Add a test field to search query objects

heyman · marijnh · commit d55469959a81 · 2026-01-13T10:46:42.000+01:00
FEATURE: Search queries now support a generic `test` field that can be
used to implement custom tests on matches.
diff --git a/src/search.ts b/src/search.ts
@@ -105,6 +105,9 @@ export class SearchQuery {
   /// @internal
   readonly unquoted: string
 
+  /// Optional test function used to filter matches.
+  readonly test: ((match: string, state: EditorState, from: number, to: number) => boolean) | undefined
+
   /// Create a query object.
   constructor(config: {
     /// The search string.
@@ -121,6 +124,10 @@ export class SearchQuery {
     replace?: string,
     /// Enable whole-word matching.
     wholeWord?: boolean
+
+    /// Optional function to filter matches. It is passed the matched
+    /// string, the editor state, and the match range.
+    test?: (match: string, state: EditorState, from: number, to: number) => boolean
   }) {
     this.search = config.search
     this.caseSensitive = !!config.caseSensitive
@@ -130,6 +137,7 @@ export class SearchQuery {
     this.valid = !!this.search && (!this.regexp || validRegExp(this.search))
     this.unquoted = this.unquote(this.search)
     this.wholeWord = !!config.wholeWord
+    this.test = config.test
   }
 
   /// @internal
@@ -142,7 +150,7 @@ export class SearchQuery {
   eq(other: SearchQuery) {
     return this.search == other.search && this.replace == other.replace &&
       this.caseSensitive == other.caseSensitive && this.regexp == other.regexp &&
-      this.wholeWord == other.wholeWord
+      this.wholeWord == other.wholeWord && this.test == other.test
   }
 
   /// @internal
@@ -178,9 +186,22 @@ abstract class QueryType<Result extends SearchResult = SearchResult> {
 const enum FindPrev { ChunkSize = 10000 }
 
 function stringCursor(spec: SearchQuery, state: EditorState, from: number, to: number) {
+  const wordTest = spec.wholeWord ? stringWordTest(state.doc, state.charCategorizer(state.selection.main.head)) : undefined
+  const test = (from: number, to: number, buffer: string, bufferPos: number) => {
+    return (
+      (wordTest ? wordTest(from, to, buffer, bufferPos) : true) &&
+      (spec.test ? spec.test(matchFromBuffer(state, from, to, buffer, bufferPos), state, from, to) : true)
+    )
+  }
   return new SearchCursor(
     state.doc, spec.unquoted, from, to, spec.caseSensitive ? undefined : x => x.toLowerCase(),
-    spec.wholeWord ? stringWordTest(state.doc, state.charCategorizer(state.selection.main.head)) : undefined)
+    wordTest || spec.test ? test : undefined)
+}
+
+function matchFromBuffer(state: EditorState, from: number, to: number, buffer: string, bufferPos: number) {
+  if (from >= bufferPos && to <= bufferPos + buffer.length)
+    return buffer.slice(from - bufferPos, to - bufferPos)
+  return state.doc.sliceString(from, to)
 }
 
 function stringWordTest(doc: Text, categorizer: (ch: string) => CharCategory) {
@@ -253,9 +274,17 @@ const enum RegExp { HighlightMargin = 250 }
 type RegExpResult = typeof RegExpCursor.prototype.value
 
 function regexpCursor(spec: SearchQuery, state: EditorState, from: number, to: number) {
+  const wordTest = spec.wholeWord ? regexpWordTest(state.charCategorizer(state.selection.main.head)) : undefined
+  const test = (from: number, to: number, match: RegExpExecArray) => {
+    return (
+      (wordTest ? wordTest(from, to, match) : true) &&
+      (spec.test ? spec.test(match[0], state, from, to) : true)
+    )
+  }
+
   return new RegExpCursor(state.doc, spec.search, {
     ignoreCase: !spec.caseSensitive,
-    test: spec.wholeWord ? regexpWordTest(state.charCategorizer(state.selection.main.head)) : undefined
+    test: wordTest || spec.test ? test : undefined
   }, from, to)
 }
 
diff --git a/test/test-query.ts b/test/test-query.ts
@@ -58,4 +58,36 @@ describe("SearchQuery", () => {
   it("can match regular expressions by word", () => {
     test(new SearchQuery({search: "a..", regexp: true, wholeWord: true}), "[aap] baap aapje [a--]w")
   })
+
+  it("can filter string matches with a test function", () => {
+    test(new SearchQuery({
+      search: "ab",
+      test(match, state, from, to) {
+        ist(match, state.doc.sliceString(from, to))
+        return from > 0
+      }
+    }), "ab a[ab] [ab]")
+  })
+
+  it("combines whole-word matches with a test function", () => {
+    test(new SearchQuery({
+      search: "ab",
+      wholeWord: true,
+      test(match, state, from, to) {
+        ist(match, state.doc.sliceString(from, to))
+        return from > 0
+      }
+    }), "ab xab [ab]")
+  })
+
+  it("can filter regexp matches with a test function", () => {
+    test(new SearchQuery({
+      search: "a.",
+      regexp: true,
+      test(match, state, from, to) {
+        ist(match, state.doc.sliceString(from, to))
+        return match != "aa"
+      }
+    }), "[ab] aa [ac]")
+  })
 })
