kodjo/Sheerka-Old
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Added simple form of concept composition

Browse Source
This commit is contained in:
Kodjo Sossouvi
2020-01-15 18:38:29 +01:00
parent 51fa9629d0
commit 8152f82c6b
22 changed files with 1105 additions and 544 deletions
Download Patch File Download Diff File Expand all files Collapse all files
core/sheerka.py
+68 -41
View File
@@ -258,6 +258,7 @@ class Sheerka(Concept):
# execution_context.log(logger or self.log, f"Parsing {debug_text}")
with execution_context.push(desc=f"Parsing using {parser.name}") as sub_context:
sub_context.add_inputs(to_parse=to_parse)
res = parser.parse(sub_context, to_parse)
if res is not None:
if hasattr(res, "__iter__"):
@@ -278,6 +279,7 @@ class Sheerka(Concept):
to_process.append(res)
if res.status:
return_value_success_found = True
sub_context.add_values(return_values=res)
if return_value_success_found:
stop_processing = True
@@ -347,51 +349,69 @@ class Sheerka(Concept):
sorted_priorities = sorted(grouped_evaluators.keys(), reverse=True)
# process
iteration = 0
while True:
simple_digest = return_values[:]
with execution_context.push(desc=f"iteration #{iteration}", iteration=iteration) as iteration_context:
simple_digest = return_values[:]
iteration_context.add_inputs(return_values=simple_digest)
for priority in sorted_priorities:
for priority in sorted_priorities:
original_items = return_values[:]
evaluated_items = []
to_delete = []
for evaluator in grouped_evaluators[priority]:
evaluator = _preprocess_evaluators(execution_context, evaluator.__class__()) # fresh copy
original_items = return_values[:]
evaluated_items = []
to_delete = []
for evaluator in grouped_evaluators[priority]:
evaluator = _preprocess_evaluators(execution_context, evaluator.__class__()) # fresh copy
# process evaluators that work on return value
from evaluators.BaseEvaluator import OneReturnValueEvaluator
if isinstance(evaluator, OneReturnValueEvaluator):
for item in original_items:
if evaluator.matches(execution_context, item):
result = evaluator.eval(execution_context, item)
if result is None:
continue
elif isinstance(result, list):
evaluated_items.extend(result)
to_delete.append(item)
elif isinstance(result, ReturnValueConcept):
evaluated_items.append(result)
to_delete.append(item)
sub_context_desc = f"Evaluating using {evaluator.name} ({priority=})"
with iteration_context.push(desc=sub_context_desc) as sub_context:
sub_context.add_inputs(return_values=original_items)
# process evaluators that work on one simple return value at the time
from evaluators.BaseEvaluator import OneReturnValueEvaluator
if isinstance(evaluator, OneReturnValueEvaluator):
debug_result = []
for item in original_items:
if evaluator.matches(sub_context, item):
result = evaluator.eval(sub_context, item)
if result is None:
debug_result.append({"input": item, "return_value": None})
continue
to_delete.append(item)
if isinstance(result, list):
evaluated_items.extend(result)
elif isinstance(result, ReturnValueConcept):
evaluated_items.append(result)
else:
error = self.new(BuiltinConcepts.INVALID_RETURN_VALUE, body=result,
evaluator=evaluator)
result = self.ret("sheerka.process", False, error, parents=[item])
evaluated_items.append(result)
debug_result.append({"input": item, "return_value": result})
else:
debug_result.append({"input": item, "return_value": "** No Match **"})
sub_context.add_values(return_values=debug_result)
# process evaluators that work on all return values
else:
if evaluator.matches(sub_context, original_items):
results = evaluator.eval(sub_context, original_items)
if results is None:
continue
if not isinstance(results, list):
results = [results]
for result in results:
evaluated_items.append(result)
to_delete.extend(result.parents)
sub_context.add_values(return_values=results)
else:
error = self.new(BuiltinConcepts.INVALID_RETURN_VALUE, body=result,
evaluator=evaluator)
evaluated_items.append(self.ret("sheerka.process", False, error, parents=[item]))
to_delete.append(item)
sub_context.add_values(return_values="** No Match **")
# process evaluators that work on all return values
else:
if evaluator.matches(execution_context, original_items):
results = evaluator.eval(execution_context, original_items)
if results is None:
continue
if not isinstance(results, list):
results = [results]
for result in results:
evaluated_items.append(result)
to_delete.extend(result.parents)
return_values = evaluated_items
return_values.extend([item for item in original_items if item not in to_delete])
return_values = evaluated_items
return_values.extend([item for item in original_items if item not in to_delete])
iteration_context.add_values(return_values=return_values[:])
# have we done something ?
to_compare = return_values[:]
@@ -399,7 +419,7 @@ class Sheerka(Concept):
break
# inc the iteration and continue
execution_context = execution_context.push(iteration=execution_context.iteration + 1)
iteration += 1
# remove all evaluation context that are not reduced
return_values = core.utils.remove_list_from_list(return_values, evaluation_return_values)
@@ -879,11 +899,11 @@ class Sheerka(Concept):
return obj
def is_unknown(self, obj):
def is_known(self, obj):
if not isinstance(obj, Concept):
return True
return obj.key == BuiltinConcepts.UNKNOWN_CONCEPT
return obj.key != str(BuiltinConcepts.UNKNOWN_CONCEPT)
def isinstance(self, a, b):
"""
@@ -1049,6 +1069,8 @@ class ExecutionContext:
self.desc = desc # human description of what is going on
self.children = []
self.preprocess = None
self.inputs = {} # what was the parameters of the execution context
self.values = {} # what was produced by the execution context
self.obj = kwargs.pop("obj", None)
@@ -1105,6 +1127,11 @@ class ExecutionContext:
self.preprocess.add(preprocess)
return self
def add_inputs(self, **kwargs):
for k, v in kwargs.items():
self.inputs[k] = v
return self
def add_values(self, **kwargs):
for k, v in kwargs.items():
self.values[k] = v
core/sheerka_transform.py
+2 -2
View File
@@ -48,7 +48,7 @@ class SheerkaTransform:
self.ids[obj] = self.id_count
if isinstance(obj, Concept):
return self.context_to_dict(obj)
return self.concept_to_dict(obj)
elif isinstance(obj, ExecutionContext):
return self.execution_context_to_dict(obj)
@@ -104,7 +104,7 @@ class SheerkaTransform:
else:
return obj
def context_to_dict(self, obj: Concept):
def concept_to_dict(self, obj: Concept):
to_dict = {
OBJ_TYPE_KEY: SheerkaTransformType.Concept,
OBJ_ID_KEY: self.id_count,
core/tokenizer.py
+58
View File
@@ -37,6 +37,8 @@ class TokenKind(Enum):
BACK_SLASH = "bslash" # \
CARAT = "carat" # ^
DOLLAR = "dollar" # $
EURO = "dollar" # €
STERLING = "steling" # £
EMARK = "emark" # !
GREATER = "greater" # >
LESS = "less" # <
@@ -128,6 +130,18 @@ class Tokenizer:
yield Token(TokenKind.MINUS, "-", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "_":
if self.i + 1 < self.text_len and self.text[self.i + 1].isalpha():
identifier = self.eat_identifier(self.i)
token_type = TokenKind.KEYWORD if identifier in self.KEYWORDS else TokenKind.IDENTIFIER
value = Keywords(identifier) if identifier in self.KEYWORDS else identifier
yield Token(token_type, value, self.i, self.line, self.column)
self.i += len(identifier)
self.column += len(identifier)
else:
yield Token(TokenKind.UNDERSCORE, "_", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "/":
yield Token(TokenKind.SLASH, "/", self.i, self.line, self.column)
self.i += 1
@@ -205,6 +219,46 @@ class Tokenizer:
yield Token(TokenKind.GREATER, ">", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "!":
yield Token(TokenKind.EMARK, "!", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "`":
yield Token(TokenKind.BACK_QUOTE, "`", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "\\":
yield Token(TokenKind.BACK_SLASH, "\\", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "^":
yield Token(TokenKind.CARAT, "^", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "$":
yield Token(TokenKind.DOLLAR, "$", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "€":
yield Token(TokenKind.EURO, "€", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "£":
yield Token(TokenKind.STERLING, "£", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "#":
yield Token(TokenKind.HASH, "#", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "°":
yield Token(TokenKind.DEGREE, "°", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "~":
yield Token(TokenKind.TILDE, "~", self.i, self.line, self.column)
self.i += 1
self.column += 1
elif c == "\n" or c == "\r":
newline = self.eat_newline(self.i)
yield Token(TokenKind.NEWLINE, newline, self.i, self.line, self.column)
@@ -234,6 +288,10 @@ class Tokenizer:
self.i += len(string)
self.column = 1 if newlines > 0 else self.column + len(string)
self.line += newlines
elif c == "_":
yield Token(TokenKind.UNDERSCORE, "_", self.i, self.line, self.column)
self.i += 1
self.column += 1
else:
raise LexerError(f"Unknown token '{c}'", self.text, self.i, self.line, self.column)