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 mux *http.ServeMux
112 // ownerRepo is the owner login as a repository name, zero when it is not
113 // a valid one (then no repository can be named like the owner).
114 ownerRepo domain.RepoName
115 // refsFan shares the fragments of the refs page between the watchers of
116 // a push; entries are few and short-lived.
117 refsFan fanout
118}
119
120// New builds the web handler.
121func New(cfg Config) (*Server, error) {
122 if cfg.Repos == nil || cfg.Browse == nil {
123 return nil, errors.New("web: the repository and browse use cases are required")
124 }
125 u, err := url.Parse(cfg.PublicURL)
126 if err != nil || u.Host == "" {
127 return nil, fmt.Errorf("web: public URL %q is not a valid base URL", cfg.PublicURL)
128 }
129 if cfg.Now == nil {
130 cfg.Now = time.Now
131 }
132 assets, err := loadAssets()
133 if err != nil {
134 return nil, fmt.Errorf("web: load assets: %w", err)
135 }
136 if cfg.Sessions != nil && cfg.Limiter == nil {
137 return nil, errors.New("web: sessions need a failure limiter")
138 }
139 s := &Server{
140 cfg: cfg, instance: u.Host, assets: assets, md: newMarkdown(),
141 pathCache: newPathCache(pathCacheBytes, pathBuilds),
142 origin: originOf(u),
143 }
144 s.ownerRepo, _ = domain.ParseRepoName(cfg.Owner)
145 s.mux = s.routes()
146 return s, nil
147}
148
149// routes is the router of the web. Any path that is not a page is the 404
150// page.
151func (s *Server) routes() *http.ServeMux {
152 mux := http.NewServeMux()
153 mux.HandleFunc("GET "+StaticPrefix, s.assets.serve)
154 mux.HandleFunc("GET /-/events", s.events)
155 if s.cfg.Sessions != nil {
156 mux.HandleFunc("GET "+loginPrefix+"{nonce}", s.loginConfirm)
157 mux.HandleFunc("POST "+loginPrefix+"{nonce}", s.loginPost)
158 mux.HandleFunc("POST /-/logout", s.logout)
159 }
160 if s.passkeysOn() {
161 s.passkeyRoutes(mux)
162 }
163 mux.HandleFunc("GET /{$}", s.page(s.home))
164 mux.HandleFunc("GET /{repo}", s.page(s.summary))
165 mux.HandleFunc("GET /{repo}/{$}", s.page(s.summary))
166 mux.HandleFunc("GET /{repo}/refs", s.page(s.refs))
167 mux.HandleFunc("GET /{repo}/commits", s.page(s.commits))
168 mux.HandleFunc("GET /{repo}/commits/{rest...}", s.page(s.commits))
169 mux.HandleFunc("GET /{repo}/commit/{sha}", s.page(s.commit))
170 mux.HandleFunc("GET /{repo}/tree", s.page(s.tree))
171 mux.HandleFunc("GET /{repo}/tree/{rest...}", s.page(s.tree))
172 mux.HandleFunc("GET /{repo}/blob/{rest...}", s.page(s.blob))
173 mux.HandleFunc("GET /{repo}/raw/{rest...}", s.raw)
174 mux.HandleFunc("GET /{repo}/paths/{sha}", s.paths)
175 mux.HandleFunc("GET /{repo}/context/{sha}", s.page(s.diffContext))
176 if s.cfg.MergeRequests != nil {
177 mux.HandleFunc("GET /{repo}/merge-requests", s.page(s.mrList))
178 mux.HandleFunc("GET /{repo}/merge-requests/{$}", s.page(s.mrList))
179 mux.HandleFunc("GET /{repo}/merge-requests/{n}", s.page(s.mrShow))
180 mux.HandleFunc("GET /{repo}/merge-requests/{n}/since", s.page(s.mrSince))
181 mux.HandleFunc("GET /{repo}/merge-requests/{n}/files", s.page(s.mrFiles))
182 mux.HandleFunc("GET /{repo}/compare/{rest...}", s.page(s.compare))
183 // The writes: POST only, behind the gate (ServeHTTP).
184 mux.HandleFunc("POST /{repo}/merge-requests", s.mrAction(s.mrCreate))
185 mux.HandleFunc("POST /{repo}/merge-requests/{n}/comments", s.mrAction(s.mrComment))
186 mux.HandleFunc("POST /{repo}/merge-requests/{n}/comments/{id}/resolve", s.mrAction(s.mrResolve))
187 mux.HandleFunc("POST /{repo}/merge-requests/{n}/merge", s.mrAction(s.mrMerge))
188 mux.HandleFunc("POST /{repo}/merge-requests/{n}/close", s.mrAction(s.mrClose))
189 mux.HandleFunc("POST /{repo}/merge-requests/{n}/ready", s.mrAction(s.mrReady))
190 mux.HandleFunc("POST /{repo}/merge-requests/{n}/read", s.mrAction(s.mrRead))
191 }
192 if s.cfg.Runs != nil {
193 mux.HandleFunc("GET /{repo}/actions", s.page(s.actions))
194 mux.HandleFunc("GET /{repo}/actions/{$}", s.page(s.actions))
195 mux.HandleFunc("GET /{repo}/actions/runs/{id}", s.page(s.run))
196 mux.HandleFunc("GET /{repo}/actions/runs/{id}/job/{job}", s.page(s.job))
197 mux.HandleFunc("GET /{repo}/actions/runs/{id}/job/{job}/raw", s.rawLog)
198 mux.HandleFunc("GET /{repo}/actions/runs/{id}/job/{job}/stream", s.logStream)
199 mux.HandleFunc("POST /{repo}/actions/runs/{id}/cancel", s.ciAction(s.runCancel))
200 mux.HandleFunc("POST /{repo}/actions/runs/{id}/rerun", s.ciAction(s.runRerun(false)))
201 mux.HandleFunc("POST /{repo}/actions/runs/{id}/rerun-failed", s.ciAction(s.runRerun(true)))
202 if s.cfg.Artifacts != nil {
203 mux.HandleFunc("GET /{repo}/actions/runs/{id}/artifacts/{artifact}", s.artifactDownload)
204 mux.HandleFunc("POST /{repo}/actions/runs/{id}/artifacts/{artifact}/delete", s.ciAction(s.artifactDelete))
205 }
206 }
207 mux.HandleFunc(catchAll, s.page(s.notFound))
208 return mux
209}
210
211// ServeHTTP implements http.Handler.
212func (s *Server) ServeHTTP(w http.ResponseWriter, r *http.Request) {
213 // Whatever the context carried, the caller is anonymous until a valid
214 // session says otherwise (withSession).
215 r = r.WithContext(domain.ContextWithPrincipal(r.Context(), domain.Anonymous()))
216 r = s.withSession(w, r)
217 if !s.gate(w, r) {
218 return
219 }
220 if to, ok := s.ownerRedirect(r); ok {
221 http.Redirect(w, r, to, http.StatusPermanentRedirect)
222 return
223 }
224 s.mux.ServeHTTP(w, r)
225}
226
227// catchAll is the pattern of the paths that are no page.
228const catchAll = "GET /{repo}/{rest...}"
229
230// ownerRedirect maps /<owner>/<repo>[.git][/rest] to /<repo>[.git][/rest]: the
231// URLs that the github context of CI shows (html_url, commit and compare
232// URLs, clone_url) have the owner of the instance in front, like GitHub's.
233//
234// The rule is by repository, not by route: when no repository named like the
235// owner is visible to the caller, every GET or HEAD of /<owner>/<rest> is
236// redirected, so /<owner>/actions reaches the repository named "actions". A
237// repository named like the owner keeps its own pages. A private one is seen
238// as missing by an anonymous caller (the use case answers ErrNotFound), so the
239// redirect is the same as for a name nobody has and tells nothing.
240//
241// The target is built from a segment that parses as a repository name, which
242// has no slash, backslash, percent sign or dot-dot: it can only be a path on
243// this origin, never a "//host" one.
244func (s *Server) ownerRedirect(r *http.Request) (string, bool) {
245 if s.cfg.Owner == "" || (r.Method != http.MethodGet && r.Method != http.MethodHead) {
246 return "", false
247 }
248 first, rest, ok := strings.Cut(strings.TrimPrefix(r.URL.EscapedPath(), "/"), "/")
249 if !ok || !domain.OwnerMatches(s.cfg.Owner, first) || rest == "" {
250 return "", false
251 }
252 repo, _, _ := strings.Cut(rest, "/")
253 if _, err := domain.ParseRepoPath(repo); err != nil {
254 return "", false
255 }
256 if s.ownerNamesARepo(r) {
257 return "", false
258 }
259 to := "/" + rest
260 if r.URL.RawQuery != "" {
261 to += "?" + r.URL.RawQuery
262 }
263 return to, true
264}
265
266// ownerNamesARepo tells whether the caller can see a repository named like
267// the owner. A lookup that fails otherwise counts as yes: the request is then
268// served as it was before, without a redirect.
269func (s *Server) ownerNamesARepo(r *http.Request) bool {
270 if s.ownerRepo.IsZero() {
271 return false // the owner login is not a repository name
272 }
273 ctx := r.Context()
274 _, err := s.cfg.Repos.Get(ctx, principal(ctx), s.ownerRepo)
275 switch {
276 case err == nil:
277 return true
278 case errors.Is(err, domain.ErrNotFound):
279 return false
280 }
281 log := zerowrap.FromCtx(ctx)
282 log.Debug().Err(err).Msg("owner redirect: the lookup of the repository named like the owner failed")
283 return true
284}
285
286// handler builds a page. It returns the component and its status, or an
287// error, which becomes the 404 or 500 page.
288type handler func(ctx context.Context, r *http.Request) (templ.Component, int, error)
289
290// page adapts a handler: it bounds the use case calls, renders the component into
291// a buffer (so that a failure never leaves half a page), and maps errors to
292// the state pages.
293func (s *Server) page(h handler) http.HandlerFunc {
294 return func(w http.ResponseWriter, r *http.Request) {
295 ctx, cancel := context.WithTimeout(r.Context(), requestTimeout)
296 defer cancel()
297 c, status, err := h(ctx, r)
298 if err != nil {
299 s.fail(ctx, w, r, err)
300 return
301 }
302 s.write(ctx, w, r, status, c)
303 }
304}
305
306// write renders c with the given status.
307func (s *Server) write(ctx context.Context, w http.ResponseWriter, r *http.Request, status int, c templ.Component) {
308 var buf bytes.Buffer
309 if err := c.Render(ctx, &buf); err != nil {
310 // The error page itself failed: a bare answer is all that is left.
311 log := zerowrap.FromCtx(ctx)
312 log.Error().Err(err).Str("path", logPath(r.URL.Path)).Msg("render page")
313 http.Error(w, "internal error", http.StatusInternalServerError)
314 return
315 }
316 h := w.Header()
317 h.Set("Content-Type", "text/html; charset=utf-8")
318 // Links on this origin are prefetched on hover, so a click shows the page
319 // at once. Every GET page is safe to load (nothing is marked read on GET).
320 h.Set("Speculation-Rules", `"`+s.assets.urls.Rules+`"`)
321 // Public pages: the browser and any proxy revalidate each time. A page
322 // rendered for the owner is private and never stored; a handler that set
323 // its own Cache-Control keeps it.
324 if h.Get("Cache-Control") == "" {
325 cacheHeaders(ctx, h, "no-cache", sessionCache)
326 }
327 body := buf.Bytes()
328 // Whether the body is sent gzipped follows from its size and the request
329 // alone, so that the tag is known before the compression is spent.
330 gz := len(body) >= minCompress && acceptsGzip(r.Header) &&
331 !noTransform(h) && h.Get("Content-Encoding") == ""
332 addVary(h, "Accept-Encoding")
333 readOnly := r.Method == http.MethodGet || r.Method == http.MethodHead
334 // A page that may be stored gets a validator: the browser then
335 // revalidates with If-None-Match and a page that did not change costs a
336 // 304. A no-store page (the owner's) is never revalidated.
337 if readOnly && status == http.StatusOK && h.Get("ETag") == "" && !hasCacheDirective(h, noStore) {
338 tag := pageETag(body, SignedIn(ctx), gz)
339 h.Set("ETag", tag)
340 if etagMatches(strings.Join(r.Header.Values("If-None-Match"), ","), tag) {
341 h.Del("Content-Type")
342 w.WriteHeader(http.StatusNotModified)
343 return
344 }
345 }
346 if gz {
347 if z, err := gzipPage(body); err == nil && len(z) < len(body) {
348 body = z
349 h.Set("Content-Encoding", "gzip")
350 }
351 }
352 h.Set("Content-Length", itoa(len(body)))
353 w.WriteHeader(status)
354 if r.Method != http.MethodHead {
355 _, _ = w.Write(body)
356 }
357}
358
359// fail answers an error of a use case: not found (and a name or revision the
360// use case refuses, which no page can show either) gives the 404 page, a
361// visitor that left gives nothing, anything else gives the 500 page with an id
362// to find the cause in the log.
363func (s *Server) fail(ctx context.Context, w http.ResponseWriter, r *http.Request, err error) {
364 switch {
365 case errors.Is(err, domain.ErrNotFound), domain.IsInvalid(err):
366 s.write(ctx, w, r, http.StatusNotFound, s.notFoundPage())
367 case errors.Is(err, context.Canceled):
368 // The client is gone; there is nobody to answer.
369 default:
370 id := requestID()
371 log := zerowrap.FromCtx(ctx)
372 log.Error().Err(err).Str("request_id", id).Str("path", logPath(r.URL.Path)).Msg("web request failed")
373 s.write(ctx, w, r, http.StatusInternalServerError, s.errorPage(id))
374 }
375}
376
377// errNotFound is returned by handlers for a resource that does not exist
378// although the use case did not say so (a ref or a path that is not there).
379var errNotFound = domain.ErrNotFound
380
381// requestID is a short random id, shown on the 500 page and logged.
382func requestID() string {
383 var b [4]byte
384 _, _ = rand.Read(b[:])
385 return hex.EncodeToString(b[:])
386}
387
388// notFound is the handler of the paths without a page.
389func (s *Server) notFound(context.Context, *http.Request) (templ.Component, int, error) {
390 return nil, 0, errNotFound
391}
392
393// base is the Layout every page starts from.
394func (s *Server) base() Layout {
395 return Layout{
396 Owner: s.cfg.Owner,
397 Version: s.cfg.Version,
398 SourceURL: s.cfg.SourceURL,
399 Assets: s.assets.urls,
400 Passkeys: s.passkeysOn(),
401 }
402}
403
404// stateLayout is the layout of a state page: only the way home.
405func (s *Server) stateLayout(title string, crumbs []Crumb) Layout {
406 l := s.base()
407 l.Title = title + " · " + s.instance
408 l.Crumbs = crumbs
409 l.Cmd = []CmdItem{{Kind: "go", Label: "home", Href: "/"}}
410 return l
411}
412
413func (s *Server) notFoundPage() templ.Component {
414 l := s.stateLayout("404", []Crumb{{Label: "not found", Current: true}})
415 return State(l, StatePage{
416 Code: "404", Title: "not found",
417 Lines: []string{"no such page, ref or path.", "check the URL, or start from the repo list."},
418 Back: []Link{{Label: "home", Href: "/"}},
419 })
420}
421
422func (s *Server) errorPage(id string) templ.Component {
423 l := s.stateLayout("error", []Crumb{{Label: "error", Current: true}})
424 return State(l, StatePage{
425 Code: "500", Title: "internal error",
426 Lines: []string{
427 "the server failed to answer this request.",
428 "request id: " + id,
429 "retry in a moment; if it persists, the owner has the log.",
430 },
431 Back: []Link{{Label: "home", Href: "/"}},
432 })
433}
434
435// originOf is the origin of a URL as a browser writes it in the Origin
436// header: scheme and host in lower case, the port only when it is not the
437// default of the scheme.
438func originOf(u *url.URL) string {
439 scheme, host, port := strings.ToLower(u.Scheme), strings.ToLower(u.Hostname()), u.Port()
440 if strings.Contains(host, ":") {
441 host = "[" + host + "]"
442 }
443 defaultPort := (scheme == "http" && port == "80") || (scheme == "https" && port == "443")
444 if port != "" && !defaultPort {
445 host += ":" + port
446 }
447 return scheme + "://" + host
448}