Refactored Mike Adriano scraper. Changed logo and favicon. Added style methods to qu.

This commit is contained in:
DebaucheryLibrarian
2020-07-16 03:47:07 +02:00
parent 6584a46d53
commit 6adfded074
63 changed files with 577 additions and 332 deletions

17
src/utils/jsdom-url.js Normal file
View File

@@ -0,0 +1,17 @@
'use strict';
const { JSDOM } = require('jsdom');
const el = new JSDOM(`
<span id="url" style="background-image: url(https://w.wallhaven.cc/full/dg/wallhaven-dg7y23.jpg);">
<span id="urlQuotes" style="background-image: url('https://w.wallhaven.cc/full/dg/wallhaven-dg7y23.jpg');">
<span id="color" style="color: rgb(255, 0, 0);">
<span id="urlSpaces" style="background-image: url( https://w.wallhaven.cc/full/md/wallhaven-mdvmrm.jpg );">
<span id="colorSpaces" style="color: rgb( 255, 0, 0 );">
`).window.document;
console.log(el.querySelector('#url').style.backgroundImage);
console.log(el.querySelector('#urlQuotes').style.backgroundImage);
console.log(el.querySelector('#color').style.color);
console.log(el.querySelector('#urlSpaces').style.backgroundImage);
console.log(el.querySelector('#colorSpaces').style.color);

View File

@@ -4,9 +4,11 @@ const { JSDOM } = require('jsdom');
const moment = require('moment');
const http = require('./http');
const { window: globalWindow } = new JSDOM('');
function trim(str) {
if (typeof str !== 'string') {
return null;
return str;
}
return str.trim().replace(/\s+/g, ' ');
@@ -55,9 +57,9 @@ function q(context, selector, attrArg, applyTrim = true) {
if (attr) {
const value = selector
? context.querySelector(selector)?.[attr] || context.querySelector(selector)?.attributes[attr]?.value
: context[attr] || context.attributes[attr]?.value;
: context[attr] || context.getAttribute(attr);
return applyTrim && value ? trim(value) : value;
return applyTrim && typeof value === 'string' ? trim(value) : value;
}
return selector ? context.querySelector(selector) : context;
@@ -77,6 +79,14 @@ function exists(context, selector) {
return !!q(context, selector);
}
function content(context, selector, applyTrim = true) {
return q(context, selector, 'textContent', applyTrim);
}
function contents(context, selector, applyTrim) {
return all(context, selector, 'textContent', applyTrim);
}
function html(context, selector) {
const el = q(context, selector, null, true);
@@ -103,6 +113,33 @@ function text(context, selector, applyTrim = true) {
return applyTrim ? trim(textValue) : textValue;
}
function removeStyleFunctionSpaces(el) {
// jsdom appears to have a bug where it ignores inline CSS attributes set to a function() containing spaces, e.g. url( image.png )
el.setAttribute('style', el.getAttribute('style').replace(/\(\s+(.*)\s+\)/g, (match, cssArgs) => `(${cssArgs})`));
}
function style(context, selector, styleAttr) {
const el = q(context, selector);
if (el) {
removeStyleFunctionSpaces(el);
return styleAttr ? el.style[styleAttr] : el.style;
}
return null;
}
function styles(context, selector, styleAttr) {
const elStyles = Array.from(context.querySelectorAll(selector), (el) => {
removeStyleFunctionSpaces(el);
return styleAttr ? el.style[styleAttr] : el.style;
});
return elStyles;
}
function number(context, selector, attr = true) {
const value = q(context, selector, attr);
@@ -236,6 +273,10 @@ const legacyFuncs = {
const quFuncs = {
all,
html,
content,
contents,
cnt: content,
cnts: contents,
date,
dur: duration,
duration,
@@ -250,6 +291,8 @@ const quFuncs = {
num: number,
poster,
q,
style,
styles,
text,
texts,
trailer: video,
@@ -265,7 +308,7 @@ function init(element, window) {
const legacyContextFuncs = Object.entries(legacyFuncs) // dynamically attach methods with context
.reduce((acc, [key, func]) => ({
...acc,
[key]: (...args) => (window && args[0] instanceof window.HTMLElement // allow for different context
[key]: (...args) => (args[0] instanceof globalWindow.HTMLElement // allow for different context
? func(...args)
: func(element, ...args)),
}), {});
@@ -273,7 +316,7 @@ function init(element, window) {
const quContextFuncs = Object.entries(quFuncs) // dynamically attach methods with context
.reduce((acc, [key, func]) => ({
...acc,
[key]: (...args) => (window && args[0] instanceof window.HTMLElement // allow for different context
[key]: (...args) => (args[0] instanceof globalWindow.HTMLElement // allow for different context
? func(...args)
: func(element, ...args)),
}), {});
@@ -319,7 +362,7 @@ function extractAll(htmlValue, selector) {
}
async function get(urlValue, selector, headers, options, queryAll = false) {
const res = await http.get(urlValue, headers);
const res = await http.get(urlValue, headers, options);
if (res.statusCode === 200) {
const item = queryAll