pages-server/server/handler.go

293 lines
11 KiB
Go
Raw Normal View History

2021-12-05 13:45:17 +00:00
package server
import (
"bytes"
"strings"
2021-12-05 14:02:44 +00:00
"github.com/rs/zerolog/log"
"github.com/valyala/fasthttp"
2021-12-05 13:45:17 +00:00
"codeberg.org/codeberg/pages/html"
2021-12-05 14:02:44 +00:00
"codeberg.org/codeberg/pages/server/cache"
2021-12-05 14:21:05 +00:00
"codeberg.org/codeberg/pages/server/dns"
2021-12-05 13:47:33 +00:00
"codeberg.org/codeberg/pages/server/upstream"
2021-12-03 02:44:21 +00:00
"codeberg.org/codeberg/pages/server/utils"
2021-12-05 13:45:17 +00:00
)
// Handler handles a single HTTP request to the web server.
2021-12-05 14:02:44 +00:00
func Handler(mainDomainSuffix, rawDomain []byte,
2021-12-05 17:18:05 +00:00
giteaRoot, rawInfoPage, giteaAPIToken string,
2021-12-05 14:02:44 +00:00
blacklistedPaths, allowedCorsDomains [][]byte,
2021-12-05 14:53:46 +00:00
dnsLookupCache, canonicalDomainCache, branchTimestampCache, fileResponseCache cache.SetGetKey) func(ctx *fasthttp.RequestCtx) {
2021-12-05 13:45:17 +00:00
return func(ctx *fasthttp.RequestCtx) {
log := log.With().Str("Handler", string(ctx.Request.Header.RequestURI())).Logger()
ctx.Response.Header.Set("Server", "Codeberg Pages")
// Force new default from specification (since November 2020) - see https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Referrer-Policy#strict-origin-when-cross-origin
ctx.Response.Header.Set("Referrer-Policy", "strict-origin-when-cross-origin")
// Enable browser caching for up to 10 minutes
ctx.Response.Header.Set("Cache-Control", "public, max-age=600")
2021-12-03 02:44:21 +00:00
trimmedHost := utils.TrimHostPort(ctx.Request.Host())
2021-12-05 13:45:17 +00:00
// Add HSTS for RawDomain and MainDomainSuffix
if hsts := GetHSTSHeader(trimmedHost, mainDomainSuffix, rawDomain); hsts != "" {
ctx.Response.Header.Set("Strict-Transport-Security", hsts)
}
// Block all methods not required for static pages
if !ctx.IsGet() && !ctx.IsHead() && !ctx.IsOptions() {
ctx.Response.Header.Set("Allow", "GET, HEAD, OPTIONS")
ctx.Error("Method not allowed", fasthttp.StatusMethodNotAllowed)
return
}
// Block blacklisted paths (like ACME challenges)
for _, blacklistedPath := range blacklistedPaths {
if bytes.HasPrefix(ctx.Path(), blacklistedPath) {
2021-12-05 13:47:33 +00:00
html.ReturnErrorPage(ctx, fasthttp.StatusForbidden)
2021-12-05 13:45:17 +00:00
return
}
}
// Allow CORS for specified domains
if ctx.IsOptions() {
allowCors := false
for _, allowedCorsDomain := range allowedCorsDomains {
if bytes.Equal(trimmedHost, allowedCorsDomain) {
allowCors = true
break
}
}
if allowCors {
ctx.Response.Header.Set("Access-Control-Allow-Origin", "*")
ctx.Response.Header.Set("Access-Control-Allow-Methods", "GET, HEAD")
}
ctx.Response.Header.Set("Allow", "GET, HEAD, OPTIONS")
ctx.Response.Header.SetStatusCode(fasthttp.StatusNoContent)
return
}
// Prepare request information to Gitea
var targetOwner, targetRepo, targetBranch, targetPath string
2021-12-05 13:47:33 +00:00
var targetOptions = &upstream.Options{
2021-12-05 13:45:17 +00:00
ForbiddenMimeTypes: map[string]struct{}{},
TryIndexPages: true,
}
// tryBranch checks if a branch exists and populates the target variables. If canonicalLink is non-empty, it will
// also disallow search indexing and add a Link header to the canonical URL.
var tryBranch = func(repo string, branch string, path []string, canonicalLink string) bool {
if repo == "" {
return false
}
// Check if the branch exists, otherwise treat it as a file path
2021-12-05 17:18:05 +00:00
branchTimestampResult := upstream.GetBranchTimestamp(targetOwner, repo, branch, giteaRoot, giteaAPIToken, branchTimestampCache)
2021-12-05 13:45:17 +00:00
if branchTimestampResult == nil {
// branch doesn't exist
return false
}
// Branch exists, use it
targetRepo = repo
targetPath = strings.Trim(strings.Join(path, "/"), "/")
2021-12-05 13:47:33 +00:00
targetBranch = branchTimestampResult.Branch
2021-12-05 13:45:17 +00:00
2021-12-05 13:47:33 +00:00
targetOptions.BranchTimestamp = branchTimestampResult.Timestamp
2021-12-05 13:45:17 +00:00
if canonicalLink != "" {
// Hide from search machines & add canonical link
ctx.Response.Header.Set("X-Robots-Tag", "noarchive, noindex")
ctx.Response.Header.Set("Link",
strings.NewReplacer("%b", targetBranch, "%p", targetPath).Replace(canonicalLink)+
"; rel=\"canonical\"",
)
}
return true
}
log.Debug().Msg("preparations")
if rawDomain != nil && bytes.Equal(trimmedHost, rawDomain) {
// Serve raw content from RawDomain
log.Debug().Msg("raw domain")
targetOptions.TryIndexPages = false
targetOptions.ForbiddenMimeTypes["text/html"] = struct{}{}
targetOptions.DefaultMimeType = "text/plain; charset=utf-8"
pathElements := strings.Split(string(bytes.Trim(ctx.Request.URI().Path(), "/")), "/")
if len(pathElements) < 2 {
// https://{RawDomain}/{owner}/{repo}[/@{branch}]/{path} is required
ctx.Redirect(rawInfoPage, fasthttp.StatusTemporaryRedirect)
return
}
targetOwner = pathElements[0]
targetRepo = pathElements[1]
// raw.codeberg.org/example/myrepo/@main/index.html
if len(pathElements) > 2 && strings.HasPrefix(pathElements[2], "@") {
log.Debug().Msg("raw domain preparations, now trying with specified branch")
if tryBranch(targetRepo, pathElements[2][1:], pathElements[3:],
2021-12-03 02:05:38 +00:00
giteaRoot+"/"+targetOwner+"/"+targetRepo+"/src/branch/%b/%p",
2021-12-05 13:45:17 +00:00
) {
log.Debug().Msg("tryBranch, now trying upstream")
2021-12-05 17:17:28 +00:00
tryUpstream(ctx, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
2021-12-05 17:18:05 +00:00
giteaRoot, giteaAPIToken,
2021-12-05 17:17:28 +00:00
canonicalDomainCache, branchTimestampCache, fileResponseCache)
2021-12-05 13:45:17 +00:00
return
}
log.Debug().Msg("missing branch")
2021-12-05 13:47:33 +00:00
html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
2021-12-05 13:45:17 +00:00
return
} else {
log.Debug().Msg("raw domain preparations, now trying with default branch")
tryBranch(targetRepo, "", pathElements[2:],
2021-12-03 02:05:38 +00:00
giteaRoot+"/"+targetOwner+"/"+targetRepo+"/src/branch/%b/%p",
2021-12-05 13:45:17 +00:00
)
log.Debug().Msg("tryBranch, now trying upstream")
2021-12-05 17:17:28 +00:00
tryUpstream(ctx, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
2021-12-05 17:18:05 +00:00
giteaRoot, giteaAPIToken,
2021-12-05 17:17:28 +00:00
canonicalDomainCache, branchTimestampCache, fileResponseCache)
2021-12-05 13:45:17 +00:00
return
}
} else if bytes.HasSuffix(trimmedHost, mainDomainSuffix) {
// Serve pages from subdomains of MainDomainSuffix
log.Debug().Msg("main domain suffix")
pathElements := strings.Split(string(bytes.Trim(ctx.Request.URI().Path(), "/")), "/")
targetOwner = string(bytes.TrimSuffix(trimmedHost, mainDomainSuffix))
targetRepo = pathElements[0]
targetPath = strings.Trim(strings.Join(pathElements[1:], "/"), "/")
if targetOwner == "www" {
// www.codeberg.page redirects to codeberg.page
ctx.Redirect("https://"+string(mainDomainSuffix[1:])+string(ctx.Path()), fasthttp.StatusPermanentRedirect)
return
}
// Check if the first directory is a repo with the second directory as a branch
// example.codeberg.page/myrepo/@main/index.html
if len(pathElements) > 1 && strings.HasPrefix(pathElements[1], "@") {
if targetRepo == "pages" {
// example.codeberg.org/pages/@... redirects to example.codeberg.org/@...
ctx.Redirect("/"+strings.Join(pathElements[1:], "/"), fasthttp.StatusTemporaryRedirect)
return
}
log.Debug().Msg("main domain preparations, now trying with specified repo & branch")
if tryBranch(pathElements[0], pathElements[1][1:], pathElements[2:],
"/"+pathElements[0]+"/%p",
) {
log.Debug().Msg("tryBranch, now trying upstream")
2021-12-05 17:17:28 +00:00
tryUpstream(ctx, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
2021-12-05 17:18:05 +00:00
giteaRoot, giteaAPIToken,
2021-12-05 17:17:28 +00:00
canonicalDomainCache, branchTimestampCache, fileResponseCache)
2021-12-05 13:45:17 +00:00
} else {
2021-12-05 13:47:33 +00:00
html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
2021-12-05 13:45:17 +00:00
}
return
}
// Check if the first directory is a branch for the "pages" repo
// example.codeberg.page/@main/index.html
if strings.HasPrefix(pathElements[0], "@") {
log.Debug().Msg("main domain preparations, now trying with specified branch")
if tryBranch("pages", pathElements[0][1:], pathElements[1:], "/%p") {
log.Debug().Msg("tryBranch, now trying upstream")
2021-12-05 17:17:28 +00:00
tryUpstream(ctx, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
2021-12-05 17:18:05 +00:00
giteaRoot, giteaAPIToken,
2021-12-05 17:17:28 +00:00
canonicalDomainCache, branchTimestampCache, fileResponseCache)
2021-12-05 13:45:17 +00:00
} else {
2021-12-05 13:47:33 +00:00
html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
2021-12-05 13:45:17 +00:00
}
return
}
// Check if the first directory is a repo with a "pages" branch
// example.codeberg.page/myrepo/index.html
// example.codeberg.page/pages/... is not allowed here.
log.Debug().Msg("main domain preparations, now trying with specified repo")
if pathElements[0] != "pages" && tryBranch(pathElements[0], "pages", pathElements[1:], "") {
log.Debug().Msg("tryBranch, now trying upstream")
2021-12-05 17:17:28 +00:00
tryUpstream(ctx, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
2021-12-05 17:18:05 +00:00
giteaRoot, giteaAPIToken,
2021-12-05 17:17:28 +00:00
canonicalDomainCache, branchTimestampCache, fileResponseCache)
2021-12-05 13:45:17 +00:00
return
}
// Try to use the "pages" repo on its default branch
// example.codeberg.page/index.html
log.Debug().Msg("main domain preparations, now trying with default repo/branch")
if tryBranch("pages", "", pathElements, "") {
log.Debug().Msg("tryBranch, now trying upstream")
2021-12-05 17:17:28 +00:00
tryUpstream(ctx, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
2021-12-05 17:18:05 +00:00
giteaRoot, giteaAPIToken,
2021-12-05 17:17:28 +00:00
canonicalDomainCache, branchTimestampCache, fileResponseCache)
2021-12-05 13:45:17 +00:00
return
}
// Couldn't find a valid repo/branch
2021-12-05 13:47:33 +00:00
html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
2021-12-05 13:45:17 +00:00
return
} else {
trimmedHostStr := string(trimmedHost)
// Serve pages from external domains
2021-12-05 14:21:05 +00:00
targetOwner, targetRepo, targetBranch = dns.GetTargetFromDNS(trimmedHostStr, string(mainDomainSuffix), dnsLookupCache)
2021-12-05 13:45:17 +00:00
if targetOwner == "" {
2021-12-05 13:47:33 +00:00
html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
2021-12-05 13:45:17 +00:00
return
}
pathElements := strings.Split(string(bytes.Trim(ctx.Request.URI().Path(), "/")), "/")
canonicalLink := ""
if strings.HasPrefix(pathElements[0], "@") {
targetBranch = pathElements[0][1:]
pathElements = pathElements[1:]
canonicalLink = "/%p"
}
// Try to use the given repo on the given branch or the default branch
log.Debug().Msg("custom domain preparations, now trying with details from DNS")
if tryBranch(targetRepo, targetBranch, pathElements, canonicalLink) {
2021-12-05 17:18:05 +00:00
canonicalDomain, valid := upstream.CheckCanonicalDomain(targetOwner, targetRepo, targetBranch, trimmedHostStr, string(mainDomainSuffix), giteaRoot, giteaAPIToken, canonicalDomainCache)
2021-12-05 13:45:17 +00:00
if !valid {
2021-12-05 13:47:33 +00:00
html.ReturnErrorPage(ctx, fasthttp.StatusMisdirectedRequest)
2021-12-05 13:45:17 +00:00
return
} else if canonicalDomain != trimmedHostStr {
// only redirect if the target is also a codeberg page!
2021-12-05 14:21:05 +00:00
targetOwner, _, _ = dns.GetTargetFromDNS(strings.SplitN(canonicalDomain, "/", 2)[0], string(mainDomainSuffix), dnsLookupCache)
2021-12-05 13:45:17 +00:00
if targetOwner != "" {
ctx.Redirect("https://"+canonicalDomain+string(ctx.RequestURI()), fasthttp.StatusTemporaryRedirect)
return
} else {
2021-12-05 13:47:33 +00:00
html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
2021-12-05 13:45:17 +00:00
return
}
}
log.Debug().Msg("tryBranch, now trying upstream")
2021-12-05 17:17:28 +00:00
tryUpstream(ctx, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
2021-12-05 17:18:05 +00:00
giteaRoot, giteaAPIToken,
2021-12-05 17:17:28 +00:00
canonicalDomainCache, branchTimestampCache, fileResponseCache)
2021-12-05 13:45:17 +00:00
return
} else {
2021-12-05 13:47:33 +00:00
html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
2021-12-05 13:45:17 +00:00
return
}
}
}
}