refactor narsese module

Experiments/Pong/main.py

@@ -20,7 +20,7 @@
 from pynars.NARS.DataStructures._py.Memory import Memory
 
 from pynars.Narsese import Task, Judgement, Term, Stamp, Base
-from pynars import Global
+from pynars.Global import Global
 from pynars.NARS import Reasoner
 from pynars.Console import handle_lines, run_file
 from pynars.Narsese._py.Compound import Compound

Experiments/Pong/main2.py

@@ -20,7 +20,7 @@
 from pynars.NARS.DataStructures._py.Memory import Memory
 
 from pynars.Narsese import Task, Judgement, Term, Stamp, Base
-from pynars import Global
+from pynars.Global import Global
 from pynars.NARS import Reasoner
 from pynars.Console import handle_lines, run_file
 from pynars.Narsese._py.Compound import Compound

pynars/Config.py

@@ -1,201 +1,2 @@
-'''
-Configure some hyper-parameters and some other settings via the file `config.json`.
-'''
-from pathlib import Path
-
-try:
-    import jstyleson as json # json with C/C++ style comments
-except:
-    # import json
-    raise "please install the module by `pip install jstyleson`"
-
-class Enable:
-    temporal_reasoning = False
-    variable = True
-    anticipation = False
-    operation = False
-    debug = False
-
-class Config:
-    r_top_level_attention_adjust = 0.5
-
-    priority: float=0.8
-    durability: float=0.8
-    quality: float=0.5
-    num_buckets: int = 100
-    max_duration: int = 10000
-    f: float=1.0
-    c: float=0.9
-    c_judgement: float=0.9
-    c_goal: float=0.9
-    k: int=1
-    p_judgement: float=0.8
-    d_judgement: float=0.5
-    p_question: float=0.9
-    d_question: float=0.9
-    p_quest: float=0.9
-    d_quest: float=0.9
-    p_goal: float=0.9
-    d_goal: float=0.9
-
-    p_feedback: float = 0.9
-    d_feedback: float = 0.5
-
-    budget_thresh: float=0.01
-
-    nlevels_task_link: int=10
-    capacity_task_link: int=100
-    nlevels_term_link: int=10
-    capacity_term_link: int=100
-    capacity_table: int=100
-
-
-    quality_min: float=0.3
-    cycles_per_duration: int=5
-    n_forget_durations: int=2
-    cycles_forget = cycles_per_duration*n_forget_durations
-
-    revision_max_occurence_distance: int=10
-
-    truth_epsilon = 0.01
-    budget_epsilon = 0.0001
-
-    r_term_complexity_unit = 0.5
-    t_sentence_directness_unit = 1.0
-
-
-    variable_repr_normalized = False
-
-    rate_discount_c = 0.5
-
-    rate_discount_p_internal_exp = 0.1
-    rate_discount_d_internal_exp = 0.1
-
-    # parameters for updating the Budget of Concept.
-    # Lower values means it is harder to change the budget, higher values means it is easier to change the budget
-    concept_update_durability_weight = 0.1
-    concept_update_quality_weight = 0.1
-
-    # temporal parameters
-    temporal_duration = 5
-    n_sequence_attempts = 10
-    n_op_condition_attempts = 10
-
-    projection_decay = 0.99
-    Td_decision_threshold = 0.51
-
-    # what this value represents was originally equal to the termlink record length (10), but we may want to adjust it or make it scaled according to duration since it has more to do with time than # of records.  it can probably be increased several times larger since each item should remain in the recording queue for longer than 1 cycle
-    novelty_horizon = 100000
-    term_link_record_length = 10
-
-    maximum_evidental_base_length = 20000
-
-    @classmethod
-    def check(cls):
-        '''Check if each parameter is valid'''
-        pass
-
-    @classmethod
-    def apply(cls):
-        '''Apply setting values of hyper-parameters'''
-        # from pynars.NARS import DataStructures
-        # # Budget
-        # DataStructures.Budget.priority = cls.priority
-        # DataStructures.Budget.durability = cls.durability
-        # DataStructures.Budget.quality = cls.quality
-
-
-def load(file_path: str):
-
-    file_path: Path = Path(file_path)
-
-    valid = True
-    if not file_path.exists():
-        if Enable.debug:
-            print(f"The file `{file_path}` does not exist.")
-        valid = False
-    if file_path.suffix != '.json':
-        if Enable.debug:
-            print(f"The file `{file_path}` should be `*.json`.")
-        valid = False
-    if not valid:
-        file_path = Path(__file__).parent/'config.json'
-        if Enable.debug:
-            print(f'Loaded config file: {file_path}')
-    try:
-        with open(file_path, 'r') as f:
-            content = json.load(f)
-    except:
-        raise f"Error when openning the file `{file_path}`."
-    # set driver mode (py/pyx/cypy/cpp)
-    try:
-        pass # TODO
-    except:
-        pass # TODO
-
-    # set hyper-parameters
-    try:
-        defaults: dict = content['HYPER-PARAMS']['DEFAULT']
-        budget = defaults.get('BUDGET', None)
-        if budget is not None:
-            budget: dict
-            Config.p_judgement = budget.get('PRIORITY_JUDGEMENT', Config.p_judgement)
-            Config.d_judgement = budget.get('DURABILITY_JUDGEMENT', Config.d_judgement)
-            Config.p_question = budget.get('PRIORITY_QUESTION', Config.p_question)
-            Config.d_question = budget.get('DURABILITY_QUESTION', Config.d_question)
-            Config.p_quest = budget.get('PRIORITY_QUEST', Config.p_quest)
-            Config.d_quest = budget.get('DURABILITY_QUEST', Config.d_quest)
-            Config.p_goal = budget.get('PRIORITY_GOAL', Config.p_goal)
-            Config.d_goal = budget.get('DURABILITY_GOAL', Config.d_goal)
-            Config.p_feedback = budget.get('PRIORITY_FEEDBACK', Config.p_feedback)
-            Config.d_feedback = budget.get('DURABILITY_FEEDBACK', Config.d_feedback)
-            Config.budget_thresh = budget.get('THRESHOLD', Config.budget_thresh)
-
-        num_buckets = defaults.get('NUM_BUCKETS', None)
-        if num_buckets is not None:
-            Config.num_buckets = 100
-        truth = defaults.get('TRUTH', None)
-        if truth is not None:
-            Config.f = truth['FREQUENCY']
-            Config.c = truth['CONFIDENCE']
-            Config.c_judgement = truth['CONFIDENCE_JUDGEMENT']
-            Config.c_goal = truth['CONFIDENCE_GOAL']
-            Config.k = truth['K']
-
-        max_duration = defaults.get('MAX_DURATION', None)
-        if max_duration is not None:
-            Config.max_duration = max_duration
-
-        concept: dict = defaults.get('CONCEPT', None)
-        if concept is not None:
-            Config.nlevels_task_link = concept.get('NUM_LEVELS_TASKLINK_BAG', Config.nlevels_task_link)
-            Config.capacity_task_link = concept.get('CAPACITY_TASKLINK_BAG', Config.capacity_task_link)
-            Config.nlevels_term_link = concept.get('NUM_LEVELS_TERMLINK_BAG', Config.nlevels_term_link)
-            Config.capacity_term_link = concept.get('CAPACITY_TERMLINK_BAG', Config.capacity_term_link)
-            Config.capacity_table = concept.get('CAPACITY_TABLE', Config.capacity_table)
-        Config.r_term_complexity_unit = defaults.get('COMPLEXITY_UNIT', Config.r_term_complexity_unit)
-        Config.quality_min = defaults.get('QUALITY_MIN', Config.quality_min)
-        Config.cycles_per_duration = defaults.get('CYCLES_PER_DURATION', Config.cycles_per_duration)
-        Config.n_forget_durations = defaults.get('NUM_FORGET_DURATIONS', Config.n_forget_durations)
-        Config.cycles_forget = Config.cycles_per_duration * Config.n_forget_durations
-        Config.revision_max_occurence_distance = defaults.get('REVISION_MAX_OCCURRENCE_DISTANCE', Config.revision_max_occurence_distance)
-
-        Config.rate_discount_c = defaults.get('RATE_DISCOUNT_CONFIDENCE', Config.rate_discount_c)
-
-        Config.rate_discount_p_internal_exp = defaults.get('RATE_DISCOUNT_PRIORITY_INTERNAL_EXPERIENCE', Config.rate_discount_p_internal_exp)
-        Config.rate_discount_d_internal_exp = defaults.get('RATE_DISCOUNT_DURABILITY_INTERNAL_EXPERIENCE', Config.rate_discount_d_internal_exp)
-
-        hyperparams: dict = content['HYPER-PARAMS']
-        Config.truth_epsilon = hyperparams.get('TRUTH_EPSILON', Config.truth_epsilon)
-        Config.budget_epsilon = hyperparams.get('BUDGET_EPSILON', Config.budget_epsilon)
-        Config.r_term_complexity_unit = hyperparams.get('COMPLEXITY_UNIT', Config.r_term_complexity_unit)
-        Config.temporal_duration = hyperparams.get('TEMPORAL_DURATION', Config.temporal_duration)
-        Config.n_sequence_attempts = hyperparams.get('TEMPORAL_DURATION', Config.n_sequence_attempts)
-        Config.n_op_condition_attempts = hyperparams.get('NUM_OP_CONDITION_ATTEMPTS', Config.n_op_condition_attempts)
-
-        pass # TODO
-    except:
-        pass # TODO
-
-    Config.check()
-    Config.apply()
+# TODO: move NAL and NARS related configurations to pynars/Narsese/Config.py, and keep Narsese related configurations in pynars/Narsese/Config.py
+from pynars.Narsese.Config import *

pynars/Global.py

@@ -1,14 +1,7 @@
 import inspect
+from pynars.Narsese import Global as _Global
 
-time = 0
-_input_id = 0
-def get_input_id():
-    global _input_id
-    input_id = _input_id
-    _input_id += 1
-    return input_id
-
-class States:
+class StatesCls:
     task = None
     belief = None
     concept = None
@@ -19,29 +12,43 @@ def reset(cls):
         ''''''
         for name in vars(cls):
             attr = getattr(cls, name)
-            if inspect.ismethod(attr): continue
-            if name.startswith('_'): continue
-            if isinstance(attr, property): continue
+            if inspect.ismethod(attr):
+                continue
+            if name.startswith('_'):
+                continue
+            if isinstance(attr, property):
+                continue
             # if isinstance(attr, classmethod): continue
             setattr(cls, name, None)
 
     @classmethod
     def record_premises(cls, task=None, belief=None):
         cls.task = task
         cls.belief = belief
-    
+
     @classmethod
     def record_concept(cls, concept=None):
         cls.concept = concept
 
     @classmethod
     def record_rules(cls, rules=None):
         cls.rules = rules
-    
+
     @property
     def time(self):
-        return time
-
+        return _Global.time
+
+    @time.setter
+    def time(self, value):
+        _Global.time = value
+
+    @staticmethod
+    def get_input_id():
+        return _Global.get_input_id()
+
     @classmethod
     def __repr__(cls):
-        return f'<States: time={time}\n\tconcept: {cls.concept}\n\ttask: {cls.task}\n\tbelief: {cls.belief}\n\trules: {cls.rules}\n>.'
+        return f'<States: time={cls.time}\n\tconcept: {cls.concept}\n\ttask: {cls.task}\n\tbelief: {cls.belief}\n\trules: {cls.rules}\n>.'
+
+
+Global = StatesCls()

pynars/NAL/Functions/Tools.py

@@ -3,7 +3,7 @@
 # def compound_remove_components
 
 from typing import Union
-from pynars import Global
+from pynars.Global import Global
 from pynars.Config import Config, Enable
 from pynars.NAL.Functions.TemporalFunctions import eternalize, project
 from pynars.Narsese import Budget

pynars/NAL/MentalOperation/_aware.py

@@ -14,7 +14,7 @@
 from pynars.Narsese import Term
 from ..Functions.Tools import truth_from_term, truth_to_quality, truth_to_term
 from pynars.Narsese import Base
-from pynars import Global
+from pynars.Global import Global
 
 
 def _aware(statement: Statement, stamp: Stamp, budget_task: Budget=None):

pynars/NAL/MentalOperation/_execute.py

@@ -10,7 +10,7 @@
 from pynars.Narsese import Term
 from ..Functions.Tools import truth_from_term, truth_to_quality
 from pynars.Narsese import Base
-from pynars import Global
+from pynars.Global import Global
 
 def execute(task: Task):
     ''''''

pynars/NARS/Control/Reasoner.py

@@ -15,7 +15,7 @@
 from pynars.Config import Enable
 from typing import Callable, List, Tuple, Union
 import pynars.NARS.Operation as Operation
-from pynars import Global
+from pynars.Global import Global
 from time import time
 from pynars.NAL.Functions.Tools import project_truth, project
 from ..GlobalEval import GlobalEval
@@ -77,7 +77,7 @@ def input_narsese(self, text, go_cycle: bool = False) -> Tuple[bool, Union[Task,
 
     def cycle(self):
         """Everything to do by NARS in a single working cycle"""
-        Global.States.reset()
+        Global.reset()
         tasks_derived: List[Task] = []
 
         judgement_revised, goal_revised, answers_question, answers_quest = None, None, None, None

pynars/NARS/InferenceEngine/GeneralEngine/GeneralEngine.py

@@ -12,7 +12,7 @@
 from typing import Callable, List, Tuple
 from ...RuleMap import RuleCallable, RuleMap
 from pynars.NAL.Inference import local__revision
-from pynars import Global
+from pynars.Global import Global
 from ..Engine import Engine
 from .extract_feature import extract_feature
 from pathlib import Path
@@ -340,7 +340,7 @@ def step(self, concept: Concept):
         '''One step inference.'''
         tasks_derived = []
 
-        Global.States.record_concept(concept)
+        Global.record_concept(concept)
 
         # Based on the selected concept, take out a task and a belief for further inference.
         task_link_valid: TaskLink = concept.task_links.take(remove=True)
@@ -352,8 +352,8 @@ def step(self, concept: Concept):
         # inference for single-premise rules
         is_valid, _, rules_immediate = GeneralEngine.match(task, None, None, task_link_valid, None)
         if is_valid:
-            Global.States.record_premises(task)
-            Global.States.record_rules(rules_immediate)
+            Global.record_premises(task)
+            Global.record_rules(rules_immediate)
             tasks = self.inference(task, None, None, task_link_valid, None, rules_immediate)
             tasks_derived.extend(tasks)
 
@@ -389,8 +389,8 @@ def step(self, concept: Concept):
 
 
         if is_valid:
-            Global.States.record_premises(task, belief)
-            Global.States.record_rules(rules)
+            Global.record_premises(task, belief)
+            Global.record_rules(rules)
             tasks = self.inference(task, belief, term_belief, task_link_valid, term_link_valid, rules)
             if term_link_valid is not None: # TODO: Check here whether the budget updating is the same as OpenNARS 3.0.4.
                 for task in tasks: TermLink.update_budget(term_link_valid.budget, task.budget.quality, belief.budget.priority if belief is not None else concept_target.budget.priority)

pynars/NARS/Operation/Execution.py

@@ -12,7 +12,7 @@
 from pynars.Narsese import Term
 from pynars.NAL.Functions.Tools import truth_from_term, truth_to_quality
 from pynars.Narsese import Base
-from pynars import Global
+from pynars.Global import Global
 
 def executed_task(task: Task):
     '''