\
**Category:** Help & Support\
**Created:** [April 15, 2026, 8:57pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885 "2026-04-15T20:57:49Z")\
**Posts on this page:** 10\
**Page:** 1
**Author:**  [@Xen](https://forum.crystal-lang.org/u/Xen)\
**Post date:** [April 15, 2026, 8:57pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/1 "2026-04-15T20:57:49Z")
As it seems the forum beats my googling skills every time (I must be getting old), I thought I’d ask:
Is there any HTML parser shard that works on a stream rather than chewing through the whole thing in one go?
I’ve always been fascinated by templating systems that work directly with the HTML structure ever since I stumbled across Amrita in Ruby.
---
**Author:**  [@kostya](https://forum.crystal-lang.org/u/kostya)\
**Post date:** [April 15, 2026, 9:29pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/2 "2026-04-15T21:29:40Z")
lexbor, streaming parse method: [lexbor/src/lexbor/parser.cr at master · kostya/lexbor · GitHub](https://github.com/kostya/lexbor/blob/master/src/lexbor/parser.cr#L117)
---
**Author:**  [@MarioAriasC](https://forum.crystal-lang.org/u/MarioAriasC)\
**Post date:** [April 16, 2026, 1:55am UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/3 "2026-04-16T01:55:12Z")
Google is getting worse by the day
---
**Author:**  [@yxhuvud](https://forum.crystal-lang.org/u/yxhuvud)\
**Post date:** [April 16, 2026, 4:37pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/4 "2026-04-16T16:37:04Z")
Oh my, if that existed for XML I’d be quite exited. We have this service at work that would be absolutely perfect for that.
---
**Author:**  [@Blacksmoke16](https://forum.crystal-lang.org/u/Blacksmoke16)\
**Post date:** [April 16, 2026, 4:46pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/5 "2026-04-16T16:46:54Z")
> [@yxhuvud](#):
>
> Oh my, if that existed for XML I’d be quite exited.
Wouldn’t this just be [XML::Reader - Crystal 1.20.0-dev](https://crystal-lang.org/api/master/XML/Reader.html)?
---
**Author:**  [@yxhuvud](https://forum.crystal-lang.org/u/yxhuvud)\
**Post date:** [April 16, 2026, 4:51pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/6 "2026-04-16T16:51:25Z")
I’d need to be able to modify the structure as I go. I don’t think that is possible with`Reader`? But thanks for the suggestion, there may be use cases for just parsing too.
---
**Author:**  [@Blacksmoke16](https://forum.crystal-lang.org/u/Blacksmoke16)\
**Post date:** [April 16, 2026, 4:54pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/7 "2026-04-16T16:54:00Z")
Ahh gotcha. might be able to pair it with a `XML::Builder` where you create a reader using the source IO then the builder with the dest IO, then as you read call related write methods, passing things thru directly, or if you need to make a change then do that instead of the existing node.
EDIT: Did something similar with `XML::Reader`/`JSON::Builder` and `XML::Builder`/`JSON::PullParser` for [oq](https://github.com/Blacksmoke16/oq/blob/master/src/converters/xml.cr).
---
**Author:**  [@Xen](https://forum.crystal-lang.org/u/Xen)\
**Post date:** [April 16, 2026, 6:31pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/8 "2026-04-16T18:31:10Z")
> [@kostya](#):
>
> lexbor, streaming parse method
Oh, great, now I just need to figure out how to use that. 😁
> [@MarioAriasC](#):
>
> Google is getting worse by the day
Well, I lied a bit, it was DuckDuckGo, but I don’t think the difference is major.
---
**Author:**  [@naqvis](https://forum.crystal-lang.org/u/naqvis)\
**Post date:** [April 17, 2026, 12:59pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/9 "2026-04-17T12:59:57Z")
> [@Xen](#):
>
> Is there any HTML parser shard that works on a stream rather than chewing through the whole thing in one go?
[crystal-html5](https://github.com/naqvis/crystal-html5) also supports parse HTML stream at two different levels.
**Token-level** — no tree built, constant memory: Useful when one only need tokens without building the tree.
```crystal
HTML5.each_token(io) do |token|
puts token.data if token.type.start_tag?
end
```
Also available as `HTML5.token_iterator(io)` if you want an `Iterator`.
**SAX-style** — event callbacks during tree construction: Useful when one need a tree-aware events
```crystal
class MyHandler
include HTML5::StreamingHandler
def on_element_open(tag, attrs, namespace)
puts "<#{tag}>"
end
end
doc = HTML5.stream(io, MyHandler.new)
```
HIH
//Ali
---
**Author:**  [@Xen](https://forum.crystal-lang.org/u/Xen)\
**Post date:** [April 17, 2026, 5:46pm UTC](https://forum.crystal-lang.org/t/streaming-html-parser/8885/10 "2026-04-17T17:46:24Z")
> [@naqvis](#):
>
> [crystal-html5](https://github.com/naqvis/crystal-html5) also supports parse HTML stream at two different levels.
Oooh, I’m getting spoilt for options here.
That’s why the forum beats google…