From 6a6078c5d393bffda15e682994811468ff86963e Mon Sep 17 00:00:00 2001
From: lyg <1543117173@qq.com>
Date: 星期二, 30 七月 2024 01:40:34 +0800
Subject: [PATCH] 增加数据库图书ISBN补全脚本
---
src/book-list-download2.mjs | 104 +++++++++++++++++++++++++++++++++++----------------
1 files changed, 71 insertions(+), 33 deletions(-)
diff --git a/src/book-list-download2.mjs b/src/book-list-download2.mjs
index e43cc2f..fea054e 100644
--- a/src/book-list-download2.mjs
+++ b/src/book-list-download2.mjs
@@ -53,6 +53,19 @@
}
}
+function addBooks(books) {
+ db.run("begin transaction");
+ for (const book of books) {
+ db.run("INSERT INTO t_books (Title, Author, Year, Publisher, ISBN) VALUES (?,?,?,?,?)",
+ [book.title, book.author, book.year, book.publisher, book.isbn], (err) => {
+ if (!err) {
+ downloadCnt++;
+ }
+ });
+ }
+ db.run("commit");
+}
+
function addBook(book) {
db.run("INSERT INTO t_books (Title, Author, Year, Publisher, ISBN) VALUES (?,?,?,?,?)",
[book.title, book.author, book.year, book.publisher, book.isbn], (err) => {
@@ -192,7 +205,7 @@
const html = cheerio.load(resp.data);
const bookDivs = html('#search-results-list > div > div.col-md-8.col-xs-9.div-o');
for (const _bookDiv of bookDivs) {
- const bookDiv=cheerio.load(_bookDiv)
+ const bookDiv = cheerio.load(_bookDiv)
const book = {};
const h3 = bookDiv('h3');
if (h3) {
@@ -256,6 +269,26 @@
return Math.random() * (max - min) + min;
}
+function importFromExcel() {
+ initDb();
+ const file = './76w.xlsx';
+ const workSheets = xlsx.parse(file);
+ const sheet = workSheets[0];
+ sheet.data.shift();
+ const books = [];
+ sheet.data.forEach((row) => {
+ const title = row[0];
+ const author = row[1]
+ const year = row[2];
+ const publisher = row[3];
+ const isbn = row[4].split(',').sort((a, b) => b.length - a.length)[0];
+
+ books.push({ title, author, year, publisher, isbn });
+ });
+ addBooks(books);
+ closeDb();
+}
+
// 寮�濮嬫椂闂�
const startTime = Date.now();
// 鍥句功鏁伴噺
@@ -263,7 +296,7 @@
// chrome椹卞姩
/** @type {WebDriver} */
let driver;
-function main() {
+function startTask() {
initLogger();
getBook()
.catch(e => {
@@ -283,38 +316,43 @@
fs.mkdirSync('D:\\book-list-crawler-cache', { recursive: true });
}
-// 澶氳繘绋嬫墽琛�
-if (isMainThread) {
- console.log(`绾跨▼鏁帮細${config.threadSize}`);
- initDb();
- let finishCnt = 0;
- const threadSize = config.threadSize;
- const bookNames = fs.readFileSync('./bookNames.txt', 'utf8').replace(/\r/, '').split('\n');
- for (let i = 0; i < threadSize; i++) {
- const worker = new Worker("./src/book-list-download2.mjs", { workerData: {} });
- worker.on("message", (message) => {
- if (message.type === 'book') {
- addBook(message.data);
- }
- else if (message.type === 'getBookName') {
- const bookName = bookNames.shift();
- if (bookName)
- console.log(bookName, `鍓╀簬锛�${bookNames.length}`);
- worker.postMessage({ type: "bookName", data: bookName, threadId: message.threadId });
- } else if (message.type === 'finish') {
- finishCnt++;
- if (finishCnt == threadSize) {
- closeDb();
- console.log(`鍏变笅杞�${downloadCnt}鏈紝鑰楁椂锛� ${msFormat(Date.now() - startTime)}銆俙);
+function main() {
+ // 澶氳繘绋嬫墽琛�
+ if (isMainThread) {
+ console.log(`绾跨▼鏁帮細${config.threadSize}`);
+ initDb();
+ let finishCnt = 0;
+ const threadSize = config.threadSize;
+ const bookNames = fs.readFileSync('./bookNames.txt', 'utf8').replace(/\r/, '').split('\n');
+ for (let i = 0; i < threadSize; i++) {
+ const worker = new Worker("./src/book-list-download2.mjs", { workerData: {} });
+ worker.on("message", (message) => {
+ if (message.type === 'book') {
+ addBook(message.data);
}
- }
+ else if (message.type === 'getBookName') {
+ const bookName = bookNames.shift();
+ if (bookName)
+ console.log(bookName, `鍓╀簬锛�${bookNames.length}锛屽凡鑾峰彇${downloadCnt}鏈琡);
+ worker.postMessage({ type: "bookName", data: bookName, threadId: message.threadId });
+ } else if (message.type === 'finish') {
+ finishCnt++;
+ if (finishCnt == threadSize) {
+ closeDb();
+ console.log(`鍏变笅杞�${downloadCnt}鏈紝鑰楁椂锛� ${msFormat(Date.now() - startTime)}銆俙);
+ }
+ }
+ });
+ }
+ process.on('SIGINT', () => {
+ closeDb();
+ console.log(`杩涚▼琚墜鍔ㄧ粨鏉燂紝鍏变笅杞�${downloadCnt}鏈紝鑰楁椂锛� ${msFormat(Date.now() - startTime)}銆俙);
+ process.exit(0);
});
+ } else {
+ startTask();
}
- process.on('SIGINT', () => {
- closeDb();
- console.log(`杩涚▼琚墜鍔ㄧ粨鏉燂紝鍏变笅杞�${downloadCnt}鏈紝鑰楁椂锛� ${msFormat(Date.now() - startTime)}銆俙);
- process.exit(0);
- });
-} else {
- main();
}
+
+// importFromExcel();
+main();
\ No newline at end of file
--
Gitblit v1.9.1