A HTML Extraction API using RegEx or HTML Parser - php

Question

I am aware that it is public opinion to not use RegEx for parsing HTML; however I do not see how it would be harmful to use RegEx (alike functions have been added in previous Scripting Languages using RegEx such as _StringBetween( ) in AutoIt3) for what I want to achieve.
I am also aware that _StringBetween( ) was not specifically written for HTML but I have been using it without any problem on HTML content for the past 8 years along with other folks.
For my HTML Extraction API I would like to present the following piece of HTML
<div class="video" id="video-91519"><!-- The value of the identifier is dynamic-->
<a href="about:blank"><img src="silly.jpg"><!-- So is the href and src in a, img -->
</div>
The reason for the API I am trying to write is to make extraction of the video_url and thumbnail extremely easy and therefore a HTML parser seems out of reach. I would like to be able to extract it using something amongst the lines of
<div class="video" id="video-{{unknown}}">{{unknown}}<a href="{{video_url}}"><img src="{{thumbnail}}">{{unknown}}</div>
Of course, in the previous piece of HTML you could do it much easier such as
<a href="{{video_url}}"><img src="{{thumbnail}}">
but I was trying to present a perfect example to avoid confusion.
How does RegEx come into play? Well, I was going to replace {{video_url}}, {{thumbnail}} and {{unknown}} with (.*?), (.*?) and .* using /s and of course making sure that there are no multiple occurences of {{video_url}} and {{thumbnail}} in the provided input (not the HTML).
So, is there any reason for me not to use RegEx or still go for a HTML parser incl. proof of concepts of either acceptable RegEx and/ or using a HTML parser? I cannot personally see how to make this happen using a HTML parser

Answer 1

I think the way you have framed the problem pre-supposes the solution: if you want to be able to specify a pattern to match, then you have to use a pattern-matching language, such as regex. But if you frame the question as allowing searches for content in the document, then other options might be available, such as a path-based input that compiled to XPath expressions, or CSS selectors as used very successfully by the likes of jQuery.
What you are building here is not really an HTML extraction API as such, but a regex processing API - you are inventing a simplified pattern-matching language which can be compiled to a regex, and that regex applied to any string.
That in itself is not a bad thing, but if the users of that pattern-matching API try to use it to parse a more complex or unpredictable document, they will have all the same problems that everyone has when they try to match HTML using regexes, plus additional limitations imposed by your pre-processor. Those limitations are an inevitable consequence of simplifying the language: you are trading some of the power of the regex engine in order to make your patterns more "user friendly".
To return to the idea of reframing the question, here is an example of a simplified matching API that could compile to CSS expressions (e.g. used with SimpleHTMLDOM):
Find: div (class:video)
Must-Contain: a, img
Match: id Against video-{{video_id}}
Child: a
Extract: href Into video_url
Child: img
Extract: src Into thumbnail
Note that this language is a lot more abstracted away from the HTML; this has advantages and disadvantages. On the one hand, the simple matching pattern in your question is easy to create based on a single example. On the other hand, it is much more fragile to variations in the HTML, either due to changes in the site, or in-page variations such as adding an extra CSS class of "featured-video" to a small number of videos. The selector-based example requires the user to learn more specifics of the API, but if they don't know HTML and pattern-matching to begin with, a verbose syntax may be more helpful to them than one involving lots of cryptic punctuation.