User:Opencooper/showKanji-dev.js: Difference between revisions

Revision as of 01:41, 5 October 2019

Latest revision as of 07:02, 15 June 2022

(48 intermediate revisions by the same user not shown)

Line 2:

// It then calls another script, bindKana.js, to clean up the display of ruby

// For configuration, please see the documentation

// TODO: Reject if any base/reading has too low/high of a ratio.

// TODO: Reject if unbalanced parenthesis count

// TODO: <rb> is not actually in the whatcg standard...

// License: CC0

Line 47 ⟶ 51:

https://en.wikipedia.org/wiki/Ninjō - No interlanguage, but wiktionary

https://en.wikipedia.org/wiki/Seiza - Interlanguage failed, but wiktionary

https://en.wikipedia.org/wiki/Epsomite - No interlanguage, but wiktionary "see" Table

https://en.wikipedia.org/wiki/Bakayaro!_I%27m_Plenty_Mad - only part of parenthesis extracted

Line 57 ⟶ 62:

https://en.wikipedia.org/wiki/Love_Live!_The_School_Idol_Movie - interpunct in reading

https://en.wikipedia.org/wiki/Lupin_the_Third:_The_Woman_Called_Fujiko_Mine - hyphen in kanji

https://en.wikipedia.org/wiki/Sunscreen

https://en.wikipedia.org/wiki/Flag_of_China

https://en.wikipedia.org/wiki/W3m

https://en.wikipedia.org/wiki/Magnum_Collection_1999_%22Dear%22

https://en.wikipedia.org/wiki/EC_Comics

https://en.wikipedia.org/wiki/CJK_characters

https://en.wikipedia.org/wiki/My_Girlfriend_is_Shobitch

https://en.wikipedia.org/wiki/Immaculate_Conception_Cathedral,_Nagasaki - partial match

https://en.wikipedia.org/wiki/USA-224 - または

https://en.wikipedia.org/wiki/Milk - bad match

https://en.wikipedia.org/wiki/Not_invented_here

*/

Line 134 ⟶ 150:

// Need to add hyphen escaped since it has special behavior in regex classes

// TODO: Just escape $kanji early instead, like we did before?

if (/-/.test(kanji)) { kanjiStripped += "\\-"; }

var kanjiAuxillary = kanjiStripped.replace(/\w/g, "");

kanjiRegexes.latinOnly = /^[A-Za-z0-9\-.?!/,:;@#$%&+=*'"・ ]+$/;

kanjiRegexes.kanaOnly = new RegExp("^[ぁ-ゔァ-ヴー" + kanjiAuxillary + "]+$");

kanjiRegexes.hiraganaOnly = new RegExp("^[ぁ-ゔーA-Za-z" + kanjiAuxillary + "]+$");

Line 147 ⟶ 163:

var leadReBase = "([ぁ-ゔァ-ヴー" + kanjiStripped + "]+)";

var kanjiEscaped = mw.util.escapeRegExp(kanji);

// Account for spaces, but ignore backslash and other misc characters

var reKanjiKanaLatin = /([\u3400-\u4DB5\u4E00-\u9FCB\uF900-\uFA6Aぁ-ゔァ-ヴーA-Za-z0-9])/g;

var kanjiSpaced = kanjiEscaped.replace(/ /g, " ?");

kanjiSpaced = kanjiSpaced.replace(reKanjiKanaLatin, "$1 ?");

Line 155 ⟶ 171:

// Add kanji to regex to make sure we're not getting the reading of some

// other term

kanjiRegexes.leadUnspaced = new RegExp(kanjiEscaped + ".[^(\n)]*?\\(" + leadReBase);

kanjiRegexes.lead = new RegExp(kanjiSpaced + ".[^(\n)]*?\\(" + leadReBase, "i"); // brittle

}

Line 198 ⟶ 214:

function parseKanaClaim(response) {

var kana;

ifvar (response.claims.P1814)properties = {

// name in kana title: "P1476",

nativeLabel: "P1705",

kana = response.claims.P1814[0].mainsnak.datavalue.value;

officialName: "P1448",

} else if (response.claims.P1476 && response.claims.P1476[0].qualifiers

nameInNativeLanguage: "P1559"

&& response.claims.P1476[0].qualifiers.P1814) {

// title w/ name in kana };

var nameInKana = "P1814";

kana = response.claims.P1476[0].qualifiers.P1814[0].datavalue.value;

} else if (response.claims.P1705 && response.claims.P1705[0].qualifiers

// Try getting nameInKana as a qualifier to some properties

&& response.claims.P1705[0].qualifiers.P1814) {

for (var prop in properties) {

// native label w/ name in kana

var pnum = properties[prop];

kana = response.claims.P1705[0].qualifiers.P1814[0].datavalue.value;

} else {

if (response.claims[pnum]) {

var kanji &&= response.claims.P1705[pnum][0].qualifiersmainsnak.P1814) {datavalue.value.text;

if (kanji.replace(/ /g, "") == wikidataKanji.replace(/ /g, "")

} else if (response.claims.P1476 && response.claims.P1476[pnum][0].qualifiers

&& response.claims.P1476[pnum][0].qualifiers.P1814[nameInKana]) {

kana = response.claims.P1476[pnum][0].qualifiers.P1814[nameInKana][0].datavalue.value;

break;

}

}

}

// Try getting nameInKana as a general claim

if (!kana && response.claims[nameInKana]) {

prop = "nameInKana";

kana = response.claims.P1814[nameInKana][0].mainsnak.datavalue.value;

}

// We couldn't find nameInKana

if (!kana) {

getInterlanguage();

return;

Line 217 ⟶ 252:

displayKana(kana);

$("#kanjiInfo").addClass("kanjiInfo-wikidata");

$("#kanjiInfo").addClass("kanjiInfo-wikidata-" + prop);

}

Line 233 ⟶ 269:

success: function(response) {

var pageId = mw.config.get( 'wgArticleId' );

var langlinkspage = response.query.pages[pageId].langlinks;

var langlinks = page ? page.langlinks : undefined;

var jaLabel;

if (langlinks) {

Line 298 ⟶ 335:

kana = kana.replace(/[・、 ]$/, "");

// Abort if our reading is only katakana (for non-Latin) or katakanaLatin

if (/^\w+$/(!kanjiRegexes.latinOnly.test(kanawikidataKanji) ||&& /^[ァ-ヴー]+$/kanjiRegexes.katakanaOnly.test(kana)) {

if (! || kanjiRegexes.kanaOnlylatinOnly.test(kana)) { return; }

console.log("showKanji-dev.js: throwing away reading: " + kana);

getWiktionary();

return;

Line 373 ⟶ 412:

var html = response.parse.text["*"];

var parsed = $($.parseHTML(html));

// Equivalent: $(".headword:lang(ja)").siblings(".Jpan.form-of:lang(ja)").first().text()

// Wiktionary tags itsadds readings with the "form-of"as classfurigana

var headword = parsed.find(".headword:lang(ja)").first();

var seeTable = parsed.find(".Jpan ruby").first();

if (headword.text() != wikidataKanji) { return; }

// Wiktionary tags its readings with the "form-of" class

var kana = headword.siblings(".Jpan.form-of:lang(ja)").first().text();

var kanji = "";

if (kana) {

var kana = kana.toHalfWidth()"";

if (headword.length) {

if (!kanjiRegexes.kanaOnly.test(kana)) { return; }

// Wiktionary already binds their kana, so we have to undo the process to get

// the constituent parts, at least with the current markup

var childNodes = headword[0].childNodes;

for (let i = 0; i < childNodes.length; i++) {

if (childNodes[i].nodeName == "RUBY") {

var ruby = $(childNodes[i]); // convert back to JQuery for convenience

ruby.children("rp").remove();

kana += ruby.children("rt").detach().text();

kanji += ruby.text();

} else if (childNodes[i].nodeType == 3) { // "#text"

kanji += childNodes[i].nodeValue;

kana += childNodes[i].nodeValue;

}

}

if (headword.text()kanji != wikidataKanji) { return; }

} else if (seeTable.length) {

kanji = seeTable.children("rb").text();

kana = seeTable.children("rt").text();

} else {

return;

}

if (kana) {

displayKana(kana);

$("#kanjiInfo").addClass("kanjiInfo-wiktionary");

// Extra stuff just for fun

var definition = headword.parent().siblings("ol").children("li").first().text();

definition = definition.split('\n', 1)[0];

definition = definition.replace(/\[[0-9]{1,2}\]/g, "");

$("#kanjiInfo").prop("title", definition);

}