Table of Contents generated with DocToc

• cnd
  • CND Interval Tree
  • CND Shim
  • CND TSort
    • TSort API
    • Some TDOP Links
  • XJSON
  • CND.TEXT.to_width
  • ToDo

cnd

a grab-bag NodeJS package mainly for functionalities that used to live in coffeenode-trm, coffeenode-bitsnpieces, and coffeenode-types

CND Interval Tree

• https://www.youtube.com/watch?v=q0QOYtSsTg4

• https://www.youtube.com/watch?v=PHlTuCVxJz4

• https://github.com/mikolalysenko/functional-red-black-tree

• Simplified API (compared to functional-red-black-tree)

• possible to dynamically add nodes

• possible to access underlying red/black tree as tree[ '%self' ], root node as tree[ '%self' ][ 'root' ]

• possible to access

1CND       = require 'cnd'
2ITREE     = CND.INTERVALTREE
3tree      = ITREE.new_tree()
4intervals = [
5  [ 3, 7, 'A', ]
6  [ 5, 7, 'B', ]
7  [ 8, 12, 'C', ]
8  [ 2, 14, 'D', ]
9  [ 4, 4, 'E', ]
10  ]
11ITREE.add_interval tree, interval for interval in intervals
12for n in [ 0 .. 15 ]
13  console.log n
14  for node in ITREE.find tree, n
15    console.log node[ 'key' ], node[ 'value' ]

CND Shim

CND now includes https://github.com/es-shims/es6-shim and https://github.com/es-shims/es7-shim; in order to use these, say

1CND = require 'cnd'
2CND.shim()

which will then polyfill you current runtime to the point where you approximately reach compatibility with NodeJS v0.12 run with the --harmony switch. Note that not everything can be polyfilled; in particular WeakMaps and yield cannot be provided this way.

CND TSort

CND TSort implements a simple tool to perform the often-needed requirement of doing a topological sorting over an DAG (directed acyclic graph). It is an adaption of the npm: tsort module.

Here is the basic usage: you instantiate a graph by saying e.g.

1CND       = require 'cnd'
2TS        = CND.TSORT
3settings  =
4  strict:   yes
5  prefixes: [ 'f|', 'g|', ]
6graph     = TS.new_graph settings

The settings object itself and its members are optional. The strict setting (true by default) will check for the well-formedness on the graph each time a relationship is entered; this means you get early errors at the cost of decreased performance. We come back to the prefixes later; the default is not to use prefixes.

Topological sorting is all about finding a total, linear ordering of actions, values, objects or whatever from a series of statements about the relative ordering of pairs of such objects. This is used in many fields, for example in package management, project management, programming language grammars and so on.

So let's say you have a todo list:

1'buy books'
2'buy food'
3'buy food'
4'cook'
5'do some reading'
6'eat'
7'fetch money'
8'go home'
9'go to bank'
10'go to exam'
11'go to market'

Now you don't have money, foods or books at home right now, but you want to eat, do some reading, and go to the exam after that, so clearly there are certain orderings of events that would make more sense than other orderings. It's perhaps not immediately clear how you can organize all these jobs when you look at the entire list, but given any two jobs, it's often easy to see which one should preced the other one.

Once we have instantiated a graph g, we can add dual relationships piecemeal, the convention being that we imagine arrows or links pointing from preconditions 'down' to consequences, which is why the corresponding method is named link_down. Calling TS.link_down g, 'buy food', 'cook' means: 'Add a link to graph g to indicate that before I can 'cook', I have to 'buy food' first', and so on. It is customary to symbolically write 'buy food' > 'cook' to indicate the dependency.

In case the graph has been instantiated with a strict: yes setting, CND TSort will validate the graph for each single new relationship; as a side effect, a list of entries is produced and returned that reflects one of the possible linearizations of the graph which satisfies all requirements so far. For the purpose of demonstration, we can take advantage of that list and print it out; we can see that the ordering of jobs will sometimes take a dramatic turn when new requirements are added. In case you've started the graph with strict: no, you'll have to call TS.sort g yourself to perform a validation and obtain a serialization.—Let's try that for some obvious dependencies on our list:

1console.log ( TS.link_down g, 'buy food',          'cook'                ).join ' > '
2console.log ( TS.link_down g, 'fetch money',       'buy food'            ).join ' > '
3console.log ( TS.link_down g, 'do some reading',   'go to exam'          ).join ' > '
4console.log ( TS.link_down g, 'cook',              'eat'                 ).join ' > '
5console.log ( TS.link_down g, 'go to bank',        'fetch money'         ).join ' > '
6console.log ( TS.link_down g, 'fetch money',       'buy books'           ).join ' > '
7console.log ( TS.link_down g, 'buy books',         'do some reading'     ).join ' > '
8console.log ( TS.link_down g, 'go to market',      'buy food'            ).join ' > '

The output from the above will be (re-arranged for readability):

1                                          buy food > cook
2             fetch money >                buy food > cook
3             fetch money >                buy food > cook >             do some reading > go to exam
4             fetch money >                buy food > cook >             do some reading > go to exam > eat
5go to bank > fetch money >                buy food > cook >             do some reading > go to exam > eat
6go to bank > fetch money >                buy food > cook >             do some reading > go to exam > eat > buy books
7go to bank > fetch money >                buy food > cook > buy books > do some reading > go to exam > eat
8go to bank > fetch money > go to market > buy food > cook > buy books > do some reading > go to exam > eat

Observe how the requirement 'fetch money' > 'buy books' made 'buy books' appear at the very end of the list, and how only the additonal requirement 'buy books' > 'do some reading' managed to put the horse before the cart, as it were. It still looks as though we're not quite done here yet, as we have to leave house after cooking and go to the exam with an empty stomach according to the current linearization, so some dependencies are still missing:

1console.log ( TS.link_down g, 'buy food',          'go home'             ).join ' > '
2console.log ( TS.link_down g, 'buy books',         'go home'             ).join ' > '
3console.log ( TS.link_down g, 'go home',           'cook'                ).join ' > '
4console.log ( TS.link_down g, 'eat',               'go to exam'          ).join ' > '

This makes the order of events more reasonable with each step:

1go to bank > fetch money > go to market > buy food > cook > buy books > do some reading > go to exam > eat > go home
2go to bank > fetch money > go to market > buy food > cook > buy books > do some reading > go to exam > eat > go home
3go to bank > fetch money > go to market > buy food > buy books > go home > cook > do some reading > go to exam > eat
4go to bank > fetch money > go to market > buy food > buy books > go home > cook > do some reading > eat > go to exam

The takeaway up to this point is that although you may have already entered all relevant activities, it may or or may not be the case that the relationships define a unique ordering—TSort will always give you some ordering, but not necessarily the only one. TSort remains silent about that. For this reason and because the precise ordering is dependent upon order of insertion, it may happen that a given result looks satisfying not because the constraints entered were both sufficient and complete, but because they happened to be added to the graph in a fitting sequence. In the same vein, adding more constraints may or may not change the linearization.

Furthermore, each new requirement may introduce an incompatible constraint. It is by no means unreasonable as such to require that we want to eat before going to the bank, but should we call TS.link_down g, 'eat', 'go to bank' at this point in time, TSort will throw an error (immediately if set to strict, otherwise as soon as TS.sort is called), complaining that it has detected cycle involving node 'buy food'. Had we added TS.link_down g, 'eat', 'go to bank' first, the error would have resulted upon adding the constraint TS.link_down g, 'go to bank', 'fetch money'; in this case the message would've been detected cycle involving node 'eat'.

Now for a more CS-ish application of TSort; specifically, we want to implement the 'function table' that Ravindrababu Ravula derives in his lecture on Compiler Design Lecture 9—Operator grammar and Operator precedence parser (one of the few materials about Pratt-style Top-Down Operator Parsing (TDOP) i was able to find on the web).

Operator Precedence Table

.	id	+	*	$
id	—	>	>	>
+	<	>	<	>
*	<	>	>	>
$	<	<	<	—

operator precedence table

       f|id > g|* > f|+ > g|+ > f|$
g|id > f|*  > ⤴

.	id	+	*	$
f	4	2	4	0
g	5	1	3	0

1
2CND                       = require 'cnd'
3rpr                       = CND.rpr
4badge                     = 'scratch'
5debug                     = CND.get_logger 'debug',     badge
6help                      = CND.get_logger 'help',      badge
7
8
9test_tsort = ->
10  TS = CND.TSORT
11  settings =
12    strict:   yes
13    prefixes: [ 'f|', 'g|', ]
14  graph = TS.new_graph settings
15  TS.link graph, 'id', '-', 'id'
16  TS.link graph, 'id', '>', '+'
17  TS.link graph, 'id', '>', '*'
18  TS.link graph, 'id', '>', '$'
19  TS.link graph, '+',  '<', 'id'
20  TS.link graph, '+',  '>', '+'
21  TS.link graph, '+',  '<', '*'
22  TS.link graph, '+',  '>', '$'
23  TS.link graph, '*',  '<', 'id'
24  TS.link graph, '*',  '>', '+'
25  TS.link graph, '*',  '>', '*'
26  TS.link graph, '*',  '>', '$'
27  TS.link graph, '$',  '<', 'id'
28  TS.link graph, '$',  '<', '+'
29  TS.link graph, '$',  '<', '*'
30  TS.link graph, '$',  '-', '$'
31  help nodes = TS.sort graph
32  matcher = [ 'f|id', 'g|id', 'f|*', 'g|*', 'f|+', 'g|+', 'g|$', 'f|$' ]
33  unless CND.equals nodes, matcher
34    throw new Error """is: #{rpr nodes}
35      expected:  #{rpr matcher}"""
36  try
37    TS.link graph, '$', '>', '$'
38    TS.link graph, '$', '<', '$'
39  catch error
40    { message } = error
41    if /^detected cycle involving node/.test message
42      warn error
43    else
44      throw error
45
46test_tsort()

1console.log '1',  ( TS.precedence_of graph, 'f|id' ) > ( TS.precedence_of graph, 'g|+'  ) # true
2console.log '2',  ( TS.precedence_of graph, 'f|id' ) > ( TS.precedence_of graph, 'g|*'  ) # true
3console.log '3',  ( TS.precedence_of graph, 'f|id' ) > ( TS.precedence_of graph, 'g|$'  ) # true
4console.log '4',  ( TS.precedence_of graph, 'f|+'  ) < ( TS.precedence_of graph, 'g|id' ) # true
5console.log '5',  ( TS.precedence_of graph, 'f|+'  ) > ( TS.precedence_of graph, 'g|+'  ) # true
6console.log '6',  ( TS.precedence_of graph, 'f|+'  ) < ( TS.precedence_of graph, 'g|*'  ) # true
7console.log '7',  ( TS.precedence_of graph, 'f|+'  ) > ( TS.precedence_of graph, 'g|$'  ) # true
8console.log '8',  ( TS.precedence_of graph, 'f|*'  ) < ( TS.precedence_of graph, 'g|id' ) # true
9console.log '9',  ( TS.precedence_of graph, 'f|*'  ) > ( TS.precedence_of graph, 'g|+'  ) # true
10console.log '10', ( TS.precedence_of graph, 'f|*'  ) > ( TS.precedence_of graph, 'g|*'  ) # true
11console.log '11', ( TS.precedence_of graph, 'f|*'  ) > ( TS.precedence_of graph, 'g|$'  ) # true
12console.log '12', ( TS.precedence_of graph, 'f|$'  ) < ( TS.precedence_of graph, 'g|id' ) # true
13console.log '13', ( TS.precedence_of graph, 'f|$'  ) < ( TS.precedence_of graph, 'g|+'  ) # true
14console.log '14', ( TS.precedence_of graph, 'f|$'  ) < ( TS.precedence_of graph, 'g|*'  ) # true

TSort API

• @new_graph = ( settings ) ->

• @link = ( me, f, r, g ) ->

• @link_down = ( me, precedence, consequence ) ->

• @link_up = ( me, consequence, precedence ) -> @link_down me, precedence, consequence

• @register = ( me, names... ) ->

• @sort = ( me ) ->

• @get_precedences = ( me ) ->

• @precedence_of = ( me, name ) ->

Some TDOP Links

• Pratt's original paper from 1973: http://hall.org.ua/halls/wizzard/pdf/Vaughan.Pratt.TDOP.pdf
• The same as an HTML page: http://tdop.github.io/
• Simple Top-Down Parsing in Python: http://effbot.org/zone/simple-top-down-parsing.htm
• Douglas Crockford on TDOP: http://javascript.crockford.com/tdop/tdop.html

XJSON

1  e         = new Set 'xy'
2  e.add new Set 'abc'
3  d         = [ 'A', 'B', e, ]
4  info CND.XJSON.stringify d
5  info CND.XJSON.parse CND.XJSON.stringify d

Output:

1["A","B",{"~isa":"set","%self":["x","y",{"~isa":"set","%self":["a","b","c"]}]}]
2[ 'A', 'B', Set { 'x', 'y', Set { 'a', 'b', 'c' } } ]

CND.TEXT.to_width

2
|北|2
|P |2
|Pe|2
|……|2
|P…|2
|a…|2
|x…|2
|a…|2
|a…|2
|……|2
|……|2
3
|北 |3
|P  |3
|Pe |3
|北…|3
|Pe…|3
|a …|3
|xa…|3
|àx…|4
|a⃝b…|4
|北…|3
|北…|3
4
|北  |4
|P   |4
|Pe  |4
|北京|4
|Pek…|4
|a n…|4
|xàx…|5
|àxa…|5
|a⃝b⃞c…|6
|北……|4
|北……|4
5
|北   |5
|P    |5
|Pe   |5
|北京 |5
|Peki…|5
|a ni…|5
|xàxa…|6
|àxáx…|7
|a⃝b⃞c⃟a…|8
|北京…|5
|北京…|5
10
|北        |10
|P         |10
|Pe        |10
|北京      |10
|Peking    |10
|a nice te…|10
|xàxáxâxãx…|14
|àxáxâxãxa…|14
|a⃝b⃞c⃟a⃝b⃞c⃟a⃝b⃞c…|18
|北京 (Pek…|10
|北京 (Pek…|10
15
|北             |15
|P              |15
|Pe             |15
|北京           |15
|Peking         |15
|a nice test to…|15
|xàxáxâxãxāxa̅xa…|21
|àxáxâxãxāxa̅xăx…|22
|a⃝b⃞c⃟a⃝b⃞c⃟a⃝b⃞c⃟a⃝b⃞c⃟…|25
|北京 (Peking) …|15
|北京 (Peking) …|15
20
|北                  |20
|P                   |20
|Pe                  |20
|北京                |20
|Peking              |20
|a nice test to see …|20
|xàxáxâxãxāxa̅xăxȧxax…|28
|àxáxâxãxāxa̅xăxȧxaxa…|28
|a⃝b⃞c⃟a⃝b⃞c⃟a⃝b⃞c⃟a⃝b⃞c⃟…|25
|北京 (Peking) 位於……|20
|北京 (Peking) 位於……|20
25
|北                       |25
|P                        |25
|Pe                       |25
|北京                     |25
|Peking                   |25
|a nice test to see the e…|25
|xàxáxâxãxāxa̅xăxȧxaxa̠xa̡xa…|35
|àxáxâxãxāxa̅xăxȧxaxa̠xa̡xa̢x…|36
|a⃝b⃞c⃟a⃝b⃞c⃟a⃝b⃞c⃟a⃝b⃞c⃟ |25
|北京 (Peking) 位於華北 (…|25
|北京 (Peking) 位於華北 (…|25
0         1         2         3
0123456789012345678901234567890123456789
a nice test to see the effect*
北京 (Peking) 位於華北 (North*
北京 (Peking) 位於華北 (North*

ToDo

• Add a utility method to enable catch-all listeners on event emitters (as used in kleinbild):

1  _emit = R.emit.bind R
2  R.emit = ( event_name, P... ) =>
3    _emit '*',  event_name, P...
4    _emit       event_name, P...
5    return null
6  R.on '*', ( event_name, P... ) =>
7    @U.dump event_name, P...
8    return null

• use together with node-icecream: ic = ( require 'node-icecream' ) { prefix: '', outputFunction: debug, }