pages-server/server/handler.go
2022-11-12 14:08:08 +01:00

289 lines
11 KiB
Go

package server
import (
"fmt"
"net/http"
"path"
"strings"
"github.com/rs/zerolog/log"
"codeberg.org/codeberg/pages/html"
"codeberg.org/codeberg/pages/server/cache"
"codeberg.org/codeberg/pages/server/context"
"codeberg.org/codeberg/pages/server/dns"
"codeberg.org/codeberg/pages/server/gitea"
"codeberg.org/codeberg/pages/server/upstream"
"codeberg.org/codeberg/pages/server/utils"
"codeberg.org/codeberg/pages/server/version"
)
const (
headerAccessControlAllowOrigin = "Access-Control-Allow-Origin"
headerAccessControlAllowMethods = "Access-Control-Allow-Methods"
)
// Handler handles a single HTTP request to the web server.
func Handler(mainDomainSuffix, rawDomain string,
giteaClient *gitea.Client,
rawInfoPage string,
blacklistedPaths, allowedCorsDomains []string,
dnsLookupCache, canonicalDomainCache cache.SetGetKey,
) http.HandlerFunc {
return func(w http.ResponseWriter, req *http.Request) {
log := log.With().Strs("Handler", []string{string(req.Host), req.RequestURI}).Logger()
ctx := context.New(w, req)
ctx.RespWriter.Header().Set("Server", "CodebergPages/"+version.Version)
// 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.RespWriter.Header().Set("Referrer-Policy", "strict-origin-when-cross-origin")
// Enable browser caching for up to 10 minutes
ctx.RespWriter.Header().Set("Cache-Control", "public, max-age=600")
trimmedHost := utils.TrimHostPort(req.Host)
// Add HSTS for RawDomain and MainDomainSuffix
if hsts := getHSTSHeader(trimmedHost, mainDomainSuffix, rawDomain); hsts != "" {
ctx.RespWriter.Header().Set("Strict-Transport-Security", hsts)
}
// Handle all http methods
ctx.RespWriter.Header().Set("Allow", http.MethodGet+", "+http.MethodHead+", "+http.MethodOptions)
switch ctx.Req.Method {
case http.MethodOptions:
// return Allow header
ctx.RespWriter.WriteHeader(http.StatusNoContent)
return
case http.MethodGet,
http.MethodHead:
// end switch case and handle allowed requests
break
default:
// Block all methods not required for static pages
ctx.String("Method not allowed", http.StatusMethodNotAllowed)
return
}
// Block blacklisted paths (like ACME challenges)
for _, blacklistedPath := range blacklistedPaths {
if strings.HasPrefix(ctx.Path(), blacklistedPath) {
html.ReturnErrorPage(ctx, "requested blacklisted path", http.StatusForbidden)
return
}
}
// Allow CORS for specified domains
allowCors := false
for _, allowedCorsDomain := range allowedCorsDomains {
if strings.EqualFold(trimmedHost, allowedCorsDomain) {
allowCors = true
break
}
}
if allowCors {
ctx.RespWriter.Header().Set(headerAccessControlAllowOrigin, "*")
ctx.RespWriter.Header().Set(headerAccessControlAllowMethods, http.MethodGet+", "+http.MethodHead)
}
// Prepare request information to Gitea
pathElements := strings.Split(strings.Trim(ctx.Path(), "/"), "/")
targetOptions := &upstream.Options{
TryIndexPages: true,
}
log.Debug().Msg("preparations")
if rawDomain != "" && strings.EqualFold(trimmedHost, rawDomain) {
// Serve raw content from RawDomain
log.Debug().Msg("raw domain")
targetOptions.TryIndexPages = false
targetOptions.ServeRaw = true
if len(pathElements) < 2 {
// https://{RawDomain}/{owner}/{repo}[/@{branch}]/{path} is required
ctx.Redirect(rawInfoPage, http.StatusTemporaryRedirect)
return
}
// 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")
newPath := path.Join(pathElements[3:]...)
branch := pathElements[2][1:]
repoOwner := pathElements[0]
repo := pathElements[1]
if timestampBranch, works := tryBranch(log, ctx, giteaClient, repoOwner, repo, branch, newPath, true); works {
targetOptions.BranchTimestamp = timestampBranch.Timestamp
log.Debug().Msg("tryBranch, now trying upstream 1")
tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
targetOptions, repoOwner, repo, timestampBranch.Branch, newPath, canonicalDomainCache)
return
}
log.Debug().Msg("missing branch info")
html.ReturnErrorPage(ctx, "missing branch info", http.StatusFailedDependency)
return
}
log.Debug().Msg("raw domain preparations, now trying with default branch")
repoOwner := pathElements[0]
repo := pathElements[1]
newPath := path.Join(pathElements[2:]...)
if timestampBranch, works := tryBranch(log, ctx, giteaClient, repoOwner, repo, "", newPath, true); works {
targetOptions.BranchTimestamp = timestampBranch.Timestamp
log.Debug().Msg("tryBranch, now trying upstream 2")
tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
targetOptions, repoOwner, repo, timestampBranch.Branch, newPath, canonicalDomainCache)
return
}
html.ReturnErrorPage(ctx, fmt.Sprintf("raw domain could not find repo '%s/%s' or repo is empty", repoOwner, repo), http.StatusNotFound)
return
} else if strings.HasSuffix(trimmedHost, mainDomainSuffix) {
// Serve pages from subdomains of MainDomainSuffix
log.Debug().Msg("main domain suffix")
targetOwner := strings.TrimSuffix(trimmedHost, mainDomainSuffix)
targetRepo := pathElements[0]
if targetOwner == "www" {
// www.codeberg.page redirects to codeberg.page // TODO: rm hardcoded - use cname?
ctx.Redirect("https://"+string(mainDomainSuffix[1:])+string(ctx.Path()), http.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:], "/"), http.StatusTemporaryRedirect)
return
}
log.Debug().Msg("main domain preparations, now trying with specified repo & branch")
branch := pathElements[1][1:]
newPath := path.Join(pathElements[2:]...)
repo := pathElements[0]
if timestampBranch, works := tryBranch(log, ctx, giteaClient, targetOwner, repo, branch, newPath, true); works {
targetOptions.BranchTimestamp = timestampBranch.Timestamp
log.Debug().Msg("tryBranch, now trying upstream 3")
tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, repo, timestampBranch.Branch, newPath, canonicalDomainCache)
} else {
html.ReturnErrorPage(ctx,
fmt.Sprintf("explizite set branch %q do not exist at '%s/%s'", branch, targetOwner, targetRepo),
http.StatusFailedDependency)
}
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")
branch := pathElements[0][1:]
repo := "pages"
newPath := path.Join(pathElements[1:]...)
if timestampBranch, works := tryBranch(log, ctx, giteaClient, targetOwner, targetRepo, branch, newPath, true); works {
targetOptions.BranchTimestamp = timestampBranch.Timestamp
log.Debug().Msg("tryBranch, now trying upstream 4")
tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, repo, timestampBranch.Branch, newPath, canonicalDomainCache)
} else {
html.ReturnErrorPage(ctx,
fmt.Sprintf("explizite set branch %q do not exist at '%s/%s'", branch, targetOwner, "pages"),
http.StatusFailedDependency)
}
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" {
repo := pathElements[0]
branch := "pages"
newPath := path.Join(pathElements[1:]...)
if timestampBranch, works := tryBranch(log, ctx, giteaClient, targetOwner, repo, branch, newPath, false); works {
targetOptions.BranchTimestamp = timestampBranch.Timestamp
log.Debug().Msg("tryBranch, now trying upstream 5")
tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, repo, timestampBranch.Branch, newPath, canonicalDomainCache)
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")
newPath := path.Join(pathElements...)
repo := "pages"
if timestampBranch, works := tryBranch(log, ctx, giteaClient, targetOwner, repo, "", newPath, false); works {
targetOptions.BranchTimestamp = timestampBranch.Timestamp
log.Debug().Msg("tryBranch, now trying upstream 6")
tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, repo, timestampBranch.Branch, newPath, canonicalDomainCache)
return
}
// Couldn't find a valid repo/branch
html.ReturnErrorPage(ctx,
fmt.Sprintf("couldn't find a valid repo[%s]", targetRepo),
http.StatusFailedDependency)
return
} else {
trimmedHostStr := string(trimmedHost)
// Serve pages from custom domains
targetOwner, targetRepo, targetBranch := dns.GetTargetFromDNS(trimmedHostStr, string(mainDomainSuffix), dnsLookupCache)
if targetOwner == "" {
html.ReturnErrorPage(ctx,
"could not obtain repo owner from custom domain",
http.StatusFailedDependency)
return
}
pathParts := pathElements
canonicalLink := false
if strings.HasPrefix(pathElements[0], "@") {
targetBranch = pathElements[0][1:]
pathParts = pathElements[1:]
canonicalLink = true
}
// Try to use the given repo on the given branch or the default branch
newPath := path.Join(pathParts...)
log.Debug().Msg("custom domain preparations, now trying with details from DNS")
if timestampBranch, works := tryBranch(log, ctx, giteaClient, targetOwner, targetRepo, targetBranch, newPath, canonicalLink); works {
targetOptions.BranchTimestamp = timestampBranch.Timestamp
canonicalDomain, valid := upstream.CheckCanonicalDomain(giteaClient, targetOwner, targetRepo, timestampBranch.Branch, trimmedHostStr, string(mainDomainSuffix), canonicalDomainCache)
if !valid {
html.ReturnErrorPage(ctx, "domain not specified in <code>.domains</code> file", http.StatusMisdirectedRequest)
return
} else if canonicalDomain != trimmedHostStr {
// only redirect if the target is also a codeberg page!
targetOwner, _, _ = dns.GetTargetFromDNS(strings.SplitN(canonicalDomain, "/", 2)[0], string(mainDomainSuffix), dnsLookupCache)
if targetOwner != "" {
ctx.Redirect("https://"+canonicalDomain+string(newPath), http.StatusTemporaryRedirect)
return
}
html.ReturnErrorPage(ctx, "target is no codeberg page", http.StatusFailedDependency)
return
}
log.Debug().Msg("tryBranch, now trying upstream 7")
tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
targetOptions, targetOwner, targetRepo, timestampBranch.Branch, newPath, canonicalDomainCache)
return
}
html.ReturnErrorPage(ctx, "could not find target for custom domain", http.StatusFailedDependency)
return
}
}
}