| 1 | package changelog |
| 2 | |
| 3 | import ( |
| 4 | "encoding/json" |
| 5 | "fmt" |
| 6 | "html" |
| 7 | "net/url" |
| 8 | "path" |
| 9 | "regexp" |
| 10 | "slices" |
| 11 | "sort" |
| 12 | "strconv" |
| 13 | "strings" |
| 14 | "time" |
| 15 | |
| 16 | "github.com/carlosframework/platform/internal/docsite" |
| 17 | ) |
| 18 | |
| 19 | // RepoURL is the public Amadan repo every branch, commit and discussion |
| 20 | // link points into. |
| 21 | const RepoURL = "https://amadan.net/carlos/platform" |
| 22 | |
| 23 | // NotesBase is where each published release's notes live. |
| 24 | const NotesBase = "https://github.com/carlosframework/releases/releases/tag/" |
| 25 | |
| 26 | type Tag struct { |
| 27 | Name string |
| 28 | SHA string |
| 29 | Date time.Time |
| 30 | } |
| 31 | |
| 32 | type Commit struct { |
| 33 | SHA string |
| 34 | Parents []string |
| 35 | AuthorName, AuthorEmail string |
| 36 | Date time.Time // committer date |
| 37 | Message string |
| 38 | } |
| 39 | |
| 40 | // DocsChange is one docs/site page a commit added lines to: the page's |
| 41 | // source as of the commit, and the 1-based new-file lines added. |
| 42 | type DocsChange struct { |
| 43 | Page string // slug, e.g. "cli" |
| 44 | Body string |
| 45 | Added []int |
| 46 | } |
| 47 | |
| 48 | type Git interface { |
| 49 | Tags() ([]Tag, error) |
| 50 | // FirstParent lists the commits on to's first-parent line that no |
| 51 | // excluded tag reaches, newest first. |
| 52 | FirstParent(to string, exclude []string) ([]Commit, error) |
| 53 | // Range lists the non-merge commits in from..to, newest first. |
| 54 | Range(from, to string) ([]Commit, error) |
| 55 | // Tombstone is the last tip Amadan kept for a merged branch, at |
| 56 | // refs/amadan/closed/<branch>. |
| 57 | Tombstone(branch string) (string, time.Time, bool) |
| 58 | // LiveBranch reports whether origin still carries the branch. |
| 59 | LiveBranch(branch string) bool |
| 60 | DocsChanges(c Commit) ([]DocsChange, error) |
| 61 | } |
| 62 | |
| 63 | type Releases interface { |
| 64 | // Body is a tag's published release notes; ok is false when the tag |
| 65 | // was never published. |
| 66 | Body(tag string) (body string, ok bool, err error) |
| 67 | } |
| 68 | |
| 69 | type Discussion struct { |
| 70 | ID int `json:"id"` |
| 71 | Title string `json:"title"` |
| 72 | State string `json:"state"` |
| 73 | Created int64 `json:"-"` |
| 74 | } |
| 75 | |
| 76 | type Discussions interface { |
| 77 | // Show returns nil, nil when there is no such discussion. |
| 78 | Show(id int) (*Discussion, error) |
| 79 | ForBranch(branch string) ([]Discussion, error) |
| 80 | } |
| 81 | |
| 82 | type Sources struct { |
| 83 | Git Git |
| 84 | Releases Releases |
| 85 | Discussions Discussions |
| 86 | } |
| 87 | |
| 88 | // Links is changelog-links.json: tag, then a bullet's bold lead, then the |
| 89 | // branches behind that bullet. |
| 90 | type Links map[string]map[string][]string |
| 91 | |
| 92 | type Changelog struct { |
| 93 | Note string `json:"//"` |
| 94 | Repo string `json:"repo"` |
| 95 | Releases []Release `json:"releases"` |
| 96 | } |
| 97 | |
| 98 | type Release struct { |
| 99 | Tag string `json:"tag"` |
| 100 | Date string `json:"date"` |
| 101 | SHA string `json:"sha"` |
| 102 | Published bool `json:"published"` |
| 103 | NotesURL string `json:"notes_url,omitempty"` |
| 104 | Notes *Notes `json:"notes"` |
| 105 | Changes []Change `json:"changes"` |
| 106 | } |
| 107 | |
| 108 | type Change struct { |
| 109 | SHA string `json:"sha"` |
| 110 | Date string `json:"date"` |
| 111 | Subject string `json:"subject"` |
| 112 | Branch string `json:"branch,omitempty"` |
| 113 | BranchURL string `json:"branch_url,omitempty"` |
| 114 | Commits int `json:"commits"` |
| 115 | Discussions []Discussion `json:"discussions,omitempty"` |
| 116 | Docs []string `json:"docs,omitempty"` |
| 117 | Credit []Credit `json:"credit"` |
| 118 | } |
| 119 | |
| 120 | const generatedNote = "Generated by go run ./cmd/changelog from git, the published release notes and Amadan. Do not edit; regenerate." |
| 121 | |
| 122 | // maxDocs caps the docs links one change or bullet shows. |
| 123 | const maxDocs = 5 |
| 124 | |
| 125 | // refSlack allows for clock skew between the machine that committed and |
| 126 | // the hub that stamped a discussion's creation. |
| 127 | const refSlack = time.Hour |
| 128 | |
| 129 | // Build joins every source into the changelog, newest release first. |
| 130 | func Build(src Sources, people People, links Links, site *docsite.Site) (*Changelog, error) { |
| 131 | tags, err := src.Git.Tags() |
| 132 | if err != nil { |
| 133 | return nil, err |
| 134 | } |
| 135 | sort.Slice(tags, func(i, j int) bool { return semverLess(tags[i].Name, tags[j].Name) }) |
| 136 | |
| 137 | b := &builder{src: src, people: people, site: site, discussions: map[int]*Discussion{}, |
| 138 | landed: map[string]bool{}, built: map[string]*Change{}} |
| 139 | b.cliHeadings = map[string]string{} |
| 140 | if cli := site.BySlug["cli"]; cli != nil { |
| 141 | for _, h := range cli.Headings { |
| 142 | if anchored(h) { |
| 143 | b.cliHeadings[h.Text] = h.Anchor |
| 144 | } |
| 145 | } |
| 146 | } |
| 147 | |
| 148 | out := &Changelog{Note: generatedNote, Repo: RepoURL} |
| 149 | used := map[string]map[string]bool{} |
| 150 | for _, t := range tags { |
| 151 | r, err := b.release(t, out.Releases) |
| 152 | if err != nil { |
| 153 | return nil, fmt.Errorf("%s: %w", t.Name, err) |
| 154 | } |
| 155 | used[t.Name] = applyLinks(r, links[t.Name]) |
| 156 | out.Releases = append(out.Releases, *r) |
| 157 | } |
| 158 | for tag, byLead := range links { |
| 159 | for lead, branches := range byLead { |
| 160 | if !used[tag][lead] { |
| 161 | return nil, fmt.Errorf("changelog-links.json: %s has no bullet whose lead is %q", tag, lead) |
| 162 | } |
| 163 | if err := checkLinkBranches(out, tag, branches); err != nil { |
| 164 | return nil, err |
| 165 | } |
| 166 | } |
| 167 | } |
| 168 | for i, j := 0, len(out.Releases)-1; i < j; i, j = i+1, j-1 { |
| 169 | out.Releases[i], out.Releases[j] = out.Releases[j], out.Releases[i] |
| 170 | } |
| 171 | if err := CheckPublic(out, people); err != nil { |
| 172 | return nil, err |
| 173 | } |
| 174 | return out, nil |
| 175 | } |
| 176 | |
| 177 | type builder struct { |
| 178 | src Sources |
| 179 | people People |
| 180 | site *docsite.Site |
| 181 | cliHeadings map[string]string |
| 182 | discussions map[int]*Discussion |
| 183 | // landed is every commit an earlier change already accounted for, |
| 184 | // so a branch stacked on one squashed before it does not count the |
| 185 | // earlier branch's commits again. |
| 186 | landed map[string]bool |
| 187 | // built is every change already joined, by its landing commit: an |
| 188 | // unpublished tag and the published release after it can list the |
| 189 | // same commit. |
| 190 | built map[string]*Change |
| 191 | } |
| 192 | |
| 193 | // RangeBases is the tags a release's changes are counted from. A |
| 194 | // published release lists what no earlier published release carried, so |
| 195 | // it owns the work of an unpublished tag before it (v0.22.2 shipped |
| 196 | // v0.22.1's change); an unpublished tag lists what no earlier tag of any |
| 197 | // kind carried. earlier is in semver order. |
| 198 | func RangeBases(published bool, earlier []Release) []string { |
| 199 | var out []string |
| 200 | for _, r := range earlier { |
| 201 | if r.Published || !published { |
| 202 | out = append(out, r.Tag) |
| 203 | } |
| 204 | } |
| 205 | return out |
| 206 | } |
| 207 | |
| 208 | func (b *builder) release(t Tag, earlier []Release) (*Release, error) { |
| 209 | r := &Release{Tag: t.Name, Date: day(t.Date), SHA: t.SHA, Changes: []Change{}} |
| 210 | body, ok, err := b.src.Releases.Body(t.Name) |
| 211 | if err != nil { |
| 212 | return nil, err |
| 213 | } |
| 214 | if ok { |
| 215 | r.Published = true |
| 216 | r.NotesURL = NotesBase + t.Name |
| 217 | // A release published with an empty body (v0.18.0) has a notes |
| 218 | // page to link and nothing to show here. |
| 219 | if n := ParseNotes(body); n.MD != "" || len(n.Sections) > 0 { |
| 220 | r.Notes = n |
| 221 | } |
| 222 | for si := range notesSections(r.Notes) { |
| 223 | for bi := range r.Notes.Sections[si].Bullets { |
| 224 | bl := &r.Notes.Sections[si].Bullets[bi] |
| 225 | bl.Docs = capped(CLIDocLinks(strings.Join(append([]string{bl.MD}, bl.Items...), "\n"), b.cliHeadings)) |
| 226 | } |
| 227 | } |
| 228 | } |
| 229 | commits, err := b.src.Git.FirstParent(t.Name, RangeBases(r.Published, earlier)) |
| 230 | if err != nil { |
| 231 | return nil, err |
| 232 | } |
| 233 | // Oldest first, so the landed set is complete for each change; the |
| 234 | // page lists them newest first. |
| 235 | for i := len(commits) - 1; i >= 0; i-- { |
| 236 | ch, err := b.change(commits[i]) |
| 237 | if err != nil { |
| 238 | return nil, fmt.Errorf("%s: %w", short(commits[i].SHA), err) |
| 239 | } |
| 240 | r.Changes = append(r.Changes, *ch) |
| 241 | } |
| 242 | for i, j := 0, len(r.Changes)-1; i < j; i, j = i+1, j-1 { |
| 243 | r.Changes[i], r.Changes[j] = r.Changes[j], r.Changes[i] |
| 244 | } |
| 245 | return r, nil |
| 246 | } |
| 247 | |
| 248 | func notesSections(n *Notes) []NoteSection { |
| 249 | if n == nil { |
| 250 | return nil |
| 251 | } |
| 252 | return n.Sections |
| 253 | } |
| 254 | |
| 255 | func (b *builder) change(c Commit) (*Change, error) { |
| 256 | if ch, ok := b.built[c.SHA]; ok { |
| 257 | return ch, nil |
| 258 | } |
| 259 | ch := &Change{ |
| 260 | SHA: c.SHA, |
| 261 | Date: day(c.Date), |
| 262 | Branch: BranchName(c.Message), |
| 263 | } |
| 264 | b.built[c.SHA] = ch |
| 265 | own, onAmadan, err := b.ownCommits(c, ch.Branch) |
| 266 | if err != nil { |
| 267 | return nil, err |
| 268 | } |
| 269 | ch.Subject = Redact(Describe(c.Message, ch.Branch, own)) |
| 270 | b.landed[c.SHA] = true |
| 271 | for _, o := range own { |
| 272 | b.landed[o.SHA] = true |
| 273 | } |
| 274 | if onAmadan { |
| 275 | ch.BranchURL = RepoURL + "/-/branch/" + url.PathEscape(ch.Branch) |
| 276 | } |
| 277 | ch.Commits = len(own) |
| 278 | |
| 279 | idents := make([]CommitIdent, 0, len(own)) |
| 280 | for _, o := range own { |
| 281 | idents = append(idents, CommitIdent{o.AuthorName, o.AuthorEmail, o.Message}) |
| 282 | } |
| 283 | ch.Credit, err = Tally(idents, b.people) |
| 284 | if err != nil { |
| 285 | return nil, err |
| 286 | } |
| 287 | if len(ch.Credit) == 0 { |
| 288 | // Every own commit was a skipped identity (GitHub's merge |
| 289 | // button); the landing commit is the evidence left. |
| 290 | ch.Credit, err = Tally([]CommitIdent{{c.AuthorName, c.AuthorEmail, c.Message}}, b.people) |
| 291 | if err != nil { |
| 292 | return nil, err |
| 293 | } |
| 294 | } |
| 295 | |
| 296 | ch.Discussions, err = b.discussionsFor(c, own, ch.Branch, onAmadan) |
| 297 | if err != nil { |
| 298 | return nil, err |
| 299 | } |
| 300 | |
| 301 | docs, err := b.src.Git.DocsChanges(c) |
| 302 | if err != nil { |
| 303 | return nil, err |
| 304 | } |
| 305 | ch.Docs = capped(b.docsLinks(docs)) |
| 306 | return ch, nil |
| 307 | } |
| 308 | |
| 309 | // ownCommits is the work a first-parent commit landed. onAmadan reports |
| 310 | // whether the branch has a page on Amadan to link to. |
| 311 | func (b *builder) ownCommits(c Commit, branch string) ([]Commit, bool, error) { |
| 312 | onAmadan := false |
| 313 | if branch != "" { |
| 314 | onAmadan = b.src.Git.LiveBranch(branch) || strings.Contains(c.Message, "Squashed from "+branch+" into") |
| 315 | } |
| 316 | if len(c.Parents) == 2 { |
| 317 | own, err := b.src.Git.Range(c.Parents[0], c.Parents[1]) |
| 318 | if err != nil { |
| 319 | return nil, false, err |
| 320 | } |
| 321 | own = b.unlanded(own) |
| 322 | if branch != "" { |
| 323 | if _, _, ok := b.src.Git.Tombstone(branch); ok { |
| 324 | onAmadan = true |
| 325 | } |
| 326 | } |
| 327 | return own, onAmadan, nil |
| 328 | } |
| 329 | if branch != "" && len(c.Parents) == 1 { |
| 330 | if tip, when, ok := b.src.Git.Tombstone(branch); ok { |
| 331 | onAmadan = true |
| 332 | // A tombstone newer than the squash belongs to a later |
| 333 | // branch that reused the name; its commits are not this |
| 334 | // change's. |
| 335 | if !when.After(c.Date.Add(refSlack)) { |
| 336 | own, err := b.src.Git.Range(c.Parents[0], tip) |
| 337 | if err != nil { |
| 338 | return nil, false, err |
| 339 | } |
| 340 | if own = b.unlanded(own); len(own) > 0 { |
| 341 | return own, onAmadan, nil |
| 342 | } |
| 343 | } |
| 344 | } |
| 345 | } |
| 346 | return []Commit{c}, onAmadan, nil |
| 347 | } |
| 348 | |
| 349 | func (b *builder) unlanded(cs []Commit) []Commit { |
| 350 | out := cs[:0:0] |
| 351 | for _, c := range cs { |
| 352 | if !b.landed[c.SHA] { |
| 353 | out = append(out, c) |
| 354 | } |
| 355 | } |
| 356 | return out |
| 357 | } |
| 358 | |
| 359 | // anchored reports whether the website gives a heading an id: its |
| 360 | // markdown-it-anchor configuration covers levels two to four. |
| 361 | func anchored(h docsite.Heading) bool { return h.Level >= 2 && h.Level <= 4 } |
| 362 | |
| 363 | func (b *builder) discussionsFor(c Commit, own []Commit, branch string, onAmadan bool) ([]Discussion, error) { |
| 364 | found := map[int]Discussion{} |
| 365 | consider := func(msg string, at time.Time) error { |
| 366 | for _, id := range DiscussionRefs(msg) { |
| 367 | if _, ok := found[id]; ok { |
| 368 | continue |
| 369 | } |
| 370 | d, err := b.show(id) |
| 371 | if err != nil { |
| 372 | return err |
| 373 | } |
| 374 | if d != nil && time.Unix(d.Created, 0).Before(at.Add(refSlack)) { |
| 375 | found[id] = *d |
| 376 | } |
| 377 | } |
| 378 | return nil |
| 379 | } |
| 380 | if err := consider(c.Message, c.Date); err != nil { |
| 381 | return nil, err |
| 382 | } |
| 383 | for _, o := range own { |
| 384 | if o.SHA == c.SHA { |
| 385 | continue |
| 386 | } |
| 387 | if err := consider(o.Message, o.Date); err != nil { |
| 388 | return nil, err |
| 389 | } |
| 390 | } |
| 391 | if onAmadan { |
| 392 | list, err := b.src.Discussions.ForBranch(branch) |
| 393 | if err != nil { |
| 394 | return nil, err |
| 395 | } |
| 396 | for _, d := range list { |
| 397 | d.Title = Redact(d.Title) |
| 398 | found[d.ID] = d |
| 399 | } |
| 400 | } |
| 401 | if len(found) == 0 { |
| 402 | return nil, nil |
| 403 | } |
| 404 | out := make([]Discussion, 0, len(found)) |
| 405 | for _, d := range found { |
| 406 | out = append(out, d) |
| 407 | } |
| 408 | sort.Slice(out, func(i, j int) bool { return out[i].ID < out[j].ID }) |
| 409 | return out, nil |
| 410 | } |
| 411 | |
| 412 | func (b *builder) show(id int) (*Discussion, error) { |
| 413 | if d, ok := b.discussions[id]; ok { |
| 414 | return d, nil |
| 415 | } |
| 416 | d, err := b.src.Discussions.Show(id) |
| 417 | if err != nil { |
| 418 | return nil, err |
| 419 | } |
| 420 | if d != nil { |
| 421 | d.Title = Redact(d.Title) |
| 422 | } |
| 423 | b.discussions[id] = d |
| 424 | return d, nil |
| 425 | } |
| 426 | |
| 427 | // docsLinks maps each added line to the nearest heading above it (level |
| 428 | // two or deeper) in the page as the commit left it, and keeps the link |
| 429 | // only if today's docs still have that page and heading. |
| 430 | func (b *builder) docsLinks(changes []DocsChange) []string { |
| 431 | var out []string |
| 432 | seen := map[string]bool{} |
| 433 | add := func(l string) { |
| 434 | if !seen[l] { |
| 435 | seen[l] = true |
| 436 | out = append(out, l) |
| 437 | } |
| 438 | } |
| 439 | for _, dc := range changes { |
| 440 | today := b.site.BySlug[dc.Page] |
| 441 | if today == nil { |
| 442 | continue |
| 443 | } |
| 444 | anchors := map[string]bool{} |
| 445 | for _, h := range today.Headings { |
| 446 | if anchored(h) { |
| 447 | anchors[h.Anchor] = true |
| 448 | } |
| 449 | } |
| 450 | then := docsite.Parse(dc.Page, dc.Body) |
| 451 | for _, line := range dc.Added { |
| 452 | anchor := "" |
| 453 | for _, h := range then.Headings { |
| 454 | if h.Line > line { |
| 455 | break |
| 456 | } |
| 457 | switch { |
| 458 | case anchored(h): |
| 459 | anchor = h.Anchor |
| 460 | case h.Level == 1: |
| 461 | anchor = "" |
| 462 | } |
| 463 | } |
| 464 | if anchor != "" && anchors[anchor] { |
| 465 | add(pageURL(dc.Page) + "#" + anchor) |
| 466 | } else if anchor == "" { |
| 467 | add(pageURL(dc.Page)) |
| 468 | } |
| 469 | } |
| 470 | } |
| 471 | // A page link says less than a section link on the same page. |
| 472 | kept := out[:0:0] |
| 473 | for _, l := range out { |
| 474 | if !strings.Contains(l, "#") && slices.ContainsFunc(out, func(o string) bool { return strings.HasPrefix(o, l+"#") }) { |
| 475 | continue |
| 476 | } |
| 477 | kept = append(kept, l) |
| 478 | } |
| 479 | return kept |
| 480 | } |
| 481 | |
| 482 | func pageURL(slug string) string { |
| 483 | if slug == "index" { |
| 484 | return "/docs/" |
| 485 | } |
| 486 | return "/docs/" + slug |
| 487 | } |
| 488 | |
| 489 | // applyLinks sets each bullet's branches from the hand-kept map and |
| 490 | // reports which leads it used. |
| 491 | func applyLinks(r *Release, byLead map[string][]string) map[string]bool { |
| 492 | used := map[string]bool{} |
| 493 | if r.Notes == nil { |
| 494 | return used |
| 495 | } |
| 496 | for si := range r.Notes.Sections { |
| 497 | for bi := range r.Notes.Sections[si].Bullets { |
| 498 | bl := &r.Notes.Sections[si].Bullets[bi] |
| 499 | if br, ok := byLead[bl.Lead]; ok && bl.Lead != "" { |
| 500 | bl.Branches = br |
| 501 | used[bl.Lead] = true |
| 502 | } |
| 503 | } |
| 504 | } |
| 505 | return used |
| 506 | } |
| 507 | |
| 508 | func checkLinkBranches(c *Changelog, tag string, branches []string) error { |
| 509 | for _, r := range c.Releases { |
| 510 | if r.Tag != tag { |
| 511 | continue |
| 512 | } |
| 513 | have := map[string]bool{} |
| 514 | for _, ch := range r.Changes { |
| 515 | if ch.Branch != "" { |
| 516 | have[ch.Branch] = true |
| 517 | } |
| 518 | } |
| 519 | for _, br := range branches { |
| 520 | if !have[br] { |
| 521 | return fmt.Errorf("changelog-links.json: %s links %q, which did not land in that release", tag, br) |
| 522 | } |
| 523 | } |
| 524 | } |
| 525 | return nil |
| 526 | } |
| 527 | |
| 528 | var ( |
| 529 | addressRe = regexp.MustCompile(`(?i)[a-z0-9._%+~-]+@[a-z0-9-]+(\.[a-z0-9-]+)*\.[a-z]{2,}`) |
| 530 | keymailRe = regexp.MustCompile(`(?i)[a-z0-9._-]+~[a-z0-9._-]+@[a-z0-9.-]+`) |
| 531 | // privateRe is every link into a repo a visitor cannot open: the |
| 532 | // platform's GitHub mirror is private, and so are the five source |
| 533 | // products' repos. The public ones (releases, homebrew-tap, skills, |
| 534 | // website) are allowed. |
| 535 | privateRe = regexp.MustCompile(`(?i)github\.com/carlosframework/(platform|platform-infrastructure|carloku[a-z-]*)\b`) |
| 536 | ) |
| 537 | |
| 538 | var privateURLRe = regexp.MustCompile(`(?i)(https?://)?github\.com/carlosframework/(platform|platform-infrastructure|carloku[a-z-]*)\b\S*`) |
| 539 | |
| 540 | // Redact replaces every address in a record the page quotes verbatim (a |
| 541 | // commit subject, a discussion title) with "[address]". Amadan shows the |
| 542 | // original; the changelog does not repeat it. |
| 543 | // |
| 544 | // A link into a private repo becomes "[private repo]" the same way. |
| 545 | func Redact(s string) string { |
| 546 | s = keymailRe.ReplaceAllString(s, "[address]") |
| 547 | s = addressRe.ReplaceAllString(s, "[address]") |
| 548 | return privateURLRe.ReplaceAllString(s, "[private repo]") |
| 549 | } |
| 550 | |
| 551 | // CheckPublic fails if the changelog would publish something a public |
| 552 | // page must not: an email address of any kind (every address people.json |
| 553 | // names, anything shaped like one, an agent account), or a link into a |
| 554 | // private repo. The generator runs it before writing and the gate runs |
| 555 | // it over the committed file. |
| 556 | func CheckPublic(c *Changelog, people People) error { |
| 557 | for _, r := range c.Releases { |
| 558 | if r.Notes != nil { |
| 559 | if err := checkPublicValue(r.Notes, people); err != nil { |
| 560 | return fmt.Errorf("%s notes: %w", r.Tag, err) |
| 561 | } |
| 562 | } |
| 563 | for _, ch := range r.Changes { |
| 564 | if err := checkPublicValue(ch, people); err != nil { |
| 565 | return fmt.Errorf("%s change %s (%s): %w", r.Tag, short(ch.SHA), ch.Subject, err) |
| 566 | } |
| 567 | } |
| 568 | } |
| 569 | return checkPublicValue(c, people) |
| 570 | } |
| 571 | |
| 572 | func checkPublicValue(v any, people People) error { |
| 573 | raw, err := json.Marshal(v) |
| 574 | if err != nil { |
| 575 | return err |
| 576 | } |
| 577 | var tree any |
| 578 | if err := json.Unmarshal(raw, &tree); err != nil { |
| 579 | return err |
| 580 | } |
| 581 | var texts []string |
| 582 | collectStrings(tree, &texts) |
| 583 | for _, t := range texts { |
| 584 | for _, form := range decodedForms(t) { |
| 585 | if err := checkPublicText(strings.ToLower(form), people); err != nil { |
| 586 | return err |
| 587 | } |
| 588 | } |
| 589 | } |
| 590 | return nil |
| 591 | } |
| 592 | |
| 593 | func collectStrings(v any, out *[]string) { |
| 594 | switch x := v.(type) { |
| 595 | case string: |
| 596 | *out = append(*out, x) |
| 597 | case []any: |
| 598 | for _, e := range x { |
| 599 | collectStrings(e, out) |
| 600 | } |
| 601 | case map[string]any: |
| 602 | for _, e := range x { |
| 603 | collectStrings(e, out) |
| 604 | } |
| 605 | } |
| 606 | } |
| 607 | |
| 608 | var ( |
| 609 | mdEscapeRe = regexp.MustCompile(`\\([!-/:-@\[-` + "`" + `{-~])`) |
| 610 | percentRe = regexp.MustCompile(`%[0-9a-fA-F]{2}`) |
| 611 | ) |
| 612 | |
| 613 | // decodedForms is a string as stored and as a reader would see it once |
| 614 | // markdown and the browser have decoded it: HTML entities, markdown |
| 615 | // backslash escapes and percent-encoding undone, each alone and all |
| 616 | // together. "ops@example.org" is an address on the page. |
| 617 | func decodedForms(s string) []string { |
| 618 | ent := html.UnescapeString(s) |
| 619 | md := mdEscapeRe.ReplaceAllString(s, "$1") |
| 620 | pct := percentRe.ReplaceAllStringFunc(s, func(m string) string { |
| 621 | if b, err := strconv.ParseUint(m[1:], 16, 8); err == nil { |
| 622 | return string(rune(b)) |
| 623 | } |
| 624 | return m |
| 625 | }) |
| 626 | all := percentRe.ReplaceAllStringFunc(mdEscapeRe.ReplaceAllString(html.UnescapeString(html.UnescapeString(s)), "$1"), func(m string) string { |
| 627 | if b, err := strconv.ParseUint(m[1:], 16, 8); err == nil { |
| 628 | return string(rune(b)) |
| 629 | } |
| 630 | return m |
| 631 | }) |
| 632 | // What a browser makes of markdown emphasis and dot segments: |
| 633 | // "**ops@example**.org" reads as an address, and |
| 634 | // "carlosframework/./platform" resolves to the private repo. |
| 635 | plain := githubPathRe.ReplaceAllStringFunc(emphasisRe.ReplaceAllString(all, ""), func(m string) string { |
| 636 | return "github.com" + path.Clean(m[len("github.com"):]) |
| 637 | }) |
| 638 | return []string{s, ent, md, pct, all, plain} |
| 639 | } |
| 640 | |
| 641 | var ( |
| 642 | emphasisRe = regexp.MustCompile("[*_`~]+") |
| 643 | githubPathRe = regexp.MustCompile(`(?i)github\.com/[^\s)"'<>\]]*`) |
| 644 | ) |
| 645 | |
| 646 | func checkPublicText(s string, people People) error { |
| 647 | for addr := range people { |
| 648 | if strings.Contains(addr, "@") && strings.Contains(s, strings.ToLower(addr)) { |
| 649 | return fmt.Errorf("would publish the address %s", addr) |
| 650 | } |
| 651 | } |
| 652 | if m := keymailRe.FindString(s); m != "" { |
| 653 | return fmt.Errorf("would publish the address %s", m) |
| 654 | } |
| 655 | if m := addressRe.FindString(s); m != "" { |
| 656 | return fmt.Errorf("would publish the address %s", m) |
| 657 | } |
| 658 | if m := privateRe.FindString(s); m != "" { |
| 659 | return fmt.Errorf("would link the private repo %s", m) |
| 660 | } |
| 661 | return nil |
| 662 | } |
| 663 | |
| 664 | func capped(s []string) []string { |
| 665 | if len(s) > maxDocs { |
| 666 | return s[:maxDocs] |
| 667 | } |
| 668 | return s |
| 669 | } |
| 670 | |
| 671 | var mergeSubjectRe = regexp.MustCompile(`^(Merge (branch|pull request|remote-tracking branch) |Squashed from )`) |
| 672 | |
| 673 | // Describe is the line a change shows: the landing commit's subject, |
| 674 | // unless that only names the branch ("fix-x", "Merge branch 'fix-x' |
| 675 | // into main"). Then the first line of its body that says something (a |
| 676 | // GitHub PR title, an Amadan merge's description), and failing that the |
| 677 | // subject of the branch's first commit. Every option is a verbatim line |
| 678 | // from the history. |
| 679 | func Describe(message, branch string, own []Commit) string { |
| 680 | lines := strings.Split(message, "\n") |
| 681 | informative := func(l string) bool { |
| 682 | l = strings.TrimSpace(l) |
| 683 | return l != "" && l != branch && !mergeSubjectRe.MatchString(l) && !trailerRe.MatchString(l) |
| 684 | } |
| 685 | if informative(lines[0]) { |
| 686 | return strings.TrimSpace(lines[0]) |
| 687 | } |
| 688 | for _, l := range lines[1:] { |
| 689 | if informative(l) { |
| 690 | return strings.TrimSpace(l) |
| 691 | } |
| 692 | } |
| 693 | for i := len(own) - 1; i >= 0; i-- { |
| 694 | if s := subject(own[i].Message); informative(s) { |
| 695 | return s |
| 696 | } |
| 697 | } |
| 698 | return strings.TrimSpace(lines[0]) |
| 699 | } |
| 700 | |
| 701 | var trailerRe = regexp.MustCompile(`^[A-Za-z-]+: .*<[^>]+>$`) |
| 702 | |
| 703 | func subject(msg string) string { |
| 704 | if i := strings.IndexByte(msg, '\n'); i >= 0 { |
| 705 | msg = msg[:i] |
| 706 | } |
| 707 | return strings.TrimSpace(msg) |
| 708 | } |
| 709 | |
| 710 | func day(t time.Time) string { return t.UTC().Format("2006-01-02") } |
| 711 | |
| 712 | func short(sha string) string { |
| 713 | if len(sha) > 8 { |
| 714 | return sha[:8] |
| 715 | } |
| 716 | return sha |
| 717 | } |
| 718 | |
| 719 | // semverLess orders vX.Y.Z tags numerically. |
| 720 | func semverLess(a, b string) bool { |
| 721 | pa, pb := semverParts(a), semverParts(b) |
| 722 | for i := range pa { |
| 723 | if pa[i] != pb[i] { |
| 724 | return pa[i] < pb[i] |
| 725 | } |
| 726 | } |
| 727 | return a < b |
| 728 | } |
| 729 | |
| 730 | func semverParts(tag string) [3]int { |
| 731 | var out [3]int |
| 732 | for i, p := range strings.SplitN(strings.TrimPrefix(tag, "v"), ".", 3) { |
| 733 | out[i], _ = strconv.Atoi(p) |
| 734 | } |
| 735 | return out |
| 736 | } |
| 737 | |
amadan