Files
reforger_crawler_main/controllers/indexerController.go
T
ilbinek 56d9e10621
Cross Compile Go / build (push) Canceled after 0s
admin dashboard, metrics and telemetry
- embedded React admin UI at /admin/ui (go:embed), login with ADMIN_SECRET
  via signed HttpOnly session cookie, X-ADMIN-KEY still works for scripts
- admin API: overview stats, indexing queue, addon management (priority,
  reindex, delete/restore/purge, bulk), whitelist CRUD, duplicate hashes,
  traffic, metrics explorer, services, events/audit log, system info
- built-in metrics: request latency per route, visitors, top addons and
  searches, indexer/scraper counters; minute buckets in the DB with hourly
  rollups and retention; optional Prometheus /metrics (METRICS_TOKEN)
- POST /back/telemetry for indexers/scrapers (metrics, events, heartbeat)
  and a stdlib-only Go client in telemetry/
- auth hardening: empty secrets no longer allow access, constant-time
  compare, CORS not applied to /admin and /back, TRUSTED_PROXIES setting
- Dockerfile builds the UI and no longer bakes config.yaml into the image;
  CI builds the UI, runs tests and builds arm64 with CGO
2026-10-03 18:10:57 +02:00

308 lines
8.3 KiB
Go

package controllers
import (
"bytes"
"encoding/json"
"fmt"
"net/http"
"strconv"
"time"
"gitea.tbdevent.eu/TBD/reforger_crawler_main/initializers"
"gitea.tbdevent.eu/TBD/reforger_crawler_main/metrics"
"gitea.tbdevent.eu/TBD/reforger_crawler_main/models"
"github.com/gin-gonic/gin"
"gorm.io/gorm"
)
func GetNextToBeIndexed(c *gin.Context) {
metrics.IndexerTimeoutsReset.Add(float64(ResetTimedOutIndexing(initializers.IndexTimeout)))
maxFileSize := c.Query("maxFileSize")
maxSize := 10_000_000_000
if maxFileSize != "" {
if val, err := strconv.Atoi(maxFileSize); err == nil {
maxSize = val
}
}
queue := "priority"
addon, err := claimNextAddon(maxSize, true)
if err != nil {
metrics.IndexerClaims.Inc("error")
c.JSON(500, gin.H{"error": err.Error()})
return
}
if addon == nil {
queue = "normal"
addon, err = claimNextAddon(maxSize, false)
if err != nil {
metrics.IndexerClaims.Inc("error")
c.JSON(500, gin.H{"error": err.Error()})
return
}
}
if addon != nil {
metrics.IndexerClaims.Inc(queue)
c.JSON(200, gin.H{"guid": addon.ID, "currentVersion": addon.CurrentVersionNumber})
return
}
// nothing to index
metrics.IndexerClaims.Inc("empty")
c.JSON(200, gin.H{"guid": "", "currentVersion": ""})
}
// claimNextAddon picks the oldest indexable addon and claims it with an
// optimistic UPDATE, so concurrent indexers can never claim the same job.
// Returns nil when no candidate is available.
func claimNextAddon(maxSize int, priorityOnly bool) (*models.Addon, error) {
for range 5 {
query := initializers.DB.Where("to_be_indexed = ?", true).Where("is_being_indexed = ?", false).Where("blocked = ?", false).Where("current_version_size <= ?", maxSize)
if priorityOnly {
query = query.Where("priority_indexing = ?", true)
}
var addon models.Addon
ret := query.Order("updated_at asc").First(&addon)
if ret.Error == gorm.ErrRecordNotFound {
return nil, nil
}
if ret.Error != nil {
return nil, ret.Error
}
claim := initializers.DB.Model(&models.Addon{}).
Where("id = ?", addon.ID).
Where("is_being_indexed = ?", false).
Where("to_be_indexed = ?", true).
Updates(map[string]any{
"is_being_indexed": true,
"index_start_time": time.Now(),
})
if claim.Error != nil {
return nil, claim.Error
}
if claim.RowsAffected == 1 {
return &addon, nil
}
// lost the race to another indexer - retry with the next candidate
}
return nil, nil
}
func SaveIndexingResult(c *gin.Context) {
var result struct {
GUID string `json:"guid"`
CurrentVersion string `json:"currentVersion"`
Success bool `json:"success"`
Files []models.AddonFile `json:"files"`
}
err := c.ShouldBindJSON(&result)
if err != nil {
metrics.IndexerSubmissions.Inc("invalid")
c.JSON(400, gin.H{"error": "Invalid JSON"})
return
}
var addon models.Addon
ret := initializers.DB.Where("id = ?", result.GUID).First(&addon)
if ret.Error != nil {
metrics.IndexerSubmissions.Inc("not_found")
c.JSON(404, gin.H{"error": "Addon not found"})
return
}
// get all files for this addon and delete them - old version
ret = initializers.DB.Where("addon_id = ?", addon.ID).Delete(&models.AddonFile{})
if ret.Error != nil {
metrics.IndexerSubmissions.Inc("error")
c.JSON(500, gin.H{"error": ret.Error.Error()})
return
}
ret = initializers.DB.CreateInBatches(&result.Files, 100)
if ret.Error != nil {
metrics.IndexerSubmissions.Inc("error")
c.JSON(500, gin.H{"error": ret.Error.Error()})
return
}
now := time.Now()
wasClaimed := addon.IsBeingIndexed && !addon.IndexStartTime.IsZero()
addon.CurrentVersionNumber = result.CurrentVersion
addon.IsBeingIndexed = false
addon.ToBeIndexed = false
addon.LastIndexedAt = &now
ret = initializers.DB.Save(&addon)
if ret.Error != nil {
metrics.IndexerSubmissions.Inc("error")
c.JSON(500, gin.H{"error": ret.Error.Error()})
return
}
outcome := "ok"
if !result.Success {
outcome = "unsuccessful"
}
metrics.IndexerSubmissions.Inc(outcome)
if wasClaimed {
metrics.IndexerJobDuration.Observe(now.Sub(addon.IndexStartTime).Seconds())
}
metrics.IndexerFilesPerSubmission.Observe(float64(len(result.Files)))
metrics.IndexerBytesIndexed.Add(float64(addon.CurrentVersionSize))
if initializers.DiscordWebhookURL != "" {
text := "Indexing of addon " + addon.Name + " (" + addon.ID + ") was successful."
colour := 2228479
nbr := strconv.Itoa(len(result.Files))
size := fmt.Sprintf("%.2f MB", float64(addon.CurrentVersionSize)/1_000_000)
addonsToBeIndexed := int64(0)
addonsCount := int64(0)
initializers.DB.Model(&models.Addon{}).Where("to_be_indexed = ?", true).Count(&addonsToBeIndexed)
initializers.DB.Model(&models.Addon{}).Count(&addonsCount)
percentage := 0.0
if addonsCount > 0 {
percentage = (float64(addonsCount-addonsToBeIndexed) / float64(addonsCount)) * 100
}
txt := fmt.Sprintf("%d/%d (%.2f%%).", addonsCount-addonsToBeIndexed, addonsCount, percentage)
myEmbed := models.CustomEmbed{
Title: text,
Description: fmt.Sprintf("Files Indexed: %s\nOverall Progress: %s", nbr, txt),
Color: colour,
Timestamp: time.Now(),
Image: models.CustomImage{
URL: addon.Preview,
},
Fields: []models.CustomEmbedField{
{
Name: "Size",
Value: size,
Inline: true,
},
{
Name: "Current Version",
Value: addon.CurrentVersionNumber,
Inline: true,
},
// {
// Name: "Indexed by",
// Value: c.ClientIP(),
// Inline: false,
// },
},
}
myHook := models.CustomHook{
Username: "Reforger Crawler",
Embeds: []models.CustomEmbed{myEmbed},
}
err := sendWebhook("index", initializers.DiscordWebhookURL, myHook)
if err != nil {
fmt.Println("Error sending webhook:", err)
}
}
c.JSON(200, gin.H{"status": "success"})
}
func DeleteAddon(c *gin.Context) {
guid := c.Query("guid")
if guid == "" {
c.JSON(400, gin.H{"error": "GUID is required"})
return
}
var addon models.Addon
ret := initializers.DB.Where("id = ?", guid).First(&addon)
if ret.Error != nil {
metrics.ScraperDeletes.Inc("not_found")
c.JSON(404, gin.H{"error": "Addon not found"})
return
}
err := initializers.DB.Transaction(func(tx *gorm.DB) error {
return DeleteAddonCascade(tx, &addon)
})
if err != nil {
metrics.ScraperDeletes.Inc("error")
c.JSON(500, gin.H{"error": err.Error()})
return
}
metrics.ScraperDeletes.Inc("deleted")
c.JSON(200, gin.H{"status": "addon deleted"})
}
// DeleteAddonCascade soft-deletes an addon and its files and removes its
// outgoing dependency edges.
func DeleteAddonCascade(tx *gorm.DB, addon *models.Addon) error {
// Delete associated files
if err := tx.Where("addon_id = ?", addon.ID).Delete(&models.AddonFile{}).Error; err != nil {
return err
}
// Delete outgoing dependency edges; incoming edges stay - other addons
// still declare this dependency even if the addon row is gone
if err := tx.Where("addon_id = ?", addon.ID).Delete(&models.AddonDependency{}).Error; err != nil {
return err
}
// Delete the addon
return tx.Delete(addon).Error
}
// ResetTimedOutIndexing releases claims older than olderThan so another
// indexer can pick the addon up. Returns the number of released claims.
func ResetTimedOutIndexing(olderThan time.Duration) int64 {
return initializers.DB.Model(&models.Addon{}).
Where("is_being_indexed = ?", true).
Where("index_start_time < ?", time.Now().Add(-olderThan)).
Update("is_being_indexed", false).RowsAffected
}
// sendWebhook sends a Discord webhook and records the outcome.
func sendWebhook(kind, webhookURL string, hook models.CustomHook) error {
err := SendCustomWebhook(webhookURL, hook)
result := "ok"
if err != nil {
result = "error"
}
metrics.DiscordWebhooks.Inc(kind, result)
return err
}
func SendCustomWebhook(webhookURL string, hook models.CustomHook) error {
payload, err := json.Marshal(hook)
if err != nil {
return err
}
req, err := http.NewRequest("POST", webhookURL, bytes.NewBuffer(payload))
if err != nil {
return err
}
req.Header.Set("Content-Type", "application/json")
client := &http.Client{}
resp, err := client.Do(req)
if err != nil {
return err
}
defer resp.Body.Close()
if resp.StatusCode < 200 || resp.StatusCode > 299 {
return fmt.Errorf("webhook failed with status code %d", resp.StatusCode)
}
return nil
}