🎨 PDF files larger than 128MB are not included in asset file content searching https://github.com/siyuan-note/siyuan/issues/9500

app/guide/20210808180117-6v0mkxr/20200923234011-ieuun1p/20230805231614-vqn28eh/20230805231816-h1z9mpc/20230805232134-3d6mx2k.sy

@@ -5,7 +5,7 @@
 	"Properties": {
 		"id": "20230805232134-3d6mx2k",
 		"title": "Search asset content",
-		"updated": "20231025095559"
+		"updated": "20231026085302"
 	},
 	"Children": [
 		{
@@ -215,7 +215,7 @@
 			"ListData": {},
 			"Properties": {
 				"id": "20230903112419-45ps71g",
-				"updated": "20231025095559"
+				"updated": "20231026085302"
 			},
 			"Children": [
 				{
@@ -262,12 +262,58 @@
 							"Type": "NodeParagraph",
 							"Properties": {
 								"id": "20231025095244-m8ut4h2",
-								"updated": "20231025095559"
+								"updated": "20231026085241"
 							},
 							"Children": [
 								{
 									"Type": "NodeText",
-									"Data": "PDFs larger than 128MB or 1024 pages are not supported"
+									"Data": "PDFs larger than 1024 pages are not supported"
+								}
+							]
+						}
+					]
+				},
+				{
+					"ID": "20231026085243-o2jstdb",
+					"Type": "NodeListItem",
+					"ListData": {
+						"BulletChar": 42,
+						"Marker": "Kg=="
+					},
+					"Properties": {
+						"id": "20231026085243-o2jstdb",
+						"updated": "20231026085302"
+					},
+					"Children": [
+						{
+							"ID": "20231026085243-3mtefdv",
+							"Type": "NodeParagraph",
+							"Properties": {
+								"id": "20231026085243-3mtefdv",
+								"updated": "20231026085302"
+							},
+							"Children": [
+								{
+									"Type": "NodeText",
+									"Data": "By default, PDFs larger than 128MB are not supported. You can set the environment variable "
+								},
+								{
+									"Type": "NodeTextMark",
+									"TextMarkType": "code",
+									"TextMarkTextContent": "SIYUAN_PDF_ASSET_CONTENT_INDEX_MAX_SIZE"
+								},
+								{
+									"Type": "NodeText",
+									"Data": "​. The value unit is bytes, for example: "
+								},
+								{
+									"Type": "NodeTextMark",
+									"TextMarkType": "code",
+									"TextMarkTextContent": "SIYUAN_PDF_ASSET_CONTENT_INDEX_MAX_SIZE=256000000"
+								},
+								{
+									"Type": "NodeText",
+									"Data": "​ to adjust the upper limit to 256MB"
 								}
 							]
 						}

app/guide/20210808180117-czj9bvb/20200812220555-lj3enxa/20230805222417-2lj3dvk/20230805225107-qm1m2f5/20230805230218-aea8icj.sy

@@ -5,7 +5,7 @@
 	"Properties": {
 		"id": "20230805230218-aea8icj",
 		"title": "搜索资源文件内容",
-		"updated": "20231025095556"
+		"updated": "20231026085139"
 	},
 	"Children": [
 		{
@@ -217,7 +217,7 @@
 			"ListData": {},
 			"Properties": {
 				"id": "20230903112243-pwm33kc",
-				"updated": "20231025095556"
+				"updated": "20231026085139"
 			},
 			"Children": [
 				{
@@ -264,12 +264,58 @@
 							"Type": "NodeParagraph",
 							"Properties": {
 								"id": "20231025095053-ee3etgb",
-								"updated": "20231025095556"
+								"updated": "20231026085014"
 							},
 							"Children": [
 								{
 									"Type": "NodeText",
-									"Data": "不支持大于 128MB 或 1024 页的 PDF"
+									"Data": "不支持大于 1024 页的 PDF"
+								}
+							]
+						}
+					]
+				},
+				{
+					"ID": "20231026085015-163ec6r",
+					"Type": "NodeListItem",
+					"ListData": {
+						"BulletChar": 42,
+						"Marker": "Kg=="
+					},
+					"Properties": {
+						"id": "20231026085015-163ec6r",
+						"updated": "20231026085139"
+					},
+					"Children": [
+						{
+							"ID": "20231026085015-9kucsof",
+							"Type": "NodeParagraph",
+							"Properties": {
+								"id": "20231026085015-9kucsof",
+								"updated": "20231026085139"
+							},
+							"Children": [
+								{
+									"Type": "NodeText",
+									"Data": "默认不支持大于 128MB 的 PDF，可通过环境变量 "
+								},
+								{
+									"Type": "NodeTextMark",
+									"TextMarkType": "code",
+									"TextMarkTextContent": "SIYUAN_PDF_ASSET_CONTENT_INDEX_MAX_SIZE"
+								},
+								{
+									"Type": "NodeText",
+									"Data": "​ 设置，值的单位是字节，比如："
+								},
+								{
+									"Type": "NodeTextMark",
+									"TextMarkType": "code",
+									"TextMarkTextContent": "SIYUAN_PDF_ASSET_CONTENT_INDEX_MAX_SIZE=256000000"
+								},
+								{
+									"Type": "NodeText",
+									"Data": "​ 将上限调整为 256MB"
 								}
 							]
 						}
@@ -297,7 +343,11 @@
 							"Children": [
 								{
 									"Type": "NodeText",
-									"Data": "Android/iOS 端不支持 PDF 资源文件内容搜索"
+									"Data": "Android/iOS 端不支持 PDF "
+								},
+								{
+									"Type": "NodeText",
+									"Data": "资源文件内容搜索"
 								}
 							]
 						}

app/guide/20211226090932-5lcq56f/20211226115423-d5z1joq/20230805232636-zh0adz2/20230805232719-04mqbcx/20230805232920-5fdco36.sy

@@ -5,7 +5,7 @@
 	"Properties": {
 		"id": "20230805232920-5fdco36",
 		"title": "搜索資源文件內容",
-		"updated": "20231025095553"
+		"updated": "20231026085346"
 	},
 	"Children": [
 		{
@@ -220,7 +220,7 @@
 			"ListData": {},
 			"Properties": {
 				"id": "20230903112449-mq7jbfe",
-				"updated": "20231025095553"
+				"updated": "20231026085346"
 			},
 			"Children": [
 				{
@@ -267,12 +267,58 @@
 							"Type": "NodeParagraph",
 							"Properties": {
 								"id": "20231025095319-egfy5yi",
-								"updated": "20231025095553"
+								"updated": "20231026085317"
 							},
 							"Children": [
 								{
 									"Type": "NodeText",
-									"Data": "不支援大於 128MB 或 1024 頁的 PDF"
+									"Data": "不支援大於 1024 頁的 PDF"
+								}
+							]
+						}
+					]
+				},
+				{
+					"ID": "20231026085318-8quzaxe",
+					"Type": "NodeListItem",
+					"ListData": {
+						"BulletChar": 42,
+						"Marker": "Kg=="
+					},
+					"Properties": {
+						"id": "20231026085318-8quzaxe",
+						"updated": "20231026085346"
+					},
+					"Children": [
+						{
+							"ID": "20231026085318-2mkqfy3",
+							"Type": "NodeParagraph",
+							"Properties": {
+								"id": "20231026085318-2mkqfy3",
+								"updated": "20231026085346"
+							},
+							"Children": [
+								{
+									"Type": "NodeText",
+									"Data": "預設不支援大於 128MB 的 PDF，可透過環境變數 "
+								},
+								{
+									"Type": "NodeTextMark",
+									"TextMarkType": "code",
+									"TextMarkTextContent": "SIYUAN_PDF_ASSET_CONTENT_INDEX_MAX_SIZE"
+								},
+								{
+									"Type": "NodeText",
+									"Data": "​ 設定，值的單位是字節，例如："
+								},
+								{
+									"Type": "NodeTextMark",
+									"TextMarkType": "code",
+									"TextMarkTextContent": "SIYUAN_PDF_ASSET_CONTENT_INDEX_MAX_SIZE=256000000"
+								},
+								{
+									"Type": "NodeText",
+									"Data": "​ 將上限調整為 256MB"
 								}
 							]
 						}

kernel/model/asset_content.go

@@ -480,7 +480,10 @@ func NewAssetsSearcher() *AssetsSearcher {
 const (
 	TxtAssetContentMaxSize = 1024 * 1024 * 4
 	PDFAssetContentMaxPage = 1024
-	PDFAssetContentMaxSize = 1024 * 1024 * 128
+)
+
+var (
+	PDFAssetContentMaxSize uint64 = 1024 * 1024 * 128
 )
 
 type AssetParseResult struct {
@@ -829,7 +832,18 @@ func (parser *PdfAssetParser) Parse(absPath string) (ret *AssetParseResult) {
 		return
 	}
 
-	if PDFAssetContentMaxSize < len(pdfData) {
+	if maxSizeVal := os.Getenv("SIYUAN_PDF_ASSET_CONTENT_INDEX_MAX_SIZE"); "" != maxSizeVal {
+		if maxSize, parseErr := strconv.ParseUint(maxSizeVal, 10, 64); nil == parseErr {
+			if maxSize != PDFAssetContentMaxSize {
+				PDFAssetContentMaxSize = maxSize
+				logging.LogInfof("set PDF asset content index max size to [%s]", humanize.Bytes(maxSize))
+			}
+		} else {
+			logging.LogWarnf("invalid env [SIYUAN_PDF_ASSET_CONTENT_INDEX_MAX_SIZE]: [%s], parsing failed: ", maxSizeVal, parseErr)
+		}
+	}
+
+	if PDFAssetContentMaxSize < uint64(len(pdfData)) {
 		// PDF files larger than 128MB are not included in asset file content searching https://github.com/siyuan-note/siyuan/issues/9500
 		logging.LogWarnf("ignore large PDF asset [%s] with [%s]", absPath, humanize.Bytes(uint64(len(pdfData))))
 		return