sysmlpy 0.1.0__py3-none-any.whl

sysmlpy/antlr/src/sysmlpy/grammar/antlr4/SysMLv2Lexer.tokens

@@ -0,0 +1,444 @@
+ABOUT=1
+ABSTRACT=2
+ACCEPT=3
+ACTION=4
+ACTOR=5
+AFTER=6
+ALIAS=7
+ALL=8
+ALLOCATE=9
+ALLOCATION=10
+ANALYSIS=11
+AND=12
+AS=13
+ASSERT=14
+ASSIGN=15
+ASSOC=16
+ASSUME=17
+AT=18
+ATTRIBUTE=19
+BEHAVIOR=20
+BIND=21
+BINDING=22
+BOOL=23
+BY=24
+CALC=25
+CASE=26
+CHAINS=27
+CLASS=28
+CLASSIFIER=29
+COMMENT=30
+COMPOSITE=31
+CONCERN=32
+CONJUGATE=33
+CONJUGATES=34
+CONJUGATION=35
+CONNECT=36
+CONNECTION=37
+CONNECTOR=38
+CONST=39
+CONSTANT=40
+CONSTRAINT=41
+CROSSES=42
+DATATYPE=43
+DECIDE=44
+DEF=45
+DEFAULT=46
+DEFINED=47
+DEPENDENCY=48
+DERIVED=49
+DIFFERENCES=50
+DISJOINING=51
+DISJOINT=52
+DO=53
+DOC=54
+ELSE=55
+END=56
+ENTRY=57
+ENUM=58
+EVENT=59
+EXHIBIT=60
+EXIT=61
+EXPOSE=62
+EXPR=63
+FALSE=64
+FEATURE=65
+FEATURED=66
+FEATURING=67
+FILTER=68
+FIRST=69
+FLOW=70
+FOR=71
+FORK=72
+FRAME=73
+FROM=74
+FUNCTION=75
+HASTYPE=76
+IF=77
+IMPLIES=78
+IMPORT=79
+IN=80
+INCLUDE=81
+INDIVIDUAL=82
+INOUT=83
+INTERACTION=84
+INTERFACE=85
+INTERSECTS=86
+INV=87
+INVERSE=88
+INVERTING=89
+ISTYPE=90
+ITEM=91
+JOIN=92
+LANGUAGE=93
+LIBRARY=94
+LOCALE=95
+LOOP=96
+MEMBER=97
+MERGE=98
+MESSAGE=99
+META=100
+METACLASS=101
+METADATA=102
+MULTIPLICITY=103
+NAMESPACE=104
+NEW=105
+NONUNIQUE=106
+NOT=107
+NULL=108
+OBJECTIVE=109
+OCCURRENCE=110
+OF=111
+OR=112
+ORDERED=113
+OUT=114
+PACKAGE=115
+PARALLEL=116
+PART=117
+PERFORM=118
+PORT=119
+PORTION=120
+PREDICATE=121
+PRIVATE=122
+PROTECTED=123
+PUBLIC=124
+REDEFINES=125
+REDEFINITION=126
+REF=127
+REFERENCES=128
+RENDER=129
+RENDERING=130
+REP=131
+REQUIRE=132
+REQUIREMENT=133
+RETURN=134
+SATISFY=135
+SEND=136
+SNAPSHOT=137
+SPECIALIZATION=138
+SPECIALIZES=139
+STAKEHOLDER=140
+STANDARD=141
+STATE=142
+STEP=143
+STRUCT=144
+SUBCLASSIFIER=145
+SUBJECT=146
+SUBSET=147
+SUBSETS=148
+SUBTYPE=149
+SUCCESSION=150
+TERMINATE=151
+THEN=152
+TIMESLICE=153
+TO=154
+TRANSITION=155
+TRUE=156
+TYPE=157
+TYPED=158
+TYPING=159
+UNIONS=160
+UNTIL=161
+USE=162
+VAR=163
+VARIANT=164
+VARIATION=165
+VERIFICATION=166
+VERIFY=167
+VIA=168
+VIEW=169
+VIEWPOINT=170
+WHEN=171
+WHILE=172
+XOR=173
+BANG_EQ_EQ=174
+COLON_COLON_GT=175
+COLON_GT_GT=176
+EQ_EQ_EQ=177
+BANG_EQ=178
+STAR_STAR=179
+ARROW=180
+DOT_DOT=181
+DOT_QUESTION=182
+COLON_COLON=183
+COLON_EQ=184
+COLON_GT=185
+LE=186
+EQ_EQ=187
+FAT_ARROW=188
+GE=189
+QUESTION_QUESTION=190
+AT_AT=191
+HASH=192
+DOLLAR=193
+PERCENT=194
+AMP=195
+LPAREN=196
+RPAREN=197
+STAR=198
+PLUS=199
+COMMA=200
+MINUS=201
+DOT=202
+SLASH=203
+COLON=204
+SEMI=205
+LT=206
+EQ=207
+GT=208
+QUESTION=209
+AT_SIGN=210
+LBRACK=211
+RBRACK=212
+CARET=213
+LBRACE=214
+PIPE=215
+RBRACE=216
+TILDE=217
+IDENTIFIER=218
+STRING=219
+DOUBLE_STRING=220
+INTEGER=221
+REAL=222
+REGULAR_COMMENT=223
+SINGLE_LINE_NOTE=224
+BARE_LINE_COMMENT=225
+WS=226
+'about'=1
+'abstract'=2
+'accept'=3
+'action'=4
+'actor'=5
+'after'=6
+'alias'=7
+'all'=8
+'allocate'=9
+'allocation'=10
+'analysis'=11
+'and'=12
+'as'=13
+'assert'=14
+'assign'=15
+'assoc'=16
+'assume'=17
+'at'=18
+'attribute'=19
+'behavior'=20
+'bind'=21
+'binding'=22
+'bool'=23
+'by'=24
+'calc'=25
+'case'=26
+'chains'=27
+'class'=28
+'classifier'=29
+'comment'=30
+'composite'=31
+'concern'=32
+'conjugate'=33
+'conjugates'=34
+'conjugation'=35
+'connect'=36
+'connection'=37
+'connector'=38
+'const'=39
+'constant'=40
+'constraint'=41
+'crosses'=42
+'datatype'=43
+'decide'=44
+'def'=45
+'default'=46
+'defined'=47
+'dependency'=48
+'derived'=49
+'differences'=50
+'disjoining'=51
+'disjoint'=52
+'do'=53
+'doc'=54
+'else'=55
+'end'=56
+'entry'=57
+'enum'=58
+'event'=59
+'exhibit'=60
+'exit'=61
+'expose'=62
+'expr'=63
+'false'=64
+'feature'=65
+'featured'=66
+'featuring'=67
+'filter'=68
+'first'=69
+'flow'=70
+'for'=71
+'fork'=72
+'frame'=73
+'from'=74
+'function'=75
+'hastype'=76
+'if'=77
+'implies'=78
+'import'=79
+'in'=80
+'include'=81
+'individual'=82
+'inout'=83
+'interaction'=84
+'interface'=85
+'intersects'=86
+'inv'=87
+'inverse'=88
+'inverting'=89
+'istype'=90
+'item'=91
+'join'=92
+'language'=93
+'library'=94
+'locale'=95
+'loop'=96
+'member'=97
+'merge'=98
+'message'=99
+'meta'=100
+'metaclass'=101
+'metadata'=102
+'multiplicity'=103
+'namespace'=104
+'new'=105
+'nonunique'=106
+'not'=107
+'null'=108
+'objective'=109
+'occurrence'=110
+'of'=111
+'or'=112
+'ordered'=113
+'out'=114
+'package'=115
+'parallel'=116
+'part'=117
+'perform'=118
+'port'=119
+'portion'=120
+'predicate'=121
+'private'=122
+'protected'=123
+'public'=124
+'redefines'=125
+'redefinition'=126
+'ref'=127
+'references'=128
+'render'=129
+'rendering'=130
+'rep'=131
+'require'=132
+'requirement'=133
+'return'=134
+'satisfy'=135
+'send'=136
+'snapshot'=137
+'specialization'=138
+'specializes'=139
+'stakeholder'=140
+'standard'=141
+'state'=142
+'step'=143
+'struct'=144
+'subclassifier'=145
+'subject'=146
+'subset'=147
+'subsets'=148
+'subtype'=149
+'succession'=150
+'terminate'=151
+'then'=152
+'timeslice'=153
+'to'=154
+'transition'=155
+'true'=156
+'type'=157
+'typed'=158
+'typing'=159
+'unions'=160
+'until'=161
+'use'=162
+'var'=163
+'variant'=164
+'variation'=165
+'verification'=166
+'verify'=167
+'via'=168
+'view'=169
+'viewpoint'=170
+'when'=171
+'while'=172
+'xor'=173
+'!=='=174
+'::>'=175
+':>>'=176
+'==='=177
+'!='=178
+'**'=179
+'->'=180
+'..'=181
+'.?'=182
+'::'=183
+':='=184
+':>'=185
+'<='=186
+'=='=187
+'=>'=188
+'>='=189
+'??'=190
+'@@'=191
+'#'=192
+'$'=193
+'%'=194
+'&'=195
+'('=196
+')'=197
+'*'=198
+'+'=199
+','=200
+'-'=201
+'.'=202
+'/'=203
+':'=204
+';'=205
+'<'=206
+'='=207
+'>'=208
+'?'=209
+'@'=210
+'['=211
+']'=212
+'^'=213
+'{'=214
+'|'=215
+'}'=216
+'~'=217
+'//'=225

sysmlpy/antlr_parser.py

@@ -0,0 +1,185 @@
+#!/usr/bin/env python3
+"""
+ANTLR4-based SysML v2.0 parser module.
+
+This module provides an alternative parser to textX, using ANTLR4 grammar
+generated from the OMG SysML v2 specification (2026-03 release).
+"""
+import sys
+import os
+
+from antlr4 import InputStream, CommonTokenStream
+from antlr4.error.ErrorListener import ErrorListener
+
+from sysmlpy.antlr.SysMLv2Lexer import SysMLv2Lexer
+from sysmlpy.antlr.SysMLv2Parser import SysMLv2Parser
+
+
+class SysMLSyntaxError(Exception):
+    """Exception raised for SysML syntax errors."""
+    pass
+
+
+class ANTLRErrorListener(ErrorListener):
+    """Custom error listener for ANTLR parsing errors."""
+    
+    def __init__(self):
+        self.errors = []
+    
+    def syntaxError(self, recognizer, offendingSymbol, line, column, msg, e):
+        self.errors.append(f"Syntax error at {line}:{column}: {msg}")
+    
+    def reportAmbiguity(self, recognizer, dfa, startIndex, stopIndex, exact, ambigAlts, configs):
+        pass
+    
+    def reportAttemptingFullContext(self, recognizer, dfa, startIndex, stopIndex, conflictingAlts, configs):
+        pass
+    
+    def reportContextSensitivity(self, recognizer, dfa, startIndex, stopIndex, prediction, configs):
+        pass
+
+
+def parse(source, library=None):
+    """Parse SysML v2.0 source and return a parse tree.
+    
+    Parameters
+    ----------
+    source : str or file-like
+        Either a string containing SysML v2.0 code, or a file object.
+    library : str or Path, optional
+        Path to SysML v2 library files for resolving imports.
+    
+    Returns
+    -------
+    ParseTree
+        The ANTLR4 parse tree (PackageContext).
+    
+    Raises
+    ------
+    SysMLSyntaxError
+        If the source contains syntax errors.
+    """
+    # Handle string or file input
+    if hasattr(source, 'read'):
+        content = source.read()
+    else:
+        content = source
+    
+    # TODO: If library path is provided, load library files and prepend to content
+    # For now, just parse the source directly
+    # Library loading would involve reading .sysml files from the library path
+    # and making them available for import resolution
+    
+    # Create input stream
+    input_stream = InputStream(content)
+    
+    # Create lexer
+    lexer = SysMLv2Lexer(input_stream)
+    
+    # Set up error listener
+    error_listener = ANTLRErrorListener()
+    lexer.removeErrorListeners()
+    lexer.addErrorListener(error_listener)
+    
+    # Create token stream
+    token_stream = CommonTokenStream(lexer)
+    
+    # Create parser
+    parser = SysMLv2Parser(token_stream)
+    parser.removeErrorListeners()
+    parser.addErrorListener(error_listener)
+    
+    # Parse the source - use rootNamespace to support multiple top-level packages
+    tree = parser.rootNamespace()
+    
+    # Check for errors
+    if error_listener.errors:
+        raise SysMLSyntaxError("\n".join(error_listener.errors))
+    
+    return tree
+
+
+def parse_file(filepath):
+    """Parse a SysML v2.0 file.
+    
+    Parameters
+    ----------
+    filepath : str
+        Path to the SysML v2.0 file.
+    
+    Returns
+    -------
+    ParseTree
+        The ANTLR4 parse tree.
+    
+    Raises
+    ------
+    SysMLSyntaxError
+        If the file contains syntax errors.
+    FileNotFoundError
+        If the file does not exist.
+    """
+    with open(filepath, 'r') as f:
+        return parse(f)
+
+
+def parse_to_json(source):
+    """Parse SysML v2.0 source and return JSON-serializable structure.
+    
+    Parameters
+    ----------
+    source : str or file-like
+        Either a string containing SysML v2.0 code, or a file object.
+    
+    Returns
+    -------
+    dict
+        A dictionary representation of the parse tree.
+    """
+    tree = parse(source)
+    return parse_tree_to_dict(tree)
+
+
+def parse_tree_to_dict(tree, include_text=False):
+    """Convert a parse tree to a dictionary.
+    
+    Parameters
+    ----------
+    tree : ParseTree
+        The ANTLR4 parse tree.
+    include_text : bool
+        Whether to include the text of each node.
+    
+    Returns
+    -------
+    dict
+        A dictionary representation of the parse tree.
+    """
+    result = {
+        'type': tree.__class__.__name__,
+    }
+    
+    if include_text:
+        result['text'] = tree.getText()
+    
+    # Get children
+    for i in range(tree.getChildCount()):
+        child = tree.getChild(i)
+        child_class = child.__class__.__name__
+        
+        # Skip terminal nodes (Token) unless requested
+        if hasattr(child, 'getSymbol'):
+            # This is a terminal node
+            if include_text:
+                result[f'child_{i}'] = {
+                    'type': child_class,
+                    'text': child.getText()
+                }
+        else:
+            # This is a rule context
+            result[f'child_{i}'] = parse_tree_to_dict(child, include_text)
+    
+    return result
+
+
+__all__ = ['parse', 'parse_file', 'parse_to_json', 'SysMLSyntaxError']