Documentation
¶
Index ¶
Constants ¶
const ( EventPullRequest = "PullRequestEvent" EventPullRequestReview = "PullRequestReviewEvent" EventIssueComment = "IssueCommentEvent" EventIssues = "IssuesEvent" )
GitHub event type constants.
Variables ¶
var ErrArchiveNotFound = errors.New("archive not found")
ErrArchiveNotFound is returned when a GH Archive hourly file is not yet published.
Functions ¶
Types ¶
type Aggregator ¶
type Aggregator struct {
// contains filtered or unexported fields
}
Aggregator collects events into per-contributor hourly summaries and per-PR observations for the merge graph.
func NewAggregator ¶
func NewAggregator(hour time.Time) *Aggregator
NewAggregator creates an aggregator for the given hour (truncated to the hour boundary).
func (*Aggregator) Add ¶
func (a *Aggregator) Add(ev Event)
Add processes a single event. Bot actors are skipped for the per-contributor summaries (we don't credit bots with activity) but PullRequestEvent observations are recorded in the merge graph regardless of actor — the merge action's actor is typically a CI bot, and that observation IS the merge graph's reason to exist.
func (*Aggregator) Count ¶
func (a *Aggregator) Count() int
Count returns the number of unique contributors aggregated.
func (*Aggregator) Hour ¶
func (a *Aggregator) Hour() time.Time
Hour returns the hour this aggregator is collecting for.
func (*Aggregator) PREvents ¶
func (a *Aggregator) PREvents() []PREvent
PREvents returns all collected PR-level observations for this hour. Each (repo, pr_number, action) tuple is unique within an hour; the storage layer COALESCEs across hours into a single PR row whose author / opened_at / merged_at / closed_at columns fill in as the corresponding events arrive.
func (*Aggregator) Results ¶
func (a *Aggregator) Results() []Summary
Results returns all summaries.
type ArchiveReader ¶
type ArchiveReader struct {
// contains filtered or unexported fields
}
func NewArchiveReader ¶
func NewArchiveReader(baseURL string) *ArchiveReader
func (*ArchiveReader) Stream ¶
Stream downloads the archive for the given hour and calls fn for each relevant event. Transient transport failures and 5xx responses are retried with exponential backoff; 404 short-circuits as ErrArchiveNotFound (the file is not yet published, later hours won't be either).
type Event ¶
type Event struct {
Type string
Action string
Actor string
Repo string
CreatedAt time.Time
// PRNumber is populated for PullRequestEvent only; zero for other
// event types. Used by the merge-graph aggregator to correlate
// opened/merged/closed events on the same PR (the actor differs:
// opens are attributed to the author, merges are usually attributed
// to a CI bot in modern OSS workflows).
PRNumber int
}
type PREvent ¶
type PREvent struct {
Repo string
Number int
Action string // opened, merged, closed
Author string // populated for opened action; empty otherwise
OccurAt time.Time // event timestamp
}
PREvent is one observation of a PullRequestEvent, persisted to the merge-graph table so opened/merged/closed actions on the same PR can be joined later. Without this, "PRs merged authored by X" can't be computed: GH Archive attributes the merged action to whoever clicked merge (usually a CI bot), not the PR author.
Per-event records are keyed by (repo, pr_number) at storage time; the storage layer COALESCEs fields so each action stamps only its own column without clobbering earlier observations of the same PR.
Source Files