select_one and select about soupsieve

[facelessuser]

I don't know why you use limit to control the number of the results. I think people would need one or all of the results.

Not necessarily. As internally it is using a generator, this gives you a way to terminate the search once you get all you need to work with, saving you from having to parse the entire tree. It depends on the kind of data you are working with and what your task is.

Regardless of whether it makes sense, this models BeautifulSoups API in this respect as it always had a limit as well.

With all of that said, for completeness with the original Beautiful Soup API, I can see the desire to have the familiar select_one() as well with its convenience of not having to index into the the first in the list. This seems a reasonable request.

Keep in mind, Beautiful Soup will be using Soup Sieve as its select library moving forward, so you'll get the select_one builtin into the tag objects for free there automatically. I always wrote Soup Sieve to be used internally in Beautiful Soup, so at the time select seemed like all I'd need. It wasn't until afterwards that I realized there were cases I would still like using it externally as well. Especially being able to precompile patterns, and build a document parser using match to return only the elements you care about.

If I could go back, I'd probably call the two functions select and select_all. Oh well, looks like it is select and select_one moving forward 🙂 .

[facelessuser]

And thanks for the css seletors you invented, like !=, that is what I need now.

Well, I didn't ivent it, it was something other libraries have been using for a while. I know JQuery and the lxml CSSSelect library implement it. It seemed too logical and convenient not to add here as well.

I like the precompile. It looks like the re module (re.compile).

Yeah, that was kind of my model when I wrote it. Seemed like if you had a script with known things you were looking for, instead of calling select over and over, and reparsing the selector string and building the search object, it would be best to:

1. Allow you to build that thing once and reuse it.

2. Cache them in case you didn't think to precompile it to begin with.

Both of these the re module employs.

[facelessuser]

Had a little time this morning, so I implemented the feature via #46. It will be in the 1.5.0 release. Not sure when that release is coming yet. Maybe not too far out. I imagine major releases will slow down moving forward as I feel the project has a solid feature base now, but I think I'll at least get this one out within the week.

[yjqiang]

find_all(name, attrs, recursive, string, limit, **kwargs) Just found it, and this makes sense. I like the precompile. It looks like the re module (re.compile).

[yjqiang]

And thanks for the css seletors you invented, like !=, that is what I need now.

[yjqiang]

With all of that said, for completeness with the original Beautiful Soup API, I can see the desire to have the familiar select_one() as well with its convenience of not having to index into the the first in the list. This seems a reasonable request.

So now I just need to sit down and wait for the new select_one api.

[facelessuser]

FYI, I've released 1.5.0 as I found a bug in nth selectors while I was adding the :dir() selector. I wanted to get the fix out, so I went ahead and cut the 1.5.0 release.