mirror of
https://github.com/bulwarkmail/webmail.git
synced 2026-10-31 00:12:39 +00:00
feat: detect unfilled placeholders and unfinished text in a message
This commit is contained in:
Linus Rath
@@ -0,0 +1,84 @@
|
||||
import { describe, it, expect } from 'vitest';
|
||||
import { findUnfilledPlaceholders, stripEmbeddedSignatureHtml } from '../placeholder-check';
|
||||
|
||||
describe('findUnfilledPlaceholders', () => {
|
||||
it('finds template placeholders in the subject and the body', () => {
|
||||
expect(findUnfilledPlaceholders('Follow-up: {{subject}}', 'Hi {{ recipient_name }},')).toEqual([
|
||||
'{{subject}}',
|
||||
'{{ recipient_name }}',
|
||||
]);
|
||||
});
|
||||
|
||||
it('finds upper-case fill-ins in brackets', () => {
|
||||
expect(findUnfilledPlaceholders('', 'See you at [TIME]. It costs [PRICE] per [UNIT_NAME].'))
|
||||
.toEqual(['[TIME]', '[PRICE]', '[UNIT_NAME]']);
|
||||
expect(findUnfilledPlaceholders('', 'Best regards, [YOUR NAME]')).toEqual(['[YOUR NAME]']);
|
||||
expect(findUnfilledPlaceholders('', 'Treffpunkt: [ÜBERGABEORT]')).toEqual(['[ÜBERGABEORT]']);
|
||||
});
|
||||
|
||||
it('finds fill-ins that ask the sender to write something', () => {
|
||||
expect(findUnfilledPlaceholders('', 'Best regards,\n[Your Name]\n[Insert date]')).toEqual([
|
||||
'[Your Name]',
|
||||
'[Insert date]',
|
||||
]);
|
||||
expect(findUnfilledPlaceholders('', 'Viele Grüße\n[Ihr Name], [Datum einfügen]')).toEqual([
|
||||
'[Ihr Name]',
|
||||
'[Datum einfügen]',
|
||||
]);
|
||||
});
|
||||
|
||||
it('finds unfinished-text markers', () => {
|
||||
expect(findUnfilledPlaceholders('', 'The venue is TBD. TODO: add the agenda. Room XXX')).toEqual([
|
||||
'TBD',
|
||||
'TODO',
|
||||
'XXX',
|
||||
]);
|
||||
});
|
||||
|
||||
it('reports a marker in brackets once', () => {
|
||||
expect(findUnfilledPlaceholders('', 'Agenda: [TODO]')).toEqual(['[TODO]']);
|
||||
});
|
||||
|
||||
it('lists each placeholder once', () => {
|
||||
expect(findUnfilledPlaceholders('{{name}}', 'Hi {{name}}, at [TIME] or [TIME]?')).toEqual([
|
||||
'{{name}}',
|
||||
'[TIME]',
|
||||
]);
|
||||
});
|
||||
|
||||
it('leaves ordinary text alone', () => {
|
||||
const text = [
|
||||
'See footnote [1] and tick [x].',
|
||||
'Option [A] is fine, so is [3D] and [2026].',
|
||||
'I still have a todo list and the date is tba.',
|
||||
'Read [the docs](https://example.com) and [Your account](https://example.com/me).',
|
||||
'The mastodon handle is TODOS and the size is XXXL.',
|
||||
'Signed, [Yours truly].',
|
||||
'A {single} brace and {{}} are not placeholders.',
|
||||
].join('\n');
|
||||
expect(findUnfilledPlaceholders('', text)).toEqual([]);
|
||||
});
|
||||
|
||||
it('ignores bracket tags and markers in the subject', () => {
|
||||
expect(findUnfilledPlaceholders('Re: [EXTERNAL] Venue TBD', 'Thanks!')).toEqual([]);
|
||||
});
|
||||
});
|
||||
|
||||
describe('stripEmbeddedSignatureHtml', () => {
|
||||
it('drops the bracketed signature range and keeps the rest', () => {
|
||||
const html = '<p>Hello</p><p data-signature-block="start"></p><p>[COMPANY] TODO</p><p data-signature-block="end"></p><p>After</p>';
|
||||
const stripped = stripEmbeddedSignatureHtml(html);
|
||||
expect(stripped).toContain('Hello');
|
||||
expect(stripped).toContain('After');
|
||||
expect(stripped).not.toContain('[COMPANY]');
|
||||
});
|
||||
|
||||
it('drops a signature atom without range markers', () => {
|
||||
const html = '<p>Hello [TIME]</p><div data-signature-block-node="">[COMPANY]</div>';
|
||||
expect(stripEmbeddedSignatureHtml(html)).toBe('<p>Hello [TIME]</p>');
|
||||
});
|
||||
|
||||
it('returns a body without a signature unchanged', () => {
|
||||
expect(stripEmbeddedSignatureHtml('<p>Hello</p>')).toBe('<p>Hello</p>');
|
||||
});
|
||||
});
|
||||
@@ -0,0 +1,83 @@
|
||||
// Finds text the sender meant to replace before sending: template
|
||||
// placeholders, bracketed fill-ins that AI drafts and copied templates leave
|
||||
// behind, and markers like TODO.
|
||||
|
||||
// Template syntax, as in template-utils. Spaces inside the braces are allowed
|
||||
// because templates pasted from other tools often have them.
|
||||
const TEMPLATE_PLACEHOLDER = /\{\{\s*[\w.-]+\s*\}\}/g;
|
||||
|
||||
// [TIME], [PRICE], [YOUR NAME]. Not followed by "(", which would make it a
|
||||
// Markdown link.
|
||||
const CAPS_BRACKET = /\[[\p{Lu}\d][\p{Lu}\d _./-]{0,28}[\p{Lu}\d]\](?!\()/gu;
|
||||
|
||||
// [Your Name], [Insert date], [Ihr Name], [Datum einfügen].
|
||||
const PROMPT_BRACKET = /\[(?:(?:your|insert|enter|add|ihr|ihre|ihren|dein|deine|deinen)\s[^[\]\n]{1,40}|[^[\]\n]{1,40}\seinfügen)\](?!\()/giu;
|
||||
|
||||
// Upper case only: "todo" or "tba" in a sentence is usually meant.
|
||||
const UNFINISHED_WORD = /(?<![\p{L}\d_])(?:TODO|TBD|TBA|TBC|FIXME|XXX)(?![\p{L}\d_])/gu;
|
||||
|
||||
const SIGNATURE_START_SELECTOR = '[data-signature-block="separator"], [data-signature-block="start"]';
|
||||
const SIGNATURE_END_SELECTOR = '[data-signature-block="end"]';
|
||||
|
||||
interface Match {
|
||||
start: number;
|
||||
end: number;
|
||||
text: string;
|
||||
}
|
||||
|
||||
function collect(text: string, patterns: RegExp[]): Match[] {
|
||||
const matches: Match[] = [];
|
||||
for (const pattern of patterns) {
|
||||
for (const m of text.matchAll(pattern)) {
|
||||
if (pattern === CAPS_BRACKET && (m[0].match(/\p{Lu}/gu)?.length ?? 0) < 2) continue;
|
||||
matches.push({ start: m.index, end: m.index + m[0].length, text: m[0] });
|
||||
}
|
||||
}
|
||||
// "[TODO]" is a bracket and a word; keep the outer match only.
|
||||
matches.sort((a, b) => a.start - b.start || b.end - a.end);
|
||||
const kept: Match[] = [];
|
||||
for (const match of matches) {
|
||||
const last = kept[kept.length - 1];
|
||||
if (last && match.start < last.end) continue;
|
||||
kept.push(match);
|
||||
}
|
||||
return kept;
|
||||
}
|
||||
|
||||
/**
|
||||
* The unfilled placeholders in a message, each once, subject first. The
|
||||
* subject is only checked for template placeholders: it often carries tags
|
||||
* like [EXTERNAL] or words from the original message, which the sender did
|
||||
* not write. `bodyText` should be the text the sender wrote, without quoted
|
||||
* replies and the signature.
|
||||
*/
|
||||
export function findUnfilledPlaceholders(subject: string, bodyText: string): string[] {
|
||||
const found = [
|
||||
...collect(subject, [TEMPLATE_PLACEHOLDER]),
|
||||
...collect(bodyText, [TEMPLATE_PLACEHOLDER, CAPS_BRACKET, PROMPT_BRACKET, UNFINISHED_WORD]),
|
||||
].map((m) => m.text);
|
||||
return Array.from(new Set(found));
|
||||
}
|
||||
|
||||
/**
|
||||
* A compose body without its embedded signature. A signature is the same in
|
||||
* every message, so a placeholder in it would stop every send.
|
||||
*/
|
||||
export function stripEmbeddedSignatureHtml(html: string): string {
|
||||
if (!html.includes('data-signature-block')) return html;
|
||||
const doc = new DOMParser().parseFromString(`<body>${html}</body>`, 'text/html');
|
||||
const startEl = doc.querySelector(SIGNATURE_START_SELECTOR);
|
||||
if (startEl) {
|
||||
const endEl = doc.querySelector(SIGNATURE_END_SELECTOR);
|
||||
let cursor: Node | null = startEl;
|
||||
while (cursor) {
|
||||
const next: Node | null = cursor.nextSibling;
|
||||
const isEnd = cursor === endEl;
|
||||
cursor.parentNode?.removeChild(cursor);
|
||||
if (isEnd) break;
|
||||
cursor = next;
|
||||
}
|
||||
}
|
||||
doc.querySelectorAll('[data-signature-block-node]').forEach((el) => el.remove());
|
||||
return doc.body.innerHTML;
|
||||
}
|
||||
Reference in New Issue
Block a user