1// SPDX-License-Identifier: AGPL-3.0-or-later
2
3// Package web is the HTML front of ermite: the pages of the read-only
4// showcase, rendered with templ from the data of the use cases.
5//
6// Like every other inbound adapter, the package reads and writes through the
7// use case interfaces of package ports (ADR 0024), never through the API. The
8// caller of every use case call is passed explicitly: the principal of the
9// request, which is anonymous unless the request carries a valid owner session
10// (ServeHTTP sets it in the context, principal reads it). The use cases then
11// show public repositories only to an anonymous caller; a private repository
12// is indistinguishable from a missing one (domain.ErrNotFound), here as
13// everywhere.
14//
15// Adding a page: a handler on Server, a route in routes, a templ page in
16// pages.templ and its view model in view.go.
17package web
18
19import (
20 "bytes"
21 "context"
22 "crypto/rand"
23 "encoding/hex"
24 "errors"
25 "fmt"
26 "net/http"
27 "net/netip"
28 "net/url"
29 "strings"
30 "time"
31
32 "github.com/a-h/templ"
33 "github.com/bnema/zerowrap"
34
35 "bnema.dev/ermite/internal/domain"
36 "bnema.dev/ermite/internal/ports"
37)
38
39// Patterns are the ServeMux patterns the handler serves: everything. The
40// router of the package answers the paths it does not know with its 404
41// page, and the more specific patterns of the server (health, API, Git) win
42// over this one.
43var Patterns = []string{"/"}
44
45// requestTimeout bounds the use case calls that build one page.
46const requestTimeout = 30 * time.Second
47
48// Config configures a Server.
49type Config struct {
50 // Repos and Browse are the use cases the pages are built from.
51 Repos ports.RepoService
52 Browse ports.BrowseService
53 // MergeRequests serves the merge request pages (ADR 0017). Without it
54 // there are no such pages, nor section for them. Leave it unset rather
55 // than assigning a typed nil pointer.
56 MergeRequests ports.MergeRequestService
57 // Runs serves the CI pages (/<repo>/actions) and the checks shown next to
58 // commits and merge requests. Without it there are no such pages. Leave it
59 // unset rather than assigning a typed nil pointer.
60 Runs ports.RunService
61 // Artifacts serves the artifacts of the runs: a section of the run page,
62 // their download and their deletion. It needs Runs. Without it there is
63 // neither. Leave it unset rather than assigning a typed nil pointer.
64 Artifacts ports.ArtifactService
65 // Events is the source of the event stream behind /-/events. Without
66 // it the pages are not live and the route answers 404. Leave it unset
67 // rather than assigning a typed nil pointer.
68 Events ports.EventService
69 // Heartbeat is how often an idle event stream gets a comment line;
70 // DefaultHeartbeat when zero.
71 Heartbeat time.Duration
72 // Owner is the owner login the GitHub-shaped URLs of CI show
73 // (<server>/<owner>/<repo>/...). A GET or HEAD of /<owner>/<rest> is
74 // redirected to /<rest> unless a repository named like the owner is
75 // visible to the caller. Empty: no redirect.
76 Owner string
77 // PublicURL is the base URL of the instance, without a trailing slash:
78 // the clone URLs are built on it, and its host is the instance name.
79 PublicURL string
80 // SourceURL is the link to the source of the running version (AGPL §13).
81 SourceURL string
82 // Version is the version of the running server.
83 Version string
84 // Now is the clock of relative dates; time.Now when nil.
85 Now func() time.Time
86 // Sessions signs the owner in (ADR 0018). Without it there are no login
87 // routes and every visitor is anonymous.
88 Sessions ports.WebSessionService
89 // Passkeys signs the owner in with a passkey and manages the passkeys
90 // (ADR 0026). Without it, or without Sessions, there are no passkey
91 // routes and no link to them. Leave it unset rather than assigning a typed
92 // nil pointer.
93 Passkeys ports.WebPasskeyService
94 // Limiter counts the failed logins; required with Sessions. Share it with
95 // the API and Git, so that a client has one budget of failures.
96 Limiter ports.FailureLimiter
97 // TrustedProxies are the networks allowed to set X-Forwarded-For, to find
98 // the address of the client that fails a login.
99 TrustedProxies []netip.Prefix
100}
101
102// Server serves the pages. It is an http.Handler.
103type Server struct {
104 cfg Config
105 instance string
106 origin string // scheme://host of the public URL, what a POST must come from
107 assets *assetSet
108 md *markdown
109 // pathCache holds the path lists the file finder fetches.
110 pathCache *pathCache
111 // hl holds the colours of highlighted diffs.
112 hl *hlCache
113 mux *http.ServeMux
114 // ownerRepo is the owner login as a repository name, zero when it is not
115 // a valid one (then no repository can be named like the owner).
116 ownerRepo domain.RepoName
117 // refsFan shares the fragments of the refs page between the watchers of
118 // a push; entries are few and short-lived.
119 refsFan fanout
120 // pulses holds the activity curves of the home rows.
121 pulses pulseCache
122}
123
124// New builds the web handler.
125func New(cfg Config) (*Server, error) {
126 if cfg.Repos == nil || cfg.Browse == nil {
127 return nil, errors.New("web: the repository and browse use cases are required")
128 }
129 u, err := url.Parse(cfg.PublicURL)
130 if err != nil || u.Host == "" {
131 return nil, fmt.Errorf("web: public URL %q is not a valid base URL", cfg.PublicURL)
132 }
133 if cfg.Now == nil {
134 cfg.Now = time.Now
135 }
136 assets, err := loadAssets()
137 if err != nil {
138 return nil, fmt.Errorf("web: load assets: %w", err)
139 }
140 if cfg.Sessions != nil && cfg.Limiter == nil {
141 return nil, errors.New("web: sessions need a failure limiter")
142 }
143 s := &Server{
144 cfg: cfg, instance: u.Host, assets: assets, md: newMarkdown(),
145 pathCache: newPathCache(pathCacheBytes, pathBuilds),
146 hl: newHLCache(hlCacheBytes),
147 origin: originOf(u),
148 }
149 s.ownerRepo, _ = domain.ParseRepoName(cfg.Owner)
150 s.mux = s.routes()
151 return s, nil
152}
153
154// routes is the router of the web. Any path that is not a page is the 404
155// page.
156func (s *Server) routes() *http.ServeMux {
157 mux := http.NewServeMux()
158 mux.HandleFunc("GET "+StaticPrefix, s.assets.serve)
159 mux.HandleFunc("GET /-/events", s.events)
160 mux.HandleFunc("GET "+ServiceWorkerPath, s.assets.sw.serve)
161 if s.cfg.Sessions != nil {
162 mux.HandleFunc("GET "+loginPrefix+"{nonce}", s.loginConfirm)
163 mux.HandleFunc("POST "+loginPrefix+"{nonce}", s.loginPost)
164 mux.HandleFunc("POST /-/logout", s.logout)
165 }
166 if s.passkeysOn() {
167 s.passkeyRoutes(mux)
168 }
169 mux.HandleFunc("GET /{$}", s.page(s.home))
170 mux.HandleFunc("GET /{repo}", s.page(s.summary))
171 mux.HandleFunc("GET /{repo}/{$}", s.page(s.summary))
172 mux.HandleFunc("GET /{repo}/refs", s.page(s.refs))
173 mux.HandleFunc("GET /{repo}/commits", s.page(s.commits))
174 mux.HandleFunc("GET /{repo}/commits/{rest...}", s.page(s.commits))
175 mux.HandleFunc("GET /{repo}/commit/{sha}", s.page(s.commit))
176 mux.HandleFunc("GET /{repo}/tree", s.page(s.tree))
177 mux.HandleFunc("GET /{repo}/tree/{rest...}", s.page(s.tree))
178 mux.HandleFunc("GET /{repo}/blob/{rest...}", s.page(s.blob))
179 mux.HandleFunc("GET /{repo}/raw/{rest...}", s.raw)
180 mux.HandleFunc("GET /{repo}/paths/{sha}", s.paths)
181 mux.HandleFunc("GET /{repo}/context/{sha}", s.page(s.diffContext))
182 if s.cfg.MergeRequests != nil {
183 mux.HandleFunc("GET /{repo}/merge-requests", s.page(s.mrList))
184 mux.HandleFunc("GET /{repo}/merge-requests/{$}", s.page(s.mrList))
185 mux.HandleFunc("GET /{repo}/merge-requests/{n}", s.page(s.mrShow))
186 mux.HandleFunc("GET /{repo}/merge-requests/{n}/since", s.page(s.mrSince))
187 mux.HandleFunc("GET /{repo}/merge-requests/{n}/files", s.page(s.mrFiles))
188 mux.HandleFunc("GET /{repo}/compare/{rest...}", s.page(s.compare))
189 // The writes: POST only, behind the gate (ServeHTTP).
190 mux.HandleFunc("POST /{repo}/merge-requests", s.mrAction(s.mrCreate))
191 mux.HandleFunc("POST /{repo}/merge-requests/{n}/comments", s.mrAction(s.mrComment))
192 mux.HandleFunc("POST /{repo}/merge-requests/{n}/comments/{id}/resolve", s.mrAction(s.mrResolve))
193 mux.HandleFunc("POST /{repo}/merge-requests/{n}/merge", s.mrAction(s.mrMerge))
194 mux.HandleFunc("POST /{repo}/merge-requests/{n}/close", s.mrAction(s.mrClose))
195 mux.HandleFunc("POST /{repo}/merge-requests/{n}/ready", s.mrAction(s.mrReady))
196 mux.HandleFunc("POST /{repo}/merge-requests/{n}/read", s.mrAction(s.mrRead))
197 }
198 if s.cfg.Runs != nil {
199 mux.HandleFunc("GET /{repo}/actions", s.page(s.actions))
200 mux.HandleFunc("GET /{repo}/actions/{$}", s.page(s.actions))
201 mux.HandleFunc("GET /{repo}/actions/runs/{id}", s.page(s.run))
202 mux.HandleFunc("GET /{repo}/actions/runs/{id}/job/{job}", s.page(s.job))
203 mux.HandleFunc("GET /{repo}/actions/runs/{id}/job/{job}/raw", s.rawLog)
204 mux.HandleFunc("GET /{repo}/actions/runs/{id}/job/{job}/stream", s.logStream)
205 mux.HandleFunc("POST /{repo}/actions/runs/{id}/cancel", s.ciAction(s.runCancel))
206 mux.HandleFunc("POST /{repo}/actions/runs/{id}/rerun", s.ciAction(s.runRerun(false)))
207 mux.HandleFunc("POST /{repo}/actions/runs/{id}/rerun-failed", s.ciAction(s.runRerun(true)))
208 if s.cfg.Artifacts != nil {
209 mux.HandleFunc("GET /{repo}/actions/runs/{id}/artifacts/{artifact}", s.artifactDownload)
210 mux.HandleFunc("POST /{repo}/actions/runs/{id}/artifacts/{artifact}/delete", s.ciAction(s.artifactDelete))
211 }
212 }
213 mux.HandleFunc(catchAll, s.page(s.notFound))
214 return mux
215}
216
217// ServeHTTP implements http.Handler.
218func (s *Server) ServeHTTP(w http.ResponseWriter, r *http.Request) {
219 // Whatever the context carried, the caller is anonymous until a valid
220 // session says otherwise (withSession).
221 r = r.WithContext(domain.ContextWithPrincipal(r.Context(), domain.Anonymous()))
222 r = s.withSession(w, r)
223 if !s.gate(w, r) {
224 return
225 }
226 if to, ok := s.ownerRedirect(r); ok {
227 http.Redirect(w, r, to, http.StatusPermanentRedirect)
228 return
229 }
230 s.mux.ServeHTTP(w, r)
231}
232
233// catchAll is the pattern of the paths that are no page.
234const catchAll = "GET /{repo}/{rest...}"
235
236// ownerRedirect maps /<owner>/<repo>[.git][/rest] to /<repo>[.git][/rest]: the
237// URLs that the github context of CI shows (html_url, commit and compare
238// URLs, clone_url) have the owner of the instance in front, like GitHub's.
239//
240// The rule is by repository, not by route: when no repository named like the
241// owner is visible to the caller, every GET or HEAD of /<owner>/<rest> is
242// redirected, so /<owner>/actions reaches the repository named "actions". A
243// repository named like the owner keeps its own pages. A private one is seen
244// as missing by an anonymous caller (the use case answers ErrNotFound), so the
245// redirect is the same as for a name nobody has and tells nothing.
246//
247// The target is built from a segment that parses as a repository name, which
248// has no slash, backslash, percent sign or dot-dot: it can only be a path on
249// this origin, never a "//host" one.
250func (s *Server) ownerRedirect(r *http.Request) (string, bool) {
251 if s.cfg.Owner == "" || (r.Method != http.MethodGet && r.Method != http.MethodHead) {
252 return "", false
253 }
254 first, rest, ok := strings.Cut(strings.TrimPrefix(r.URL.EscapedPath(), "/"), "/")
255 if !ok || !domain.OwnerMatches(s.cfg.Owner, first) || rest == "" {
256 return "", false
257 }
258 repo, _, _ := strings.Cut(rest, "/")
259 if _, err := domain.ParseRepoPath(repo); err != nil {
260 return "", false
261 }
262 if s.ownerNamesARepo(r) {
263 return "", false
264 }
265 to := "/" + rest
266 if r.URL.RawQuery != "" {
267 to += "?" + r.URL.RawQuery
268 }
269 return to, true
270}
271
272// ownerNamesARepo tells whether the caller can see a repository named like
273// the owner. A lookup that fails otherwise counts as yes: the request is then
274// served as it was before, without a redirect.
275func (s *Server) ownerNamesARepo(r *http.Request) bool {
276 if s.ownerRepo.IsZero() {
277 return false // the owner login is not a repository name
278 }
279 ctx := r.Context()
280 _, err := s.cfg.Repos.Get(ctx, principal(ctx), s.ownerRepo)
281 switch {
282 case err == nil:
283 return true
284 case errors.Is(err, domain.ErrNotFound):
285 return false
286 }
287 log := zerowrap.FromCtx(ctx)
288 log.Debug().Err(err).Msg("owner redirect: the lookup of the repository named like the owner failed")
289 return true
290}
291
292// handler builds a page. It returns the component and its status, or an
293// error, which becomes the 404 or 500 page.
294type handler func(ctx context.Context, r *http.Request) (templ.Component, int, error)
295
296// page adapts a handler: it bounds the use case calls, renders the component into
297// a buffer (so that a failure never leaves half a page), and maps errors to
298// the state pages.
299func (s *Server) page(h handler) http.HandlerFunc {
300 return func(w http.ResponseWriter, r *http.Request) {
301 ctx, cancel := context.WithTimeout(r.Context(), requestTimeout)
302 defer cancel()
303 c, status, err := h(ctx, r)
304 if err != nil {
305 s.fail(ctx, w, r, err)
306 return
307 }
308 s.write(ctx, w, r, status, c)
309 }
310}
311
312// write renders c with the given status.
313func (s *Server) write(ctx context.Context, w http.ResponseWriter, r *http.Request, status int, c templ.Component) {
314 var buf bytes.Buffer
315 if err := c.Render(ctx, &buf); err != nil {
316 // The error page itself failed: a bare answer is all that is left.
317 log := zerowrap.FromCtx(ctx)
318 log.Error().Err(err).Str("path", logPath(r.URL.Path)).Msg("render page")
319 http.Error(w, "internal error", http.StatusInternalServerError)
320 return
321 }
322 h := w.Header()
323 h.Set("Content-Type", "text/html; charset=utf-8")
324 // Links on this origin are prefetched on hover, so a click shows the page
325 // at once. Every GET page is safe to load (nothing is marked read on GET).
326 h.Set("Speculation-Rules", `"`+s.assets.urls.Rules+`"`)
327 // Public pages: the browser and any proxy revalidate each time. A page
328 // rendered for the owner is private and never stored; a handler that set
329 // its own Cache-Control keeps it.
330 if h.Get("Cache-Control") == "" {
331 cacheHeaders(ctx, h, "no-cache", sessionCache)
332 }
333 body := buf.Bytes()
334 // Whether the body is sent gzipped follows from its size and the request
335 // alone, so that the tag is known before the compression is spent.
336 gz := len(body) >= minCompress && acceptsGzip(r.Header) &&
337 !noTransform(h) && h.Get("Content-Encoding") == ""
338 addVary(h, "Accept-Encoding")
339 readOnly := r.Method == http.MethodGet || r.Method == http.MethodHead
340 // A page that may be stored gets a validator: the browser then
341 // revalidates with If-None-Match and a page that did not change costs a
342 // 304. A no-store page (the owner's) is never revalidated.
343 if readOnly && status == http.StatusOK && h.Get("ETag") == "" && !hasCacheDirective(h, noStore) {
344 tag := pageETag(body, SignedIn(ctx), gz)
345 h.Set("ETag", tag)
346 if etagMatches(strings.Join(r.Header.Values("If-None-Match"), ","), tag) {
347 h.Del("Content-Type")
348 w.WriteHeader(http.StatusNotModified)
349 return
350 }
351 }
352 if gz {
353 if z, err := gzipPage(body); err == nil && len(z) < len(body) {
354 body = z
355 h.Set("Content-Encoding", "gzip")
356 }
357 }
358 h.Set("Content-Length", itoa(len(body)))
359 w.WriteHeader(status)
360 if r.Method != http.MethodHead {
361 _, _ = w.Write(body)
362 }
363}
364
365// fail answers an error of a use case: not found (and a name or revision the
366// use case refuses, which no page can show either) gives the 404 page, a
367// visitor that left gives nothing, anything else gives the 500 page with an id
368// to find the cause in the log.
369func (s *Server) fail(ctx context.Context, w http.ResponseWriter, r *http.Request, err error) {
370 switch {
371 case errors.Is(err, domain.ErrNotFound), domain.IsInvalid(err):
372 s.write(ctx, w, r, http.StatusNotFound, s.notFoundPage())
373 case errors.Is(err, context.Canceled):
374 // The client is gone; there is nobody to answer.
375 default:
376 id := requestID()
377 log := zerowrap.FromCtx(ctx)
378 log.Error().Err(err).Str("request_id", id).Str("path", logPath(r.URL.Path)).Msg("web request failed")
379 s.write(ctx, w, r, http.StatusInternalServerError, s.errorPage(id))
380 }
381}
382
383// errNotFound is returned by handlers for a resource that does not exist
384// although the use case did not say so (a ref or a path that is not there).
385var errNotFound = domain.ErrNotFound
386
387// requestID is a short random id, shown on the 500 page and logged.
388func requestID() string {
389 var b [4]byte
390 _, _ = rand.Read(b[:])
391 return hex.EncodeToString(b[:])
392}
393
394// notFound is the handler of the paths without a page.
395func (s *Server) notFound(context.Context, *http.Request) (templ.Component, int, error) {
396 return nil, 0, errNotFound
397}
398
399// base is the Layout every page starts from.
400func (s *Server) base() Layout {
401 return Layout{
402 Owner: s.cfg.Owner,
403 Version: s.cfg.Version,
404 SourceURL: s.cfg.SourceURL,
405 Assets: s.assets.urls,
406 Passkeys: s.passkeysOn(),
407 }
408}
409
410// stateLayout is the layout of a state page: only the way home.
411func (s *Server) stateLayout(title string, crumbs []Crumb) Layout {
412 l := s.base()
413 l.Title = title + " · " + s.instance
414 l.Crumbs = crumbs
415 l.Cmd = []CmdItem{{Kind: "go", Label: "home", Href: "/"}}
416 return l
417}
418
419func (s *Server) notFoundPage() templ.Component {
420 l := s.stateLayout("404", []Crumb{{Label: "not found", Current: true}})
421 return State(l, StatePage{
422 Code: "404", Title: "not found",
423 Lines: []string{"no such page, ref or path.", "check the URL, or start from the repo list."},
424 Back: []Link{{Label: "home", Href: "/"}},
425 })
426}
427
428func (s *Server) errorPage(id string) templ.Component {
429 l := s.stateLayout("error", []Crumb{{Label: "error", Current: true}})
430 return State(l, StatePage{
431 Code: "500", Title: "internal error",
432 Lines: []string{
433 "the server failed to answer this request.",
434 "request id: " + id,
435 "retry in a moment; if it persists, the owner has the log.",
436 },
437 Back: []Link{{Label: "home", Href: "/"}},
438 })
439}
440
441// originOf is the origin of a URL as a browser writes it in the Origin
442// header: scheme and host in lower case, the port only when it is not the
443// default of the scheme.
444func originOf(u *url.URL) string {
445 scheme, host, port := strings.ToLower(u.Scheme), strings.ToLower(u.Hostname()), u.Port()
446 if strings.Contains(host, ":") {
447 host = "[" + host + "]"
448 }
449 defaultPort := (scheme == "http" && port == "80") || (scheme == "https" && port == "443")
450 if port != "" && !defaultPort {
451 host += ":" + port
452 }
453 return scheme + "://" + host
454}