Files
leadchat/app/javascript/dashboard/helper/editorHelper.js
Sivin Varghese 863c033699 fix: Strip unsupported markdown formatting from canned responses (#13028)
# Pull Request Template

## Description

This PR fixes, 
1. **Issue with canned response insertion** - Canned responses with
formatting (bold, italic, code, lists, etc.) were not being inserted
into channels that don't support that formatting.
Now unsupported markdown syntax is automatically stripped based on the
channel's schema before insertion.
2. **Make image node optional** - Images are now stripped while paste.
9e269fca04
3. Enable **bold** and _italic_ for API channel

Fixes
https://linear.app/chatwoot/issue/CW-6091/editor-breaks-when-inserting-canned-response

## Type of change

- [x] Bug fix (non-breaking change which fixes an issue)

## How Has This Been Tested?

### Loom video
https://www.loom.com/share/9a5215dfef2949fcaa3871f51bdec4bb


## Checklist:

- [x] My code follows the style guidelines of this project
- [x] I have performed a self-review of my code
- [x] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] I have added tests that prove my fix is effective or that my
feature works
- [x] New and existing unit tests pass locally with my changes
- [x] Any dependent changes have been merged and published in downstream
modules
2025-12-09 09:15:35 +05:30

526 lines
18 KiB
JavaScript

import {
messageSchema,
MessageMarkdownTransformer,
MessageMarkdownSerializer,
} from '@chatwoot/prosemirror-schema';
import { replaceVariablesInMessage } from '@chatwoot/utils';
import * as Sentry from '@sentry/vue';
import { FORMATTING, MARKDOWN_PATTERNS } from 'dashboard/constants/editor';
import camelcaseKeys from 'camelcase-keys';
/**
* The delimiter used to separate the signature from the rest of the body.
* @type {string}
*/
export const SIGNATURE_DELIMITER = '--';
/**
* Parse and Serialize the markdown text to remove any extra spaces or new lines
*/
export function cleanSignature(signature) {
try {
// remove any horizontal rule tokens
signature = signature
.replace(/^( *\* *){3,} *$/gm, '')
.replace(/^( *- *){3,} *$/gm, '')
.replace(/^( *_ *){3,} *$/gm, '');
const nodes = new MessageMarkdownTransformer(messageSchema).parse(
signature
);
return MessageMarkdownSerializer.serialize(nodes);
} catch (e) {
// eslint-disable-next-line no-console
console.warn(e);
Sentry.captureException(e);
// The parser can break on some cases
// for example, Token type `hr` not supported by Markdown parser
return signature;
}
}
/**
* Adds the signature delimiter to the beginning of the signature.
*
* @param {string} signature - The signature to add the delimiter to.
* @returns {string} - The signature with the delimiter added.
*/
function appendDelimiter(signature) {
return `${SIGNATURE_DELIMITER}\n\n${cleanSignature(signature)}`;
}
/**
* Check if there's an unedited signature at the end of the body
* If there is, return the index of the signature, If there isn't, return -1
*
* @param {string} body - The body to search for the signature.
* @param {string} signature - The signature to search for.
* @returns {number} - The index of the last occurrence of the signature in the body, or -1 if not found.
*/
export function findSignatureInBody(body, signature) {
const trimmedBody = body.trimEnd();
const cleanedSignature = cleanSignature(signature);
// check if body ends with signature
if (trimmedBody.endsWith(cleanedSignature)) {
return body.lastIndexOf(cleanedSignature);
}
return -1;
}
/**
* Appends the signature to the body, separated by the signature delimiter.
*
* @param {string} body - The body to append the signature to.
* @param {string} signature - The signature to append.
* @returns {string} - The body with the signature appended.
*/
export function appendSignature(body, signature) {
const cleanedSignature = cleanSignature(signature);
// if signature is already present, return body
if (findSignatureInBody(body, cleanedSignature) > -1) {
return body;
}
return `${body.trimEnd()}\n\n${appendDelimiter(cleanedSignature)}`;
}
/**
* Removes the signature from the body, along with the signature delimiter.
*
* @param {string} body - The body to remove the signature from.
* @param {string} signature - The signature to remove.
* @returns {string} - The body with the signature removed.
*/
export function removeSignature(body, signature) {
// this will find the index of the signature if it exists
// Regardless of extra spaces or new lines after the signature, the index will be the same if present
const cleanedSignature = cleanSignature(signature);
const signatureIndex = findSignatureInBody(body, cleanedSignature);
// no need to trim the ends here, because it will simply be removed in the next method
let newBody = body;
// if signature is present, remove it and trim it
// trimming will ensure any spaces or new lines before the signature are removed
// This means we will have the delimiter at the end
if (signatureIndex > -1) {
newBody = newBody.substring(0, signatureIndex).trimEnd();
}
// let's find the delimiter and remove it
const delimiterIndex = newBody.lastIndexOf(SIGNATURE_DELIMITER);
if (
delimiterIndex !== -1 &&
delimiterIndex === newBody.length - SIGNATURE_DELIMITER.length // this will ensure the delimiter is at the end
) {
// if the delimiter is at the end, remove it
newBody = newBody.substring(0, delimiterIndex);
}
// return the value
return newBody;
}
/**
* Replaces the old signature with the new signature.
* If the old signature is not present, it will append the new signature.
*
* @param {string} body - The body to replace the signature in.
* @param {string} oldSignature - The signature to replace.
* @param {string} newSignature - The signature to replace the old signature with.
* @returns {string} - The body with the old signature replaced with the new signature.
*
*/
export function replaceSignature(body, oldSignature, newSignature) {
const withoutSignature = removeSignature(body, oldSignature);
return appendSignature(withoutSignature, newSignature);
}
/**
* Extract text from markdown, and remove all images, code blocks, links, headers, bold, italic, lists etc.
* Links will be converted to text, and not removed.
*
* @param {string} markdown - markdown text to be extracted
* @returns
*/
export function extractTextFromMarkdown(markdown) {
return markdown
.replace(/```[\s\S]*?```/g, '') // Remove code blocks
.replace(/`.*?`/g, '') // Remove inline code
.replace(/!\[.*?\]\(.*?\)/g, '') // Remove images before removing links
.replace(/\[([^\]]+)\]\([^)]+\)/g, '$1') // Remove links but keep the text
.replace(/#+\s*|[*_-]{1,3}/g, '') // Remove headers, bold, italic, lists etc.
.split('\n')
.map(line => line.trim())
.filter(Boolean)
.join('\n') // Trim each line & remove any lines only having spaces
.replace(/\n{2,}/g, '\n') // Remove multiple consecutive newlines (blank lines)
.trim(); // Trim any extra space
}
/**
* Scrolls the editor view into current cursor position
*
* @param {EditorView} view - The Prosemirror EditorView
*
*/
export const scrollCursorIntoView = view => {
// Get the current selection's head position (where the cursor is).
const pos = view.state.selection.head;
// Get the corresponding DOM node for that position.
const domAtPos = view.domAtPos(pos);
const node = domAtPos.node;
// Scroll the node into view.
if (node && node.scrollIntoView) {
node.scrollIntoView({ behavior: 'smooth', block: 'center' });
}
};
/**
* Returns a transaction that inserts a node into editor at the given position
* Has an optional param 'content' to check if the
*
* @param {Node} node - The prosemirror node that needs to be inserted into the editor
* @param {number} from - Position in the editor where the node needs to be inserted
* @param {number} to - Position in the editor where the node needs to be replaced
*
*/
export function insertAtCursor(editorView, node, from, to) {
if (!editorView) {
return undefined;
}
// This is a workaround to prevent inserting content into new line rather than on the exiting line
// If the node is of type 'doc' and has only one child which is a paragraph,
// then extract its inline content to be inserted as inline.
const isWrappedInParagraph =
node.type.name === 'doc' &&
node.childCount === 1 &&
node.firstChild.type.name === 'paragraph';
if (isWrappedInParagraph) {
node = node.firstChild.content;
}
let tr;
if (to) {
tr = editorView.state.tr.replaceWith(from, to, node).insertText(` `);
} else {
tr = editorView.state.tr.insert(from, node);
}
const state = editorView.state.apply(tr);
editorView.updateState(state);
editorView.focus();
return state;
}
/**
* Determines the appropriate node and position to insert an image in the editor.
*
* Based on the current editor state and the provided image URL, this function finds out the correct node (either
* a standalone image node or an image wrapped in a paragraph) and its respective position in the editor.
*
* 1. If the current node is a paragraph and doesn't contain an image or text, the image is inserted directly into it.
* 2. If the current node isn't a paragraph or it's a paragraph containing text, the image will be wrapped
* in a new paragraph and then inserted.
* 3. If the current node is a paragraph containing an image, the new image will be inserted directly into it.
*
* @param {Object} editorState - The current state of the editor. It provides necessary details like selection, schema, etc.
* @param {string} fileUrl - The URL of the image to be inserted into the editor.
* @returns {Object|null} An object containing details about the node to be inserted and its position. It returns null if no image node can be created.
* @property {Node} node - The ProseMirror node to be inserted (either an image node or a paragraph containing the image).
* @property {number} pos - The position where the new node should be inserted in the editor.
*/
export const findNodeToInsertImage = (editorState, fileUrl) => {
const { selection, schema } = editorState;
const { nodes } = schema;
const currentNode = selection.$from.node();
const {
type: { name: typeName },
content: { size, content },
} = currentNode;
const imageNode = nodes.image.create({ src: fileUrl });
if (!imageNode) return null;
const isInParagraph = typeName === 'paragraph';
const needsNewLine =
!content.some(n => n.type.name === 'image') && size !== 0 ? 1 : 0;
return {
node: isInParagraph ? imageNode : nodes.paragraph.create({}, imageNode),
pos: selection.from + needsNewLine,
};
};
/**
* Set URL with query and size.
*
* @param {Object} selectedImageNode - The current selected node.
* @param {Object} size - The size to set.
* @param {Object} editorView - The editor view.
*/
export function setURLWithQueryAndSize(selectedImageNode, size, editorView) {
if (selectedImageNode) {
// Create and apply the transaction
const tr = editorView.state.tr.setNodeMarkup(
editorView.state.selection.from,
null,
{
src: selectedImageNode.src,
height: size.height,
}
);
if (tr.docChanged) {
editorView.dispatch(tr);
}
}
}
/**
* Strips unsupported markdown formatting from content based on the editor schema.
* This ensures canned responses with rich formatting can be inserted into channels
* that don't support certain formatting (e.g., API channels don't support bold).
*
* @param {string} content - The markdown content to sanitize
* @param {Object} schema - The ProseMirror schema with supported marks and nodes
* @returns {string} - Content with unsupported formatting stripped
*/
export function stripUnsupportedFormatting(content, schema) {
if (!content || typeof content !== 'string') return content;
if (!schema) return content;
let sanitizedContent = content;
// Get supported marks and nodes from the schema
// Note: ProseMirror uses snake_case internally (code_block, bullet_list, etc.)
// but our FORMATTING constant uses camelCase (codeBlock, bulletList, etc.)
// We use camelcase-keys to normalize node names for comparison
const supportedMarks = Object.keys(schema.marks || {});
const nodeKeys = Object.keys(schema.nodes || {});
const nodeKeysObj = Object.fromEntries(nodeKeys.map(k => [k, true]));
const supportedNodes = Object.keys(camelcaseKeys(nodeKeysObj));
// Process each formatting type in order (codeBlock before code is important!)
MARKDOWN_PATTERNS.forEach(({ type, patterns }) => {
// Check if this format type is supported by the schema
const isMarkSupported = supportedMarks.includes(type);
const isNodeSupported = supportedNodes.includes(type);
// If not supported, strip the formatting
if (!isMarkSupported && !isNodeSupported) {
patterns.forEach(({ pattern, replacement }) => {
sanitizedContent = sanitizedContent.replace(pattern, replacement);
});
}
});
return sanitizedContent;
}
/**
* Content Node Creation Helper Functions for
* - mention
* - canned response
* - variable
* - emoji
*/
/**
* Centralized node creation function that handles the creation of different types of nodes based on the specified type.
* @param {Object} editorView - The editor view instance.
* @param {string} nodeType - The type of node to create ('mention', 'cannedResponse', 'variable', 'emoji').
* @param {Object|string} content - The content needed to create the node, which varies based on node type.
* @returns {Object|null} - The created ProseMirror node or null if the type is not supported.
*/
const createNode = (editorView, nodeType, content) => {
const { state } = editorView;
switch (nodeType) {
case 'mention': {
const mentionType = content.type || 'user';
const displayName = content.displayName || content.name;
const mentionNode = state.schema.nodes.mention.create({
userId: content.id,
userFullName: displayName,
mentionType,
});
return mentionNode;
}
case 'cannedResponse': {
// Strip unsupported formatting before parsing to ensure content can be inserted
// into channels that don't support certain markdown features (e.g., API channels)
const sanitizedContent = stripUnsupportedFormatting(
content,
state.schema
);
return new MessageMarkdownTransformer(state.schema).parse(
sanitizedContent
);
}
case 'variable':
return state.schema.text(`{{${content}}}`);
case 'emoji':
return state.schema.text(content);
case 'tool': {
return state.schema.nodes.tools.create({
id: content.id,
name: content.title,
});
}
default:
return null;
}
};
/**
* Object mapping types to their respective node creation functions.
*/
const nodeCreators = {
mention: (editorView, content, from, to) => ({
node: createNode(editorView, 'mention', content),
from,
to,
}),
cannedResponse: (editorView, content, from, to, variables) => {
const updatedMessage = replaceVariablesInMessage({
message: content,
variables,
});
const node = createNode(editorView, 'cannedResponse', updatedMessage);
return {
node,
from: node.textContent === updatedMessage ? from : from - 1,
to,
};
},
variable: (editorView, content, from, to) => ({
node: createNode(editorView, 'variable', content),
from,
to,
}),
emoji: (editorView, content, from, to) => ({
node: createNode(editorView, 'emoji', content),
from,
to,
}),
tool: (editorView, content, from, to) => ({
node: createNode(editorView, 'tool', content),
from,
to,
}),
};
/**
* Retrieves a content node based on the specified type and content, using a functional approach to select the appropriate node creation function.
* @param {Object} editorView - The editor view instance.
* @param {string} type - The type of content node to create ('mention', 'cannedResponse', 'variable', 'emoji').
* @param {string|Object} content - The content to be transformed into a node.
* @param {Object} range - An object containing 'from' and 'to' properties indicating the range in the document where the node should be placed.
* @param {Object} variables - Optional. Variables to replace in the content, used for 'cannedResponse' type.
* @returns {Object} - An object containing the created node and the updated 'from' and 'to' positions.
*/
export const getContentNode = (
editorView,
type,
content,
{ from, to },
variables
) => {
const creator = nodeCreators[type];
return creator
? creator(editorView, content, from, to, variables)
: { node: null, from, to };
};
/**
* Get the formatting configuration for a specific channel type.
* Returns the appropriate marks, nodes, and menu items for the editor.
*
* @param {string} channelType - The channel type (e.g., 'Channel::FacebookPage', 'Channel::WebWidget')
* @returns {Object} The formatting configuration with marks, nodes, and menu properties
*/
export function getFormattingForEditor(channelType) {
return FORMATTING[channelType] || FORMATTING['Context::Default'];
}
/**
* Menu Positioning Helpers
* Handles floating menu bar positioning for text selection in the editor.
*/
const MENU_CONFIG = { H: 46, W: 300, GAP: 10 };
/**
* Calculate selection coordinates with bias to handle line-wraps correctly.
* @param {EditorView} editorView - ProseMirror editor view
* @param {Selection} selection - Current text selection
* @param {DOMRect} rect - Container bounding rect
* @returns {{start: Object, end: Object, selTop: number, onTop: boolean}}
*/
export function getSelectionCoords(editorView, selection, rect) {
const start = editorView.coordsAtPos(selection.from, 1);
const end = editorView.coordsAtPos(selection.to, -1);
const selTop = Math.min(start.top, end.top);
const spaceAbove = selTop - rect.top;
const onTop =
spaceAbove > MENU_CONFIG.H + MENU_CONFIG.GAP || end.bottom > rect.bottom;
return { start, end, selTop, onTop };
}
/**
* Calculate anchor position based on selection visibility and RTL direction.
* @param {Object} coords - Selection coordinates from getSelectionCoords
* @param {DOMRect} rect - Container bounding rect
* @param {boolean} isRtl - Whether text direction is RTL
* @returns {number} Anchor x-position for menu
*/
export function getMenuAnchor(coords, rect, isRtl) {
const { start, end, onTop } = coords;
if (!onTop) return end.left;
// If start of selection is visible, align to text. Else stick to container edge.
if (start.top >= rect.top) return isRtl ? start.right : start.left;
return isRtl ? rect.right - MENU_CONFIG.GAP : rect.left + MENU_CONFIG.GAP;
}
/**
* Calculate final menu position (left, top) within container bounds.
* @param {Object} coords - Selection coordinates from getSelectionCoords
* @param {DOMRect} rect - Container bounding rect
* @param {boolean} isRtl - Whether text direction is RTL
* @returns {{left: number, top: number, width: number}}
*/
export function calculateMenuPosition(coords, rect, isRtl) {
const { start, end, selTop, onTop } = coords;
const anchor = getMenuAnchor(coords, rect, isRtl);
// Calculate Left: shift by width if RTL, then make relative to container
const rawLeft = (isRtl ? anchor - MENU_CONFIG.W : anchor) - rect.left;
// Ensure menu stays within container bounds
const left = Math.min(Math.max(0, rawLeft), rect.width - MENU_CONFIG.W);
// Calculate Top: align to selection or bottom of selection
const top = onTop
? Math.max(-26, selTop - rect.top - MENU_CONFIG.H - MENU_CONFIG.GAP)
: Math.max(start.bottom, end.bottom) - rect.top + MENU_CONFIG.GAP;
return { left, top, width: MENU_CONFIG.W };
}
/* End Menu Positioning Helpers */