Merge remote-tracking branch 'origin/dev' into dev

2023-11-10 11:51:17 +08:00 · 2023-11-10 11:51:17 +08:00 · 180dce0a78
commit 180dce0a78
parent cdeeb6bec8 29f8f0fe85
5 changed files with 73 additions and 28 deletions
--- a/app/guide/20210808180117-6v0mkxr/20200923234011-ieuun1p/20210808180303-xaduj2o/20200924100744-br924ar.sy
+++ b/app/guide/20210808180117-6v0mkxr/20200923234011-ieuun1p/20210808180303-xaduj2o/20200924100744-br924ar.sy
@ -6,7 +6,7 @@
 		"id": "20200924100744-br924ar",
 		"title": "Assets",
 		"type": "doc",
-		"updated": "20230820185510"
+		"updated": "20231110112758"
 	},
 	"Children": [
 		{
@ -856,7 +856,7 @@
 			"ListData": {},
 			"Properties": {
 				"id": "20230202231731-bdh7lab",
-				"updated": "20230211103249"
+				"updated": "20231110112758"
 			},
 			"Children": [
 				{
@ -868,7 +868,7 @@
 					},
 					"Properties": {
 						"id": "20230202231732-n7z8jth",
-						"updated": "20230211103249"
+						"updated": "20231110112758"
 					},
 					"Children": [
 						{
@ -876,7 +876,7 @@
 							"Type": "NodeParagraph",
 							"Properties": {
 								"id": "20230202231732-f3jkj7p",
-								"updated": "20230211103249"
+								"updated": "20231110112758"
 							},
 							"Children": [
 								{
@ -908,7 +908,16 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": " in the kernel boot log"
+									"Data": " in the kernel boot log; If you want to disable OCR, set the environment variable "
+								},
+								{
+									"Type": "NodeTextMark",
+									"TextMarkType": "code",
+									"TextMarkTextContent": "SIYUAN\\_TESSERACT\\_ENABLED=false"
+								},
+								{
+									"Type": "NodeText",
+									"Data": ""
 								}
 							]
 						}
@ -936,7 +945,11 @@
 							"Children": [
 								{
 									"Type": "NodeText",
-									"Data": "SiYuan does not enable all installed language packs, because too many language packs will cause OCR to be slow or even timeout and return empty results, and take up too many system resources, so by default only "
+									"Data": "SiYuan"
+								},
+								{
+									"Type": "NodeText",
+									"Data": " does not enable all installed language packs, because too many language packs will cause OCR to be slow or even timeout and return empty results, and take up too many system resources, so by default only "
 								},
 								{
 									"Type": "NodeTextMark",
@ -945,7 +958,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": ", "
+									"Data": ", "
 								},
 								{
 									"Type": "NodeTextMark",
@ -954,7 +967,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": ", "
+									"Data": ", "
 								},
 								{
 									"Type": "NodeTextMark",
@ -963,7 +976,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": ", "
+									"Data": ", "
 								},
 								{
 									"Type": "NodeTextMark",
@ -972,7 +985,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": ", "
+									"Data": ", "
 								},
 								{
 									"Type": "NodeTextMark",
@ -981,7 +994,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": ", "
+									"Data": ", "
 								},
 								{
 									"Type": "NodeTextMark",
@ -990,7 +1003,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": " and "
+									"Data": " and "
 								},
 								{
 									"Type": "NodeTextMark",
@ -999,7 +1012,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": " will be enabled at most, if you need more accurate language support, you can specify the language pack by configuring the environment variable "
+									"Data": " will be enabled at most, if you need more accurate language support, you can specify the language pack by configuring the environment variable "
 								},
 								{
 									"Type": "NodeTextMark",
@ -1008,7 +1021,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": ", such as "
+									"Data": ", such as "
 								},
 								{
 									"Type": "NodeTextMark",
@ -1017,7 +1030,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": ""
+									"Data": ""
 								}
 							]
 						}
@ -1081,7 +1094,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": ", the unit of value is bytes, for example: "
+									"Data": ", the unit of value is bytes, for example: "
 								},
 								{
 									"Type": "NodeTextMark",
@ -1090,7 +1103,7 @@
 								},
 								{
 									"Type": "NodeText",
-									"Data": " adjust the upper limit to 4MB"
+									"Data": " adjust the upper limit to 4MB"
 								}
 							]
 						}
--- a/app/guide/20210808180117-czj9bvb/20200812220555-lj3enxa/20210808180321-hbvl5c2/20200915214115-42b8zma.sy
+++ b/app/guide/20210808180117-czj9bvb/20200812220555-lj3enxa/20210808180321-hbvl5c2/20200915214115-42b8zma.sy
@ -6,7 +6,7 @@
 		"id": "20200915214115-42b8zma",
 		"title": "资源文件",
 		"type": "doc",
-		"updated": "20230820185522"
+		"updated": "20231110112936"
 	},
 	"Children": [
 		{
@ -878,7 +878,7 @@
 			"ListData": {},
 			"Properties": {
 				"id": "20230202231309-pcjl7c2",
-				"updated": "20230211103308"
+				"updated": "20231110112936"
 			},
 			"Children": [
 				{
@ -890,7 +890,7 @@
 					},
 					"Properties": {
 						"id": "20230202231311-7qdk1za",
-						"updated": "20230211103308"
+						"updated": "20231110112936"
 					},
 					"Children": [
 						{
@ -898,7 +898,7 @@
 							"Type": "NodeParagraph",
 							"Properties": {
 								"id": "20230202231311-n1pf7in",
-								"updated": "20230211103308"
+								"updated": "20231110112936"
 							},
 							"Children": [
 								{
@ -928,6 +928,15 @@
 									"TextMarkType": "code",
 									"TextMarkTextContent": "tesseract-ocr enabled"
 								},
+								{
+									"Type": "NodeText",
+									"Data": "；如果要禁用 OCR，请设置环境变量 "
+								},
+								{
+									"Type": "NodeTextMark",
+									"TextMarkType": "code",
+									"TextMarkTextContent": "SIYUAN_TESSERACT_ENABLED=false"
+								},
 								{
 									"Type": "NodeText",
 									"Data": ""
--- a/app/guide/20211226090932-5lcq56f/20211226115423-d5z1joq/20211226121203-rjjngpz/20211226123038-4umgpxy.sy
+++ b/app/guide/20211226090932-5lcq56f/20211226115423-d5z1joq/20211226121203-rjjngpz/20211226123038-4umgpxy.sy
@ -5,7 +5,7 @@
 	"Properties": {
 		"id": "20211226123038-4umgpxy",
 		"title": "資料文件",
-		"updated": "20230820185516"
+		"updated": "20231110112927"
 	},
 	"Children": [
 		{
@ -861,7 +861,7 @@
 			"ListData": {},
 			"Properties": {
 				"id": "20230202231516-o6k9mj1",
-				"updated": "20230211103259"
+				"updated": "20231110112927"
 			},
 			"Children": [
 				{
@ -873,7 +873,7 @@
 					},
 					"Properties": {
 						"id": "20230202231516-pwj2ndg",
-						"updated": "20230211103259"
+						"updated": "20231110112927"
 					},
 					"Children": [
 						{
@ -881,7 +881,7 @@
 							"Type": "NodeParagraph",
 							"Properties": {
 								"id": "20230202231516-8trf08t",
-								"updated": "20230211103259"
+								"updated": "20231110112927"
 							},
 							"Children": [
 								{
@ -911,6 +911,15 @@
 									"TextMarkType": "code",
 									"TextMarkTextContent": "tesseract-ocr enabled"
 								},
+								{
+									"Type": "NodeText",
+									"Data": "；如果要停用 OCR，請設定環境變量 "
+								},
+								{
+									"Type": "NodeTextMark",
+									"TextMarkType": "code",
+									"TextMarkTextContent": "SIYUAN_TESSERACT_ENABLED=false"
+								},
 								{
 									"Type": "NodeText",
 									"Data": ""
--- a/kernel/model/ocr.go
+++ b/kernel/model/ocr.go
@ -37,8 +37,9 @@ func autoOCRAssets() {
 			util.AssetsTextsLock.Lock()
 			util.AssetsTexts[p] = text
 			util.AssetsTextsLock.Unlock()
-			util.AssetsTextsChanged = true
-
+			if "" != text {
+				util.AssetsTextsChanged = true
+			}
 			if 4 <= i { // 一次任务中最多处理 4 张图片，防止卡顿
 				break
 			}
--- a/kernel/util/tesseract.go
+++ b/kernel/util/tesseract.go
@ -69,7 +69,9 @@ func GetAssetText(asset string, force bool) string {
 	AssetsTextsLock.Lock()
 	AssetsTexts[asset] = ret
 	AssetsTextsLock.Unlock()
-	AssetsTextsChanged = true
+	if "" != ret {
+		AssetsTextsChanged = true
+	}
 	return ret
 }

@ -149,6 +151,17 @@ func InitTesseract() {
 		}
 	}

+	// Supports via environment var `SIYUAN_TESSERACT_ENABLED=false` to close OCR https://github.com/siyuan-note/siyuan/issues/9619
+	if enabled := os.Getenv("SIYUAN_TESSERACT_ENABLED"); "" != enabled {
+		if enabledBool, parseErr := strconv.ParseBool(enabled); nil == parseErr {
+			TesseractEnabled = enabledBool
+			if !enabledBool {
+				logging.LogInfof("tesseract-ocr disabled by env")
+				return
+			}
+		}
+	}
+
 	TesseractLangs = filterTesseractLangs(langs)
 	logging.LogInfof("tesseract-ocr enabled [ver=%s, maxSize=%s, langs=%s]", ver, humanize.Bytes(TesseractMaxSize), strings.Join(TesseractLangs, "+"))
 }