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

Revision as of 08:41, 27 April 2020

Latest revision as of 07:02, 15 June 2022

(27 intermediate revisions by the same user not shown)

Line 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 66 ⟶ 67:

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 143 ⟶ 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 164 ⟶ 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 207 ⟶ 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 226 ⟶ 252:

displayKana(kana);

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

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

}

Line 242 ⟶ 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 307 ⟶ 335:

kana = kana.replace(/[・、 ]$/, "");

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

if ((!kanjiRegexes.latinOnly.test(wikidataKanji) && /^[ァ-ヴー]+$/kanjiRegexes.katakanaOnly.test(kana))

|| /^\w+$/kanjiRegexes.latinOnly.test(kana)) {

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

getWiktionary();

Line 387 ⟶ 415:

// Wiktionary adds readings as furigana

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

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

var kanji = "";

var kana = "";

if (headword.length) {

// 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;var ichildNodes <= headword[0].childNodes.length; i++) {

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

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

var ruby = $ if (childNodes[i]).nodeName //== convert back to JQuery for"RUBY") convenience{

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

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

kana += ruby.children("rtrp").detach().textremove();

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

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

kanji += childNodes[i].nodeValue;

kana += childNodes[i].nodeValue;

}

}

if (kanji != wikidataKanji) { return; }

} else if (seeTable.length) {

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

rubykana = seeTable.children("rprt").removetext();

} else {

return;

}

if (kanji != wikidataKanji) { return; }

if (kana) {