API should be implemented as web page (entry point) returning results based on provided GET params. API can be secured with HTTP basic authentication.
There are 4 actions:
- check
- seed
- items
- item

Action is passed as "action" GET param to the entrypoint.


----------------------------------------------------------------------------------------------------
[entrypoint]?action=check
----------------------------------------------------------------------------------------------------
should return HTTP status code 200 providing information that entrypoint is working properly.

----------------------------------------------------------------------------------------------------
[entrypoint]?action=seed&startDate=YYYY-MM-DDTHH:mm:ssZ&endDate=YYYY-MM-DDTHH:mm:ssZ
----------------------------------------------------------------------------------------------------
parameters:
startDate - the start of time frame which should be applied to returned seeds. If this is a first run - this parameter will not be provided meaning that all documents should be returned.
endDate - the end of time frame. Always provided.

startDate and endDate parameters are encoded as YYYY-MM-DD'T'HH:mm:ss'Z'. Result should be valid XML of form:
<seeds>
   <seed id="document_id_1" />
   <seed id="document_id_2" />
   ...
</seeds>

attributes "id" are required.

----------------------------------------------------------------------------------------------------
[entrypoint]?action=items&id[]=document_id_1&id=document_id_2
----------------------------------------------------------------------------------------------------
parameters:
id[] - array of document IDs that should be returned

Result should be valid XML of form:
<items>
   <item id="document_id_1">
      <url>[http://document_uri]</url>
      <version>[document_version]</version>
	  <created>2013-11-11T21:00:00Z</created>
	  <updated>2013-11-11T21:00:00Z</updated>
	  <filename>filename.ext</filename>
	  <mimetype>mime/type</mimetype>
	  <metadata>
	     <meta name="meta_name_1">meta_value_1</meta>
	     <meta name="meta_name_2">meta_value_2</meta>
		 ...
	  </metadata>
	  <auth>
		 <token>auth_token_1</token>
		 <token>auth_token_2</token>
		 ...
	  </auth>
	  <content>Document content</content>
   </item>
   ...
</items>

id, url, version are required, the rest is optional. If "auth" tag is provided - document will be treated as non-public with defined access tokens, if it is ommited - document will be public.
if content tag is ommited - connector will ask for document content as "action=item" separate API call.

----------------------------------------------------------------------------------------------------
[entrypoint]?action=item&id=document_id
----------------------------------------------------------------------------------------------------
parameters:
id - requested document ID

Result should be the document content. It does not have to be XML - you may return binary data (PDF, DOC, etc) which represent the document.