Lightnovelpub: Refactor with state enum (#1644)

* refactor

* parseChapter
This commit is contained in:
K1ngfish3r
2025-04-22 19:20:24 +05:00
committed by GitHub
parent 0d4d850935
commit a34f5154af
+309 -168
View File
@@ -7,7 +7,7 @@ import dayjs from 'dayjs';
class LightNovelPub implements Plugin.PagePlugin { class LightNovelPub implements Plugin.PagePlugin {
id = 'lightnovelpub'; id = 'lightnovelpub';
name = 'LightNovelPub'; name = 'LightNovelPub';
version = '2.1.0'; version = '2.2.0';
icon = 'src/en/lightnovelpub/icon.png'; icon = 'src/en/lightnovelpub/icon.png';
site = 'https://www.lightnovelpub.com/'; site = 'https://www.lightnovelpub.com/';
headers = { headers = {
@@ -17,37 +17,39 @@ class LightNovelPub implements Plugin.PagePlugin {
parseNovels(html: string) { parseNovels(html: string) {
const novels: Plugin.NovelItem[] = []; const novels: Plugin.NovelItem[] = [];
let tempNovel = {} as Plugin.NovelItem; let tempNovel: Partial<Plugin.NovelItem> = {};
let isParsingNovel = false; let state: ParsingState = ParsingState.Idle;
const parser = new Parser({ const parser = new Parser({
onopentag(name, attribs) { onopentag(name, attribs) {
if (isParsingNovel) { if (attribs['class'] === 'novel-item') {
if (name === 'a') { state = ParsingState.Novel;
}
if (state !== ParsingState.Novel) return;
switch (name) {
case 'a':
tempNovel.path = attribs['href'].slice(1); tempNovel.path = attribs['href'].slice(1);
tempNovel.name = attribs['title']; tempNovel.name = attribs['title'];
} break;
if (name === 'img') { case 'img':
tempNovel.cover = attribs['data-src'] || attribs['src']; tempNovel.cover = attribs['data-src'] || attribs['src'];
} break;
if (tempNovel.path && tempNovel.cover) {
novels.push(tempNovel);
tempNovel = {} as Plugin.NovelItem;
}
}
},
onattribute(name, value) {
if (name === 'class' && value === 'novel-item') {
isParsingNovel = true;
} }
}, },
onclosetag(name) { onclosetag(name) {
if (name === 'li') { if (name === 'li') {
isParsingNovel = false; if (tempNovel.path && tempNovel.cover) {
novels.push(tempNovel as Plugin.NovelItem);
tempNovel = {};
}
state = ParsingState.Idle;
} }
}, },
}); });
parser.write(html); parser.write(html);
parser.end(); parser.end();
return novels; return novels;
} }
@@ -55,13 +57,15 @@ class LightNovelPub implements Plugin.PagePlugin {
page: number, page: number,
{ filters }: Plugin.PopularNovelsOptions<typeof this.filters>, { filters }: Plugin.PopularNovelsOptions<typeof this.filters>,
): Promise<Plugin.NovelItem[]> { ): Promise<Plugin.NovelItem[]> {
let link = `${this.site}browse/`; const linkParts = [
link += `${filters.genres.value}/`; this.site + 'browse',
link += `${filters.order.value}/`; filters.genres.value,
link += `${filters.status.value}/`; filters.order.value,
link += page; filters.status.value,
page.toString(),
];
const body = await fetchApi(link).then(r => r.text()); const body = await fetchApi(linkParts.join('/')).then(r => r.text());
return this.parseNovels(body); return this.parseNovels(body);
} }
@@ -70,186 +74,305 @@ class LightNovelPub implements Plugin.PagePlugin {
novelPath: string, novelPath: string,
): Promise<Plugin.SourceNovel & { totalPages: number }> { ): Promise<Plugin.SourceNovel & { totalPages: number }> {
const body = await fetchApi(this.site + novelPath).then(r => r.text()); const body = await fetchApi(this.site + novelPath).then(r => r.text());
const novel: Plugin.SourceNovel & { totalPages: number } = { const novel: Partial<Plugin.SourceNovel> & Partial<{ totalPages: number }> =
path: novelPath, {
name: '', path: novelPath,
summary: '', chapters: [],
totalPages: 1, };
chapters: [], let state: ParsingState = ParsingState.Idle;
}; const summaryParts: string[] = [];
let isHeaderStats = false;
let isStatus = false;
let isTotalChapters = false;
let isNovelName = false;
let isAuthorName = false;
let isSummary = false;
let isGenres = false;
let isTags = false;
const genreArray: string[] = []; const genreArray: string[] = [];
let isCover = false;
const parser = new Parser({ const parser = new Parser({
onopentag(name, attribs) { onopentag(name, attribs) {
if (name === 'h1' && attribs['class']?.includes('novel-title')) { switch (name) {
isNovelName = true; case 'h1':
} if (attribs['class']?.includes('novel-title')) {
if (name === 'div' && attribs['class']?.includes('content')) { state = ParsingState.NovelName;
isSummary = true; }
} break;
if (name === 'figure' && attribs['class'] === 'cover') { case 'figure':
isCover = true; if (attribs['class'] === 'cover') {
} state = ParsingState.Cover;
if (isCover) { }
if (name === 'img') { break;
novel.cover = attribs['data-src'] || attribs['src']; case 'img':
} if (state === ParsingState.Cover) {
} novel.cover = attribs['data-src'] || attribs['src'];
}, }
onopentagname(name) { break;
if (isHeaderStats && name === 'strong') { case 'strong':
isStatus = true; if (state === ParsingState.HeaderStats) {
} if (attribs['class']) {
if (isSummary && name === 'br') { state = ParsingState.Status;
novel.summary += '\n'; } else {
} state = ParsingState.TotalChapters;
if (isGenres && name === 'a') { }
isTags = true; }
} break;
}, case 'br':
onattribute(name, value) { if (state === ParsingState.Summary) {
if (name === 'class' && value === 'header-stats') { summaryParts.push('<LINE_BREAK>');
isTotalChapters = true; }
isHeaderStats = true; break;
} case 'a':
if (name === 'itemprop' && value === 'author') { if (state === ParsingState.Genres) {
isAuthorName = true; state = ParsingState.Tags;
} }
if (name === 'class' && value === 'categories') { break;
isGenres = true; case 'div':
if (attribs['class']) {
if (attribs['class'].includes('content')) {
state = ParsingState.Summary;
} else {
const map: Record<string, ParsingState> = {
'categories': ParsingState.Genres,
'header-stats': ParsingState.HeaderStats,
'expand': ParsingState.Idle,
};
state = map[attribs['class']] ?? state;
}
}
break;
default:
if (attribs['itemprop'] === 'author') {
state = ParsingState.AuthorName;
}
break;
} }
}, },
ontext(data) { ontext(data) {
if (isTotalChapters) { switch (state) {
novel.totalPages = Math.ceil(parseInt(data, 10) / 100); case ParsingState.TotalChapters:
} if (!novel.totalPages) {
if (isStatus) { novel.totalPages = Math.ceil(parseInt(data, 10) / 100);
novel.status = data.trim(); }
} break;
if (isNovelName) { case ParsingState.Status:
novel.name += data.trim(); novel.status = data.trim();
} break;
if (isAuthorName) { case ParsingState.NovelName:
novel.author = data; novel.name = (novel.name || '') + data.trim();
} break;
if (isSummary) { case ParsingState.AuthorName:
novel.summary += data; novel.author = data;
} break;
if (isTags) { case ParsingState.Summary:
genreArray.push(data); summaryParts.push(data);
break;
case ParsingState.Tags:
genreArray.push(data);
break;
} }
}, },
onclosetag(name) { onclosetag(name) {
if (name === 'strong') { switch (name) {
isTotalChapters = false; case 'strong':
isStatus = false; if (state === ParsingState.TotalChapters) {
} state = ParsingState.HeaderStats;
if (name === 'i') { } else if (state === ParsingState.Status) {
isStatus = false; state = ParsingState.Idle;
} }
if (name === 'h1') { break;
isNovelName = false; case 'i':
} if (state === ParsingState.Status) {
if (name === 'span') { state = ParsingState.Idle;
isAuthorName = false; }
} break;
if (name === 'div') { case 'h1':
isHeaderStats = false; if (state === ParsingState.NovelName) {
isSummary = false; state = ParsingState.Idle;
isGenres = false; }
} break;
if (name === 'a') { case 'span':
isTags = false; if (state === ParsingState.AuthorName) {
} state = ParsingState.Idle;
if (name === 'figure') { }
isCover = false; break;
case 'div':
if (
state === ParsingState.Summary ||
state === ParsingState.Genres
) {
state = ParsingState.Idle;
}
break;
case 'a':
if (state === ParsingState.Tags) {
state = ParsingState.Genres;
}
break;
case 'figure':
if (state === ParsingState.Cover) {
state = ParsingState.Idle;
}
break;
case 'p':
if (state === ParsingState.Summary) {
summaryParts.push('<PARAGRAPH_BREAK>');
}
break;
} }
}, },
onend() {
const text = summaryParts
.join('')
.replace(/<PARAGRAPH_BREAK>/g, '\n\n')
.replace(/<LINE_BREAK>/g, '\n')
.replace(/\r\n/g, '\n')
.replace(/&nbsp;/g, ' ');
const paragraphs = text
.split('\n\n')
.map(p => p.trim().replace(/[ \t]+/g, ' '))
.filter(p => p.length > 0);
novel.summary = paragraphs.join('\n\n');
summaryParts.length = 0;
novel.genres = genreArray.join(', ');
},
}); });
parser.write(body); parser.write(body);
parser.end(); parser.end();
novel.genres = genreArray.join(', ');
return novel; return novel as Plugin.SourceNovel & { totalPages: number };
} }
async parsePage(novelPath: string, page: string): Promise<Plugin.SourcePage> { async parsePage(novelPath: string, page: string): Promise<Plugin.SourcePage> {
const url = this.site + novelPath + '/chapters/page-' + page; const url = this.site + novelPath + '/chapters/page-' + page;
const body = await fetchApi(url).then(res => res.text()); const body = await fetchApi(url).then(res => res.text());
const chapter: Plugin.ChapterItem[] = []; const chapters: Plugin.ChapterItem[] = [];
let tempChapter = {} as Plugin.ChapterItem; let tempChapter: Partial<Plugin.ChapterItem> = {};
let isChapterItem = false; let state: ParsingState = ParsingState.Idle;
let isChapterList = false;
const parser = new Parser({ const parser = new Parser({
onopentag(name, attribs) { onopentag(name, attribs) {
if (isChapterList) { if (attribs['class'] === 'chapter-list') {
if (name === 'li') { state = ParsingState.ChapterList;
isChapterItem = true; return;
tempChapter.chapterNumber = Number(attribs['data-orderno']);
}
if (isChapterItem) {
if (name === 'a') {
tempChapter.name = attribs['title'];
tempChapter.path = attribs['href'].slice(1);
}
if (name === 'time') {
tempChapter.releaseTime = dayjs(
attribs['datetime'],
).toISOString();
}
if (
tempChapter.chapterNumber &&
tempChapter.path &&
tempChapter.releaseTime
) {
chapter.push(tempChapter);
tempChapter = {} as Plugin.ChapterItem;
}
}
} }
},
onattribute(name, value) { switch (state) {
if (name === 'class' && value === 'chapter-list') { case ParsingState.ChapterList:
isChapterList = true; if (name === 'li') {
state = ParsingState.ChapterItem;
tempChapter = {
chapterNumber: Number(attribs['data-orderno'] || 0),
};
}
break;
case ParsingState.ChapterItem:
switch (name) {
case 'a':
tempChapter.name = attribs['title'];
tempChapter.path = attribs['href']?.slice(1);
break;
case 'time':
tempChapter.releaseTime = dayjs(
attribs['datetime'],
).toISOString();
break;
}
break;
} }
}, },
onclosetag(name) { onclosetag(name) {
if (name === 'a') { switch (state) {
isChapterItem = false; case ParsingState.ChapterItem:
} if (name === 'li') {
if (name === 'ul') { if (
isChapterList = false; tempChapter.chapterNumber !== undefined &&
tempChapter.path &&
tempChapter.releaseTime
) {
chapters.push(tempChapter as Plugin.ChapterItem);
}
state = ParsingState.ChapterList;
}
break;
case ParsingState.ChapterList:
if (name === 'ul') {
state = ParsingState.Idle;
}
break;
} }
}, },
}); });
parser.write(body); parser.write(body);
parser.end(); parser.end();
const chapters = chapter;
return { chapters }; return { chapters };
} }
async parseChapter(chapterPath: string): Promise<string> { async parseChapter(chapterPath: string): Promise<string> {
const html = await fetchApi(this.site + chapterPath).then(r => r.text()); const html = await fetchApi(this.site + chapterPath).then(r => r.text());
const parts: string[] = [];
const regexPatterns: RegExp[] = [ let depth = 0;
/(<div id="chapter-container[^]+?)<div class="chapternav/, let state: ParsingState = ParsingState.Idle;
]; const chapterHtml: string[] = [];
const extractContent = (patterns: RegExp[]) => {
patterns.forEach(regex => { type EscapeChar = '&' | '<' | '>' | '"' | "'";
const match = html.match(regex)?.[1]; const escapeRegex = /[&<>"']/g;
if (match) parts.push(match); const escapeMap: Record<EscapeChar, string> = {
}); '&': '&amp;',
'<': '&lt;',
'>': '&gt;',
'"': '&quot;',
"'": '&#39;',
}; };
extractContent(regexPatterns); const escapeHtml = (text: string): string =>
const chapterText = parts.join(); text.replace(escapeRegex, char => escapeMap[char as EscapeChar]);
return chapterText.replace(/(?<=\/p>).*?(?=<p>)/g, '');
const parser = new Parser({
onopentag(name, attribs) {
switch (state) {
case ParsingState.Idle:
if (name === 'div' && attribs['id'] === 'chapter-container') {
state = ParsingState.Chapter;
depth++;
}
break;
case ParsingState.Chapter:
if (name === 'div') depth++;
break;
default:
return;
}
if (state === ParsingState.Chapter) {
const attr = Object.keys(attribs).map(key => {
const value = attribs[key].replace(/"/g, '&quot;');
return ` ${key}="${value}"`;
});
chapterHtml.push(`<${name}${attr.join('')}>`);
}
},
ontext(text) {
if (state === ParsingState.Chapter) {
chapterHtml.push(escapeHtml(text));
}
},
onclosetag(name) {
if (state === ParsingState.Chapter) {
if (!parser['isVoidElement'](name)) {
chapterHtml.push(`</${name}>`);
}
if (name === 'div') depth--;
if (depth === 0) {
state = ParsingState.Stopped;
}
}
},
});
parser.write(html);
parser.end();
return chapterHtml.join('');
} }
async searchNovels(searchTerm: string): Promise<Plugin.NovelItem[]> { async searchNovels(searchTerm: string): Promise<Plugin.NovelItem[]> {
@@ -356,3 +479,21 @@ class LightNovelPub implements Plugin.PagePlugin {
} }
export default new LightNovelPub(); export default new LightNovelPub();
enum ParsingState {
Idle,
Novel,
HeaderStats,
Status,
Stopped,
Chapter,
ChapterItem,
ChapterList,
TotalChapters,
NovelName,
AuthorName,
Summary,
Genres,
Tags,
Cover,
}