rtf-stream-parser/dist/src/de-encapsulate.js

Stream Transform class to tokenize RTF, and another to de-encapsulate text or HTML

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.DeEncapsulate = void 0;
const checkVersion_1 = require("./features/checkVersion");
const countTokens_1 = require("./features/countTokens");
const handleCharacterSet_1 = require("./features/handleCharacterSet");
const handleControlsAndDestinations_1 = require("./features/handleControlsAndDestinations");
const handleDeEncapsulation_1 = require("./features/handleDeEncapsulation");
const handleFonts_1 = require("./features/handleFonts");
const handleGroupState_1 = require("./features/handleGroupState");
const handleOutput_1 = require("./features/handleOutput");
const handleUnicodeSkip_1 = require("./features/handleUnicodeSkip");
const textEscapes_1 = require("./features/textEscapes");
const ProcessTokens_1 = require("./ProcessTokens");
const deEncExtraDefaultOptions = {
    htmlEncodeNonAscii: false,
    htmlFixContentType: false,
    htmlPreserveSpaces: false,
    mode: 'either',
    prefix: false,
    outlookQuirksMode: false,
};
function htmlEntityEncode(str) {
    const pieces = [];
    let ascii = true;
    for (const char of str) {
        const codepoint = char.codePointAt(0);
        if (codepoint === 0xA0) {
            ascii = false;
            pieces.push(' ');
        }
        else if (codepoint > 0x7F) {
            ascii = false;
            pieces.push('&#x' + codepoint.toString(16) + ';');
        }
        else {
            pieces.push(char);
        }
    }
    const out = ascii ? str : pieces.join('');
    return out;
}
const rxCharset = /(\bcharset=)([\w-]+)(")/i;
class DeEncapsulate extends ProcessTokens_1.ProcessTokens {
    constructor(options) {
        super(options);
        this._featureHandlers = [
            countTokens_1.countTokens,
            checkVersion_1.checkVersion,
            handleGroupState_1.handleGroupState,
            handleUnicodeSkip_1.handleUnicodeSkip,
            handleControlsAndDestinations_1.handleControlsAndDestinations,
            handleCharacterSet_1.handleCharacterSet,
            handleFonts_1.handleFonts,
            handleDeEncapsulation_1.handleDeEncapsulation,
            handleOutput_1.handleOutput,
            textEscapes_1.handleTextEscapes,
        ];
        this._rootState = { uc: 1, groupDepth: 0, destDepth: 0, destGroupDepth: 0 };
        this._state = this._rootState;
        this._fromhtml = false;
        this._fromtext = false;
        this._didHtmlCharsetReplace = false;
        this._options = Object.assign(Object.assign(Object.assign({}, ProcessTokens_1.procTokensDefaultOptions), deEncExtraDefaultOptions), options);
    }
    get isHtml() {
        return this._fromhtml;
    }
    get isText() {
        return this._fromtext;
    }
    get originalHtmlCharset() {
        return this._originalHtmlCharset;
    }
    _getOutputAsString(data, font) {
        var _a;
        let [outStr, areSymbolFontCodepoints] = super._getOutputAsString(data, font);
        if (this._fromhtml) {
            const insideHtmltag = !!((_a = this._state.allDestinations) === null || _a === void 0 ? void 0 : _a['htmltag']);
            if (insideHtmltag) {
                if (this._options.htmlFixContentType && !this._didHtmlCharsetReplace) {
                    outStr = outStr.replace(rxCharset, (match, pre, charset, post) => {
                        this._didHtmlCharsetReplace = true;
                        this._originalHtmlCharset = charset;
                        return pre + 'UTF-8' + post;
                    });
                }
            }
            else {
                outStr = outStr.replace(/<|>/g, match => {
                    return match === '<' ? '&lt;' : '&gt;';
                });
                if (this._options.htmlPreserveSpaces) {
                    if (outStr === ' ') {
                        outStr = '\u00A0';
                    }
                    else {
                        outStr = outStr
                            .replace(/  +/g, match => ' ' + '\u00A0'.repeat(match.length - 1))
                            .replace(/^ +/, match => '\u00A0'.repeat(match.length))
                            .replace(/ +$/, match => '\u00A0'.repeat(match.length));
                    }
                }
                if (this._options.htmlEncodeNonAscii) {
                    outStr = htmlEntityEncode(outStr);
                }
            }
        }
        return [outStr, areSymbolFontCodepoints];
    }
    _getCurrentFont() {
        const allDests = this._state.allDestinations || {};
        const insideHtmltag = !!allDests['htmltag'];
        return insideHtmltag ? undefined : super._getCurrentFont();
    }
}
exports.DeEncapsulate = DeEncapsulate;
exports.default = DeEncapsulate;