diff options
Diffstat (limited to 'update/episodes/index.js')
| -rw-r--r-- | update/episodes/index.js | 141 |
1 files changed, 141 insertions, 0 deletions
diff --git a/update/episodes/index.js b/update/episodes/index.js new file mode 100644 index 0000000..f071707 --- /dev/null +++ b/update/episodes/index.js @@ -0,0 +1,141 @@ +const config = require("./series.json"); +const axios = require('axios'); +const fs = require('fs'); +const WikiTextParser = require('parse-wikitext'); +const parser = new WikiTextParser("mlp.fandom.com"); + +(async () => { + let all = []; + + for (let series of config) { + let data = { + command: series.cmd, + title: series.title, + date: series.date, + link: series.page, + description: { + series: null, + plot: null, + }, + seasons: [] + }; + console.log(series.title + " (" + data.date + ")"); + + ddata = (await axios.get("https://mlp.fandom.com/api.php?format=json&action=query&prop=extracts&exlimit=max&explaintext&exintro&titles=" + encodeURI(series.page) + "&redirects=")).data; + data.description.series = ddata.query.pages[Object.keys(ddata.query.pages)[0]].extract.split("\n")[0]; + data.description.plot = ddata.query.pages[Object.keys(ddata.query.pages)[0]].extract.split("\n")[1]; + + for (let s of series.seasons) { + console.log(" Season " + s.id); + + let season = { + name: "Season " + s.id, + id: "s" + s.id, + date: null, + count: null, + episodes: [] + } + + let sdata = (await axios.get("https://mlp.fandom.com/api.php?action=query&prop=revisions&titles=" + s.table.page + "&rvslots=*&rvprop=content&formatversion=2&format=json")).data; + let sections = parser.pageToSectionObject(sdata.query.pages[0].revisions[0].slots.main.content); + let episodes = sections.Episodes[s.table.section].content + .filter(i => i.startsWith("|")) + .join("||") + .replace("|}", "") + .split("|-").map(i => { + return i.split("||").filter(i => i.trim() !== ""); + }) + .filter(i => i.length > 1) + .map(i => { + let index = 0; + return i.map(i => { + return i.replace(/^(\| |\|)(.*)/gm, "$2") + .replace(/^style="(.*)"( |)\|( |)(.*)/gm, "$4") + .replace(/\[\[((.*)\|(.*)|(.*))\]\]/gm, "$3$4"); + }).map(i => { + if (index === 0) { + index++; + return i.replace(/(.*) \((.*)\)/gm, "$1|$2"); + } else { + index++; + return i; + } + }) + }); + + index = 1; + for (let e of episodes) { + process.stdout.write(" S" + s.id + "E" + index); + + let episode = { + local: null, + global: null, + date: null, + name: null, + writer: null, + cover: null, + characters: [], + plot: null + } + + episode.local = e[s.table.fields.number].replace(/(.*)\|(.*)/gm, "$1").replace(/''(.*)''/gm, "$1").replace(/{{(.*)\|(.*)}}/gm, "$2").replace(/(.*)\|(.*)/gm, "$2"); + episode.global = e[s.table.fields.number].replace(/(.*)\|(.*)/gm, "$2"); + episode.date_pre = e[s.table.fields.date].replace(/{{(.*)\|(.*)\|(.*)\|link=(no|yes)}}/gm, "$3").replace(/{{(.*)\|(.*)}}/gm, "$2").replace(/(.*)\|(.*)/gm, "$2"); + try { + episode.date = new Date(e[s.table.fields.date].replace(/{{(.*)\|(.*)\|(.*)\|link=(no|yes)}}/gm, "$3").replace(/{{(.*)\|(.*)}}/gm, "$2").replace(/(.*)\|(.*)/gm, "$2")).toISOString(); + } catch (e) { + episode.date = null; + } + episode.name = e[s.table.fields.title]; + episode.writer = e[s.table.fields.writer].replace(/{{(.*)\|(.*)\|(.*)\|link=(no|yes)}}/gm, "$3").replace(/(]]|\[\[)/gm, "").replace(/(.*)\|(.*)/gm, "$2"); + + process.stdout.write(": " + episode.name + "\n"); + + if (episode.local.length !== 2 || episode.global.length > 3 || episode.global.length < 2 || episode.date_pre.length !== 10) { + console.log("FAILURE: " + JSON.stringify(episode)); + } + + let edata = (await axios.get("https://mlp.fandom.com/api.php?action=query&prop=revisions&titles=" + encodeURI(episode.name).replaceAll("?", "%3F").replaceAll("&", "%26") + "&rvslots=*&rvprop=content&formatversion=2&format=json")).data; + let eext = (await axios.get("https://mlp.fandom.com/api.php?format=json&action=query&prop=extracts&exlimit=max&explaintext&exintro&titles=" + encodeURI(episode.name).replaceAll("?", "%3F").replaceAll("&", "%26") + "&redirects=")).data; + + episode.plot = eext.query.pages[Object.keys(eext.query.pages)[0]].extract.replace(/(.*)\n(.*)/, "$2").replace(/(.*)\n\n(.*)/gm, "$2").replace(/([.?!])\s*(?=[A-Z])/g, "$1|").split("|").join(" "); + + let infobox = null; + try { + sections = parser.pageToSectionObject(edata.query.pages[0].revisions[0].slots.main.content); + box = parser.parseInfoBox(sections["content"]); + if (box.template === "Infobox episode") { + infobox = parser.parseInfoBox(sections["content"]).values; + } + } catch (e) { + console.log(edata); + throw e; + } + + if (infobox !== null) { + try { + if (typeof infobox.image !== "undefined") episode.cover = (await axios.head("https://mlp.fandom.com/Special:FilePath/" + encodeURI(infobox.image.replace(/<!--[\s\S]*?-->/g, "").trim()).replaceAll("?", "%3F").replaceAll("&", "%26"))).request.res.responseUrl; + } catch (e) { + try { + if (typeof infobox.image !== "undefined") episode.cover = e.request.res.responseUrl; + } catch (e2) { + console.error(e2); + throw e; + } + } + if (typeof infobox.featured !== "undefined") episode.characters = infobox.featured.replace(/\[\[(([a-zA-Z0-9 .\-# _\\\/]*)\|([a-zA-Z0-9 .\-_\\\/]*)|([a-zA-Z0-9 .\-#_\\\/]*))\]\]/gm, "$3$4").replace(/(<(\/| |)(\/| |)br(\/| |)(\/| |)>)/gm, "|||").split("|||") + } + + season.episodes.push(episode); + index++; + } + + season.count = season.episodes.length; + season.date = season.episodes[0].date; + data.seasons.push(season); + } + all.push(data); + } + + fs.writeFileSync("./data/series.json", JSON.stringify(all, null, 4)); +})();
\ No newline at end of file |