Implementing Queries with HTTP

1
Implementing Queries with HTTP

• A. Warnock
• A/WWW Enterprises
• warnock_at_clark.net

2
Why Use Open Protocols?

• Anyone can write servers
• Anyone can write clients
• Leverages off of existing development
• Uses formalized standards
• In short
• Interoperability

3
What Protocols Are Available?

• ftp
• telnet
• gopher (really telnet)
• WAIS (Z39.50-1988)
• http CGI
• Z39.50 v2, v3

4
http Capabilities

• http does only a few things
• get files (w/ MIME type)
• put parameters (from forms)
• execute scripts (CGI)
• http defines only syntax, not content
• Queries depend on content

5
CGI Scripting

• CGI runs scripts on server
• accepts parameters from client
• parameter input via command line or stdin
• pipes output to stdout
• Sample CGI Input Stream
• cgi-bin/ltscriptnamegt?ltparameter-listgt
• The parameter list is unspecified - it can be
  anything the form implementor desires

6
Search Requirements

• Search Bookkeeping Query
• Bookkeeping consists of
• User ID inf.
• What database to query?
• What elements to return?
• What format is requested?
• How many records to return?

7
What Goes Into A Query?

• A query is constructed by specifying
• Search Term(s)
• Field Names
• Words, phrases, etc.
• Word stems, wildcards
• Stems, substrings, area relations
• None of these is prescribed in http!

8
http Query Summary

• http can transmit arbitrary queries, but contains
  nothing to help in interpreting them
• We must make up and agree upon a query syntax
  from scratch
• This is called a Search Profile

9
Example Scenario

• User completes an HTML form and submits it
• Server receives a POST request - a bunch of
  information
• Where the message came from
• How long the message is
• The message text

10
Sample Query

• Step 1
• http//ltservergt/cgi-bin/query?dbInventory
• userwarnocknrecsallelementB
• formatHTMLplatformLandsat7TM
• rect42N112.5Wcenter512kmNS
• 512kmEWoperatorspatialIntersect
• resolutionMaximum
• Must also include all state information

11
Critique

• Whats so hard about this?
• Is the query case sensitive?
• What if the user is French? Russian? Chinese?
• Do I specify platform or Platform or Source
  or source?
• Is the order of parameters important?
• How do I specify other resolutions? Can you
  generalize this to the query in Step 9? Whats
  your confidence that it would be correct?

12
Summary

• http can be used to transmit queries and
  responses, but theres nothing in the protocol
  itself to make the process unambiguous
• We have to do all the work outside the protocol -
  that is, we have to write a profile!