Hello

This is a note how I upload Json documents to a couchDB using

Webclient   http://www.squeaksource.com/WebClient
and a modified version of http://www.squeaksource.com/JSON   JSON-jrd.28

The test case is

  |d r|
  d := Dictionary new. d at: 'title' put: 'The title of this card'.
  d at: 'body' put: (8820 asCharacter asString, 'aäbc', Character cr).
  r := WriteStream on: String new.
  (JsonObject newFrom: d) jsonWriteOn: r.
  "r contents"

  WebClient httpPut: 'http://192.168.0.121:5984/test/myDoc6' content:
r contents type: ''.

WebClient gives back as contents
  '{"ok":true,"id":"myDoc6","rev":"1-cef58e13534fc0fcf7f38262bc086d12"}
'


To make this work I had to patch the method

Json escapeForCharacter: aCharacter


This was necessary because for example

    Json escapeForCharacter: 228 asCharacter

gave back
    '\uE4'

instead of
     '\u00E4'


I changed the method Json escapeForCharacter to

escapeForCharacter: c
	
	| index nnnn |
	^ (index := c asciiValue + 1) <= escapeArray size
		ifTrue: [ ^ escapeArray at: index ]
		ifFalse: [nnnn := ((c asciiValue bitAnd: 16rFFFF) printStringBase: 16) .
						[nnnn size < 4] whileTrue: [nnnn := '0', nnnn].
			 	^ '\u', nnnn]


My question:

Is there a nicer way of doing

nnnn := ((c asciiValue bitAnd: 16rFFFF) printStringBase: 16) .
						[nnnn size < 4] whileTrue: [nnnn := '0', nnnn].
			 	^ '\u', nnnn

Basically I have to patch zeros in front of a ByteString which is too short.


Regards
Hannes



P.S. This note does not address the issue that it would be nice to NOT
escape characters which have a code >127 at all but rather keep them
as Unicode characters as the Json spec allows for this.
http://www.json.org/

Any Unicode character except " and \ is allowed.

This is about fixing an error which comes up when posting Json objects
to CouchDB.