feat: Pass around one big global context

internal/context/context.go

@@ -0,0 +1,30 @@
+package context
+
+import (
+	"net/http"
+
+	"git.polynom.me/rio/internal/metrics"
+	"git.polynom.me/rio/internal/repo"
+)
+
+type GlobalContext struct {
+	DefaultCSP   string
+	PagesDomain  string
+	Gitea        *repo.GiteaClient
+	MetricConfig *metrics.LokiMetricConfig
+}
+
+type Context struct {
+	Username string
+	Reponame string
+	Domain   string
+	Path     string
+
+	// HTTP Stuff
+	Referrer  string
+	UserAgent string
+	Writer    http.ResponseWriter
+
+	// Pointer to the global context
+	Global *GlobalContext
+}

internal/metrics/metrics.go

@@ -1,9 +1,11 @@
-package pages
+package metrics
 
 import (
 	"encoding/json"
+	"fmt"
 	"io/ioutil"
 	"net/http"
+	"regexp"
 	"strconv"
 	"strings"
 	"time"
@@ -12,16 +14,29 @@ import (
 )
 
 type LokiMetricConfig struct {
-	Url     string
-	Enabled bool
+	Url           string
+	BotUserAgents *[]regexp.Regexp
+	Enabled       bool
 }
 
 // Checks if we should send a metric ping to Loki based on the served path.
-func (c *LokiMetricConfig) shouldSendMetrics(path string) bool {
-	return strings.HasSuffix(path, ".html") && c.Enabled
+func (c *LokiMetricConfig) ShouldSendMetrics(path, userAgent string) bool {
+	if !strings.HasSuffix(path, ".html") || !c.Enabled {
+		return false
+	}
+
+	// Filter out bots
+	for _, pattern := range *c.BotUserAgents {
+		if pattern.MatchString(userAgent) {
+			return false
+		}
+	}
+
+	return true
 }
 
-func (c *LokiMetricConfig) sendMetricPing(domain, path string) {
+func (c *LokiMetricConfig) SendMetricPing(domain, path, referrer string) {
+	msg := fmt.Sprintf("path=\"%s\" referrer=\"%s\"", path, referrer)
 	data := map[string]interface{}{
 		"streams": []map[string]interface{}{
 			{
@@ -34,7 +49,7 @@ func (c *LokiMetricConfig) sendMetricPing(domain, path string) {
 				"values": [][]interface{}{
 					{
 						strconv.Itoa(int(time.Now().UnixNano())),
-						"path=" + path,
+						msg,
 					},
 				},
 			},
@@ -72,3 +87,30 @@ func (c *LokiMetricConfig) sendMetricPing(domain, path string) {
 		}
 	}()
 }
+
+// Reads a JSON array of bot user agents from disk and parses them
+// into regular expressions.
+func ReadBotPatterns(file string) ([]regexp.Regexp, error) {
+	content, err := ioutil.ReadFile(file)
+	if err != nil {
+		log.Warnf("Failed to read bot metrics file: %v", err)
+		return []regexp.Regexp{}, err
+	}
+
+	var payload []string
+	err = json.Unmarshal(content, &payload)
+	if err != nil {
+		log.Warnf("Failed to unmarshal file: %v", err)
+		return []regexp.Regexp{}, err
+	}
+
+	patterns := make([]regexp.Regexp, 0)
+	for _, v := range payload {
+		patterns = append(
+			patterns,
+			*regexp.MustCompile(v),
+		)
+	}
+
+	return patterns, nil
+}

internal/metrics/metrics_test.go

@@ -0,0 +1,24 @@
+package metrics
+
+import (
+	"regexp"
+	"testing"
+)
+
+func TestShouldPing(t *testing.T) {
+	cfg := LokiMetricConfig{
+		Enabled: true,
+		Url:     "",
+		BotUserAgents: &[]regexp.Regexp{
+			*regexp.MustCompile("random-bot/.*"),
+		},
+	}
+
+	if cfg.ShouldSendMetrics("/index.html", "random-bot/v23.5") {
+		t.Fatalf("Accepted bot user-agent")
+	}
+
+	if !cfg.ShouldSendMetrics("/index.html", "Firefox/...") {
+		t.Fatalf("Rejected real user-agent")
+	}
+}

internal/pages/pages.go

@@ -8,6 +8,7 @@ import (
 	"time"
 
 	"git.polynom.me/rio/internal/constants"
+	"git.polynom.me/rio/internal/context"
 	"git.polynom.me/rio/internal/repo"
 
 	"github.com/patrickmn/go-cache"
@@ -45,13 +46,14 @@ func addHeaders(csp, contentType string, contentLength int, w http.ResponseWrite
 	}
 }
 
-func ServeFile(username, reponame, path, defaultCsp, domain string, giteaClient *repo.GiteaClient, metricConfig *LokiMetricConfig, w http.ResponseWriter) {
+func ServeFile(context *context.Context) {
 	// Strip away a starting / as it messes with Gitea
+	path := context.Path
 	if path[:1] == "/" {
 		path = path[1:]
 	}
 
-	key := makePageContentCacheEntry(username, path)
+	key := makePageContentCacheEntry(context.Username, path)
 	entry, found := pageCache.Get(key)
 	var content []byte
 	var mimeType string
@@ -65,25 +67,25 @@ func ServeFile(username, reponame, path, defaultCsp, domain string, giteaClient
 		since = &sinceRaw
 	}
 
-	content, changed, err := giteaClient.GetFile(
-		username,
-		reponame,
+	content, changed, err := context.Global.Gitea.GetFile(
+		context.Username,
+		context.Reponame,
 		constants.PagesBranch,
 		path,
 		since,
 	)
-	csp := repo.GetCSPForRepository(username, reponame, "", giteaClient)
+	csp := repo.GetCSPForRepository(context.Username, context.Reponame, "", context.Global.Gitea)
 
 	if err != nil {
 		if !found {
-			log.Errorf("Failed to get file %s/%s/%s (%s)", username, reponame, path, err)
-			addHeaders(csp, "text/html", 0, w)
-			w.WriteHeader(404)
+			log.Errorf("Failed to get file %s/%s/%s (%s)", context.Username, context.Reponame, path, err)
+			addHeaders(csp, "text/html", 0, context.Writer)
+			context.Writer.WriteHeader(404)
 		} else {
 			log.Debugf("Request failed but page %s is cached in memory", path)
-			addHeaders(csp, mimeType, len(content), w)
-			w.WriteHeader(200)
-			w.Write(content)
+			addHeaders(csp, mimeType, len(content), context.Writer)
+			context.Writer.WriteHeader(200)
+			context.Writer.Write(content)
 		}
 
 		return
@@ -91,9 +93,9 @@ func ServeFile(username, reponame, path, defaultCsp, domain string, giteaClient
 
 	if found && !changed {
 		log.Debugf("Page %s is unchanged and cached in memory", path)
-		addHeaders(csp, mimeType, len(content), w)
-		w.WriteHeader(200)
-		w.Write(content)
+		addHeaders(csp, mimeType, len(content), context.Writer)
+		context.Writer.WriteHeader(200)
+		context.Writer.Write(content)
 		return
 	}
 
@@ -113,12 +115,12 @@ func ServeFile(username, reponame, path, defaultCsp, domain string, giteaClient
 	)
 
 	log.Debugf("Page %s requested from Gitea and cached in memory at %v", path, now)
-	addHeaders(csp, mimeType, len(content), w)
-	w.WriteHeader(200)
-	w.Write(content)
+	addHeaders(csp, mimeType, len(content), context.Writer)
+	context.Writer.WriteHeader(200)
+	context.Writer.Write(content)
 
 	// Tell Loki about if, if desired
-	if metricConfig.shouldSendMetrics(path) {
-		metricConfig.sendMetricPing(domain, path)
+	if context.Global.MetricConfig.ShouldSendMetrics(path, context.UserAgent) {
+		context.Global.MetricConfig.SendMetricPing(context.Domain, path, context.Referrer)
 	}
 }