Changes in natlinkutils.py

January, 2014: a change with the python statement import

The following statement is changed, which can cause problems in NatLink grammar files:

from natlink import *

has been changed to:

import natlink

Therefore variables, functions or exceptions from natlink (the natlink.pyd file) could be (often were) called in a confusing way:

## previous:
import natlinkutils
# and use functions or exceptions from natlink as
# if they were taken from natlinkutils:
    natlinkutils.recognitionMimic(....)
    try:
        ...
    except natlinkutils.BadWindow:
        ...

or even names from natlink could be called unqualified, residing in the global namespace:

## previous:
from natlinkutils import *

# and use functions or exceptions from natlink as
# if they were in the global namespace
    recognitionMimic(....)
    try:
        ...
    except BadWindow:
        ...

The preferred way to call these names:

As the use of from ... import * is now considered bad programming practice, we advise the following way:

import natlinkutils
import natlink
    natlink.recognitionMimic(....)
    try:
        ...
    except natlink.BadWindow:
        ...

Earlier changes:

Only a few minor changes, as addition to Scott Weinstein's installation are made.

"moddle" has been changed to "middle"
in the function activateSet a [:] is added, in order to make copy of the list of activeRules.

This file should come in the core folder of the NatLink system.

Change January 2012 (Quintijn)

at callback time now also self.wordsByRule is provided. This dictionary is identical to the output of the utility function convertResults (in natlinkutils).

Example (also see unittestNatlink.py, around line 2530):

Consider the grammar (the list {furniture} containing "table" and "chair".

<run> exported = test [<optional>+] {colors}+ <extra>;
<optional> = very | small | big;
<extra> = {furniture};

The recognition test very big blue chair will go through the rules:

rule	words	prevRule	nextRule	prevWords	nextWords
run	[test]		optional	[]	["very", "big"]
optional	["very", "big"]	run	run	["test"]	["blue"]
run	["blue"]	optional	extra	["very", "big"]	["chair"]
extra	["chair"]	run		["blue"]	[]

The additional helper variables are:

variable	contents
self.fullResults	[('test', 'run'), ('very', 'optional'), ('big', 'optional'), ('blue', 'run'), ('chair', 'extra')]
self.seqsAndRules	[(['test'], 'run'), (['very', 'big'], 'optional'), (['blue'], 'run'), (['chair'], 'extra')]
self.wordsByRule	{'optional': ['very', 'big'], 'run': ['test', 'blue'], 'extra': ['chair']}

Changes March 2010 (Quintijn)

added deactivateSet() function in GrammarBase
added exceptlist optional variable to activateAll method of GrammarBase
added callRuleResultsFunctions in resultsCallback, so the calling of the rule result functions can be overloaded (for DocstringGrammar)
through this function the instance variables self.prevRule, self.prevWords, self.nextRule and self.nextWords are maintained, to facilitate looking forward and looking back one step in a recognition.
Preventing the detailed gotResults functions to execute: In a rare case it was needed to prevent the rule functions to execute. This is the case if self.doOnlyGotResultsObject is set to 1 or True. Only the gotResultsObject function is called in this case.

Subclasses

For the more practical subclasses of GrammarBase see the Grammar classes section.