Theory/Magic behind creating API

Im a newbie PHP programmer and I have a few questions about creating a REST based API service.

Basically, its going to be a github open source project which is going to scrape various data off the web and offer it as an API in XML. Now here are my questions in terms of how should I or how can I do this.

1) Since there isnt a robust/same pattern for getting various data through scraping, what is the best way to actually output the xml?

I mean the PHP file would have various lines of extracting data from various points in the code and the file would be a lot of lines. Is it a good idea to type the code to output the result in there?

2) Is there a way to organize the scraping code in a sort of class?
I cant think of a way that would work besides linear approach where not ev开发者_Python百科en a function is created and you just apply functions (in general).

3) If theres a way to do that ^^ , how can it you output it?

Is there any other approach besides using another file and getting the contents from the main file and displaying the code through the secondary file.

4) If I were to offer the API in XML and JSON, is there a way to port from one result to another or will I have to manually create the fields in json or xml and place the content in there?

I might have more questions that might arise after these have been answered but I hope I get everything cleared up. Also, this is assuming that the results are not fetched from a DB so the data has to be scraped/tabulated on every request. (even though caching will be implemented later)

Thanks

This question is probably more appropriate on https://codereview.stackexchange.com/

Not to be rude, but a newbie programmer developing an API is like a first-year med student offering to do open-heart transplants for free. I understand that you believe that you can program, but if you intend to release publicly accessible code, you probably need more experience. Otherwise guys like me will muck through it and file bug reports ridiculing your code.

That said, if you want theory of good API design you should probably check out Head First Object Oriented Analysis and Design. You'll want to focus on these key concepts

1. Program to an Interface, not an Implementation
2. Encapsulate what varies

...and follow other good design principles.

...honestly, there's a lot to cover to good interface and good systems design. You can use this as a learning exercise, but let people know they shouldn't rely on your code. Though they should know that screen scraping is far more brittle and instable than web service API requests anyway, but many don't.

That said, to provide some initial guidance:

1. Yes, use OOP. Encapsulate the part that actually does the scraping (presumably using cURL) in a class. This will allow you to switch scraping engines transparently to the end user. Encapsulate your outputs in classes, which will allow for easy extension (i.e. if JSON output is in a Single Responsibility class and XML Output is in another, I can add RSS Output easily by making a new class without breaking your old code)
2. Think about the contracts your code must live up to. That will drive the interface. If you are scraping a particular type of data (say, sports scores for a given day), those should drive the types of operations available (i.e. function getSportsScoresForDate(date toGet))
3. Start with your most abstract/general operations at a top level interface, then use other interfaces that extend that interface. This allows users to have interfaces at different levels of granularity (i.e. class SensorDataInterface has a method getData(). HeartRateMonitorInterface extends SensorDataInterface and adds getDataForTimeInterval())