It makes me wonder why I've never seen it before
-
Why ? Because it is probably more complicated than it seems.
It's not really. I'm almost done with it.
Real programmers use butterflies
-
It's not really. I'm almost done with it.
Real programmers use butterflies
Yes, I was merely talking about normal mortals like us, not wizards :)
-
Yes, I was merely talking about normal mortals like us, not wizards :)
That'd be a witch. Wizards are a different thing altogether. :-D
Real programmers use butterflies
-
On IoT you don't have a lot of luxuries. You simply learn to do without. Well, one area where I have to do without is HTML and XML well formedness checking and validation. That might be an issue where data interchange is concerned, but not so much where rendering HTML or XHTML content is concerned. What do you do on an error? You fail. You can either stop, or continue to render, possibly having some bad content displayed as a result, but this is still a better case than failing outright halfway through the parse because the document forgot a
</b>. In fact, this is what commercial browsers do. Here's the thing. If this is what you're doing, you don't need a DTD. You don't need an XSD Schema. You don't even need a heckin stack! The result is much faster and lighter with a smaller binary footprint. So why haven't I seen a pull reader with minimal validation/well formedness checking in the open source pool? You'd think such a beast would be incredibly useful for building web browsers - even tiny ones - especially tiny ones! *cracks knuckles* I shouldn't have to be writing this. It's one of those things that leaves me wondering why it doesn't exist already.Real programmers use butterflies
I preprocess / strip out all the HTML and insert my own markup directives. I don't need a "paragraph" keyword to tell me where a paragraph should start or end; etc. But as you say, it's assumed to be valid HTML in the first place (or made to be).
It was only in wine that he laid down no limit for himself, but he did not allow himself to be confused by it. ― Confucian Analects: Rules of Confucius about his food
-
I preprocess / strip out all the HTML and insert my own markup directives. I don't need a "paragraph" keyword to tell me where a paragraph should start or end; etc. But as you say, it's assumed to be valid HTML in the first place (or made to be).
It was only in wine that he laid down no limit for himself, but he did not allow himself to be confused by it. ― Confucian Analects: Rules of Confucius about his food
As far as preprocessing I don't have the memory or nvs storage to do that on my device, and it wouldn't really buy me anything much even if i did. Much better in my scenario at least, to just read with a pull parser in a loop, get tag names back and set values on a context structure i use while rendering. The context structure has the current position and flags like "bold" or "italic", styles and font faces, that sort of thing. I mean, if I knew at compile time what my HTML was going to be I'd just generate C++ code from it that renders it, and that sort of "preprocessing" would be a huge win, but in my scenario I have to read arbitrary HTML.
Real programmers use butterflies
-
On IoT you don't have a lot of luxuries. You simply learn to do without. Well, one area where I have to do without is HTML and XML well formedness checking and validation. That might be an issue where data interchange is concerned, but not so much where rendering HTML or XHTML content is concerned. What do you do on an error? You fail. You can either stop, or continue to render, possibly having some bad content displayed as a result, but this is still a better case than failing outright halfway through the parse because the document forgot a
</b>. In fact, this is what commercial browsers do. Here's the thing. If this is what you're doing, you don't need a DTD. You don't need an XSD Schema. You don't even need a heckin stack! The result is much faster and lighter with a smaller binary footprint. So why haven't I seen a pull reader with minimal validation/well formedness checking in the open source pool? You'd think such a beast would be incredibly useful for building web browsers - even tiny ones - especially tiny ones! *cracks knuckles* I shouldn't have to be writing this. It's one of those things that leaves me wondering why it doesn't exist already.Real programmers use butterflies
-
honey the codewitch wrote:
*cracks knuckles*
Ye-aaaaah! Rock on Witchy Poo! Looking forward to the article(s) :thumbsup:
Thanks for the vote of confidence. I'm getting there but it's a bit of a bear. For starters, everything has to stream, because you don't have a ton of RAM. 1kB is a big deal, so i let you specify as little as 128 bytes for a buffer. I can't stream attribute and element names, but I can stream attribute values and element content,
Nbytes at a time (depending on what you had setNto) So if you have a long attribute value while you're doingwhile(reader.read()) {you'll get multiplereader.node_type()==ml_node_type::attribute_contentresults back before gettingreader.node_type()==ml_node_type::attribute_endNot only are there a zillion html entities like © (©) but I had to make a state machine to decode all of them efficiently off a unicode stream. Also this:<span class="foo">this is valid
<span class='foo'>and this
<input disabled>
<div class=you_thought_this_would_be easy id=but_no_because_html_hates_you_this_is_also_valid>So this is kinda rough sometimes, but I'm making progress. Fortunately I don't have to care about custom entity references, namespace declarations, or even well formedness (balanced tags, etc) which makes some of it pretty easy.
Real programmers use butterflies
-
Thanks for the vote of confidence. I'm getting there but it's a bit of a bear. For starters, everything has to stream, because you don't have a ton of RAM. 1kB is a big deal, so i let you specify as little as 128 bytes for a buffer. I can't stream attribute and element names, but I can stream attribute values and element content,
Nbytes at a time (depending on what you had setNto) So if you have a long attribute value while you're doingwhile(reader.read()) {you'll get multiplereader.node_type()==ml_node_type::attribute_contentresults back before gettingreader.node_type()==ml_node_type::attribute_endNot only are there a zillion html entities like © (©) but I had to make a state machine to decode all of them efficiently off a unicode stream. Also this:<span class="foo">this is valid
<span class='foo'>and this
<input disabled>
<div class=you_thought_this_would_be easy id=but_no_because_html_hates_you_this_is_also_valid>So this is kinda rough sometimes, but I'm making progress. Fortunately I don't have to care about custom entity references, namespace declarations, or even well formedness (balanced tags, etc) which makes some of it pretty easy.
Real programmers use butterflies
-
You bugger. The bit of my brain I need to use to make an intelligent response is entirely consumed with the act of killing myself laughing at your code-snippet. Fan-friggen-tastic :laugh: :laugh: :thumbsup:
I've got it running. I even wrote most of the article, but I should probably test it more.
Real programmers use butterflies
-
I've got it running. I even wrote most of the article, but I should probably test it more.
Real programmers use butterflies
