After years of lack of a proper maintainer for one of why's jewels, it has been decided to finally close the book on hpricot. Most users have migrated to alternatives and there is simply no time or energy to continue with the current codebase.
If you feel that you have the time and wish to take it over, I suggest you instead think about making the hpricot-like API within nokogiri 100% compatible, that is a better use of your time.
But if you still feel like "No damnit, I wanna work on hpricot itself still!" then fork this repo and start work. Send @evanphx or @nicksieger a message if you feel like you want to take over the gem name with new releases under the hpricot name.
Thanks to _why for all the fun. We'll never forget it.
Hpricot is a fast, flexible HTML parser written in C. It's designed to be very accommodating (like Tanaka Akira's HTree) and to have a very helpful library (like some JavaScript libs -- JQuery, Prototype -- give you.) The XPath and CSS parser, in fact, is based on John Resig's JQuery.
Also, Hpricot can be handy for reading broken XML files, since many of the same techniques can be used. If a quote is missing, Hpricot tries to figure it out. If tags overlap, Hpricot works on sorting them out. You know, that sort of thing.
Please read this entire document before making assumptions about how this software works.
Let's clear up what Hpricot is.
- Hpricot is a standalone library. It requires no other libraries. Just Ruby!
- While priding itself on speed, Hpricot works hard to sort out bad HTML and pays a small penalty in order to get that right. So that's slightly more important to me than speed.
- If you can see it in Firefox, then Hpricot should parse it. That's how it should be! Let me know the minute it's otherwise.
- Primarily, Hpricot is used for reading HTML and tries to sort out troubled HTML by having some idea of what good HTML is. Some people still like to use Hpricot for XML reading, but remember to use the Hpricot::XML() method for that!
First, here are all the links you need to know:
- http://wiki.github.com/hpricot/hpricot is the Hpricot wiki and http://github.com/hpricot/hpricot/issues is the bug tracker. Go there for news and recipes and patches. It's the center of activity.
- http://github.com/hpricot/hpricot is the main Git repository for Hpricot. You can get the latest code there.
- See COPYING for the terms of this software. (Spoiler: it's absolutely free.)
If you have any trouble, don't hesitate to contact the author. As always, I'm not going to say "Use at your own risk" because I don't want this library to be risky. If you trip on something, I'll share the liability by repairing things as quickly as I can. Your responsibility is to report the inadequacies.
You may get the latest stable version from Rubyforge. Win32 binaries, Java binaries (for JRuby), and source gems are available.
$ gem install hpricot
We're going to run through a big pile of examples to get you jump-started. Many of these examples are also found at http://wiki.github.com/hpricot/hpricot/hpricot-basics, in case you want to add some of your own.
You have probably got the gem, right? To load Hpricot:
require 'rubygems'
require 'hpricot'
If you've installed the plain source distribution, go ahead and just:
require 'hpricot'
The Hpricot() method takes a string or any IO object and loads the contents into a document object.
doc = Hpricot("<p>A simple <b>test</b> string.</p>")
To load from a file, just get the stream open:
doc = open("index.html") { |f| Hpricot(f) }
To load from a web URL, use open-uri, which comes with Ruby:
require 'open-uri'
doc = open("http://qwantz.com/") { |f| Hpricot(f) }
Hpricot uses an internal buffer to parse the file, so the IO will stream properly and large documents won't be loaded into memory all at once. However, the parsed document object will be present in memory, in its entirety.
Use Doc.search:
doc.search("//p[@class='posted']")
#=> #<Hpricot:Elements[{p ...}, {p ...}]>
Doc.search can take an XPath or CSS expression. In the above example, all paragraph
elements are grabbed which have a class attribute of "posted".
A shortcut is to use the divisor:
(doc/"p.posted")
#=> #<Hpricot:Elements[{p ...}, {p ...}]>
If you're looking for a single element, the at method will return the first element matched by the expression. In this case, you'll get back the element itself rather than the Hpricot::Elements array.
doc.at("body")['onload']
The above code will find the body tag and give you back the onload attribute. This is the most common reason to use the element directly: when reading and writing HTML attributes.
Just as with browser scripting, the inner_html property can be used to get the inner contents of an element.
(doc/"#elementID").inner_html
#=> "..contents.."
If your expression matches more than one element, you'll get back the contents of ''all the matched elements''. So you may want to use first to be sure you get back only one.
(doc/"#elementID").first.inner_html
#=> "..contents.."
If you want the HTML for the whole element (not just the contents), use to_html:
(doc/"#elementID").to_html
#=> "<div id='elementID'>...</div>"
All searches return a set of Hpricot::Elements. Go ahead and loop through them like you would an array.
(doc/"p/a/img").each do |img|
puts img.attributes['class']
end
Searches can be continued from a collection of elements, in order to search deeper.
# find all paragraphs.
elements = doc.search("/html/body//p")
# continue the search by finding any images within those paragraphs.
(elements/"img")
#=> #<Hpricot::Elements[{img ...}, {img ...}]>
Searches can also be continued by searching within container elements.
# find all images within paragraphs.
doc.search("/html/body//p").each do |para|
puts "== Found a paragraph =="
pp para
imgs = para.search("img")
if imgs.any?
puts "== Found #{imgs.length} images inside =="
end
end
Of course, the most succinct ways to do the above are using CSS or XPath.
# the xpath version
(doc/"/html/body//p//img")
# the css version
(doc/"html > body > p img")
# ..or symbols work, too!
(doc/:html/:body/:p/:img)
You may certainly edit objects from within your search loops. Then, when you spit out the HTML, the altered elements will show.
(doc/"span.entryPermalink").each do |span|
span.attributes['class'] = 'newLinks'
end
puts doc
This changes all span.entryPermalink elements to span.newLinks. Keep in mind that there are often more convenient ways of doing this. Such as the set method:
(doc/"span.entryPermalink").set(:class => 'newLinks')
Every element can tell you its unique path (either XPath or CSS) to get to the element from the root tag.
The css_path method:
doc.at("div > div:nth(1)").css_path
#=> "div > div:nth(1)"
doc.at("#header").css_path
#=> "#header"
Or, the xpath method:
doc.at("div > div:nth(1)").xpath
#=> "/div/div:eq(1)"
doc.at("#header").xpath
#=> "//div[@id='header']"
When loading HTML documents, you have a few settings that can make Hpricot more or less intense about how it gets involved.
Really, there are so many ways to clean up HTML and your intentions may be to keep the HTML as-is. So Hpricot's default behavior is to keep things flexible. Making sure to open and close all the tags, but ignore any validation problems.
As of Hpricot 0.4, there's a new :fixup_tags option which will attempt to shift the document's tags to meet XHTML 1.0 Strict.
doc = open("index.html") { |f| Hpricot f, :fixup_tags => true }
This doesn't quite meet the XHTML 1.0 Strict standard, it just tries to follow the rules a bit better. Like: say Hpricot finds a paragraph in a link, it's going to move the paragraph below the link. Or up and out of other elements where paragraphs don't belong.
If an unknown element is found, it is ignored. Again, :fixup_tags.
So, let's go beyond just trying to fix the hierarchy. The :xhtml_strict option really tries to force the document to be an XHTML 1.0 Strict document. Even at the cost of removing elements that get in the way.
doc = open("index.html") { |f| Hpricot f, :xhtml_strict => true }
What measures does :xhtml_strict take?
- Shift elements into their proper containers just like :fixup_tags.
- Remove unknown elements.
- Remove unknown attributes.
- Remove illegal content.
- Alter the doctype to XHTML 1.0 Strict.
The last option is the :xml option, which makes some slight variations on the standard mode. The main difference is that :xml mode won't try to output tags which are friendlier for browsers. For example, if an opening and closing br tag is found, XML mode won't try to turn that into an empty element.
XML mode also doesn't downcase the tags and attributes for you. So pay attention to case, friends.
The primary way to use Hpricot's XML mode is to call the Hpricot.XML method:
doc = open("http://redhanded.hobix.com/index.xml") do |f|
Hpricot.XML(f)
end
Also, :fixup_tags is canceled out by the :xml option. This is because :fixup_tags makes assumptions based how HTML is structured. Specifically, how tags are defined in the XHTML 1.0 DTD.