Friday, December 14, 2012

About the anatomy of ArchiveLink


I just played arround with ArchiveLink, and so I decided to read the specification to inform myself how it internally works. What you can use in order to archive your SAP documents (invoices, ...) is an ArchiveLink speaking HTTP Content Server. So here a short summary of what I understood from the ArchiveLink specification so far.


We want to store documents inside a content repository. I guess it is not required to explain this term further. ArchiveLink knows the term 'component', whereby a component represents a content unit on a administrative level. Several component types are used (for instance 'data'),  Components are summarized via a document header. So here a scenario:

  • One Content Repository contains multiple document headers
  • One document header references multiple components
  • One component contains one content unit
A document header (id, status, date, ...) and a component (content type, status, ...) has some adminstrative meta data attached.

So an archive needs to reflect that terms somehow. The most intuitive way seems to reflect it by using folders. So the content repository contains a folder of the type header which contains component folders those are containing documents to keep the content. Important is that ArchiveLink does by design not transfer any SAP business object related meta data. Only administrative meta data is transfered to the archive.

The protocol

HTTP is used to exchange date on a lower level. So we are speaking about a kind of RESTFul service access. The format is 'http://${server}:${port}/${service name}/${command}?${command parameters}. If we take security into account (more about it later) and take a spot on the get command by retrieving the data component then the URL has the following format:

http://${host}:${port}/${service name}?get&contRep=${repo id}&docId=${doc id}&compId=data&accessMode=r&authId=${user id}&expiration=${time}?secKey=${base64 encoded security key}.

The following commands are available:

  • get: get a content unit
  • info: get info about a document
  • docGet: get the whole content
  • create: Create a new document
  • update: Modify an existing document
  • append: Append data to a content unit
  • search: Search by using a pattern inside a content unit (full text search)
  • attrSearch: Search for a specific attribute value
  • mCreate: Create new documents
  • serverInfo: Retrieve information about the content server
  • putCert: Transfer the client certificate
The resonse is a little bit old scool (the specification document is from 2001) and so does not return JSON, but Multipart Form Data.

Here an example response from the publically available specification document:

HTTP/1.1 200 (OK)
Server: Microsoft-IIS/4.0
Date: Wed, 04 Nov 1998 07:41:03 GMT
Content-Type: multipart/form-data; boundary=A495ukjfasdfddrg4hztzu...
...some more header informations...
Content-Length: 32413
X-dateC: 1998-10-07
X-timeC: 07:55:57
X-dateM: 1998-10-07
X-timeM: 07:55:57
X-contRep: K1
X-numComps: 2
X-docId: ID
X-docStatus: online
X-pVersion: 0045
Content-Type: application/x-alf; charset=
Content-Length: 2591
X-compId: descr
X-Content-Length: 2591
X-compDateC: 1998-10-07
X-compTimeC: 07:55:57
X-compDateM: 1998-10-07
April 2001 265
X-compTimeM: 07:55:57
X-compStatus: online
X-pVersion: 0045
...component data ...
Content-Type: application/x-alf; charset=
Content-Length: 29313
X-compId: data
X-Content-Length: 29213
X-compDateC: 1998-10-07
X-compTimeC: 07:55:57
X-compDateM: 1998-10-07
X-compTimeM: 07:55:57
X-compStatus: online
X-compStatus: online
X-pVersion: 0045
...component data ...
Beim docGet-Kommando auf ein leeres Dokument steht im Response-Body beispielsweise:

This is the furst article of several ones those are focusing on ArchiveLink. What we should have learned is the following: An ArchiveLink archive can be realized by using a HTTP Content Server. The Content Server then could speak with a simple File System or even better an DMS. The content server provides several commands and functions to store and access content items. There are several kinds of content items, wrapped by components. A specific URL pattern is used on side of the HTTP Content Server. The response of the HTTP Content Server has the Multipart Form Data format.

No comments:

Post a Comment