unified

Project: unified-doc/unified-doc

Package: unified-doc-util-mark@1.1.4

  1. unified-doc hast util to mark text nodes.
  1. unified 180
  2. util 145
  3. unist 132
  4. hast 75
  5. file 33
  6. content 20
  7. document 10
  8. doc 8
  9. highlight 7
  10. nlp 6
  11. mark 4
  12. annotate 4

unified-doc-util-mark

unified-doc hast utility to mark text nodes.


Install

npm install unified-doc-util-mark

Use

Given a hast tree parsed from some HTML content:

import mark from 'unified-doc-util-mark';

// html: '<blockquote><strong>some</strong>\ncontent</blockquote>';
const hast = {
  type: 'root',
  children: [
    {
      type: 'element',
      tagName: 'blockquote',
      children: [
        {
          type: 'element',
          tagName: 'strong',
          children: [
            {
              type: 'text',
              value: 'some',
            },
          ],
        },
        {
          type: 'text',
          value: '\ncontent',
        },
      ],
    },
  ],
};

const marks = [
  { id: 'a', start: 1, end: 4, classNames: ['a', 'b'] },
];

expect(mark(hast, marks)).toEqual({
  type: 'root',
  children: [
    {
      type: 'element',
      tagName: 'blockquote',
      children: [
        {
          type: 'element',
          tagName: 'strong',
          children: [
            {
              type: 'text',
              value: 's',
            },
            {
              type: 'element',
              tagName: 'mark',
              properties: {
                dataMarkId: 'a',
                id: 'a',
                classNames: ['a', 'b'],
              },
              children: [
                {
                  type: 'text',
                  value: 'ome',
                },
              ],
            },
          ],
        },
        {
          type: 'text',
          value: '\ncontent',
        },
      ],
    },
  ],
});

API

mark(hast, marks)

Interface

function mark(hast: Hast, marks: Mark[]): Hast

Accepts a hast tree and applies marks to overlapping text nodes. Returns a new tree.

mark also supports overlapping marks, and applying custom properties to marked nodes (classNames, style, dataset).

const marks = [
  {
    id: 'a',
    start: 3,
    end: 8,
    classNames: ['a', 'b'],
    dataset: { category: 'A' }
  },
  {
    id: 'b',
    start: 6,
    end: 10,
    style: { background: 'red' }
  },
];

expect(mark(hast, marks)).toEqual({
  type: 'root',
  children: [
    {
      type: 'element',
      tagName: 'blockquote',
      children: [
        {
          type: 'element',
          tagName: 'strong',
          children: [
            {
              type: 'text',
              value: 'som',
            },
            {
              type: 'element',
              tagName: 'mark',
              properties: {
                id: 'a',
                classNames: ['a', 'b'],
                dataMarkId: 'a',
                dataCategory: 'A',
              },
              children: {
                type: 'text',
                value: 'e',
              },
            },
          ],
        },
        {
          type: 'element',
          tagName: 'mark',
          properties: {
            classNames: ['a', 'b'],
            dataMarkId: 'a',
            dataCategory: 'A',
          },
          children: [
            {
              type: 'text',
              value: '\nc',
            },
            {
              type: 'element',
              tagName: 'mark',
              properties: {
                id: 'b',
                dataMarkId: 'b',
                style: 'background: red',
              },
              children: [
                {
                  type: 'text',
                  value: 'on',
                },
              ],
            },
          ],
          type: 'text',
          value: '\ncontent',
        },
        {
          type: 'element',
          tagName: 'mark',
          properties: {
            dataMarkId: 'b',
            style: 'background: red; color white',
          },
          children: [
            {
              type: 'text',
              value: 'te',
            },
          ],
          {
            type: 'text',
            value: 'nt',
          },
        },
      ],
    },
  ],
});
interface Mark {
  /** unique ID for mark (required for mark algorithm to work) */
  id: string;
  /** start offset of the mark relative to `textContent` of the `hast` */
  start: number;
  /** end offset of the mark relative to `textContent` of the `hast` */
  end: number;
  /** apply optional CSS classnames to marked nodes */
  classNames?: string[];
  /** apply optional dataset attributes (i.e. `data-*`) to marked nodes */
  dataset?: Record<string, any>;
  /** contextual data can be stored here */
  data?: Record<string, any>;
  /** apply optional styles to marked nodes */
  style?: Record<string, any>;
}

A Mark is an object that contains at least the id, start and end properties, which provides positional information on segments of text nodes that can be marked. Marks are useful in various document features:

mark nodes can be visually customized by specifying the optional classNames, style and dataset properties. All other data maybe organized under the data property.

The start and end properties of a Mark indicate text offset values relative to the textContent of the provided hast tree. The mark algorithm uses these offsets to determine how to insert mark nodes in the tree only within the specified offset range, while preserving the semantic structure of the tree.

The following pseudocode helps visualize this behavior:

const html = '<blockquote><strong>some</strong>\ncontent</blockquote>'jj
const textContent = 'some\ncontent';
const textNodes = ['some', '\ncontent'];
const textOffsets = [
  { start: 0, end: 4 }, // "[some]\ncontent"
  { start: 4, end: 12 }, // "some[\ncontent]"
];

const marks = [
  { id: 'a', classNames: ['a', 'b'], start: 3, end: 8 },
  { id: 'b', style: { background: 'red' }, start: 6, end: 10 },
];
const markedTextSegments = [
  {
    value: 'som', // "[som]e" text node
    textOffset: { start: 0, end: 3},
    markIds: [], // not marked
  },
  {
    value: 'e', // "som[e]" text node
    textOffset: { start: 3, end: 4},
    markIds: ['a'], // marked by 'a'
  },
  {
    value: '\nc', // "[\nc]ontent" text node
    textOffset: { start: 4, end: 6},
    markIds: ['a'], // marked by 'a'
  },
  {
    value: 'on', // "\nc[on]tent" text node
    textOffset: { start: 6, end: 8},
    markIds: ['a', 'b'], // marked by 'a' + 'b'
  },
  {
    value: 'te', // "\ncon[te]nt" text node
    textOffset: { start: 8, end: 10},
    markIds: ['b'], // marked by 'b'
  },
  {
    value: 'nt', // "\nconte[nt]" text node
    textOffset: { start: 10, end: 12},
    markIds: [], // not marked
  },
];

// marked hast tree based on markedTextSegments computed above
const marked = mark(hast, marks);