Add local_search_enabled flag to classifier (#291)

* Add local_search_enabled flag

* Add comment

* Update documentation

Author: mgdigital

bitmagnet.io/guides/reprocess-reclassify.md

@@ -25,6 +25,7 @@ The `reprocess` command will re-queue torrents to allow the latest updates to be
 To reprocess all torrents in your index, simply run `bitmagnet reprocess`. If you've indexed a lot of torrents, this will take a while, so there are a few options available to control exactly what gets reprocessed:
 
 - `apisDisabled`: Disable API calls during classification. This makes the classifier run a _lot_ faster, but disables identification with external services such as TMDB (metadata already gathered from external APIs is not lost).
+- `localSearchDisabled`: Disable the local search query on the content table for matching torrents to known content. This should be tried before any external API call is attempted, but it's an expensive query and so it's useful to be able to disable it using this flag.
 - `contentType`: Only reprocess torrents of a certain content type. For example, `bitmagnet reprocess --contentType movie` will only reprocess movies. Multiple content types can be comma separated, and `null` refers to torrents of unknown content type.
 - `orphans`: Only reprocess torrents that have no content record.
 - `classifyMode`: This controls how already matched torrents are handled.

internal/app/cmd/processcmd/command.go

@@ -39,6 +39,11 @@ func New(p Params) (Result, error) {
 				Value: false,
 				Usage: "disable API calls for the classifier workflow",
 			},
+			&cli.BoolFlag{
+				Name:  "localSearchDisabled",
+				Value: false,
+				Usage: "disable local search queries for the classifier workflow",
+			},
 		},
 		Action: func(ctx *cli.Context) error {
 			pr, err := p.Processor.Get()
@@ -53,6 +58,9 @@ func New(p Params) (Result, error) {
 			if ctx.Bool("apisDisabled") {
 				flags["apis_enabled"] = false
 			}
+			if ctx.Bool("localSearchDisabled") {
+				flags["local_search_enabled"] = false
+			}
 			var infoHashes []protocol.ID
 			for _, infoHash := range ctx.StringSlice("infoHash") {
 				id, err := protocol.ParseID(infoHash)

internal/app/cmd/reprocesscmd/command.go

@@ -66,6 +66,11 @@ func New(p Params) (Result, error) {
 				Value: false,
 				Usage: "disable API calls for the classifier workflow",
 			},
+			&cli.BoolFlag{
+				Name:  "localSearchDisabled",
+				Value: false,
+				Usage: "disable local search queries for the classifier workflow",
+			},
 		},
 		Action: func(ctx *cli.Context) error {
 			var classifyMode processor.ClassifyMode
@@ -85,6 +90,9 @@ func New(p Params) (Result, error) {
 			if ctx.Bool("apisDisabled") {
 				flags["apis_enabled"] = false
 			}
+			if ctx.Bool("localSearchDisabled") {
+				flags["local_search_enabled"] = false
+			}
 			var contentTypes []model.NullContentType
 			for _, contentType := range ctx.StringSlice("contentType") {
 				if contentType == "null" {

internal/classifier/classifier.core.yaml

@@ -89,7 +89,10 @@ workflows:
         condition: "!result.hasAttachedContent && result.hasBaseTitle"
         if_action:
           find_match:
-            - attach_local_content_by_search
+            - if_else:
+                condition: "flags.local_search_enabled"
+                if_action: attach_local_content_by_search
+                else_action: unmatched
             - if_else:
                 condition: "flags.apis_enabled && flags.tmdb_enabled"
                 if_action: attach_tmdb_content_by_search
@@ -226,11 +229,13 @@ keywords:
     - yvm
     - (#|10|11|12|13|14|15|16|17) ?y ?o
 flag_definitions:
+  local_search_enabled: bool
   apis_enabled: bool
   tmdb_enabled: bool
   delete_content_types: content_type_list
   delete_xxx: bool
 flags:
+  local_search_enabled: true
   apis_enabled: true
   tmdb_enabled: true
   delete_content_types: []

internal/processor/batch/queue/handler.go

@@ -72,7 +72,7 @@ func New(p Params) Result {
 				priority := 10
 				// prioritise jobs where API calls are disabled as they will run faster:
 				if msg.ApisDisabled() {
-					priority = 5
+					priority = 4
 				}
 				maxInfoHash := msg.InfoHashGreaterThan
 				chunkSize := uint(0)

internal/processor/hook_0_9_0/hook.go

@@ -62,7 +62,8 @@ func execHook(d *dao.Query) error {
 		job, jobErr := batch.NewQueueJob(batch.MessageParams{
 			UpdatedBefore: time.Now(),
 			ClassifierFlags: classifier.Flags{
-				"apis_enabled": false,
+				"apis_enabled":         false,
+				"local_search_enabled": false,
 			},
 			ChunkSize: 10_000,
 			BatchSize: 100,

internal/processor/queue/handler.go

@@ -34,6 +34,13 @@ func New(p Params) Result {
 				if err := json.Unmarshal([]byte(job.Payload), msg); err != nil {
 					return err
 				}
+				// The following is somewhat of a hack to alter the `local_search_enabled` flag for jobs queued by the upgrade hook between 0.9.0 and 0.9.3.
+				// It should be removed at a later date.
+				if job.Priority == 5 && msg.ClassifierFlags != nil {
+					if _, ok := msg.ClassifierFlags["local_search_enabled"]; !ok {
+						msg.ClassifierFlags["local_search_enabled"] = false
+					}
+				}
 				return pr.Process(ctx, *msg)
 			}, handler.JobTimeout(time.Second*60*10), handler.Concurrency(int(p.Config.Concurrency))), nil
 		}),