From 92c1b3abe15b82486427ef2e9e2455524e0c6c84 Mon Sep 17 00:00:00 2001
From: lyg <1543117173@qq.com>
Date: 星期五, 14 六月 2024 23:01:48 +0800
Subject: [PATCH] 回写图书信息到Excel
---
src/main.mjs | 43 +++++++++++--------------------------------
1 files changed, 11 insertions(+), 32 deletions(-)
diff --git a/src/main.mjs b/src/main.mjs
index 4e7605b..bbf02b6 100644
--- a/src/main.mjs
+++ b/src/main.mjs
@@ -361,10 +361,9 @@
}
}
- /* if (pdfUrl) {
+ if (pdfUrl) {
return pdfUrl;
- } else */
- if (textUrl) {
+ } else if (textUrl) {
return textUrl;
} else {
book.state = "娌℃湁text鏂囦欢";
@@ -395,7 +394,7 @@
async function downloadFile(book, url) {
console.log(`涓嬭浇鏂囦欢: ${url}`);
const ext = url.split(".").pop().toLowerCase();
- const filepath = `./downloads/${book.id} ${book.isbn}.txt`;
+ const filepath = `./downloads/${book.id} ${book.isbn}.${ext}`;
if (fs.existsSync(filepath)) {
book.state = `涓嬭浇瀹屾垚`;
book.format = ext;
@@ -432,27 +431,6 @@
book.format = ext;
book.file = filepath;
book.url = url;
- console.log(`涓嬭浇瀹屾垚锛�${filepath}`);
- setTimeout(() => {
- if (ext === "gz" || ext === "zip") {
- unzip(_filepath, filepath);
- fs.unlinkSync(_filepath);
- }
- let text = fs.readFileSync(filepath, 'utf-8');
- text = getTextFromHtml(text);
- fs.writeFileSync(filepath, text, 'utf-8');
- try {
- fs.writeFileSync(filepath + '.result.txt', cleanText(text), 'utf-8');
- } catch (e) {
- reject(e);
- try {
- out.close();
- fs.unlink(filepath, (e) => console.error(e));
- } catch (e) {
- console.error(e);
- }
- }
- }, 1000);
resolve(true);
});
stream.on("error", (err) => {
@@ -504,13 +482,10 @@
function getBookInfo(book) {
return retry(async () => {
- const publisher = await driver.executeScript(`return document.querySelector("span[itemprop=publisher]").textContent`);
- const datePublished = await driver.executeScript(`return document.querySelector("span[itemprop=datePublished]").textContent`);
- let pages = await driver.executeScript(`return document.querySelector("span[data-id=resultsCount]").textContent`);
- pages = pages.split(' / ')[1];
- book.publisher = publisher;
- book.pubDate = datePublished;
- book.pages = pages;
+ book.publisher = await driver.executeScript(`return document.querySelector("span[itemprop=publisher]").textContent`).catch(e => 0);
+ book.pubDate = await driver.executeScript(`return document.querySelector("span[itemprop=datePublished]").textContent`).catch(e => 0);
+ let pages = await driver.executeScript(`return document.querySelector("span[data-id=resultsCount]").textContent`).catch(e => 0);
+ if (pages) { book.pages = pages.split(' / ')[1]; }
});
}
@@ -573,6 +548,7 @@
try {
await downloadFile(book, url);
console.log(`涓嬭浇瀹屾垚: ${book.id} ${book.title}`);
+ console.log('finish: ' + JSON.stringify(book));
} catch (e) { }
successCount++;
// 绛変竴娈垫椂闂村啀涓嬩竴涓�
@@ -588,6 +564,9 @@
for (const book of books) {
const index = data.findIndex((row) => row[0] === book.id);
if (index > -1) {
+ data[index][5] = book.publisher;
+ data[index][6] = book.pubDate;
+ data[index][11] = book.pages;
data[index][12] = book.state;
data[index][13] = book.format;
data[index][14] = book.file;
--
Gitblit v1.9.1