Document search and add engine web application

Question

I want to develop a asp.net web application which should do the following task a) user should be able to add content to the document. Content to be added can include text as well as image, screen shots etc. b) user should be able to search based on some keywords. when searching with the keyword appropriate content along with images(if any) should be shown to user.

I am not sure what should be the proper approach for this. One way i think is to store text content in some xml file and later search for keywords by going though each node of xml and displaying. but i am not sure how to attach image content with xml. Also this method doesn't seem to be nice and efficient if with time document size increases a lot.

Anyone please suggest some proper way to do above requirement. Any hint would be appreciated.

Answer 1

Split it to two tasks. Editation and search.

Full text search is solved problem. Simply use Sphinx Search and you are done. Sphinx is simple to use and can do everything you will need. It has MySQL interface (your app connects to sphinx the same way as to second MySQL database).

Editation is a bit more complicated. If I understand correctly, you want multiple users to edit single document concurrently.

I recommend using websockets to notify other clients about changes in document. Long-polling and Server Sent Events have ugly side effects, like stopping browser from making another requests to server. To implement client side in Javascript, I would use React, Angular or similar framework to make updates as easy as possible.

Server side requires modification-friendly representation of a document, so if one user changes one part, and another user another part, your app should be able to merge changes. Changing completely different parts is easy, but it may be tricky to change the same paragraph or document node. Exact representation of each change depends on format of your document.

I do not see much benefits of using XML rather than any other format. It may be practical for document representation, but it will not help with merging of colliding modifications. I would start with plain array of strings, each representing a single paragraph. Extending it to full XML document is the easy part, once two users can edit the same paragraph.

To store images in XML, simply store files using their hash as a file name and then use such name to link the file in XML. Git does the same thing and it works nicely. You may want to count references to identify unused files.