kjhf/Splashtag Titles.html

## Splashtag Titles.html

<!DOCTYPE html>
<html lang="en">
  <head>
    <meta charset="utf-8">
    <title>Splashtag Titles</title>
    <style>

:root {
    color      : #000000;
    font-family: Arial, sans-serif;
    font-size  : 16px;
}

body {
    background: #ffffff;
}

    </style>
    <script>

// Load all files from an <input type="file"> element
async function loadFiles(id) {
    let files  = document.getElementById(id).files;
    let loaded = [];
    let errors = [];

    // No files selected
    if (files.length == 0)
        return { loaded: [], errors: [ "No file selected for " + id ] };

    // Load all files
    for (let file of files) {
        try {
            let data = JSON.parse(await file.text());
            data.locale = file.name.replace(".json", "");
            loaded.push(data);
        }
        catch(err) { errors.push(`Could not load file ${file.name} because: ${err}`); }
    }

    return { loaded: loaded, errors: errors };
}

// Parse the gendered components of an adjective
function parseAdjective(text) {
    let ret = [ "", "", "", "" ];

    // Null-transform text
    let alone = text.replace(/\[.*?\]/g, "");

    // Process all meta commands in the text
    while (text.indexOf("[") != -1) {
        let x      = text.indexOf("[") + 1;
        let y      = text.indexOf("]", x);
        let params = text.substring(x, y);

        // Append everything before the opening square bracket
        let part = text.substring(0, x - 1);
        for (let z = 0; z < 4; z++)
            ret[z] += part;

        // Retain only the part after the closing square bracket
        text = text.substring(y + 1);

        // The meta command is not a gendered set
        if (
            !params.startsWith("group=00c9 type=0005") ||
            params.endsWith("params=")
        ) continue;

        // Convert the parameter text into 16-bit numbers
        let shorts = new Uint16Array(new Uint8Array(params
            .substring(params.indexOf("params=") + 7)
            .split(" ").map(b=>parseInt(b, 16))
        ).buffer);

        // Process gendered components
        params = [];
        for (x = 0; x < shorts.length;) {
            let length = shorts[x++] / 2;
            params.push(String.fromCharCode(
                ... shorts.slice(x, x + length)));
            x += length;
        }
        if (params.length != 4)
            throw "Parameter length mismatch";

        // Append gendered components
        for (x = 0; x < 4; x++)
            ret[x] += params[x];
    }

    // Append any remaining text
    for (let x = 0; x < 4; x++)
        ret[x] += text;

    // Remove unused variants
    for (let x = 0; x < 4; x++) {
        if (
            x != 0 && ret[x] == alone ||
            x == 0 && ret[3] != alone && ret[x] == alone
        ) ret[x] = "";
    }

    // Neutral variant if only gendered is available
    if (!ret[2] && ret[3])
        ret[2] = alone;

    return ret;
}

// Process locales file for title strings
function parseLocales(locales) {
    let ret = {
        adjectives: {},
        subjects  : {}
    };

    // Process all locales
    for (let locale of locales) {

        // Process adjectives
        for (let adj of Object.entries(locale["CommonMsg/Byname/BynameAdjective"])) {
            let id   = parseInt(adj[0]);
            let item = ret.adjectives[id];
            if (!item)
                item = ret.adjectives[id] = {};
            item[locale.locale] = parseAdjective(adj[1]);
        }

        // Process subjects
        for (let sub of Object.entries(locale["CommonMsg/Byname/BynameSubject"])) {
            let id = sub[0].split("_");
            id[0] = parseInt(id[0]);
            let item = ret.subjects[id[0]];
            if (!item)
                item = ret.subjects[id[0]] = {};
            if (!item[locale.locale])
                item[locale.locale] = new Array(2);
            item[locale.locale][id[1]] = sub[1].replace(/\[.*?\]/g, "");

            // Meta command patterns. Left here for reference if needed.
            //if (sub[1].indexOf("[") == -1) continue;
            //let x = sub[1].indexOf("[") + 1;
            //let params = sub[1].substring(x, sub[1].indexOf("]", x));
            //if (
            //    (
            //        params.indexOf("group=00c9") == -1 ||
            //        params.indexOf("type=0005" ) == -1
            //    ) &&
            //    !(
            //        params.startsWith("ruby=") ||
            //        params == "/ruby"
            //    ) &&
            //    !params.endsWith("params=") &&
            //    !params.endsWith("params=01 ff ff 00") &&
            //    !params.endsWith("params=02 ff ff 00") &&
            //    !params.endsWith("params=03 ff ff 00") &&
            //    !params.endsWith("params=04 ff ff 00")
            //) console.log(locale.locale, sub[0], sub[1]);
        }

        // Eliminate duplicate subjects
        for (let sub of Object.values(ret.subjects)) {
            for (let lang of Object.values(sub)) {
                if (lang[1] == lang[0]) {
                    let dupe = DUPE_SUBJECTS[locale.locale] || DUPE_SUBJECTS.USen;
                    lang[0] += dupe[0];
                    lang[1] += dupe[1];
                }
            }
        }

    }

    return ret;
}

// Duplicate subject suffixes
let DUPE_SUBJECTS = {
    "EUde": [ " (m)", " (w)" ],
    "USen": [ " (m)", " (f)" ]
};

// Sorting comparator
function sortOrder(a, b) {
    return a.USen[0].toLowerCase().localeCompare(b.USen[0].toLowerCase());
}

// Produce output to the user
function output(id, list, useEnglishOnly, oldFormatObtain) {
    let out = [];
    if (useEnglishOnly) {
        /*
            !Title
            !How to obtain
            !Added (season)
        */
        for (let item of list) {
            if (oldFormatObtain) {
                out.push("|-\n" +
                    "|" + item.USen.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                    "|" + item.HowToGet + ",<br>Season " + item.Season
                );
            } else {
                out.push("|-\n" +
                    "|" + item.USen.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                    "|" + item.HowToGet + "\n" +
                    "|" + item.Season
                );
            }
        }
    }
    else {
        /*
            !English
            !Japanese
            !Dutch
            !French<br>(Canada)
            !French<br>(France)
            !German
            !Italian*
            !Russian
            !Spanish<br>(Latin America)
            !Spanish<br>(Spain)
            !Chinese<br>(Simplified)
            !Chinese<br>(Traditional)
            !Korean
        */
        for (let item of list) {
            out.push("|-\n" +
                "|" + item.USen.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.JPja.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.EUnl.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.USfr.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.EUfr.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.EUde.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.EUit.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.EUru.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.USes.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.EUes.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.CNzh.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.TWzh.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
                "|" + item.KRko.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>")
            );
            if (oldFormatObtain) {
                out.push("|" + item.HowToGet + ",<br>Season " + item.Season);
            }
        }
    }
    document.getElementById(id).value = out.join("\n");
}

// Process all selected files
async function process() {

    let useEnglishOnly = document.getElementById("englishOnly").checked;
    let useOtherLanguages = document.getElementById("otherLanguages").checked;
    let oldFormatObtain = document.getElementById("oldFormatObtain").checked;

    if (!(useEnglishOnly ^ useOtherLanguages)) {
        throw new Error(`Bad state for the checkboxes. useEnglishOnly=${useEnglishOnly} useOtherLanguages=${useOtherLanguages}`);
    }

    // Load files into memory
    let locales    = await loadFiles("locales");
    let adjectives = await loadFiles("adjectives");
    let subjects   = await loadFiles("subjects");
    let errors     = [ ... locales.errors, ... adjectives.errors, ... subjects.errors ];
    if (errors.length != 0) {
        console.log(errors.join("\n"));
        alert("One or more errors occurred. See the console output.");
        return;
    }

    // Process locales into titles
    locales = locales.loaded;
    let titles = parseLocales(locales);
    console.log(titles); // For debugging

    // Associate unlock data
    for (let unlock of adjectives.loaded[0])
        Object.assign(titles.adjectives[unlock.Id], unlock);
    for (let unlock of subjects  .loaded[0])
        Object.assign(titles.subjects  [unlock.Id], unlock);

    // Process wikitext output
    adjectives = Object.values(titles.adjectives).filter(a=>a.USen[0]);
    adjectives.sort(sortOrder);
    output("adjectivesOut", adjectives, useEnglishOnly, oldFormatObtain);
    subjects   = Object.values(titles.subjects  ).filter(s=>s.USen[0]);
    subjects  .sort(sortOrder);
    output("subjectsOut", subjects, useEnglishOnly, oldFormatObtain);
}

    </script>
  </head>
  <body>

<div style="display: inline-grid; grid-template-columns: auto auto; gap: 4px; align-items: center;">
  <div>Languages:</div>
  <input type="file" id="locales" multiple>
  <div>Adjectives:</div>
  <input type="file" id="adjectives" multiple>
  <div>Subjects:</div>
  <input type="file" id="subjects" multiple>
  <div>
    <input type="checkbox" id="oldFormatObtain" name="oldFormatObtain">
    <label for="oldFormatObtain">Use old format for Obtain, Season</label>
  </div>
  <br>
  <div>
    <input type="radio" id="englishOnly" name="outputMode" value="englishOnly" checked>
    <label for="englishOnly">English only</label><br>
    <input type="radio" id="otherLanguages" name="outputMode" value="otherLanguages">
    <label for="otherLanguages">In other languages</label><br>
  </div>
  <div style="grid-area: span 1 / span 2;">
    <button onclick="process();">Process</button>
  </div>
</div>
<div style="display: flex; flex-direction: column; row-gap: 4px;">
  <div style="margin-top: 16px;">Adjectives</div>
  <textarea style="height: 250px; border: 1px solid #cccccc;" id="adjectivesOut"></textarea>
  <div style="margin-top: 16px;">Subjects</div>
  <textarea style="height: 250px; border: 1px solid #cccccc;" id="subjectsOut"></textarea>
</div>

  </body>
</html>

	<!DOCTYPE html>
	<html lang="en">
	<head>
	<meta charset="utf-8">
	<title>Splashtag Titles</title>
	<style>

	:root {
	color : #000000;
	font-family: Arial, sans-serif;
	font-size : 16px;
	}

	body {
	background: #ffffff;
	}

	</style>
	<script>

	// Load all files from an <input type="file"> element
	async function loadFiles(id) {
	let files = document.getElementById(id).files;
	let loaded = [];
	let errors = [];

	// No files selected
	if (files.length == 0)
	return { loaded: [], errors: [ "No file selected for " + id ] };

	// Load all files
	for (let file of files) {
	try {
	let data = JSON.parse(await file.text());
	data.locale = file.name.replace(".json", "");
	loaded.push(data);
	}
	catch(err) { errors.push(`Could not load file ${file.name} because: ${err}`); }
	}

	return { loaded: loaded, errors: errors };
	}

	// Parse the gendered components of an adjective
	function parseAdjective(text) {
	let ret = [ "", "", "", "" ];

	// Null-transform text
	let alone = text.replace(/\[.*?\]/g, "");

	// Process all meta commands in the text
	while (text.indexOf("[") != -1) {
	let x = text.indexOf("[") + 1;
	let y = text.indexOf("]", x);
	let params = text.substring(x, y);

	// Append everything before the opening square bracket
	let part = text.substring(0, x - 1);
	for (let z = 0; z < 4; z++)
	ret[z] += part;

	// Retain only the part after the closing square bracket
	text = text.substring(y + 1);

	// The meta command is not a gendered set
	if (
	!params.startsWith("group=00c9 type=0005") \|\|
	params.endsWith("params=")
	) continue;

	// Convert the parameter text into 16-bit numbers
	let shorts = new Uint16Array(new Uint8Array(params
	.substring(params.indexOf("params=") + 7)
	.split(" ").map(b=>parseInt(b, 16))
	).buffer);

	// Process gendered components
	params = [];
	for (x = 0; x < shorts.length;) {
	let length = shorts[x++] / 2;
	params.push(String.fromCharCode(
	... shorts.slice(x, x + length)));
	x += length;
	}
	if (params.length != 4)
	throw "Parameter length mismatch";

	// Append gendered components
	for (x = 0; x < 4; x++)
	ret[x] += params[x];
	}

	// Append any remaining text
	for (let x = 0; x < 4; x++)
	ret[x] += text;

	// Remove unused variants
	for (let x = 0; x < 4; x++) {
	if (
	x != 0 && ret[x] == alone \|\|
	x == 0 && ret[3] != alone && ret[x] == alone
	) ret[x] = "";
	}

	// Neutral variant if only gendered is available
	if (!ret[2] && ret[3])
	ret[2] = alone;

	return ret;
	}

	// Process locales file for title strings
	function parseLocales(locales) {
	let ret = {
	adjectives: {},
	subjects : {}
	};

	// Process all locales
	for (let locale of locales) {

	// Process adjectives
	for (let adj of Object.entries(locale["CommonMsg/Byname/BynameAdjective"])) {
	let id = parseInt(adj[0]);
	let item = ret.adjectives[id];
	if (!item)
	item = ret.adjectives[id] = {};
	item[locale.locale] = parseAdjective(adj[1]);
	}

	// Process subjects
	for (let sub of Object.entries(locale["CommonMsg/Byname/BynameSubject"])) {
	let id = sub[0].split("_");
	id[0] = parseInt(id[0]);
	let item = ret.subjects[id[0]];
	if (!item)
	item = ret.subjects[id[0]] = {};
	if (!item[locale.locale])
	item[locale.locale] = new Array(2);
	item[locale.locale][id[1]] = sub[1].replace(/\[.*?\]/g, "");

	// Meta command patterns. Left here for reference if needed.
	//if (sub[1].indexOf("[") == -1) continue;
	//let x = sub[1].indexOf("[") + 1;
	//let params = sub[1].substring(x, sub[1].indexOf("]", x));
	//if (
	// (
	// params.indexOf("group=00c9") == -1 \|\|
	// params.indexOf("type=0005" ) == -1
	// ) &&
	// !(
	// params.startsWith("ruby=") \|\|
	// params == "/ruby"
	// ) &&
	// !params.endsWith("params=") &&
	// !params.endsWith("params=01 ff ff 00") &&
	// !params.endsWith("params=02 ff ff 00") &&
	// !params.endsWith("params=03 ff ff 00") &&
	// !params.endsWith("params=04 ff ff 00")
	//) console.log(locale.locale, sub[0], sub[1]);
	}

	// Eliminate duplicate subjects
	for (let sub of Object.values(ret.subjects)) {
	for (let lang of Object.values(sub)) {
	if (lang[1] == lang[0]) {
	let dupe = DUPE_SUBJECTS[locale.locale] \|\| DUPE_SUBJECTS.USen;
	lang[0] += dupe[0];
	lang[1] += dupe[1];
	}
	}
	}

	}

	return ret;
	}

	// Duplicate subject suffixes
	let DUPE_SUBJECTS = {
	"EUde": [ " (m)", " (w)" ],
	"USen": [ " (m)", " (f)" ]
	};

	// Sorting comparator
	function sortOrder(a, b) {
	return a.USen[0].toLowerCase().localeCompare(b.USen[0].toLowerCase());
	}

	// Produce output to the user
	function output(id, list, useEnglishOnly, oldFormatObtain) {
	let out = [];
	if (useEnglishOnly) {
	/*
	!Title
	!How to obtain
	!Added (season)
	*/
	for (let item of list) {
	if (oldFormatObtain) {
	out.push("\|-\n" +
	"\|" + item.USen.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.HowToGet + ",<br>Season " + item.Season
	);
	} else {
	out.push("\|-\n" +
	"\|" + item.USen.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.HowToGet + "\n" +
	"\|" + item.Season
	);
	}
	}
	}
	else {
	/*
	!English
	!Japanese
	!Dutch
	!French<br>(Canada)
	!French<br>(France)
	!German
	!Italian*
	!Russian
	!Spanish<br>(Latin America)
	!Spanish<br>(Spain)
	!Chinese<br>(Simplified)
	!Chinese<br>(Traditional)
	!Korean
	*/
	for (let item of list) {
	out.push("\|-\n" +
	"\|" + item.USen.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.JPja.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.EUnl.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.USfr.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.EUfr.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.EUde.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.EUit.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.EUru.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.USes.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.EUes.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.CNzh.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.TWzh.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>") + "\n" +
	"\|" + item.KRko.filter(x=>x).map(x=>"'''"+x+"'''").join(" /<br>")
	);
	if (oldFormatObtain) {
	out.push("\|" + item.HowToGet + ",<br>Season " + item.Season);
	}
	}
	}
	document.getElementById(id).value = out.join("\n");
	}

	// Process all selected files
	async function process() {

	let useEnglishOnly = document.getElementById("englishOnly").checked;
	let useOtherLanguages = document.getElementById("otherLanguages").checked;
	let oldFormatObtain = document.getElementById("oldFormatObtain").checked;

	if (!(useEnglishOnly ^ useOtherLanguages)) {
	throw new Error(`Bad state for the checkboxes. useEnglishOnly=${useEnglishOnly} useOtherLanguages=${useOtherLanguages}`);
	}

	// Load files into memory
	let locales = await loadFiles("locales");
	let adjectives = await loadFiles("adjectives");
	let subjects = await loadFiles("subjects");
	let errors = [ ... locales.errors, ... adjectives.errors, ... subjects.errors ];
	if (errors.length != 0) {
	console.log(errors.join("\n"));
	alert("One or more errors occurred. See the console output.");
	return;
	}

	// Process locales into titles
	locales = locales.loaded;
	let titles = parseLocales(locales);
	console.log(titles); // For debugging

	// Associate unlock data
	for (let unlock of adjectives.loaded[0])
	Object.assign(titles.adjectives[unlock.Id], unlock);
	for (let unlock of subjects .loaded[0])
	Object.assign(titles.subjects [unlock.Id], unlock);

	// Process wikitext output
	adjectives = Object.values(titles.adjectives).filter(a=>a.USen[0]);
	adjectives.sort(sortOrder);
	output("adjectivesOut", adjectives, useEnglishOnly, oldFormatObtain);
	subjects = Object.values(titles.subjects ).filter(s=>s.USen[0]);
	subjects .sort(sortOrder);
	output("subjectsOut", subjects, useEnglishOnly, oldFormatObtain);
	}

	</script>
	</head>
	<body>

	<div style="display: inline-grid; grid-template-columns: auto auto; gap: 4px; align-items: center;">
	<div>Languages:</div>
	<input type="file" id="locales" multiple>
	<div>Adjectives:</div>
	<input type="file" id="adjectives" multiple>
	<div>Subjects:</div>
	<input type="file" id="subjects" multiple>
	<div>
	<input type="checkbox" id="oldFormatObtain" name="oldFormatObtain">
	<label for="oldFormatObtain">Use old format for Obtain, Season</label>
	</div>
	<br>
	<div>
	<input type="radio" id="englishOnly" name="outputMode" value="englishOnly" checked>
	<label for="englishOnly">English only</label><br>
	<input type="radio" id="otherLanguages" name="outputMode" value="otherLanguages">
	<label for="otherLanguages">In other languages</label><br>
	</div>
	<div style="grid-area: span 1 / span 2;">
	<button onclick="process();">Process</button>
	</div>
	</div>
	<div style="display: flex; flex-direction: column; row-gap: 4px;">
	<div style="margin-top: 16px;">Adjectives</div>
	<textarea style="height: 250px; border: 1px solid #cccccc;" id="adjectivesOut"></textarea>
	<div style="margin-top: 16px;">Subjects</div>
	<textarea style="height: 250px; border: 1px solid #cccccc;" id="subjectsOut"></textarea>
	</div>

	</body>
	</html>