Alternatively, you could use a tool or extension that does full-page screenshots and then run image optimisation on them. I do this a lot for local Pinterest-type inspiration store. At the moment I use [Nimbus](https://chrome.google.com/webstore/detail/nimbus-screenshot-...) but it seems like evert few months the extension I'm using starts to fail with certain websites (scrolljackers, mostly), and I switch to a different, newer extension.
Alternatively again, but back to saving websites, surely someone's created a nice tool that will download a page to store it as an archive that won't be broken? Pinboard for example has archiving for an extra fee, so I wonder how he does it at his scale.
Related to your problem grep'ing, I'm slowly working on a small idea to have a local tagging/metadata approach for finding things.
Wallabag[0] does that. It's a self-hosted Pocket-like read-it-later service that strips the page and saves the text of the article in a local database, therefore allowing you full text search right from your own server. It even adds some additional neat features like adding notes to the articles.
The only downside: damn those two currently available themes are awful.
"Ask HN: Do you still use browser bookmarks?" (19 days ago, 451 comments):
I would agree that bookmarks, generally, are a poor fit to current needs or requirements. When a typical hard drive was, say, 100 - 500 MB, the idea of only saving the URL, and not the content, could be argued. With mobile devices having 128 GB - 1 TB of microSD storage, there's no reason you cannot store everything you've read, or at least everything you're interested in, locally, on desktop, laptop, or mobile.
Which is what you've done.
But you're running into the underlying problem (as am I): a pile of randomly-titled, poorly-metadata'd PDFs isn't particularly useful.
There are tools -- Zotero and ... some others -- which manage references, but IMO do so quite poorly. The problem is that what they introduce is a metadata vetting and management problem, and one that IMO GUI tools manage quite poorly. The fact that the tools aren't available on Mobile (I use an Android tablet almost exclusively, because reasons, and yes, it sucks in a great many ways), makes this problem all the more nontractable.
I have the same problem, it turns out, with bookreaders. I use two, mostly: PocketBook and FBReader, with ~2,000 or so references. Unfortunately, other than title and author search, I've little by way of organisation of these, which is ... a major problem.
I'm using Pocket, The Article Management Tool that Gets Worse The More You Use It[tm] (https://redd.it/5x2sfx), which ... suffers many of the same problems, and adds a few more of its own.
I suppose the next logical step would be to save the page with assets so it can be made available for posterity. Not sure what impact that would have on storage use but it would at least enable full text search.
When I need the context I check the original link. If it's not there I try Google Cache or archive.org. If it's lost, I find an alternative (thanks to the title and notes fields in buku). That's more or less my workflow when in comes to the 8K odd bookmarks I have.
I can link my dotfiles if you're interested.
Instead, if I like a page I want to re-visit, I simply Print it to PDF. Then, I move all the PDF's from my Desktop every week, into their own permanent storage location .. meaning that I have every interesting web page I've ever read since 2000.
Trouble is, now I have a large PDF collection to manage. I get along fine with "ls -l | grep <something>" this and "pdf2txt <blah.pdf> | grep <something>" that .. but of course, this is not as 'clean' as if I had a Bookmark Manager to do all my searching/grep'ing/grok'ing/etc.