TODO list

[zmwangx]

Rolling TODO list thread No. 2. Previous thread is #69.

---

Possible improvements we'd like to see:

• Document APIs in NumPy format [owner @jarun]
• Refactor and write unit tests

And moonshots:

• Support DDG (possibly in a separate project?)

PRs welcome!

---

Archive:

• Customizable colors — implemented in Customizable colors #84.

[zmwangx]

@jarun I would like to add the ability to open Google's cached results. Pretty useful when you encounter things like this:

which honestly appears quite often, or when you hit servers located god-knows-where that just won't serve the page.

The parsing strategy is to watch for li.action-menu-item inside div.s (abstract scope), and for a inside such an li, test for https://webcache.googleusercontent.com/search?q=cache: at the beginning of href. The parsing load is minimal. Result will receive one additional kwarg as a result.

The CLI will be designed to be c index, with c for cache.

Thoughts/objections?

[zmwangx]

It may be too niche of a feature and we might as well leave it to o. Not sure.

[jarun]

I'm fine with c. However, we need to indicate if cache is available or not, maybe with a (C) along with the rest of the result?

[zmwangx]

It may be too niche of a feature and we might as well leave it to o.

I mean, cache will be available if you open the search in Google. What I wasn't sure about is whether we should add this feature at all.

However, we need to indicate if cache is available or not, maybe with a (C) along with the rest of the result?

Just print an error message if cache is not available.

[jarun]

Please gimme some time to think on this feature.

[jarun]

I believe we can skip the google cache feature. It's an add-on feature that doesn't add much value.

[Gilepo]

I would love to see a cached page option, it would useful in seeing pages that i may not trust.

Another suggestion would be an option to allow omitted results, particularly handy in site searches or filetype searches.

an example of this is the search

site:blog.checkpoint.com/wp-content/uploads/ filetype:pdf

which returns 1 result or 17 results.

[zmwangx]

Re cache: @jarun Thoughts on fulfilling user request?

Re omitted: the latest parser logic doesn't omit any potentially valid result. I've seen the same query returning different numbers of results across runs too, but didn't look into it. Could be a bug (although I can't see any race condition), or could be Google returning different results. I'm not at the computer now, but I'll look deeper asap.

[jarun]

@Gilepo

it would useful in seeing pages that i may not trust

I didn't think from this perspective earlier, but you can always use the o key at the omniprompt to open the search in the browser. One additional step in the workflow but what is the percentage of pages you do not trust? I'ld say it's a rare case and we already have a way.

@zmwangx

the latest parser logic doesn't omit any potentially valid result

You are probably chasing another issue. @Gilepo is talking about

In order to show you the most relevant results, we have omitted some entries very similar to the 46 already displayed.
If you like, you can repeat the search with the omitted results included."

We can support this at the omniprompt if our parser finds it. We can show a short message and provide a omniprompt key.

[Gilepo]

Sorry, im new and may be imprecise.
By omitted results I was referring to results that google omits with the following message.

In order to show you the most relevant results, we have omitted some entries very similar to the 1 already displayed. If you like, you can repeat the search with the omitted results included.

The link included to the additional results had the var
&filter=0
Which i believe includes the full set.

I had not experienced results changing with googler until after you mentioned it.

googler -w 'blog.checkpoint.com/wp-content/uploads' filetype:pdf
returns 1 result
If I then input 2 into the omniprompt returns 10 results (one of which is not a pdf)

Perhaps i am misunderstanding this

@jarun
That is easy enough to use as a solution, my initial mention was prompted as a use case not really a request. I have yet to actually need the feature.

[zmwangx]

You are probably chasing another issue.

Yes I was. I was on my phone, and I knew a symptom that did match the description, so I didn't check.

We can support this at the omniprompt if our parser finds it.

TL;DR: I need more samples.

It's definitely in #extrares, but #extrares is also used for "searches related to..." (sample query: google) so that's not definitive. In the case above, the "omitted search" is also in #ofr, but I need more samples to decide if that's the key.

Turning filtering off is trivial, it's a matter of adding a filter=0 query.

But this brings us to something bigger: should we allow dynamic option changes within a single invocation (e.g., dynamically enabling/changing time limit in the prompt), just like Google's search tools? The UI could be as simple as entering the options as is at the prompt, and replace - with + to turn off an option. It would be great in a sense, but it might or might not be confusing to the user, and either case it will be quite some work for us.

[zmwangx]

I believe #ofr is the correct selector. I reached the conclusion by searching for very long sentences from Wikipedia. The last page usually has this "omitted results" thing, always enclosed by #ofr.

Interestingly, I also discovered that results taken down by DMCA are enclosed by #mfr. Try this: https://goo.gl/DYcHD2