based on https://github.com/talis/rdfquads.js
_ = (window? and window._) or require('underscore')
toNative = require('./xsd2native').toNative
uriRegex = /<([^>]*)>/
literalRegex = /^([-]?\d.*)|(?:\"(.*)\")(?:\@([a-z]+)|\^\^(.*)|(?:))$/
class RdfUri
constructor: (url) ->
match = url.match(uriRegex)
if match
@raw = match[1]
else
@raw = url
return
toString: ->
@raw
class RdfObject
constructor: (val) ->
if typeof val isnt undefined and val.match
uriMatch = val.match(uriRegex)console.log “WE ARE HERE”,val
if uriMatchconsole.log “uriMatch”,uriMatch
@raw = uriMatch[0]
@value = uriMatch[1]
@type = "uri"
@where = "A"
else
literalMatch = val.match(literalRegex)console.log “literalMatch:”, literalMatch, “>#{val}<”
if literalMatch
@raw = literalMatch[0]
@type = 'literal'
@where = "B"
if literalMatch[1] and literalMatch[1].indexOf('-') < 1 # starts with digit
@isNum = true
@value = literalMatch[1]
if literalMatch[2] # wrapped in quotes
@value = literalMatch[2]
if literalMatch[3] # @lang
@where += "_lang"
@literal_lang = literalMatch[3]
if literalMatch[4] # ^^literalType
@where += "_type"
@literal_type = literalMatch[4]
elseconsole.log “literalRegex failed for”, val
@raw = val
@type = 'FAILED'
@value = val
@where = "C_literalRegex failed"
else
@raw = val # maybe an integer? TODO tighten this up big time!
@type = "literal"
getNativeValue: ->
if not @ntval?
@ntval = toNative(@value,@literal_type,@isNum,@raw)
if not @ntval?
return @value
return @ntval
toString: ->
@raw
repr: ->
console.log "NO, THIS DOES NOT GET CALLED!"
if @type is 'literal'
if _.isString(@raw)
return "\"#{@raw}\""
else if _.isNumber(@raw)
return @raw
else
throw new Error "RdfObject(#{@raw}) is a literal which is neither a String nor a Number"
else if @type is 'uri'
return "<#{@raw}>"
else
throw new Error "RdfObject(#{@raw}).type is neither literal nor uri"
isUri: ->
@type is "uri"
isLiteral: ->
@type is "literal"
getLiteralType: ->
@literal_type
uri_or_literal: ->
if @isLiteral()
return [@raw] # TODO remove this and the complementary unwrapping of the array
else
return @raw # "<#{@raw}>"
getRawOrValue: ->
if @isLiteral()
return @raw
else
return @value
class Quad # TODO change to SPOGI
constructor: (subject, pred, obj, graph, id) ->
@s = new RdfUri(subject)
@p = new RdfUri(pred)
@o = new RdfObject(obj)
@g = new RdfUri(graph)
@i = id
return
toString: ->
id = @i? and " # #{@i}" or ""
"<" + @s + "> <" + @p + "> " + @o.repr() + " <" + @g + "> .#{id}"
repr: ->
@toString()
asLine: ->
@toString() + "\n"
spogiRegex = ///
\s* # zero or more blank spaces
the subject
(<[^>]*> # an uri
|_:[A-Za-z][A-Za-z0-9]*) # 'blank' identifier
\s+ # at least one blank spacethe predicate
(<[^>]*> # an uri TODO require content
|_:[A-Za-z][A-Za-z0-9]*) # 'blank' identifier
\s+ # at least one blank spacethe object
(
(?:[-]?\d[^\s]*) # a numbery-type thingy or date
|(?:\"(?:.*)\") # a stringy-type thingy
(?:\@(?:[a-z_]+) # with language (optional)
|\^\^(?:.*) # or type (optional)
|(?:) # or neither
)
|(?:<[^>]*>|) # or uri
)
\s+ # at least one blank space
(<[^>]*>|) # the (optional) graph
\s*\. # the quad-concluding period (.)
the optional nooron-style id for the spogi
\s*\#*\s*(
.* # TODO replace this wildcard with the following, once debugged [A-Za-z0-9]*_ # user_symbol (actually should NOT have leading digits)
[A-Za-z0-9]*_ # session_no
[A-Za-z0-9]*_ # sec since unix epoch
[A-Za-z0-9]* # nsec part of time
)$ # TODO permit subsequent comments after a space
///
https://regex101.com/r/A48eDR/1/ this captures everything into a single group with (?:
ttlObjectRegex = ///
(
(\" # the start of a quoted region
.* # anything (but what about quote escapes or CR escapes?)
\" # the end of a quoted region
( # optionally, followed by
(\@[a-z][a-z]) # a language spec eg @en or @de
|
(\^\^ # a datatype spec, a pair of carets followed by a curie or url
([\w]+\:[\w]* # a curie (with a possible blank local part)
|<[^>]+> # an uri
)
)
| # or nothing
)
)
|
([\w]+\:[\w]* # a curie (with a possible blank local part)
|<[^>]+> # an uri
)
)
///n5Regex = ///
^ # the beginning of the line
\s* # zero or more blank spaces
the subject
([\w]+\:[\w\-]* # a curie (with a possible blank local part)
|<[^>]+> # an uri
)
\s+ # at least one blank spacethe predicate
([\w]+\:[\w\-]* # a curie (with a possible blank local part)
|<[^>]+> # an uri
)
\s+ # at least one blank spacethe object (consisting of either a quoted and either typed or untyped string or an URI)
(
(\" # the start of a quoted region GROUP 3
.* # anything (but what about quote escapes or CR escapes?)
\" # the end of a quoted region
( # optionally, followed by
(\@[a-z][a-z]) # a language spec eg @en or @de GROUP 6
|
(\^\^ # a datatype spec, a pair of carets followed by a curie or url GROUP 7
([\w]+\:[\w\-]* # a curie (with a possible blank local part) GROUP 8
|<[^>]+> # an uri
)
)
| # or nothing
)
)
|
(?:[\w]+\:[\w\-]* # a curie (with a possible blank local part) (?: match everything enclosed
|<[^>]+> # an uri
)
)
\s+ # at least one blank space
the graph (required)
([\w]+\:[\w\-]* # a curie (with a possible blank local part) GROUP 9
|<[^>]+> # an uri
)
\s+ # at least one blank space
\. # the trailing period
\s+ # at least one blank space
\# # the TTL/TRIG comment character
\s+ # at least one blank space
([\w]+) # a mandatory nooron-style id like SFM_crm114_bfg995_pdq357 GROUP 10TODO replace the above wildcard with the following, once debugged [A-Za-z0-9]_ # user_symbol (actually should NOT have leading digits) [A-Za-z0-9]_ # session_no [A-Za-z0-9]_ # sec since unix epoch [A-Za-z0-9] # nsec part of time
.* # permit subsequent comments after a space
$ # the end of the line
///
isComment = /^\s*\/\//
parseQuadLine = (line) ->
if not line? or line is "" or line.match(isComment)
null
else
match = line.match(spogiRegex)
if match
retval =
s: new RdfUri(match[1].trim()).raw
p: new RdfUri(match[2].trim()).raw
o: new RdfObject(match[3].trim()).uri_or_literal()
g: new RdfUri(match[4].trim()).raw
i: match[5].trim()
return retval
else
console.log "spogiRegex FAILED:", line
parseQuintLineToPenta = (line) ->
if not line? or line is "" or line.match(isComment)
null
else
match = line.match(spogiRegex)
if match
retval =
s: new RdfUri(match[1].trim())
p: new RdfUri(match[2].trim())
o: new RdfObject(match[3].trim())
g: new RdfUri(match[4].trim())
i: match[5].trim()
return retval
else
console.log "spogiRegex FAILED:", line
parseQuadLineToQuint = (line) ->
if not line? or line is "" or line.match(isComment)
null
else
match = line.match(spogiRegex)
if match
retval = [
new RdfUri(match[1].trim()).raw
new RdfUri(match[2].trim()).raw
new RdfObject(match[3].trim()).getRawOrValue()
new RdfUri(match[4].trim()).raw
match[5].trim()]
return retval
else
console.log "spogiRegex FAILED:", line
parseN5LineToQuint = (line) ->
if not line? or line is "" or line.match(isComment)
null
else
match = line.match(n5Regex)
if match
TODO make full use of the hints in these other groups to perfect persistence group 3 is the object URI or CURIE (when present) group 4 is the object string value (when present) group 5 is the (when present) group 6 is the language spec (when present) eg ‘@en’ group 7 is ^^ (when present) eg ‘^^’ group 8 is the datatype spec (when present) eg ‘xsd:int’ | http://oink.ly
noodbId = match[10].trim()
retval = [
new RdfUri(match[1].trim()).raw
new RdfUri(match[2].trim()).raw
new RdfObject(match[3].trim()).getRawOrValue()
new RdfUri(match[9].trim()).raw
noodbId]
return retval
else
console.log "n5Regex FAILED:", line
(exports ? this).parseQuadLine = parseQuadLine # deprecated
(exports ? this).parseQuintLineToPenta = parseQuintLineToPenta # deprecated
(exports ? this).parseQuadLineToQuint = parseQuadLineToQuint
(exports ? this).parseN5LineToQuint = parseN5LineToQuint
(exports ? this).ttlObjectRegex = ttlObjectRegex
(exports ? this).Quad = Quad
(exports ? this).RdfObject = RdfObject