kouhais-mirrors
/
safer-indie-wiki-buddy
mirror of https://github.com/KevinPayravi/indie-wiki-buddy.git
1
0
Fork 0
Code

Improvements to search engine link handling 

* Handling when Google results use Google's middleman (google.com/url?q=)
* When handling middleman links, instead of replacing search result hrefs with updated links, we now add the updated link in a custom data-iwb-href attribute
* Better handling of special characters in page titles
safari
Kevin Payravi 2024-06-04 00:03:07 -04:00
parent af752726ab
commit 5995b28efa
4 changed files with 49 additions and 37 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
data/sitesEN.json
View File

@ -2529,14 +2529,14 @@
"origin": "Fear & Hunger Fandom Wiki", "origin": "Fear & Hunger Fandom Wiki",
"origin_base_url": "fearandhunger.fandom.com", "origin_base_url": "fearandhunger.fandom.com",
"origin_content_path": "/wiki/", "origin_content_path": "/wiki/",
"origin_main_page": "%22Fear_and_Hunger:_the_Tormentpedia%22_Wiki" "origin_main_page": "\"Fear_and_Hunger:_the_Tormentpedia\"_Wiki"
} }
], ],
"destination": "Fear & Hunger Wiki", "destination": "Fear & Hunger Wiki",
"destination_base_url": "fearandhunger.wiki.gg", "destination_base_url": "fearandhunger.wiki.gg",
"destination_platform": "mediawiki", "destination_platform": "mediawiki",
"destination_icon": "fearhungerwiki.png", "destination_icon": "fearhungerwiki.png",
"destination_main_page": "%22Fear_and_Hunger:_the_Tormentpedia%22_Wiki", "destination_main_page": "\"Fear_and_Hunger:_the_Tormentpedia\"_Wiki",
"destination_search_path": "/index.php", "destination_search_path": "/index.php",
"destination_content_path": "/wiki/", "destination_content_path": "/wiki/",
"tags": [ "tags": [

22
scripts/common-functions.js
View File

@ -160,8 +160,15 @@ async function commonFunctionFindMatchingSite(site, crossLanguageSetting, dest =
function commonFunctionGetOriginArticle(originURL, matchingSite) { function commonFunctionGetOriginArticle(originURL, matchingSite) {
let url = new URL('https://' + originURL.replace(/.*https?:\/\//, '')); let url = new URL('https://' + originURL.replace(/.*https?:\/\//, ''));
const path = decodeURIComponent(decodeURIComponent(String(url.pathname.split('&')[0]).split(matchingSite['origin_content_path'])[1] || '')); let article = String(url.pathname).split(matchingSite['origin_content_path'])[1] || '';
return path;
// If a Fextralife wiki, replace plus signs with spaces
// When there are multiple plus signs together, this regex will only replace only the first
if (originURL.includes('.wiki.fextralife.com')) {
article = article.replace(/(?<!\+)\+/g, ' ');
}
return article;
} }
function commonFunctionGetDestinationArticle(matchingSite, article) { function commonFunctionGetDestinationArticle(matchingSite, article) {
@ -190,15 +197,10 @@ function commonFunctionGetNewURL(originURL, matchingSite) {
// Replace underscores with spaces as that performs better in search // Replace underscores with spaces as that performs better in search
destinationArticle = destinationArticle.replaceAll('_', ' '); destinationArticle = destinationArticle.replaceAll('_', ' ');
// If a Fextralife wiki, replace plus signs with spaces
// When there are multiple plus signs together, this regex will only replace only the first
if (matchingSite['origin_base_url'].includes('.wiki.fextralife.com')) {
destinationArticle = destinationArticle.replace(/(?<!\+)\+/g, ' ');
}
// Encode article // Encode article
destinationArticle = encodeURIComponent(destinationArticle); // We decode + encode to ensure we don't double-encode,
// in the event a string is already encoded
destinationArticle = encodeURIComponent(decodeURIComponent(destinationArticle));
let searchParams = ''; let searchParams = '';
switch (matchingSite['destination_platform']) { switch (matchingSite['destination_platform']) {

4
scripts/content-banners.js
View File

@ -315,7 +315,7 @@ function displayRedirectBanner(newUrl, id, destinationName, destinationLanguage,
} }
function main() { function main() {
extensionAPI.runtime.sendMessage({action: 'getStorage'}, (storage) => { extensionAPI.runtime.sendMessage({ action: 'getStorage' }, (storage) => {
// Check if extension is on: // Check if extension is on:
if ((storage.power ?? 'on') === 'on') { if ((storage.power ?? 'on') === 'on') {
// Check if there is a pathname, to ensure we're looking at an article // Check if there is a pathname, to ensure we're looking at an article
@ -328,7 +328,7 @@ function main() {
if (currentURL.hostname.match(breezewikiRegex) || (storage.breezewikiHost === 'CUSTOM' && storage.breezewikiCustomHost?.includes(currentURL.hostname))) { if (currentURL.hostname.match(breezewikiRegex) || (storage.breezewikiHost === 'CUSTOM' && storage.breezewikiCustomHost?.includes(currentURL.hostname))) {
origin = String(currentURL.pathname).split('/')[1] + '.fandom.com/wiki/'; origin = String(currentURL.pathname).split('/')[1] + '.fandom.com/wiki/';
if (currentURL.search.includes('?q=')) { if (currentURL.search.includes('?q=')) {
origin = 'https://' + origin + currentURL.search.substring(3).split('&')[0]; origin = 'https://' + origin + currentURL.search.substring(3);
} else { } else {
origin = 'https://' + origin + currentURL.pathname.split('/')[3]; origin = 'https://' + origin + currentURL.pathname.split('/')[3];
} }

56
scripts/content-search-filtering.js
View File

@ -154,11 +154,6 @@ function stringToId(string) {
return string.replaceAll(' ', '-').replaceAll("'", '').replace(/\W/g, '').toLowerCase(); return string.replaceAll(' ', '-').replaceAll("'", '').replace(/\W/g, '').toLowerCase();
} }
// Function to escape string to use in regex
function escapeRegex(string) {
return string.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
}
function removeSubstringIfAtEnd(str, sub) { function removeSubstringIfAtEnd(str, sub) {
if (sub && str.endsWith(sub)) { if (sub && str.endsWith(sub)) {
return str.slice(0, -sub.length); return str.slice(0, -sub.length);
@ -193,16 +188,14 @@ function replaceSearchResults(searchResultContainer, site, link) {
let indieResultText = document.createElement('span'); let indieResultText = document.createElement('span');
if (originArticle && originArticle !== site['origin_main_page']) { if (originArticle && originArticle !== site['origin_main_page']) {
let destinationArticleTitle = removeSubstringIfAtEnd(destinationArticle, site['destination_content_suffix']).replace(site['destination_content_prefix'], '').replaceAll('_', ' '); let destinationArticleTitle = removeSubstringIfAtEnd(destinationArticle, site['destination_content_suffix']).replace(site['destination_content_prefix'], '').replaceAll('_', ' ');
// If a Fextralife wiki, replace plus signs with spaces
// When there are multiple plus signs together, this regex will only replace only the first // Decode article
if (link.includes('.wiki.fextralife.com')) { destinationArticleTitle = decodeURIComponent(destinationArticleTitle);
destinationArticleTitle = destinationArticleTitle.replace(/(?<!\+)\+/g, ' ');
}
if (site['language'] === 'EN' && link.match(/fandom\.com\/[a-z]{2}\/wiki\//)) { if (site['language'] === 'EN' && link.match(/fandom\.com\/[a-z]{2}\/wiki\//)) {
indieResultText.innerText = 'Look up "' + decodeURIComponent(destinationArticleTitle) + '" on ' + site.destination + ' (EN)'; indieResultText.innerText = 'Look up "' + destinationArticleTitle + '" on ' + site.destination + ' (EN)';
} else { } else {
indieResultText.innerText = 'Look up "' + decodeURIComponent(destinationArticleTitle) + '" on ' + site.destination; indieResultText.innerText = 'Look up "' + destinationArticleTitle + '" on ' + site.destination;
} }
} else { } else {
if (site['language'] === 'EN' && link.match(/fandom\.com\/[a-z]{2}\/wiki\//)) { if (site['language'] === 'EN' && link.match(/fandom\.com\/[a-z]{2}\/wiki\//)) {
@ -480,16 +473,18 @@ async function filterSearchResult(matchingSite, searchResult, searchEngine, coun
if (searchResultContainer) { if (searchResultContainer) {
// If this page from Fandom is the same as a re-ordered page, filter it out // If this page from Fandom is the same as a re-ordered page, filter it out
let searchResultLink = searchResult.href; let searchResultLink = searchResult.getAttribute('data-iwb-href') || searchResult.href;
let originArticle = commonFunctionGetOriginArticle(searchResultLink, matchingSite); let originArticle = commonFunctionGetOriginArticle(searchResultLink, matchingSite);
let destinationArticle = commonFunctionGetDestinationArticle(matchingSite, originArticle); let destinationArticle = commonFunctionGetDestinationArticle(matchingSite, originArticle);
if (reorderedHrefs.find((href) => href.match( if (reorderedHrefs.find((href) => href.match(
// Match for destination URL with content path and article name
new RegExp( new RegExp(
`http(s)*://${matchingSite['destination_base_url']}${matchingSite['destination_content_path']}${encodeURIComponent(destinationArticle)}$` `http(s)?://${matchingSite['destination_base_url']}${matchingSite['destination_content_path']}${decodeURI(destinationArticle)}$`
) )
))) { ))) {
countFiltered += hideSearchResults(searchResultContainer, searchEngine, matchingSite, 'off'); countFiltered += hideSearchResults(searchResultContainer, searchEngine, matchingSite, 'off');
console.debug(`Indie Wiki Buddy has hidden a result matching ${searchResult.href} because we re-ordered an indie wiki result with a matching article`); console.debug(`Indie Wiki Buddy has hidden a result matching ${searchResultLink} because we re-ordered an indie wiki result with a matching article`);
} else if (searchFilterSetting !== 'disabled') { } else if (searchFilterSetting !== 'disabled') {
if (searchFilterSetting === 'hide') { if (searchFilterSetting === 'hide') {
// Else, if the user has the preference set to hide search results, hide it indiscriminately // Else, if the user has the preference set to hide search results, hide it indiscriminately
@ -540,11 +535,11 @@ async function reorderSearchResults(searchResults, searchEngine, storage) {
document.querySelector('#main div[data-hveid]'); document.querySelector('#main div[data-hveid]');
// Get the first Fandom/Fextralife/Neoseeker result, if it exists // Get the first Fandom/Fextralife/Neoseeker result, if it exists
const firstNonIndieResult = document.querySelector(` const nonIndieResults = document.querySelectorAll(`
div[data-hveid] a[href*='.fandom.com/'][href*='/wiki/']:first-of-type:not([role='button']):not([target='_self']), div[data-hveid] a[href*='.fandom.com/'][href*='/wiki/']:first-of-type:not([role='button']):not([target='_self']),
div[data-hveid] a[href*='.wiki.fextralife.com/']:first-of-type:not([role='button']):not([target='_self']), div[data-hveid] a[href*='.wiki.fextralife.com/']:first-of-type:not([role='button']):not([target='_self']),
div[data-hveid] a[href*='.neoseeker.com/wiki/']:first-of-type:not([role='button']):not([target='_self'])`); div[data-hveid] a[href*='.neoseeker.com/wiki/']:first-of-type:not([role='button']):not([target='_self'])`);
const firstNonIndieResult = Array.from(nonIndieResults).filter((e) => !e.closest('g-section-with-header, div[aria-expanded], div[data-q], div[data-minw], div[data-num-cols], div[data-docid], div[data-lpage]'))[0];
if (!resultsFirstChild || !firstNonIndieResult) return; if (!resultsFirstChild || !firstNonIndieResult) return;
searchResults.some((result, i) => { searchResults.some((result, i) => {
@ -567,7 +562,7 @@ async function reorderSearchResults(searchResults, searchEngine, storage) {
continue; continue;
} }
const searchResultLink = searchResult.href || ''; const searchResultLink = searchResult.getAttribute('data-iwb-href') || searchResult.href || '';
// Handle re-ordering of results to move destination results up the page // Handle re-ordering of results to move destination results up the page
let matchingDest = await commonFunctionFindMatchingSite(searchResultLink, crossLanguageSetting, true); let matchingDest = await commonFunctionFindMatchingSite(searchResultLink, crossLanguageSetting, true);
@ -597,7 +592,7 @@ async function reorderSearchResults(searchResults, searchEngine, storage) {
for (const searchResult of resultsToSort) { for (const searchResult of resultsToSort) {
try { try {
await reorderDestinationSearchResult(firstNonIndieResult, searchResult); await reorderDestinationSearchResult(firstNonIndieResult, searchResult);
reorderedHrefs.push(searchResult.href); reorderedHrefs.push(searchResult.getAttribute('data-iwb-href') || searchResult.href);
} catch (e) { } catch (e) {
console.log('Indie Wiki Buddy failed to properly re-order search results with error: ' + e); console.log('Indie Wiki Buddy failed to properly re-order search results with error: ' + e);
} }
@ -614,7 +609,7 @@ async function filterSearchResults(searchResults, searchEngine, storage, reorder
try { try {
// Check that result isn't within another result // Check that result isn't within another result
if (!searchResult.closest('.iwb-detected') || !searchResult.closest('.iwb-detected')?.querySelector('.iwb-new-link')) { if (!searchResult.closest('.iwb-detected') || !searchResult.closest('.iwb-detected')?.querySelector('.iwb-new-link')) {
let searchResultLink = searchResult.href || ''; let searchResultLink = searchResult.getAttribute('data-iwb-href') || searchResult.href || '';
if (!searchResultLink) { if (!searchResultLink) {
continue; continue;
@ -670,6 +665,22 @@ function startFiltering(searchEngine, storage, mutations = null, observer = null
// Determine which search engine we're on // Determine which search engine we're on
switch (searchEngine) { switch (searchEngine) {
case 'google': case 'google':
// Query Google results and rewrite HREFs when Google uses middleman links (i.e. google.com/url?q=)
let searchResults = document.querySelectorAll("div[data-hveid] a:first-of-type:not([role='button']):not([target='_self'])");
searchResults.forEach((searchResult) => {
if (searchResult.href) {
const link = new URL(searchResult.href);
if (link.href.includes('https://www.google.com/url')) {
try {
const destinationLink = link.searchParams.get('url') || link.searchParams.get('q');
searchResult.setAttribute('data-iwb-href', destinationLink);
} catch (e) {
console.log('Indie Wiki Buddy failed to parse Bing link with error: ', e);
}
}
}
});
// Function to filter search results in Google // Function to filter search results in Google
function filterGoogle(reorderedHrefs) { function filterGoogle(reorderedHrefs) {
let searchResults = document.querySelectorAll(` let searchResults = document.querySelectorAll(`
@ -682,7 +693,6 @@ function startFiltering(searchEngine, storage, mutations = null, observer = null
async function reorderGoogle() { async function reorderGoogle() {
let searchResults = document.querySelectorAll("div[data-hveid] a:first-of-type:not([role='button']):not([target='_self'])"); let searchResults = document.querySelectorAll("div[data-hveid] a:first-of-type:not([role='button']):not([target='_self'])");
// Remove any matches that are not "standard" search results - this could've been done with :has() but limited browser support right now // Remove any matches that are not "standard" search results - this could've been done with :has() but limited browser support right now
searchResults = Array.from(searchResults).filter((e) => !e.closest('g-section-with-header, div[aria-expanded], div[data-q], div[data-minw], div[data-num-cols], div[data-docid], div[data-lpage]')); searchResults = Array.from(searchResults).filter((e) => !e.closest('g-section-with-header, div[aria-expanded], div[data-q], div[data-minw], div[data-num-cols], div[data-docid], div[data-lpage]'));
@ -715,7 +725,7 @@ function startFiltering(searchEngine, storage, mutations = null, observer = null
try { try {
let decodedLink = base64Decode(encodedLink.searchParams.get('u').replace(/^a1/, '')); let decodedLink = base64Decode(encodedLink.searchParams.get('u').replace(/^a1/, ''));
if (decodedLink.includes('.fandom.com') || decodedLink.includes('.wiki.fextralife.com') || decodedLink.includes('.neoseeker.com/wiki/')) { if (decodedLink.includes('.fandom.com') || decodedLink.includes('.wiki.fextralife.com') || decodedLink.includes('.neoseeker.com/wiki/')) {
searchResult.href = decodedLink; searchResult.setAttribute('data-iwb-href', decodedLink);
searchResults.push(searchResult); searchResults.push(searchResult);
} }
} catch (e) { } catch (e) {
@ -804,7 +814,7 @@ function startFiltering(searchEngine, storage, mutations = null, observer = null
const extractedURL = decodeURIComponent(match && match[1]); const extractedURL = decodeURIComponent(match && match[1]);
if (extractedURL.includes('.fandom.com') || extractedURL.includes('.wiki.fextralife.com') || extractedURL.includes('.neoseeker.com/wiki/')) { if (extractedURL.includes('.fandom.com') || extractedURL.includes('.wiki.fextralife.com') || extractedURL.includes('.neoseeker.com/wiki/')) {
searchResult.href = extractedURL; searchResult.setAttribute('data-iwb-href', extractedURL);
searchResults.push(searchResult); searchResults.push(searchResult);
} }
} catch (e) { } catch (e) {