,7 +230,8 @@ func (i *iterator) NextTimeline() bool {
return true
}
-func (i *iterator) TimelineValue() timelineItem {
+// TimelineItemValue return the actual timeline item value
+func (i *iterator) TimelineItemValue() timelineItem {
return i.timeline.query.Repository.Issues.Nodes[0].Timeline.Edges[i.timeline.index].Node
}
@@ -253,9 +253,20 @@ func (i *iterator) queryIssueEdit() bool {
i.issueEdit.index = 0
i.timeline.issueEdit.index = -2
+ return i.nextValidIssueEdit()
+}
+
+func (i *iterator) nextValidIssueEdit() bool {
+ // issueEdit.Diff == nil happen if the event is older than early 2018, Github doesn't have the data before that.
+ // Best we can do is to ignore the event.
+ if issueEdit := i.IssueEditValue(); issueEdit.Diff == nil || string(*issueEdit.Diff) == "" {
+ return i.NextIssueEdit()
+ }
return true
}
+// NextIssueEdit return true if there is a next issue edit and increments the index by one.
+// It is used iterates over all the issue edits. Extra queries are made if it is necessary.
func (i *iterator) NextIssueEdit() bool {
if i.err != nil {
return false
@@ -266,7 +277,7 @@ func (i *iterator) NextIssueEdit() bool {
if i.timeline.issueEdit.index == -2 {
if i.issueEdit.index < min(i.capacity, len(i.issueEdit.query.Repository.Issues.Nodes[0].UserContentEdits.Nodes))-1 {
i.issueEdit.index++
- return true
+ return i.nextValidIssueEdit()
}
if !i.issueEdit.query.Repository.Issues.Nodes[0].UserContentEdits.PageInfo.HasPreviousPage {
@@ -297,7 +308,7 @@ func (i *iterator) NextIssueEdit() bool {
// loop over them timeline comment edits
if i.timeline.issueEdit.index < min(i.capacity, len(i.timeline.query.Repository.Issues.Nodes[0].UserContentEdits.Nodes))-1 {
i.timeline.issueEdit.index++
- return true
+ return i.nextValidIssueEdit()
}
if !i.timeline.query.Repository.Issues.Nodes[0].UserContentEdits.PageInfo.HasPreviousPage {
@@ -311,6 +322,7 @@ func (i *iterator) NextIssueEdit() bool {
return i.queryIssueEdit()
}
+// IssueEditValue return the actual issue edit value
func (i *iterator) IssueEditValue() userContentEdit {
// if we are using issue edit query
if i.timeline.issueEdit.index == -2 {
@@ -337,9 +349,19 @@ func (i *iterator) queryCommentEdit() bool {
i.commentEdit.index = 0
i.timeline.commentEdit.index = -2
+ return i.nextValidCommentEdit()
+}
+
+func (i *iterator) nextValidCommentEdit() bool {
+ // if comment edit diff is a nil pointer or points to an empty string look for next value
+ if commentEdit := i.CommentEditValue(); commentEdit.Diff == nil || string(*commentEdit.Diff) == "" {
+ return i.NextCommentEdit()
+ }
return true
}
+// NextCommentEdit return true if there is a next comment edit and increments the index by one.
+// It is used iterates over all the comment edits. Extra queries are made if it is necessary.
func (i *iterator) NextCommentEdit() bool {
if i.err != nil {
return false
@@ -350,7 +372,7 @@ func (i *iterator) NextCommentEdit() bool {
if i.commentEdit.index < min(i.capacity, len(i.commentEdit.query.Repository.Issues.Nodes[0].Timeline.Nodes[0].IssueComment.UserContentEdits.Nodes))-1 {
i.commentEdit.index++
- return true
+ return i.nextValidCommentEdit()
}
if !i.commentEdit.query.Repository.Issues.Nodes[0].Timeline.Nodes[0].IssueComment.UserContentEdits.PageInfo.HasPreviousPage {
@@ -372,7 +394,7 @@ func (i *iterator) NextCommentEdit() bool {
// loop over them timeline comment edits
if i.timeline.commentEdit.index < min(i.capacity, len(i.timeline.query.Repository.Issues.Nodes[0].Timeline.Edges[i.timeline.index].Node.IssueComment.UserContentEdits.Nodes))-1 {
i.timeline.commentEdit.index++
- return true
+ return i.nextValidCommentEdit()
}
if !i.timeline.query.Repository.Issues.Nodes[0].Timeline.Edges[i.timeline.index].Node.IssueComment.UserContentEdits.PageInfo.HasPreviousPage {
@@ -392,6 +414,7 @@ func (i *iterator) NextCommentEdit() bool {
return i.queryCommentEdit()
}
+// CommentEditValue return the actual comment edit value
func (i *iterator) CommentEditValue() userContentEdit {
if i.timeline.commentEdit.index == -2 {
return i.commentEdit.query.Repository.Issues.Nodes[0].Timeline.Nodes[0].IssueComment.UserContentEdits.Nodes[i.commentEdit.index]
@@ -407,3 +430,9 @@ func min(a, b int) int {
return a
}
+
+func reverseEdits(edits []userContentEdit) {
+ for i, j := 0, len(edits)-1; i < j; i, j = i+1, j-1 {
+ edits[i], edits[j] = edits[j], edits[i]
+ }
+}