@stack.thefennec.dev/telegram-export-parser/dist/parsers/text-entities.js

TypeScript library for parsing Telegram Desktop's data export with full type safety

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.processTextEntities = exports.cleanupMarkdown = exports.ENTITY_PARSERS = void 0;
const types_1 = require("../types");
const parser_1 = require("../core/parser");
const base_1 = require("./base");
const actors_1 = require("./actors");
// =====================================================
// UTILITIES - IMMUTABLE PURE FUNCTIONS
// =====================================================
const getUrlForType = (type, text) => {
    switch (type) {
        case types_1.TEXT_ENTITY_TYPES.HASHTAG:
        case types_1.TEXT_ENTITY_TYPES.CASHTAG:
        case types_1.TEXT_ENTITY_TYPES.BOT_COMMAND:
            return `#${text}`;
        case types_1.TEXT_ENTITY_TYPES.EMAIL:
            return `mailto:${text}`;
        case types_1.TEXT_ENTITY_TYPES.PHONE:
            return `tel:${text.replace(/\s+/g, '')}`;
        case types_1.TEXT_ENTITY_TYPES.MENTION:
            return `https://t.me/${text}`;
        case types_1.TEXT_ENTITY_TYPES.MENTION_NAME:
            return `tg://user?id=${text}`;
        case types_1.TEXT_ENTITY_TYPES.LINK:
        default:
            return text;
    }
};
const escapeHtml = (text) => text
    .replace(/&/g, '&')
    .replace(/</g, '&lt;')
    .replace(/>/g, '&gt;')
    .replace(/"/g, '&quot;')
    .replace(/'/g, '&#39;');
const createBaseEntity = (raw, converter) => ({
    text: raw.text,
    toMarkdown: () => converter.toMarkdown(raw),
    toHTML: () => converter.toHTML(raw)
});
// =====================================================
// TEXT ENTITY PARSERS
// =====================================================
const plainTextEntityParser = (0, parser_1.createParser)({
    name: 'plain-text-entity',
    priority: 10, // Lowest priority - fallback
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.PLAIN,
    parse: (raw) => createBaseEntity(raw, {
        toMarkdown: (entity) => entity.text,
        toHTML: (entity) => escapeHtml(entity.text)
    })
});
const boldTextEntityParser = (0, parser_1.createParser)({
    name: 'bold-text-entity',
    priority: 80,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.BOLD,
    parse: (raw) => ({
        ...createBaseEntity(raw, {
            toMarkdown: (entity) => `**${entity.text}**`,
            toHTML: (entity) => `<strong>${escapeHtml(entity.text)}</strong>`
        }),
        type: types_1.TEXT_ENTITY_TYPES.BOLD
    })
});
const italicTextEntityParser = (0, parser_1.createParser)({
    name: 'italic-text-entity',
    priority: 75,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.ITALIC,
    parse: (raw) => ({
        ...createBaseEntity(raw, {
            toMarkdown: (entity) => `*${entity.text}*`,
            toHTML: (entity) => `<em>${escapeHtml(entity.text)}</em>`
        }),
        type: types_1.TEXT_ENTITY_TYPES.ITALIC
    })
});
const underlineTextEntityParser = (0, parser_1.createParser)({
    name: 'underline-text-entity',
    priority: 70,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.UNDERLINE,
    parse: (raw) => ({
        ...createBaseEntity(raw, {
            toMarkdown: (entity) => `__${entity.text}__`,
            toHTML: (entity) => `<u>${escapeHtml(entity.text)}</u>`
        }),
        type: types_1.TEXT_ENTITY_TYPES.UNDERLINE
    })
});
const strikethroughTextEntityParser = (0, parser_1.createParser)({
    name: 'strikethrough-text-entity',
    priority: 65,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.STRIKETHROUGH,
    parse: (raw) => ({
        ...createBaseEntity(raw, {
            toMarkdown: (entity) => `~~${entity.text}~~`,
            toHTML: (entity) => `<del>${escapeHtml(entity.text)}</del>`
        }),
        type: types_1.TEXT_ENTITY_TYPES.STRIKETHROUGH
    })
});
const spoilerTextEntityParser = (0, parser_1.createParser)({
    name: 'spoiler-text-entity',
    priority: 60,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.SPOILER,
    parse: (raw) => createBaseEntity(raw, {
        toMarkdown: (entity) => `||${entity.text}||`,
        toHTML: (entity) => `<span class="spoiler">${escapeHtml(entity.text)}</span>`
    })
});
const codeTextEntityParser = (0, parser_1.createParser)({
    name: 'code-text-entity',
    priority: 85,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.CODE,
    parse: (raw) => ({
        ...createBaseEntity(raw, {
            toMarkdown: (entity) => `\`${entity.text}\``,
            toHTML: (entity) => `<code>${escapeHtml(entity.text)}</code>`
        }),
        type: types_1.TEXT_ENTITY_TYPES.CODE,
        language: raw.language
    })
});
const preTextEntityParser = (0, parser_1.createParser)({
    name: 'pre-text-entity',
    priority: 90,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.PRE,
    parse: (raw) => ({
        ...createBaseEntity(raw, {
            toMarkdown: (entity) => {
                const language = entity.language?.trim() ?? '';
                return language
                    ? `\`\`\`${language}\n${entity.text}\n\`\`\``
                    : `\`\`\`\n${entity.text}\n\`\`\``;
            },
            toHTML: (entity) => {
                const language = entity.language?.trim() ?? '';
                const className = language ? ` class="language-${language}"` : '';
                return `<pre><code${className}>${escapeHtml(entity.text)}</code></pre>`;
            }
        }),
        type: types_1.TEXT_ENTITY_TYPES.PRE,
        language: raw.language
    })
});
const customEmojiEntityParser = (0, parser_1.createParser)({
    name: 'custom-emoji-entity',
    priority: 55,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.CUSTOM_EMOJI,
    parse: (raw) => ({
        ...createBaseEntity(raw, {
            toMarkdown: (entity) => entity.document_id
                ? `${entity.text} <!-- custom emoji: ${entity.document_id} -->`
                : entity.text,
            toHTML: (entity) => entity.document_id
                ? `<span class="custom-emoji" data-document-id="${escapeHtml(entity.document_id)}">${escapeHtml(entity.text)}</span>`
                : escapeHtml(entity.text)
        }),
        type: types_1.TEXT_ENTITY_TYPES.CUSTOM_EMOJI,
        documentURL: (0, base_1.parseExportedFile)(raw.document_id)
    })
});
const blockquoteEntityParser = (0, parser_1.createParser)({
    name: 'blockquote-entity',
    priority: 50,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.BLOCKQUOTE,
    parse: (raw) => ({
        ...createBaseEntity(raw, {
            toMarkdown: (entity) => {
                const quoteLines = entity.text.split('\n');
                const quotePrefix = entity.collapsed ? '**> ' : '> ';
                return quoteLines
                    .map((line, index) => index === 0 ? `${quotePrefix}${line}` : `> ${line}`)
                    .join('\n');
            },
            toHTML: (entity) => {
                const className = entity.collapsed ? 'blockquote collapsed' : 'blockquote';
                return `<blockquote class="${className}">${escapeHtml(entity.text)}</blockquote>`;
            }
        }),
        type: types_1.TEXT_ENTITY_TYPES.BLOCKQUOTE,
        collapsed: raw.collapsed ?? false
    })
});
const bankCardEntityParser = (0, parser_1.createParser)({
    name: 'bank-card-entity',
    priority: 45,
    canHandle: (raw) => raw.type === types_1.TEXT_ENTITY_TYPES.BANK_CARD,
    parse: (raw) => ({
        ...createBaseEntity(raw, {
            toMarkdown: (entity) => {
                const formattedCard = entity.text.replace(/(\d{4})(?=\d)/g, '$1 ');
                return `\`${formattedCard}\``;
            },
            toHTML: (entity) => {
                const formattedCard = entity.text.replace(/(\d{4})(?=\d)/g, '$1 ');
                return `<code class="bank-card">${escapeHtml(formattedCard)}</code>`;
            }
        }),
        type: types_1.TEXT_ENTITY_TYPES.BANK_CARD
    })
});
// =====================================================
// LINK ENTITY PARSERS
// =====================================================
const linkConverter = {
    toMarkdown: (entity) => {
        const url = entity.href ?? getUrlForType(entity.type, entity.text);
        return `[${entity.text}](${url})`;
    },
    toHTML: (entity) => {
        const url = escapeHtml(entity.href ?? getUrlForType(entity.type, entity.text));
        return `<a href="${url}">${escapeHtml(entity.text)}</a>`;
    }
};
const createLinkEntityParser = (name, type, priority) => (0, parser_1.createParser)({
    name,
    priority,
    canHandle: (raw) => raw.type === type,
    parse: (raw) => ({
        ...createBaseEntity(raw, linkConverter),
        type: raw.type,
        url: raw.href ?? getUrlForType(raw.type, raw.text)
    })
});
const textLinkEntityParser = createLinkEntityParser('text-link-entity', types_1.TEXT_ENTITY_TYPES.TEXT_LINK, 95);
const linkEntityParser = createLinkEntityParser('link-entity', types_1.TEXT_ENTITY_TYPES.LINK, 40);
const hashtagEntityParser = createLinkEntityParser('hashtag-entity', types_1.TEXT_ENTITY_TYPES.HASHTAG, 35);
const cashtagEntityParser = createLinkEntityParser('cashtag-entity', types_1.TEXT_ENTITY_TYPES.CASHTAG, 30);
const botCommandEntityParser = createLinkEntityParser('bot-command-entity', types_1.TEXT_ENTITY_TYPES.BOT_COMMAND, 25);
const emailEntityParser = createLinkEntityParser('email-entity', types_1.TEXT_ENTITY_TYPES.EMAIL, 20);
const phoneEntityParser = createLinkEntityParser('phone-entity', types_1.TEXT_ENTITY_TYPES.PHONE, 15);
// =====================================================
// MENTION ENTITY PARSERS
// =====================================================
const createMentionEntityParser = (name, type, priority) => (0, parser_1.createParser)({
    name,
    priority,
    canHandle: (raw) => raw.type === type,
    parse: (raw) => ({
        ...createBaseEntity(raw, linkConverter),
        type: raw.type,
        mention: (0, actors_1.parseUsernameMention)(raw.text, raw.user_id)
    })
});
const mentionEntityParser = createMentionEntityParser('mention-entity', types_1.TEXT_ENTITY_TYPES.MENTION, 85);
const mentionNameEntityParser = createMentionEntityParser('mention-name-entity', types_1.TEXT_ENTITY_TYPES.MENTION_NAME, 80);
// =====================================================
// EXPORTS
// =====================================================
exports.ENTITY_PARSERS = [
    // High priority - specific formatting
    textLinkEntityParser,
    preTextEntityParser,
    mentionEntityParser,
    mentionNameEntityParser,
    codeTextEntityParser,
    boldTextEntityParser,
    italicTextEntityParser,
    underlineTextEntityParser,
    strikethroughTextEntityParser,
    spoilerTextEntityParser,
    customEmojiEntityParser,
    blockquoteEntityParser,
    bankCardEntityParser,
    // Medium priority - links and commands
    linkEntityParser,
    hashtagEntityParser,
    cashtagEntityParser,
    botCommandEntityParser,
    emailEntityParser,
    phoneEntityParser,
    // Lowest priority - fallback
    plainTextEntityParser
];
// =====================================================
// UTILITY FUNCTIONS FOR PROCESSING
// =====================================================
const cleanupMarkdown = (text) => text
    .replace(/[ \t]+/g, ' ')
    .replace(/\n{4,}/g, '\n\n\n')
    .replace(/`\s+/g, '` ')
    .replace(/\s+`/g, ' `')
    .replace(/\*\*\s+/g, '**')
    .replace(/\s+\*\*/g, '**')
    .replace(/\*\s+/g, '*')
    .replace(/\s+\*/g, '*')
    .split('\n')
    .map(line => line.trim())
    .join('\n')
    .trim();
exports.cleanupMarkdown = cleanupMarkdown;
const processTextEntities = (entities) => ({
    markdown: (0, exports.cleanupMarkdown)(entities.map(entity => entity.toMarkdown()).join('')),
    html: entities.map(entity => entity.toHTML()).join('')
});
exports.processTextEntities = processTextEntities;
//# sourceMappingURL=text-entities.js.map