xgettext-html

#!/usr/bin/env node
/*
 * xgettext-html: HTML gettext parser
 * Copyright (C) 2016 Pierre Ossman
 * Licensed under MPL 2.0 (see LICENSE.txt)
 */

var getopt = require('node-getopt');

var jsdom = require("jsdom");
var fs = require("fs");

opt = getopt.create([
    ['o' , 'output=FILE'      , 'write output to specified file'],
    ['h' , 'help'             , 'display this help'],
]).bindHelp().parseSystem();

var strings = {};

function addString(str, location) {
    if (str.length == 0) {
        return;
    }

    if (strings[str] === undefined) {
        strings[str] = {}
    }
    strings[str][location] = null;
}

// See https://html.spec.whatwg.org/multipage/dom.html#attr-translate
function process(elem, locator, enabled) {
    function isAnyOf(searchElement, items) {
        return items.indexOf(searchElement) !== -1;
    }

    if (elem.hasAttribute("translate")) {
        if (isAnyOf(elem.getAttribute("translate"), ["", "yes"])) {
            enabled = true;
        } else if (isAnyOf(elem.getAttribute("translate"), ["no"])) {
            enabled = false;
        }
    }

    if (enabled) {
        if (elem.hasAttribute("abbr") &&
            elem.tagName === "TH") {
            addString(elem.getAttribute("abbr"), locator(elem));
        }
        if (elem.hasAttribute("alt") &&
            isAnyOf(elem.tagName, ["AREA", "IMG", "INPUT"])) {
            addString(elem.getAttribute("alt"), locator(elem));
        }
        if (elem.hasAttribute("download") &&
            isAnyOf(elem.tagName, ["A", "AREA"])) {
            addString(elem.getAttribute("download"), locator(elem));
        }
        if (elem.hasAttribute("label") &&
            isAnyOf(elem.tagName, ["MENUITEM", "MENU", "OPTGROUP",
                                   "OPTION", "TRACK"])) {
            addString(elem.getAttribute("label"), locator(elem));
        }
        if (elem.hasAttribute("placeholder") &&
            isAnyOf(elem.tagName in ["INPUT", "TEXTAREA"])) {
            addString(elem.getAttribute("placeholder"), locator(elem));
        }
        if (elem.hasAttribute("title")) {
            addString(elem.getAttribute("title"), locator(elem));
        }
        if (elem.hasAttribute("value") &&
            elem.tagName === "INPUT" &&
            isAnyOf(elem.getAttribute("type"), ["reset", "button"])) {
            addString(elem.getAttribute("value"), locator(elem));
        }
    }

    for (var i = 0;i < elem.childNodes.length;i++) {
        node = elem.childNodes[i];
        if (node.nodeType === node.ELEMENT_NODE) {
            process(node, locator, enabled);
        } else if (node.nodeType === node.TEXT_NODE && enabled) {
            addString(node.data.trim(), locator(node));
        }
    }
}

for (var i = 0;i < opt.argv.length;i++) {
    var file;

    fn = opt.argv[i];
    file = fs.readFileSync(fn, "utf8");
    dom = new jsdom.JSDOM(file, { includeNodeLocations: true });
    body = dom.window.document.body;

    locator = function (elem) {
        offset = dom.nodeLocation(elem).startOffset;
        line = file.slice(0, offset).split("\n").length;
        return fn + ":" + line;
    };

    process(body, locator, true);
}

var output = "";

for (str in strings) {
    output += "#:";
    for (location in strings[str]) {
        output += " " + location;
    }
    output += "\n";

    output += "msgid " + JSON.stringify(str) + "\n";
    output += "msgstr \"\"\n";
    output += "\n";
}

fs.writeFileSync(opt.options.output, output);