From eb84c1f90f923aaaa6505e963e41ad6ce4f66c2a Mon Sep 17 00:00:00 2001
From: Daniel <845765@qq.com>
Date: Sat, 21 Sep 2024 17:29:01 +0800
Subject: [PATCH] :art: Highlight regular expression search results
 https://github.com/siyuan-note/siyuan/issues/11112

---
 kernel/model/file.go      |  8 ++++--
 kernel/model/search.go    | 22 ++++++++++++++---
 kernel/sql/block_query.go | 52 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 77 insertions(+), 5 deletions(-)

diff --git a/kernel/model/file.go b/kernel/model/file.go
index 320eba633..c316786f1 100644
--- a/kernel/model/file.go
+++ b/kernel/model/file.go
@@ -804,12 +804,16 @@ func GetDoc(startID, endID, id string, index int, query string, queryTypes map[s
 	subTree := &parse.Tree{ID: rootID, Root: &ast.Node{Type: ast.NodeDocument}, Marks: tree.Marks}
 
 	var keywords []string
-	if "" != query && (0 == queryMethod || 1 == queryMethod) { // 只有关键字搜索和查询语法搜索才支持高亮
+	if "" != query && (0 == queryMethod || 1 == queryMethod || 3 == queryMethod) { // 只有关键字、查询语法和正则表达式搜索支持高亮
 		if 0 == queryMethod {
 			query = stringQuery(query)
 		}
 		typeFilter := buildTypeFilter(queryTypes)
-		keywords = highlightByQuery(query, typeFilter, rootID)
+		if 0 == queryMethod || 1 == queryMethod {
+			keywords = highlightByFTS(query, typeFilter, rootID)
+		} else {
+			keywords = highlightByRegexp(query, typeFilter, rootID)
+		}
 	}
 
 	for _, n := range nodes {
diff --git a/kernel/model/search.go b/kernel/model/search.go
index 7a0949ae1..5a60cff96 100644
--- a/kernel/model/search.go
+++ b/kernel/model/search.go
@@ -1290,8 +1290,8 @@ func fullTextSearchByRegexp(exp, boxFilter, pathFilter, typeFilter, orderBy stri
 	stmt := "SELECT * FROM `blocks` WHERE " + fieldFilter + " AND type IN " + typeFilter
 	stmt += boxFilter + pathFilter
 	stmt += " " + orderBy
-	stmt += " LIMIT " + strconv.Itoa(pageSize) + " OFFSET " + strconv.Itoa((page-1)*pageSize)
-	blocks := sql.SelectBlocksRawStmtNoParse(stmt, Conf.Search.Limit)
+	regex := regexp.MustCompile(exp)
+	blocks := sql.SelectBlocksRegex(stmt, regex, Conf.Search.Name, Conf.Search.Alias, Conf.Search.Memo, Conf.Search.IAL, page, pageSize)
 	ret = fromSQLBlocks(&blocks, "", beforeLen)
 	if 1 > len(ret) {
 		ret = []*Block{}
@@ -1354,7 +1354,7 @@ func fullTextSearchByFTS(query, boxFilter, pathFilter, typeFilter, orderBy strin
 	return
 }
 
-func highlightByQuery(query, typeFilter, id string) (ret []string) {
+func highlightByFTS(query, typeFilter, id string) (ret []string) {
 	const limit = 256
 	table := "blocks_fts"
 	if !Conf.Search.CaseSensitive {
@@ -1383,6 +1383,22 @@ func highlightByQuery(query, typeFilter, id string) (ret []string) {
 	return
 }
 
+func highlightByRegexp(query, typeFilter, id string) (ret []string) {
+	fieldFilter := fieldRegexp(query)
+	stmt := "SELECT * FROM `blocks` WHERE " + fieldFilter + " AND type IN " + typeFilter
+	stmt += " AND root_id = '" + id + "'"
+	regex := regexp.MustCompile(query)
+	sqlBlocks := sql.SelectBlocksRegex(stmt, regex, Conf.Search.Name, Conf.Search.Alias, Conf.Search.Memo, Conf.Search.IAL, 1, 256)
+	for _, block := range sqlBlocks {
+		keyword := gulu.Str.SubstringsBetween(block.Content, search.SearchMarkLeft, search.SearchMarkRight)
+		if 0 < len(keyword) {
+			ret = append(ret, keyword...)
+		}
+	}
+	ret = gulu.Str.RemoveDuplicatedElem(ret)
+	return
+}
+
 func fullTextSearchCount(query, boxFilter, pathFilter, typeFilter string) (matchedBlockCount, matchedRootCount int) {
 	query = filterQueryInvisibleChars(query)
 	if ast.IsNodeIDPattern(query) {
diff --git a/kernel/sql/block_query.go b/kernel/sql/block_query.go
index e6d461f0b..d267395c5 100644
--- a/kernel/sql/block_query.go
+++ b/kernel/sql/block_query.go
@@ -20,6 +20,7 @@ import (
 	"bytes"
 	"database/sql"
 	"math"
+	"regexp"
 	"sort"
 	"strconv"
 	"strings"
@@ -619,6 +620,57 @@ func SelectBlocksRawStmt(stmt string, page, limit int) (ret []*Block) {
 	return
 }
 
+func SelectBlocksRegex(stmt string, exp *regexp.Regexp, name, alias, memo, ial bool, page, pageSize int) (ret []*Block) {
+	rows, err := query(stmt)
+	if err != nil {
+		logging.LogErrorf("sql query [%s] failed: %s", stmt, err)
+		return
+	}
+	defer rows.Close()
+	count := 0
+	for rows.Next() {
+		count++
+		if count <= (page-1)*pageSize {
+			continue
+		}
+
+		var block Block
+		if err := rows.Scan(&block.ID, &block.ParentID, &block.RootID, &block.Hash, &block.Box, &block.Path, &block.HPath, &block.Name, &block.Alias, &block.Memo, &block.Tag, &block.Content, &block.FContent, &block.Markdown, &block.Length, &block.Type, &block.SubType, &block.IAL, &block.Sort, &block.Created, &block.Updated); err != nil {
+			logging.LogErrorf("query scan field failed: %s\n%s", err, logging.ShortStack())
+			return
+		}
+
+		hitContent := exp.MatchString(block.Content)
+		hitName := name && exp.MatchString(block.Name)
+		hitAlias := alias && exp.MatchString(block.Alias)
+		hitMemo := memo && exp.MatchString(block.Memo)
+		hitIAL := ial && exp.MatchString(block.IAL)
+		if hitContent || hitName || hitAlias || hitMemo || hitIAL {
+			if hitContent {
+				block.Content = exp.ReplaceAllString(block.Content, "__@mark__${0}__mark@__")
+			}
+			if hitName {
+				block.Name = exp.ReplaceAllString(block.Name, "__@mark__${0}__mark@__")
+			}
+			if hitAlias {
+				block.Alias = exp.ReplaceAllString(block.Alias, "__@mark__${0}__mark@__")
+			}
+			if hitMemo {
+				block.Memo = exp.ReplaceAllString(block.Memo, "__@mark__${0}__mark@__")
+			}
+			if hitIAL {
+				block.IAL = exp.ReplaceAllString(block.IAL, "__@mark__${0}__mark@__")
+			}
+
+			ret = append(ret, &block)
+			if len(ret) >= pageSize {
+				break
+			}
+		}
+	}
+	return
+}
+
 func selectBlocksRawStmt(stmt string, limit int) (ret []*Block) {
 	rows, err := query(stmt)
 	if err != nil {