feat: add getText() and generateText() methods (fix #1428) (#1875)

* move getTextBetween method

* add getText method

* refactoring

* refactoring

* refactoring

* move renderText to schema, add generateText method

* add GenerateText demo

* docs: update

* remove demo from html page

packages/core/src/helpers/generateText.ts

@@ -0,0 +1,29 @@
+import { Node } from 'prosemirror-model'
+import getSchema from './getSchema'
+import { Extensions, JSONContent, TextSerializer } from '../types'
+import getTextSeralizersFromSchema from './getTextSeralizersFromSchema'
+import getText from './getText'
+
+export default function generateText(
+  doc: JSONContent,
+  extensions: Extensions,
+  options?: {
+    blockSeparator?: string,
+    textSerializers?: Record<string, TextSerializer>,
+  },
+): string {
+  const {
+    blockSeparator = '\n\n',
+    textSerializers = {},
+  } = options || {}
+  const schema = getSchema(extensions)
+  const contentNode = Node.fromJSON(schema, doc)
+
+  return getText(contentNode, {
+    blockSeparator,
+    textSerializers: {
+      ...textSerializers,
+      ...getTextSeralizersFromSchema(schema),
+    },
+  })
+}

packages/core/src/helpers/getSchemaByResolvedExtensions.ts

@@ -77,6 +77,12 @@ export default function getSchemaByResolvedExtensions(extensions: Extensions): S
       })
     }
 
+    const renderText = getExtensionField<NodeConfig['renderText']>(extension, 'renderText', context)
+
+    if (renderText) {
+      schema.toText = renderText
+    }
+
     return [extension.name, schema]
   }))
 

packages/core/src/helpers/getText.ts

@@ -0,0 +1,18 @@
+import { TextSerializer } from '../types'
+import { Node as ProseMirrorNode } from 'prosemirror-model'
+import getTextBetween from './getTextBetween'
+
+export default function getText(
+  node: ProseMirrorNode,
+  options?: {
+    blockSeparator?: string,
+    textSerializers?: Record<string, TextSerializer>,
+  },
+) {
+  const range = {
+    from: 0,
+    to: node.content.size,
+  }
+
+  return getTextBetween(node, range, options)
+}

packages/core/src/helpers/getTextBetween.ts

@@ -0,0 +1,45 @@
+import { Range, TextSerializer } from '../types'
+import { Node as ProseMirrorNode } from 'prosemirror-model'
+
+export default function getTextBetween(
+  startNode: ProseMirrorNode,
+  range: Range,
+  options?: {
+    blockSeparator?: string,
+    textSerializers?: Record<string, TextSerializer>,
+  },
+): string {
+  const { from, to } = range
+  const {
+    blockSeparator = '\n\n',
+    textSerializers = {},
+  } = options || {}
+  let text = ''
+  let separated = true
+
+  startNode.nodesBetween(from, to, (node, pos, parent, index) => {
+    const textSerializer = textSerializers?.[node.type.name]
+
+    if (textSerializer) {
+      if (node.isBlock && !separated) {
+        text += blockSeparator
+        separated = true
+      }
+
+      text += textSerializer({
+        node,
+        pos,
+        parent,
+        index,
+      })
+    } else if (node.isText) {
+      text += node?.text?.slice(Math.max(from, pos) - pos, to - pos)
+      separated = false
+    } else if (node.isBlock && !separated) {
+      text += blockSeparator
+      separated = true
+    }
+  })
+
+  return text
+}

packages/core/src/helpers/getTextSeralizersFromSchema.ts

@@ -0,0 +1,9 @@
+import { Schema } from 'prosemirror-model'
+import { TextSerializer } from '../types'
+
+export default function getTextSeralizersFromSchema(schema: Schema): Record<string, TextSerializer> {
+  return Object.fromEntries(Object
+    .entries(schema.nodes)
+    .filter(([, node]) => node.spec.toText)
+    .map(([name, node]) => [name, node.spec.toText]))
+}