Enhanced complex concepts handling

2020-01-11 08:03:35 +01:00
parent a62c1f0f13
commit 40416ac337
24 changed files with 1647 additions and 961 deletions
@@ -0,0 +1,116 @@
+from core.builtin_concepts import BuiltinConcepts
+from parsers.BaseParser import BaseParser
+from parsers.ConceptLexerParser import UnrecognizedTokensNode, ConceptNode
+from parsers.PythonParser import PythonParser
+
+
+class PythonWithConceptsParser(BaseParser):
+    def __init__(self, **kwargs):
+        super().__init__("PythonWithConcepts", 20)
+        self.identifiers = None
+        self.identifiers_key = None
+
+    @staticmethod
+    def sanitize(identifier):
+        res = ""
+        for c in identifier:
+            res += c if c.isalnum() else "0"
+        return res
+
+    def parse(self, context, text):
+        sheerka = context.sheerka
+        if not sheerka.isinstance(text, BuiltinConcepts.PARSER_RESULT):
+            return None
+
+        nodes = text.body
+        if not isinstance(nodes, list):
+            return None
+
+        if len(nodes) == 0:
+            return None
+
+        if not isinstance(nodes[0], (ConceptNode, UnrecognizedTokensNode)):
+            return None
+
+        source = ""
+        to_parse = ""
+        identifiers = {}
+        identifiers_key = {}
+        python_ids_mappings = {}
+
+        def _get_identifier(c):
+            """
+            Internal function because I don't want identifiers, identifiers_key and python_ids_mappings
+            to be instance variables
+            I would like to keep this parser as stateless as possible
+            :param c:
+            :return:
+            """
+            if id(c) in identifiers:
+                return identifiers[id(c)]
+
+            identifier = "__C__" + self.sanitize(c.key or c.name)
+            if c.id:
+                identifier += "__" + c.id
+
+            if identifier in identifiers_key:
+                identifiers_key[identifier] += 1
+                identifier += f"_{identifiers_key[identifier]}"
+            else:
+                identifiers_key[identifier] = 0
+
+            identifier += "__C__"
+
+            identifiers[id(c)] = identifier
+            return identifier
+
+        for node in nodes:
+            if isinstance(node, ConceptNode):
+                source += node.source
+                if to_parse:
+                    to_parse += " "
+                concept = node.concept
+                python_id = _get_identifier(concept)
+                to_parse += python_id
+                python_ids_mappings[python_id] = concept
+            else:
+                source += node.source
+                to_parse += node.source
+
+        with context.push(self, "Trying Python for '" + to_parse + "'") as sub_context:
+            python_parser = PythonParser()
+            result = python_parser.parse(sub_context, to_parse)
+
+        if result.status:
+            python_node = result.body.body
+            python_node.source = source
+            python_node.concepts = python_ids_mappings
+
+            return sheerka.ret(
+                self.name,
+                True,
+                sheerka.new(
+                    BuiltinConcepts.PARSER_RESULT,
+                    parser=self,
+                    source=source,
+                    body=result.body.body,
+                    try_parsed=None))
+
+        else:
+
+            return sheerka.ret(
+                self.name,
+                False,
+                result.body)
+
+    def concept_identifier(self, concept):
+        if id(concept) in self.identifiers:
+            return self.identifiers[id(concept)]
+
+
+        identifier = "__C__" + (concept.key or concept.name)
+        if concept.id:
+            identifier += "__" + concept.id
+        identifier += "__C__"
+
+        return identifier