為 Google Search 添加 AI 答案（由 Google Gemma + GPT-4o 提供支援！）
I still want this feature...
Also, I just installed the script to try it again, just to check if my suggestion was implemented or not, and I already got the error message "ChatGPT is flooded with too many requests. Try switching on Proxy Mode in toolbar" right on my first google try...
After I changed it as requested, it's stuck on waiting for reply....
ok @hacker09 I'm solving ChatGPT Widescreen new UI fix because many users are waiting for that right now, then after that, do you want this or the Autoclear front-end history clearing more?
I would likely use both together
I was using another script available on greasyfork that does the same as this one, but that one is older... And now has the same issue as this script.
I liked finding the auto-clear script because it cleared all the random stuff I googled after I logged in on chatgpt. The issue that old script has is that sometimes (maybe after a cookie expires or something like that) I had to keep logging in on chatgpt back again.
If there is a way to make the GoogleGPT script work without requiring an account login I would prefer to have this script fixed first.
My browser auto-clears everything (history/cookies etc) when closed anyway.
ok then I will prioritize this one, also auto-clear already gets rid of chats in back-end so if your concern is privacy, functionally it solved it already
Thanks!
I'm concerned about privacy, but not about that script. I only use that script to save time mass deleting chats, which makes it look better and not cluttered, but currently, it is really tricky because I can chat in deleted chats, so I get an error message saying that the chat no longer exists...
let me know if that is possible or not and when you update it
@hacker09 I just updated it to work with Proxy Mode again no login
Perfect!!!
if you could remove the little ad footer that would be nice... Sometimes instead of an ad I get an inspirational quote or a random joke https://imgur.com/a/u3RkUr6
Also, sometimes the background grey bar is much larger than it needs to be, it should stop when the last suggestion is shown, but it keeps going till the end of the page. https://imgur.com/a/NWcZXFN
I'm using pagetual as well, and the more I scroll down and new pages load, the longer the grey background bar goes.
Not sure if this is something you could try to figure out, but most of the contributors listed can't be seen after the first contributors line https://chatgpt.js.org/#/
@hacker09 the footer is my way to continue to work on it for free. For background bar, does it stretch before pagetual's first next-page-load? Also does it stretch with pagetual off? Also what is your browser and userscript manager
Not sure if this is something you could try to figure out, but most of the contributors listed can't be seen after the first contributors line https://chatgpt.js.org/#/
It looks like this to me:
How about you? What is your browser?
Yes it does it stretch with pagetual off.
edge/tampermonkey
Maybe you could convert the markdown that chatgpt uses, like ** into a <b> html tag
Maybe you could convert the markdown that chatgpt uses, like ** into a
<b>html tag
Ok I added it
Thanks!
It shouldn't run on the Google Images page, but it keeps running on this page
ok @hacker09 I fixed it thanks for heads up
Thanks!
That was so quick!
Occasionally the answer I get is messed up like
Hey @hacker09 I fixed it, the reason was for that specific API, I never wrote the logic to parse it as a stream, because originally the priority was get as many proxy APIs going as possible, and during testing, that one always skipped via tryDiffAPI() since it was temporarily down for the day so I never caught that. But now I wrote the logic, so all the APIs stream perfectly now
Thanks
I am not sure why, but seems like your script has something against Google developer coding queries, haha.
I searched for https://www.google.com/search?q=join+2+arrays+online+javascript&safe=off&sca_upv=1#ip=1 and it made your script menu page stay in the bottom of the page, and it also makes the whole site keep focusing on your element until the answer finished, which makes pagetual load the next page forever until the AI finishes the answer.
Can we have the script stop auto-focusing on the chat bar?
@hacker09 is this with the latest version where Auto-Scroll is now a mode (and off by default)?
(in which case it only auto-focuses if app is fully contained within the viewport)
Oh ok, not it wasn't I guess it was the version released 4 hrs ago
Now with the latest version it no longer scrolls down or auto-focuses, but on my link above I'm still seeing it below everything, at the bottom of the page.
Oh ok, not it wasn't I guess it was the version released 4 hrs ago
Now with the latest version it no longer scrolls down or auto-focuses, but on my link above I'm still seeing it below everything, at the bottom of the page.
wdym can you screenshot
Wow never encountered imgur down before but I shall investigate whenever they recover
I've never seen that before either https://streamable.com/gcg0fz
btw if you sign up for GitHub, you could make contributions directly to the code @ https://github.com/KudoAI/googlegpt since you're a programmer
I've never seen that before either https://streamable.com/gcg0fz
That only happens w/ Pagetual right?
nvm I see it happens without Pagetual in Edge only, I will investigate now
Nope.
Funny story, the issue is https://chromewebstore.google.com/detail/anti-anti-debug/mnmnmcmdkigakhlfkcdimghndnmomfeo
Funny story, the issue is https://chromewebstore.google.com/detail/anti-anti-debug/mnmnmcmdkigakhlfkcdimghndnmomfeo
I don't understand what you mean, also I found a way to fix, but it will hide their showing of native AI and I don't want to do this (they stalk me and my BraveGPT made them copy search.brave.com way of inserting ai answer, but they use a flex way that is incompatible unless I hide them)
"their showing of native AI"? what do you mean?
I mean that I tested all my scripts and extensions off, and when I had only https://chromewebstore.google.com/detail/anti-anti-debug/mnmnmcmdkigakhlfkcdimghndnmomfeo enabled, your script had that issue.
Ohhh i get it. But actually I'm having that issue too without your extension. It's because they added a new CSS property that occasionally triggers to fit their native way of showing AI (copied from my AI apps) but actually I figured a way around it. All I have to do is remove and re-insert my app div so it no longer shoves to the bottom. But I have to smoke first then will finalize the details of this solution
good...
smoking kills though...
this is usually true but it saves me from COVID (makes discomfort go away)
I figured out, it is the markdown rendering causing the bug
omg I finally fixed it, it took forever and much testing of different techniques, but the bug comes from the markdown renderer `pedantic` option is `false` by default causing it to do unwanted stuff triggering weird behavior, so I set it to `true` and it finally went away
It causes code to look like this now instead though:
```javascript
code
```
... which is actually beneficial since syntax is not highlighted for code (yet) so there's no use for the language word (yet), so at least now it informs what language the code is
Thanks
That would be a nice future update to format the code
It also shouldn't run on the videos pages https://imgur.com/a/jm8vX0s
Thanks
That would be a nice future update to format the code
I added it look
It also shouldn't run on the videos pages
https://imgur.com/a/jm8vX0s
For this, I noticed before it does, but did not block it since that page has empty sidebar so GoogleGPT does not intrude IMO. Do you find it bad to be there? For example, user could be searching for video of some recent event, and GoogleGPT could provide more context
Cool, that was fast!
But javascript const is wrong...
I guess the title in chatgpt would be javascript, but it was added along the code as it if were part of the code... That should be fixed.
Also, could you add a "copy code block" button?
Isn't bad, but with pagetual is, the bug is in the image I shared...
It leaves page 1, and goes to page 2 for no reason, messing up the whole next pages view.
Actually, I'm still having the same issue of the video I sent you earlier with pagetual when I google "matrix background HTML" Also, now the menu is white even if my whole google page is black.
It's broken so badly and crashing my page, as well as loading the next pages forever and in an unreadable way that I had to disable the script
https://imgur.com/a/8b7Hkr1
The site and AI answer content are no longer readable, so the script is unusable.
Cool, that was fast!
But
javascript constis wrong... I guess the title in chatgpt would be javascript, but it was added along the code as it if were part of the code... That should be fixed. Also, could you add a "copy code block" button?
Google is such a pain, I added special logic to fix formatting code after that screenshot, no other seach engine required such special treatment for my BraveGPT and DuckDuckGPT
Also, now the menu is white even if my whole google page is black.
Again Google is very painful and uncooperative when it comes to scheme detection, I have to constantly update isDarkMode() (they don't use a simple flag like everyone else)
Actually, I'm still having the same issue of the video I sent you earlier with pagetual when I google "matrix background HTML"
I never actually installed Pagetual to test what you mean and since you didn't say anything I let it go, I'll install and check it out after I eat (I often take hours to eat though)
It's broken so badly and crashing my page, as well as loading the next pages forever and in an unreadable way that I had to disable the scripthttps://imgur.com/a/8b7Hkr1The site and AI answer content are no longer readable, so the script is unusable.
It crashes your page with Pagetual you mea? And how are the next pages unreadable?
For the darkmode, it's because there are two ways it is detected, one is on a element that doesn't always exist, and the other is on the logo being white/blacck that doesn't always exist. But I noticed since Google stalks me, after I added logo detection (just like first time then I removed it since they updated logo to toy with me) they immediately changed the logo again
Also I don't use Google because they are a very unprivate company/search engine making it so much extra work monitoring their site for changes to accomodate their unnecessary and relentless need to make updates that don't really add value (similar to OpenAI to chatgpt.com at at times)
How it gets
https://imgur.com/a/8b7Hkr1
How should be
https://imgur.com/a/VFU5p2o
How it gets
https://imgur.com/a/keqkh4e
How it should be
https://imgur.com/a/B3y02t9
https://imgur.com/a/kpETjey
Normal behavior until it reaches 0:16
0:17 glitches and the next page is called and attached to the page (this shouldn't happen unless the user scrolls nearly all the whole page to the bottom)
0:06 glitches and leaves the top right side of the first page, and is attached to the next page that was just loaded (The links are moved to the left side of the screen, making it impossible to fully see them, I can only see parts of the right end side of the links)
0:02 the glitch above gets fixed because I manually scrolled down, now I can finally fully see the whole link (the left part that was hidden)
Thinking about it now, maybe if you use // @noframes that would be fixed.
Hey thanks for the video, I added // @noframes also a scheme selection popup you can pick from the userscript manager toolbar menu:
The way it works, it will default to Auto and try to sync w/ Google's scheme displayed, but if you pick 'Dark' or 'Light' it will always show in that mode (so should fix the dark scheme fails to auto-detect for you)
For the moving-around glitch, is that with Pagetual on or off?
This service has been suspended.
@hacker09 I removed the proxy API that got suspended by its creator's host so it works 100% of the time again
Fast as always!
😎
Send Reply, is always auto focused a billion times until the answer ends, so it's impossible to click on other input places in the search page/bar, there is also no way to edit whatever I search for, so I have to keep waiting for the answer to finish so that I can finally edit whatever I searched for.
I'm guessing that this is the line that should be removed
appDiv.querySelector('#app-chatbar').focus()
Send Reply, is always auto focused a billion times until the answer ends
hey sorry I just read this, I changed it to focus only one time
there is also no way to edit whatever I search for, so I have to keep waiting for the answer to finish so that I can finally edit whatever I searched for
How would you like to edit it? What about a stop button for answer to stop generating? Or allowing typing new query in chatbar and send to interrupt? Or a cancel button when "Waiting for reply..." is newly shown in footer after your reply is sent, restoring the chatbar?
No worries.
That doesn't sound good.
I'm not sure I get your second suggestion.
I'm only talking about the google search bar, not your script "search/input" bar.
I would rather never have your script input box focused, or only focused once like 3 seconds after the page fully loaded.
How come? Google de-focuses search bar currently, and GoogleGPT focuses GoogleGPT bar when answer starts loading. Is it because you're often typing a new query in Google Search bar, and then GoogleGPT takes focus away?
Basically yes.
If I search for a huge chunk of text, and click on it (after all results showed up), then click on where I want to modify my text, your script focuses on its input bar, so I have to click on the search bar again, find where I was at/which word I wanted to modify, then click on it again, then I'm finally able to modify that specific part of my long search text
Ok I added a menu toggle to disable it
Thanks.
Btw recently your script makes Google impossible to use
https://imgur.com/a/ZfbCbpV
The page crashes right after the answer ends
This seems to be a very known Chrome bug ongoing for many years "happening for multiple websites" https://www.reddit.com/r/chrome/comments/m3d7w8/getting_status_access_violation_error/?rdt=49814
Some solutions posted are:
.
I solved it by just resetting settings on chrome resigned in and it worked
.
I was able to solve this by accessing chrome://flags and resetting them all. I tinker with these flags before and I guess one of the updates has a conflict with one of the flags I set.
.
Settings > On Startup (on the left side) > Advanced > Use Hardware Acceleration (set to OFF)
@hacker09 can you try disabling Related Queries for a while and tell me if the Chrome bug ever occurs? (So I can pinpoint if the transition fade-in fx are the cause)
I did it, but nothing changed It's hard to change settings now, since after google finishes loading your script crashes the tab, and all tampermonkey scripts and options as well.
It's much easier to change what I already have in the storage
{
"googleGPT_autoget": true,
"googleGPT_fontSize": 16,
"googleGPT_notFirstRun": true,
"googleGPT_proxyAPIenabled": true,
"googleGPT_replyLanguage": "en-US",
"googleGPT_scheme": "dark",
"googleGPT_stickySidebar": false,
"googleGPT_widerSidebar": false,
"googleGPT_rqDisabled": true
}
I thought you said it was after GoogleGPT finishes loading answer
I was able to replicate the bug in Edge!
Yes I know. This is a very common bug for Chromium/Edge userss so I have to dig thru pages of complaints on https://answers.microsoft.com/en-us/microsoftedge/forum/all/status-access-violation-issues/1fd4a2ef-6736-441f-8421-6ed167105093 for hints on how to fix
Is it? I've never heard about or seen it until your latest updates.
Maybe you could undo whatever you did recently.
I doubt users would want to use/install your script if a whole long tutorial and regedit/etc change has to be done in their system.
I do not intend for users to read long tutorial or make regedit changes. I said I'm combing thru pages for how to fix GoogleGPT to work despite this Microsoft bug. I did pinpoint it only occurs in Streaming Mode though, so I just have to play around and make it work despite Microsoft's common bug is not fixed by Microsoft for years
I will probably temporarily disable Streaming Mode for Edge users if I can't fix it fast so at least it works 100% of the time again
@hacker09 I fixed it! Can you confirm it works in Streaming Mode now?
Cool
Could you explain what the issue was?
From my research, the STATUS_ACCESS_VIOLATION happens because Google added renderer code integrity years ago, then Chromium browsers (like Edge) adopted it. It produces lots of false positives though, so Google probably fixed it just enough so fewer Chrome users encounter/complain, but other publishers like Microsoft did not.
So while my text stream renderer in dataProcess.stream() ran smooth in every other browser, in Edge it triggered false positive (maybe thinking DDOS is occuring for going so fast) so I staggered the chunk processing every 50ms to conquer this glitch
I see, Thanks.
The timeout isn't long enough yet, I still get that error sometimes
@hacker09 can you double it to 100 in the line when you Ctrl+F for 'Edge delay' ? Then if it still happens, can you double it againn? Then can you let me know the lowest number that makes the error never occur?
You've already done it, so I'll test it for a while
same...
ok it is doubled to 200ms
same...
After testing other userscripts in Edge that also process streams, the same error kept happening. It only goes away if the delay is 250ms, but it's too choppy to be an enjoyable "Streaming Mode" so I auto-disabled it for Edge users until Microsoft patches this common false positive (it just makes no sense I spend so much time trying to obliterate this roadblock when I can be delivering functional new features)
I opened an issue @ https://github.com/KudoAI/googlegpt/issues/35 in case someone can make a workaround, in the meantime it's really not that bad in regular mode (just shows the answer after all chunks are received)
Yeah, I was about to recommend you to do that. Thanks
When shopping for stuff, your script menu stays over items
hey @hacker09 I can't get your version of Shopping page to appear, can you paste the URL?
https://www.google.com/search?q=adapter+macbook+to+hdmi&ie=UTF-8&safe=off
I'm still getting Error code: STATUS_ACCESS_VIOLATION message too
I had to manually disable stream mode to make it work
thanks for Shopping link I fixed it, re: STATUS_ACCESS_VIOLATION can you confirm it never appears after your manual disabling Stream Mode? If so, I will investigate it only after I finish working on this new popup menu to allow highlighting text on any site to launch GoogleGPT
Yes, I think it was fixed when I disabled it.
But honestly, this option should be hidden for edge users
It's not supposed to be enablable in Edge but show a warning:
I prefer the warning over hiding because it encourages users to switch browsers if they truly want the mode, also potentially gets them to complain in the Microsoft bug thread so Edge can finally support it properly
I don't think that users would switch, but I agree with that last part.
Can we at least have it hidden on the tampermonkey menu then?
I think it's better to have a single settings button on the menu, than a whole lot of options there and also on tampermonkey
I also can still enable it, and I'm not seeing that message
Why would users not switch? Also The Tampermonkey menu is the only spot it can be viewed so hiding it makes it impossible to know about the mode otherwise. I will make a single settings modal it's on the tasklist. Also does this return true for you in console:
navigator.userAgent.includes('Edg')
if not, what does console.log(navigator.userAgent) log?
Check it out my tasks list is long (look at size of scroll thumb) and grows every day
Because I wouldn't...
Yh I noticed that, but hide everything on TM and add it inside this single menu
It does return true
Edg/126.0.0.0
Yh that is the same for my tasklist.
I've quite a few tasklists actually
I like to work from bottom-to-top though this is where I'm at:
Because I wouldn't...
But you are one person you cannot possibly speak for what others who are not you would do, everyone prefers different things, some may want a stream response and would sacrifice Edge for the gain
Yh I noticed that, but hide everything on TM and add it inside this single menu
The next task shall be a single settings modal accessible from settings menu item in Tampermonkey menu and app corner button
It does return true
For thiis line
if (isEdge || getUserscriptManager() != 'Tampermonkey') saveSetting('streamingDisabled', true) // disable streaming if Edge or not TM
Can you add lines above/below it so it is like this
config.streamingDisabled = false
if (isEdge || getUserscriptManager() != 'Tampermonkey') saveSetting('streamingDisabled', true) // disable streaming if Edge or not TM
console.log('config.streamingDisabled is', config.streamingDisabled)
Then does this log show true. Also for the line
registerMenu() // create browser toolbar menu
Before it can you add console.log('isEdge is', isEdge) and reveal what it logs
That works, better 1 line saying "settings" than a lot of lines with a lot of options.
I feel like it would be easier if you give me the full script with the code you want me to test, so I just click whatever button "TEST" and give you the output.
you can use raw.github or pastebin.com
Thanks
I got Error code: STATUS_ACCESS_VIOLATION
Is there a "TEST" button somewhere?
No can you click again, this time the app will not even show in sidebar, the console will log isEdge is true/false and config.streamingDisabled is true/false
After my tab crashes, so does the console and everything is cleared
We can try alert or local/session/TM storage
Ok I changed it to alert and disabled more of the main routine https://github.com/KudoAI/googlegpt/raw/test-ver-for-hacker09/greasemonkey/googlegpt.user.js
isEdge is false and config.streamingDisabled is false
>:( Edge is beating me up I have to take a Korean noodle break to recover, it even completely derailed me from my productive flow building the new highlight-to-search from any site feature
What's the alert say now? (It is changed to alert(navigator.userAgent)) https://github.com/KudoAI/googlegpt/raw/test-ver-for-hacker09/greasemonkey/googlegpt.user.js
Safari? hahaha
Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36
But this works
https://www.whatismybrowser.com/
https://www.whatismybrowser.com/detect/what-is-my-user-agent/ (I get Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36 Edg/126.0.0.0)
Ok I wrapped the alert in window.onload how about now https://github.com/KudoAI/googlegpt/raw/test-ver-for-hacker09/greasemonkey/googlegpt.user.js
same
Ok I changed it to alert(JSON.stringify(navigator.userAgentData?.brands)) how about now
This makes sense to me
https://community.atlassian.com/t5/App-Central-questions/User-agent-string-value-not-showing-EDGE-browser-info/qaq-p/1625685
I tested some stuff on other sites, and it always worked, but nothing worked on google
[{"brand":"Not/A)Brand","version":"8"},{"brand":"Chromium","version":"126"},{"brand":"Google Chrome","version":"126"}]
But earlier you said
It does return true
Edg/126.0.0.0
This was on Google?
I hate Edge with a passion it has sucked at least 12 hours out of my life for nothing in return. Microsoft programmers are just bad
I don't remember, I guess it has here actually...
Hmm, it may be google.com fault though, because that code works on other sites.
I changed it to `alert(navigator.appVersion)` what about now https://github.com/KudoAI/googlegpt/raw/test-ver-for-hacker09/greasemonkey/
Also in console on google.com, does pasting this in console show Microsoft Edge in list
JSON.stringify(navigator.userAgentData.brands)
JSON.stringify(navigator.userAgentData.brands) '[{"brand":"Not/A)Brand","version":"8"},{"brand":"Chromium","version":"126"},{"brand":"Google Chrome","version":"126"}]'
Ok and the new test alert?
no change
but it was changed to appVersion obj and my last link was wrong https://github.com/KudoAI/googlegpt/raw/test-ver-for-hacker09/greasemonkey/googlegpt.user.js
Also you must admit you hate Google/Microsoft too, you are only testing a simple alert and I can tell it is very irritating to you (otherwise you would respond faster) that if only an employee just properly coded the browser detection, neither of us would have to waste our unpaid time on this unnecessary task cleaning up after highly paid employees
We've been doing this for literally 3 hours
A Microsoft/Google employee would get paid like $1000+ in that time
My uncle's best friend works for Microsoft he's not smart either. Also why Windows always crashes
alert = 5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36
Since its impossible we can just give up and you can leave a notice on your script/greasyfork for edge users to never enable stream mode
HhHhaahaaa
That's why they ditched Internet Explorer and old Edge, the code was so bad it would take too many hours to fix and employees get paid a lot, so a boss said nahhh we using Chromium
I delayed the alert 5s because I am very curiuos now does it ever show right (since my Edge works) https://github.com/KudoAI/googlegpt/raw/test-ver-for-hacker09/greasemonkey/googlegpt.user.js
Microsoft funded me with like $100000 in service credits btw but I still speak truth on programmer quality, nothing to do with financial generosity (especially true since their bad programmers get paid ridiculously)
I added
// @match https://*/*
on your script
and I get
[{"brand":"Not/A)Brand","version":"8"},{"brand":"Chromium","version":"126"},{"brand":"Microsoft Edge","version":"126"}]
On any website that isn't google
I'm not sure what google is doing, but it has hard coded navigator.userAgent etc inside script tags, maybe it's possible that google is changing my navigator from edge to safari for whatever reason?
Either that or edge is doing that on google.com, likely it does the same to be able to make edge install chrome extensions?
I wonder if any of those options are doable
Can you disable pagetual and all other scripts/extensions, since Google doesn't do that to me? And just in console, can you ever get it to log navigator showing anything with Edge in it
I did it, I created a whole new profile just to test your script, and nothing changed, neither on private mode
After disabling all extensions, just type navigator and hit enter in console on Google.com, then expand the obj, can you find anything w/ Edge
I did it, I just browsed as guest before installing TM and your script, and navigator.userAgent returned safari
forget about TM and script just disable them, also forget userAgent does appVersion show Edge
If userAgent does not show Edge appVersion also won't....
It does not show edge as I said...
I disabled Tracking prevention
Updated my browser
restarted it
and nothing changed either...
sorry, 您的ip已由于触发防滥用检测而被封禁,请勿滥用本站,本服务网址是https://chat18.aichatos8.com 或者 https://cat.chatavx.com/ 如果你不在本网站，请前往本网站使用即可 如需合作接口调用请联系微信chatkf123 或者前往 https://binjie09.shop 自助购买key, 认为是误封需要解封的请前往https://www.ip.cn/ 查询ip信息,并发送信息至邮件 gpt33@binjie.site ，站长会定期看邮件并处理解封和合作问题，如需调用接口请见接口文档https://apifox.com/apidoc/shared-803d9df6-a071-4b3e-9d69-ea1281614d82 如需合作接口调用请联系微信chatkf123 或者前往 https://cat.chatavx.com/ 注册使用（可付费使用gpt4 注册可免费使用3.5）
Ok I made it skip on this msg, also added X-Real-IP to request headers to increase the chances the generated fake IP is detected instead of real one
If userAgent does not show Edge appVersion also won't....
It does not show edge as I said...
Not necessarily, a script can be hacking userAgent and not appVersion since userAgent is the common one to mess with
Thanks.
Oh I see...
If your script is now meant to run on all sites, there's no need/point or rule making you have to add a bunch of sites to match.
That is not what the rule means.
if you want to do this for whatever reason, why not use 1 single regex include line?
I told the Greasy Fork owner the same thing, the reason is he discourages single match all and punishes those who do by hiding the script from pages like /by-site/google.com.... https://github.com/greasyfork-org/greasyfork/issues/1281
@hacker09 why don't you open an issue to complain about this, he is very stubborn and doesn't believe single me when I ever mention this or that is bad UX about this website
(he gets mad easily though, notice he muted all public conversation on my productive issue thread because it made him look like his reasoning was poor to encourage longer match lines thus larger filesizes, but the stubbornness is like he doesn't want any more convincing or else he might have to change what he's stubborn to not change)
It's hard to get him to change anything you must be very persuasive, believe me I joined this site around the time it launched 11 years ago, the design did not change at all lol
I often use money to effectively persuade him though (mentioning ad revenue will drop/rise etc.) if that's helpful
Oh I see, well I agree with both of you actually.
Most of the time I don't want to see 1000 search result pages just because match all is default, but occasionally? yes, I would do.
Like in your case, right now you are really concerned and wanting your script to show when searched by site even when/if you use match all, I would do that if I am crazily happy or getting money or whatever from a script I made (which often happens), even if by site "google.com" showed 1000 search pages I would take my time to slowly go through all these pages and learn from other scripts, install/test/report them or whatever else.
But usually a single search return "most" scripts, that have what I/most users are looking for.
A good solution for both sides would be a quick mini toggle "show match all scripts" kind of a thing.
I'm sure some devs and crazy users looking to know about every single script ever made "potentially" to google.com /etc would be interested on.
Does using 1 single regex include line make the script not show up when searched by site? I've to test this further to know for sure.
Maybe some regex would make it happen, while other kind of regex/symbols or whatever would make the script be hidden.
Another option you have is use match all if you don't want regex, or 1 single regex include line, but then use an extra single match google.com just to make your script show up when searched by site.
even if by site "google.com" showed 1000 search pages I would take my time to slowly go through all these pages and learn from other scripts, install/test/report them or whatever else.
idk what you mean
Does using 1 single regex include line make the script not show up when searched by site?
// @match *://*/* matches all (discouraged by GF owner) and hides script from /for-site pages
But usually a single search return "most" scripts, that have what I/most users are looking for.
...and GoogleGPT absolutely belongs on /for-site/google.com so until GF owner fixes above punitive behavior, I will not be matching all
Does using 1 single regex include line make the script not show up when searched by site? I've to test this further to know for sure.
IDK what regex, but if you test some single line to not hide from /for-site/google.com then I shall use it
Another option you have is use match all if you don't want regex, or 1 single regex include line, but then use an extra single match google.com just to make your script show up when searched by site.
That didn't work
When I'm really into a script I created, I want to know what other devs did recently and in the past, so I check those pages.
I report old stuff that breaks the new GF rules mostly because they don't work either.
I don't want to make/copy someone else's script, or see that someone did it with one of my scripts, which is another reason to check.
Or if I just love using a website a lot, then it would be helpful knowing all scripts that apply to it.
I know, // @match *://*/* does not mean 1 single regex include line.
I got that, it's your choice, but if you use 1 single regex include line or if you use // @match *://*/* and // @match *:google.com/*, does it not show up when searched by site google.com? (NVM I got your answer on the next line)
The solution then is to try figuring out a regex that works with 1 include to add many sites to it while making it show up when searched by site google.com
`// @match *://*/*` absolutely is 1 single regex line and since it makes it say 'all sites' then probably all regex will disable any individual /for-site page
This is outside my control you need to open an issue on greasyfork-org/greasyfork github to fix. Suggest your ideas then lemme know what happens
He doesn't like me because I bring great shame to his perceived intelligence when I productively criticize UI things. You are another user though, mentioning issue as someone else might make him take this problem more seriously
He takes great pride in his version of this UI, that's why it remained unchanged for over ten years, urging action from me will do nothing to cchange this only he can do it, you as another impartial user may change his mind
// @match does not support regex, only // @include does
Normal users can try publishing scripts with match or include and use regex, I/you can test it without his help or without him changing anything.
that's just not true
I have no interest in testing to circumvent site restrictions, you can if you want and if it behaves exactly as I know it will then you have to contact him to change it to work the way you are explaining you want it to
Weird
https://stackoverflow.com/a/31819287
According to the documentation on https://www.tampermonkey.net/documentation.php?locale=en#meta:match that is https://developer.chrome.com/docs/extensions/mv2/match-patterns and stackoverflow I believe that match is just much more limited than include at least.
// ==UserScript==
// @name Abc New Userscript
// @namespace http://tampermonkey.net/
// @version 2024-06-21
// @description try to take over the world!
// @author You
// @match https://myanimelist.net/anime/\d+
// @match https://myanimelist.net/anime/d+
// @match https://myanimelist.net/anime/[0-9]+
// @grant none
// ==/UserScript==
alert() //supposed to alert on https://myanimelist.net/anime/345
ChatGPT
The issue with your current @match directives in the userscript is due to the fact that they do not use regular expressions as you might expect from JavaScript contexts. In userscripts (like those for Tampermonkey or Greasemonkey), @match directives do not support regular expressions directly. Instead, they use a simpler pattern matching syntax that resembles shell-style wildcards.
Why it doesn't match https://myanimelist.net/anime/345:
None of the patterns you've defined (https://myanimelist.net/anime/\d+, https://myanimelist.net/anime/d+, https://myanimelist.net/anime/[0-9]+) actually match the URL https://myanimelist.net/anime/345 because they are either incorrectly formatted or don't utilize wildcards correctly.
Correct Approach:
Since @match directives in userscripts don't support regular expressions but use basic wildcard matching, you should specify the exact URL format or use the appropriate wildcard syntax:
To match URLs like https://myanimelist.net/anime/345, you should use:
// @match https://myanimelist.net/anime/*
Anyway, that doesn't matter.
I'm not trying to "to circumvent site restrictions", using regex in include to match multiple websites in 1 line is allowed, many scripts do that, my question is if you use that would it show up when searched by site google.com?
It's just a suggestion that could help you out and decrease your script line codes and size, as you wanted.
I may test it myself, or not...
Well ChatGPT is often wrong, but even if it's not, match is recommended over include for security, but I doubt it will show up on /for-site/google.com if you use include w/ regex because this site is doing 'For all sites' for match with wildcards, why would it be different for regex
Yes, you are right. But after my google search I still think that include is more powerful.
You are right, I only use it when I need to use complex regexes.
Tested and worked
// @include /^(https?:\/\/)(www\.)?(google|anotherfakesite|example)(\.com)(\/.*)?/
Also if you want to help me out you need to tell the owner to make it work the way you expect, that's the only way I'm goign to use regex, otherwise I am paying a penalty of not showing up on /for-site/google.com and you cannot convince me to pay this hefty price for the gain of reduce KB in size
Ok I will use regex!
This was something I did and used for years in my script(s) and never got banned/reported over that, so I believe it won't be an issue either at the moment.
I'm not trying to convince you, I just would like to help.
If you happened to deny help and deny using include/regex I wouldn't care.
Wait you used include. Why not test match? I don't think include will bbe supported in future Tampermonkey versions
I did not deny help I sought help and told you exactly how (tell owner to make match wildstars not exclude script from by-site/ pages but you won't do it for some reason)
To make sure it will work you are allowed to create another account and post the script using include+regex on it, then take the script down after a few minutes, or you can do that on this account too. I do this all the time for testing purposes as well.
I don't care to use include though it's deprecated
I've tested match as your image said \d+ to match numbers, but much is too secure and dumb, so you can't use it, the only thing I know that can be used is *, which don't help me, and don't allow you to use or | in regex, as include does.
> I don't care to use include though it's deprecated
Me neither, so I will use it as long as possible, TM started saying it will stop working in the beginning of 2022 I think, the message is still saying that will stop working on "Early 2023 due to Manifest V3" hahaha, I'm just waiting.
Devs hate google because of manifest v3, so google has been delaying the update for at least 2 years now
It's not deprecated, but may not be supported on Manifest V3, no one knows.
Yh that post is what I said earlier...
> I don't care to use include though it's deprecated
Me neither, so I will use it as long as possible, TM started saying it will stop working in the beginning of 2022 I think, the message is still saying that will stop working on "Early 2023 due to Manifest V3" hahaha, I'm just waiting.
Devs hate google because of manifest v3, so google has been delaying the update for at least 2 years now
Tampermonkey already doesn't work for new users because of this change without awkwardly toggling on Dev mode. I don't even advise them to use it anymore for this reason
If you don't believe me, uninstall Tampermonkey, then re-install it. None of your sccripts will work
So the point is Tampermonkey's author is very obedient to Google and MV3 so include will mos definitely stop working too
I'm not sure it matters, and having to enable Dev mode isn't related to include I think.
Every script manager that finally moves to V3 will have to do that from what I understood.
you pasted a message saying include will top working "due to MV3" and enabling Dev mode was also from moving to MV3 so that is how they are related
i.e. the more the Tampermonkey author conforms to MV3 standards the more pre-MV3 things like include will stop working. So past users (of which I have hundreds of thousands) will not even see a menu to update if I don't beat up this obstacle now using match
I do believe you, I just had to do that on another laptop I got. I'm aware of this change/request of TM to user allow the dev mode.
I think that TM is just ahead of others when it comes to migrating to V3, that is why TM already requests this.
If there was a way to bypass this, I bet TM would do that, and so would any other script manager.
TM won't have include obsolete just to obey google, google requires every extension to move to V3, therefore include is automatically obsolete on any script manager that starts using V3. "If there was a way to bypass this, I bet TM would do that, and so would any other script manager."
Either way, it's your choice.
For now I would (you could) use include, and wait until google/TM finally moves to V3, this may be a year ahead or more, or less, no one knows either.
At least for now and for a while your script would have less lines of code and take less size.
include is automatically obsolete on any script manager that starts using V3
Then it doesn't work on new Tampermonkey (5.2.0+)? That's more reason to stay far away with a ten foot pole
You can leave a match on your notes, and use it when the time comes.
If you install new Tampermonkey it says it uses MV3 now is the reason Dev mode is required
I am not messing with include that's like driving on a bridge that's collapsing
I'm on 5.1.1 that is the latest available on stores, and it works
Are you scared of the site owner?
Your call, just wanted to let you know that this option is available, and is the only one...
I'm on 5.1.1 that is the latest available on stores, and it works
Nahh that's Edge, in Chrome (which most of my users use) it's already MV3
Your call, just wanted to let you know that this option is available, and is the only one...
Not only option, you can lobby site owner to not restrict loose matches from showing on /by-site pages
I'm not, but I can see that I will have the exact same experience as you, and even if I don't, I can see that this will take years for him to do (most likely), so there's no point in suggesting that or anything.
How come years, also I'm 1 guy that complains to him a lot, you are new guy bringing fresh vote to make him realize multiple users find it annoying to instill urgency
I'm not sure about an exact timeframe
Unless your native language is English, I'm pretty sure that "Click green Install this script button at top of page (will load in userscript manager installed above)" is incorrect English.
It should be.
"Click on the green Install this script button at the top of the page (will load in the userscript manager installed above)"
Indeed English is my native language, omitting 'the's makes it shorter and is not ungrammatical
Good to know
I thought it would help, but nvm
I know your PT BR is wrong, and extremely likely the PT as well
It should be either
1. alimentado
2. distribuído
instead of desenvolvido
Cna you correct it on GitHub? If you refer to a UI message you can edit https://github.com/KudoAI/googlegpt/blob/main/greasemonkey/_locales/pt/messages.json or if you mean in a meta description you can edit https://github.com/KudoAI/googlegpt/blob/main/greasemonkey/googlegpt.user.js
I prefer to just give you the suggestions here
"state_on": { "message": "Activado" },
should be
"state_on": { "message": "Ativado" },
"buttonLabel_moreApps": { "message": "Mais aplicações ChatGPT" },
should be
"buttonLabel_moreApps": { "message": "Mais aplicativos ChatGPT" },
"alert_tooManyRequests": { "message": "O ChatGPT é inundado com muitas solicitações" },
should be
"alert_tooManyRequests": { "message": "O ChatGPT está sobrecarregado" },
"alert_thenRefreshPage": { "message": "Em seguida, actualizar esta página" },
should be
"alert_thenRefreshPage": { "message": "Em seguida, recarregue esta página" },
"alert_waitingResponse": { "message": "A aguardar resposta do ChatGPT" },
should be
"alert_waitingResponse": { "message": "Esperando resposta do ChatGPT" },
Everywhere saying "actual..." whatever is wrong and should be "atual..." something (just remove the c)
Distribuído pela should be either
1. Alimentado por
2. Distribuído por
Ok thanks for the corrections, look how many lines got changed (across 8 apps) thanks to your vigilance:
/autoheal
I prefer to just give you the suggestions here
"state_on": { "message": "Activado" },
should be
"state_on": { "message": "Ativado" },
"buttonLabel_moreApps": { "message": "Mais aplicações ChatGPT" },
should be
"buttonLabel_moreApps": { "message": "Mais aplicativos ChatGPT" },
"alert_tooManyRequests": { "message": "O ChatGPT é inundado com muitas solicitações" },
should be
"alert_tooManyRequests": { "message": "O ChatGPT está sobrecarregado" },
"alert_thenRefreshPage": { "message": "Em seguida, actualizar esta página" },
should be
"alert_thenRefreshPage": { "message": "Em seguida, recarregue esta página" },
"alert_waitingResponse": { "message": "A aguardar resposta do ChatGPT" },
should be
"alert_waitingResponse": { "message": "Esperando resposta do ChatGPT" },
Everywhere saying "actual..." whatever is wrong and should be "atual..." something (just remove the c)
Distribuído pela should be either
1. Alimentado por
2. Distribuído por
Glad I could help
@(UwU)
What's your problem?
thanks @hacker09 it is fixed
@hacker09 how come you disable the bg animations? Also without the stars, do you prefer the background be a single solid color like before, or the gradient like now?
I care about content, not themes, it's easy to read stuff without dots moving all over.
I don't really care, solid, or gradient either is fine.
Ok
Is it possible to remove the requirement to login?
ChatGPT no longer requires login to work...