Reworked Interpreter WIP

.vscode/launch.json

@@ -8,7 +8,7 @@
             "name": "Python: Aktuelle Datei",
             "type": "python",
             "request": "launch",
-            "program": "run.py",
+            "program": "debug.py",
             "console": "integratedTerminal"
         }
     ]

interpreter/NassiShneidermann.py

@@ -6,7 +6,7 @@ import os.path
 import secrets
 
 from draw.Iinstruction import Iinstruction
-from interpreter import interpret_source as itp
+from interpreter.interpret_source import JavaInterpreter
 from draw.code_to_image_wrapper import NSD_writer
 import draw.code_to_image as cti
 
@@ -70,4 +70,5 @@ class NassiShneidermanDiagram:
                     raise
 
     def load_from_file(self, filepath:str):
-        self.function_scopes = itp.load_scoped_instructions(filepath)
+        itp = JavaInterpreter(filepath)
+        self.function_scopes = itp.load_instruction_scopes()

interpreter/interpret_source.py

@@ -1,8 +1,6 @@
 import logging
-from os import remove
 import re
-from re import split
-from typing import List, Match, Tuple
+from typing import Dict, List, Match, Tuple
 
 from errors.custom import InterpreterException, JavaSyntaxError, ScopeNotFoundException
 from draw.Iinstruction import *
@@ -28,215 +26,281 @@ class Function_scope(Iterable):
     def __iter__(self):
         return self.contents.__iter__()
 
-COMMENT_PATTERN = re.compile(r"""^(//|--|#).*""")
-REMOVE_KEYWORDS = [' ', ';', "public", "private", "final", "protected"]
-VARIABLE_TAGS = ["byte", "short", "int", "long", "float", "double", "boolean", "char", "String"]
-FUNCTION_IDENTIFIERS = ["void"]
-FUNCTION_IDENTIFIERS.extend(VARIABLE_TAGS)
+class JavaInterpreter:
 
-WHILE_TAG = "solange " #german for 'while'. Change this depending on your language
+    def __init__(self, filepath: str) -> None:
+        with open(filepath) as file:
+            if not file.readable():
+                raise InterpreterException(f"Unable to read input file {filepath}")
+            self.src = file.read()
+            self.lines = [] # to be filled in later
+        self._init_regex()
+        self._recompile_regex()
 
-REPLACE = dict((re.escape(k), '') for k in REMOVE_KEYWORDS)
-remove_pattern = re.compile("|".join(REPLACE.keys()), flags=re.MULTILINE)
+    def reset_tags(self, additional_tags: Dict[str, List[str]]=None):
+        """Reset the Interpreters internal tags to include the new tags given in {additional_tags}
+        the key for comment tags is "comments"
+        the key for tags that should be ignored is "ignore"
+        the key for additional type tags is "types"
+        """
 
-variable_regex = "^("
-for kw in VARIABLE_TAGS:
-    variable_regex += fr"""{kw}|"""
-variable_pattern = re.compile(variable_regex[0:-1]+")(.*)=?(.*)", flags=re.MULTILINE)
+        if additional_tags:
+            if "comments" in additional_tags.keys():
+                self.comment_tags.extend(additional_tags["comments"])
+            if "ignore" in additional_tags.keys():
+                self.remove_tags.extend(additional_tags["ignore"])
+            if "types" in additional_tags.keys():
+                self.type_tags.extend(additional_tags["types"])
+                self.function_tags.extend(additional_tags["types"])
+        else:
+            self._init_regex()
+        self._recompile_regex()
 
-function_regex = "^("
-for kw in FUNCTION_IDENTIFIERS:
-    function_regex += fr"""{kw}|"""
-function_regex = function_regex[0:-1]+ ").*([(].*[)].*)"
+    def _init_regex(self):
+        """Initialize all tag lists to their default state"""
 
-function_pattern = re.compile(function_regex)
+        self.comment_tags = ["//", "#", "--"] #dont ask why a java interpreter supports lua and python style comments
+        self.remove_tags = ['\n', "public", "private", "protected", "final"]
+        self.type_tags = ["byte", "short", "int", "long", "float", "double", "boolean", "char", "String"]
+        self.function_tags = ["void"]
+        self.function_tags.extend(self.type_tags)
 
-def replace_all_tags(org: str):
-    return remove_pattern.sub(lambda m: REPLACE[re.escape(m.group(0))], org)
+    def _recompile_regex(self):
+        """recompile all regex's using the new tag lists"""
+        comment_regex = r"""^("""
+        for tag in self.comment_tags:
+            comment_regex += fr"""{re.escape(tag)}|"""
+        self.comment_pattern = re.compile(comment_regex[:-1]+").*", flags=re.MULTILINE)
 
-def load_src(filepath: str) -> List[str]:
-    lines: List[str] = []
-    brace_open_count, brace_closed_count = 0,0
-    try:
-        with open(filepath, encoding="utf-8") as file:
-            for _line in file:
-                line = replace_all_tags(_line.strip())
-                if line and not COMMENT_PATTERN.match(line):
-                    lines.append(line)
-                if '{' in line:
-                    brace_open_count += 1
-                if '}' in line:
-                    brace_closed_count += 1
-    except:
-        raise FileNotFoundError(f"File {filepath} was not found!")
+        remove_regex = r"""^("""
+        for tag in self.remove_tags:
+            remove_regex += fr"""{re.escape(tag)}|"""
+        self.remove_pattern = re.compile(remove_regex[:-1]+')', flags=re.MULTILINE)
 
-    if brace_open_count != brace_closed_count:
-        raise JavaSyntaxError("Number of opened braces does not match number of closed ones. Program is ill-formed!")
+        variable_regex = r"""^("""
+        for tag in self.type_tags:
+            variable_regex += fr"""{re.escape(tag)}|"""
+        self.variable_pattern = re.compile(variable_regex[:-1]+")(.*=|.*;)(.*)", flags=re.MULTILINE)
+
+        function_regex = r"""^(?P<return_type>"""
+        for tag in self.function_tags:
+            function_regex += fr"""{re.escape(tag)}|"""
+        self.function_pattern = re.compile(function_regex[:-1]+")(?P<name>.*)[(](?P<args>.*)[)][^;]", flags=re.MULTILINE)
+
+    def _check_src(self, idx: int, tag: str) -> bool:
+        if idx >= len(self.lines):
+            return False
+        return tag in self.lines[idx]
+
+    def _check_line_start(self, idx: int, tag: str) -> bool:
+        if idx >= len(self.lines):
+            return False
+        return self.lines[idx].startswith(tag)
+
+    def _get_scope_start_offset(self, i: int) -> int:
+        if self._check_src(i, "{"):
+            return 1
+        elif self._check_src(i+1, "{"):      
+            return 2
+        raise ScopeNotFoundException("Unable to find scope start. Is the program ill-formed?")
+
+    def _handle_while(self, line: str, idx: int):
+        bracket_idx = line.rindex(')') # throws if while contruct is illformed
+
+        instruction_txt = line[6:bracket_idx]
+        brace_offset = self._get_scope_start_offset(idx)
+        child_instructions, idx = self._get_instructions_in_scope(idx+brace_offset)
+        return while_instruction_front(("while" + instruction_txt), child_instructions), idx
     
-    return lines
+    def _handle_else(self, idx: int):
+        brace_offset = self._get_scope_start_offset(idx)
+        return self._get_instructions_in_scope(idx+brace_offset)
 
-def check_src(src: List[str], line_index: int, tag: str) -> bool:
-    if line_index >= len(src):
-        return False
-    return tag in src[line_index]
+    def _get_else_scope(self, idx:int):
+        instructions = None
+        if self._check_line_start(idx, "}else"):
+            logging.debug("found else construct in line: %i", idx+1)
+            instructions, idx = self._handle_else(idx)
+        
+        elif self._check_line_start(idx+1, "else"):
+            logging.debug("found else construct in line: %i", idx+2)
+            instructions, idx = self._handle_else(idx+1)
+        return instructions
+        
 
-def check_line_start(src: List[str], line_index: int, tag: str) -> bool:
-    if line_index >= len(src):
-        return False
-    return src[line_index].startswith(tag)
+    def _handle_if(self, line: str, idx: int):
+        bracket_idx = line.rindex(')') # throws if the contruct is illformed
+        instruction_txt = line[3:bracket_idx]
 
-def get_next_occurence_of(src: List[str], start_idx:int, tag:str) -> int:
-    """Returns the index of the next occurence of tag in src from start_idx"""
-    i = start_idx
-    while i < len(src):
-        if check_src(src, i, tag):
-            break
-        i += 1
-    return i
+        brace_offset = self._get_scope_start_offset(idx)
+        true_instructions, idx = self._get_instructions_in_scope(idx+brace_offset)
 
-def get_scope_start_offset(src: List[str], i: int) -> int:
-    if check_src(src, i, "{"):
-        return 1
-    elif check_src(src, i+1, "{"):
-        return 2
-    raise ScopeNotFoundException("Unable to find scope start. Is the program ill-formed?")
+        false_instructions = self._get_else_scope(idx)
+        
+        return if_instruction(instruction_txt, true_instructions, false_instructions), idx
 
+    def _handle_do_while(self, line: str, idx: int):
+        brace_offset = self._get_scope_start_offset(idx)
+        child_instructions, idx = self._get_instructions_in_scope(idx+brace_offset)
 
-def handle_while(line: str, src: List[str], i: int) -> Tuple[Iinstruction, int]:
-    bracket_idx = line.rindex(')') # throws if while contruct is illformed
+        instruction_txt = None
+        if self._check_line_start(idx, "}while("):
+            end_line = self.lines[idx]
+            bracket_index = end_line.rindex(')') #throws if contruct id ill-formed
+            instruction_txt = end_line[7:bracket_index]
+        elif self._check_line_start(idx+1, "while("):
+            idx += 1
+            end_line = self.lines[idx]
+            bracket_index = end_line.rindex(')')
+            instruction_txt = end_line[6:bracket_index]
+        else:
+            raise JavaSyntaxError("Ill-formed do-while construct!")
 
-    instruction_txt = line[6:bracket_idx]
-    brace_offset = get_scope_start_offset(src, i)
-    child_instructions, i = get_instructions_in_scope(src, i+brace_offset)
-    return while_instruction_front((WHILE_TAG + instruction_txt), child_instructions), i
+        return while_instruction_back(instruction_txt, child_instructions), idx
 
-def handle_else(src: List[str], i: int) -> Tuple[List[Iinstruction], int]:
-    brace_offset = get_scope_start_offset(src, i)
-    return get_instructions_in_scope(src, i+brace_offset)
+    def _handle_variable(self, line: str, idx: int):
+        groups = self.variable_pattern.match(line).groups()
+        var_type = groups[0]
+        var_name = groups[1][:-1]
+        var_value = groups[2]
+        if var_value == '':
+            return generic_instruction(f"declare variable '{var_name}' of type {var_type}"), idx
+        return generic_instruction(f"declare variable '{var_name}' of type {var_type} with value {var_value}"), idx
 
-def handle_if(line: str, src: List[str], i: int) -> Tuple[Iinstruction, int]:
-    bracket_idx = line.rindex(')') # throws if the contruct is illformed
-    instruction_txt = line[3:bracket_idx]
+    def _handle_instruction(self, line: str, idx:int) -> Tuple[Iinstruction, int]:
+        if line.startswith("while("):
+            logging.debug("Found while construct in line: %i", idx+1)
+            return self._handle_while(line, idx)
+        
+        elif line.startswith("if("):
+            logging.debug("Found if construct in line: %i", idx+1)
+            return self._handle_if(line, idx)
 
-    brace_offset = get_scope_start_offset(src, i)
-    true_instructions, i = get_instructions_in_scope(src, i+brace_offset)
+        elif line.startswith("do"):
+            logging.debug("Found do-while construct in line: %i", idx+1)
+            return self._handle_do_while(line, idx)
 
-    false_instructions = None
+        elif self.variable_pattern.match(line):
+            logging.debug("Found variable in line %i", idx+1)
+            return self._handle_variable(line, idx)
 
-    #check for else statements
-    if check_line_start(src, i, "}else"):
-        logging.debug("found else construct in line: %i", i+1)
-        false_instructions, i = handle_else(src, i)
-    
-    elif check_line_start(src, i+1, "else"):
-        logging.debug("found else construct in line: %i", i+2)
-        false_instructions, i = handle_else(src, i+1)
-    
-    return if_instruction(instruction_txt, true_instructions, false_instructions), i
+        else:
+            logging.debug("Found generic instruction in line %i", idx+1)
+            return generic_instruction(line), idx
 
-def handle_do_while(line: str, src: List[str], i: int) -> Tuple[Iinstruction, int]:
-    brace_offset = get_scope_start_offset(src, i)
-    child_instructions, i = get_instructions_in_scope(src, i+brace_offset)
-
-    instruction_txt = None
-    if check_line_start(src, i, "}while("):
-        end_line = src[i]
-        bracket_index = end_line.rindex(')') #throws if contruct id ill-formed
-        instruction_txt = end_line[7:bracket_index]
-    elif check_line_start(src, i+1, "while("):
-        i += 1
-        end_line = src[i]
-        bracket_index = end_line.rindex(')')
-        instruction_txt = end_line[6:bracket_index]
-    else:
-        raise JavaSyntaxError("Ill-formed do-while construct!")
-
-    return while_instruction_back(instruction_txt, child_instructions), i
-
-def handle_variable(line:str, src: List[str], i: int) -> Tuple[Iinstruction, int]:
-    groups = variable_pattern.match(line).groups()
-    var_type = groups[0]
-    var_name = groups[1][:-1]
-    var_value = groups[2]
-    if var_value == "":
-        return generic_instruction(f"declare variable '{var_name}' of type {var_type}"), i
-    return generic_instruction(f"declare variable '{var_name}' of type {var_type} with value {var_value}"), i
-
-
-def handle_instruction(line: str, src: List[str], i: int) -> Tuple[Iinstruction, int]:
-    if line.startswith("while("):
-        logging.debug("Found while construct in line: %i", i+1)
-        return handle_while(line, src, i)
-    
-    elif line.startswith("if("):
-        logging.debug("Found if construct in line: %i", i+1)
-        return handle_if(line, src, i)
-
-    elif line.startswith("do"):
-        logging.debug("Found do-while construct in line: %i", i+1)
-        return handle_do_while(line, src, i)
-
-    elif variable_pattern.match(line):
-        logging.debug("Found variable in line %i", i+1)
-        print("found variable in line ", i+1)
-        return handle_variable(line, src, i)
-    
-    else:
-        logging.debug("found generic instruction in line %i", i+1)
-        return generic_instruction(line), i
-
-def get_instructions_in_scope(src: List[str], start_idx: int = 0) -> Tuple[List[Iinstruction], int]:
-    outer_scope: List[Iinstruction] = []
-    i = start_idx
-    while i < len(src):
-
-        line = src[i]
-        try:
-            if check_src(src, i, '}'): #We exited this scope, return it
+    def _get_instructions_in_scope(self, idx: int=0) -> Tuple[List[Iinstruction], int]:
+        scope: List[Iinstruction] = []
+        i = idx
+        while i < len(self.lines):
+            line = self.lines[i]
+            
+            if self._check_src(i, '}'):
                 break
 
-            instruction, i = handle_instruction(line, src, i)
-            outer_scope.append(instruction)
-            
-        except Exception as e:
-            logging.error("Encountered error in line %i: %s", i+1, str(e))
-            raise
-        except:
-            logging.fatal("Encountered unexpected error in line: %i", i+1)
-            raise
-        i += 1
+            instruction, i = self._handle_instruction(line, i)
+            scope.append(instruction)
+
+            i += 1
+        return scope, i
     
-    return outer_scope, i
+    def _remove_keywords(self):
+        self.src = self.src.replace(' ', '')
+        self.src = self.comment_pattern.sub('', self.src)
+        self.src = self.remove_pattern.sub('', self.src)
+        self.lines = self.src.splitlines(True)
 
-def get_function_info(match: Match[str], line: str) -> Tuple[str, str, str]:
-    groups = match.groups()
-    ftype = groups[0]
-    fargs = groups[1]
-    fname = line.removeprefix(ftype).removesuffix(fargs) #remove return type and argument list to get the function name
-    return ftype, fname, fargs
+    def _get_function_info(self, match: Match[str]) -> Tuple[str, str, List[str]]:
+        groups = match.groupdict()
+        ftype = groups["return_type"]
+        fargs = groups["args"]
+        fname = groups["name"]
+        return ftype, fname, fargs.split(',')
 
-def get_function_scopes(src: List[str]) -> List[Function_scope]:
-    functions = []
+    def _get_function_instructions(self, function_header: str) -> List[Iinstruction]:
+        idx = self.lines.index(function_header)
+        brace_offset = self._get_scope_start_offset(idx)
+        return self._get_instructions_in_scope(idx+brace_offset)[0]
 
-    i = 0
-    while i < len(src):
-        line = src[i]
-        if match:=function_pattern.match(line):
-            function_return_type, function_name, function_args = get_function_info(match, line)
+    def _get_function_scopes(self) -> List[Function_scope]:
+        scopes = []
+        for match in self.function_pattern.finditer(self.src):
+            span = match.span()
+            header = self.src[span[0]:span[1]]
 
-            brace_offset = get_scope_start_offset(src, i)
-            child_instructions, i = get_instructions_in_scope(src, i+brace_offset)
-            functions.append(Function_scope(child_instructions, function_name, function_return_type, function_args.split(',')))
-        i+=1
-    return functions
+            rtype, name, args = self._get_function_info(match)
 
-def remove_keywords(src: str) -> str:
-    without_comments = ""
-    for line in src.splitlines(True):
-        if line != "\n" and not COMMENT_PATTERN.match(line):
-            without_comments += line
-    return remove_pattern.sub('', without_comments)
+            child_instructions = self._get_function_instructions(header)
 
-def load_scoped_instructions(filepath: str) -> List[Function_scope]:
-    src = load_src(filepath)
-    return get_function_scopes(src)
+            scopes.append(Function_scope(child_instructions, name, rtype, args))
+        return scopes
+
+
+    def load_instruction_scopes(self):
+        self._remove_keywords()
+        return self._get_function_scopes()
+
+        
+# COMMENT_PATTERN = re.compile(r"""^(//|#|--)""")
+# REMOVE_KEYWORDS = [' ', "public", "private", "final", "protected"]
+# VARIABLE_TAGS = ["byte", "short", "int", "long", "float", "double", "boolean", "char", "String"]
+# FUNCTION_IDENTIFIERS = ["void"]
+# FUNCTION_IDENTIFIERS.extend(VARIABLE_TAGS)
+
+# WHILE_TAG = "solange " #german for 'while'. Change this depending on your language
+
+# REPLACE = dict((re.escape(k), '') for k in REMOVE_KEYWORDS)
+# remove_pattern = re.compile("|".join(REPLACE.keys()), flags=re.MULTILINE)
+
+# variable_regex = "^("
+# for kw in VARIABLE_TAGS:
+#     variable_regex += fr"""{kw}|"""
+# variable_pattern = re.compile(variable_regex[0:-1]+")(.*=|.*;)(.*)", flags=re.MULTILINE)
+
+# function_regex = "^(?P<return_type>"
+# for kw in FUNCTION_IDENTIFIERS:
+#     function_regex += fr"""{kw}|"""
+# function_regex = function_regex[0:-1]+ ")(?P<name>.*)[(](?P<args>.*)[)][^;]"
+
+# function_pattern = re.compile(function_regex, flags=re.MULTILINE)
+
+# def check_src(src: List[str], line_index: int, tag: str) -> bool:
+#     if line_index >= len(src):
+#         return False
+#     return tag in src[line_index]
+
+# def check_line_start(src: List[str], line_index: int, tag: str) -> bool:
+#     if line_index >= len(src):
+#         return False
+#     return src[line_index].startswith(tag)
+
+# def get_next_occurence_of(src: List[str], start_idx:int, tag:str) -> int:
+#     """Returns the index of the next occurence of tag in src from start_idx"""
+#     i = start_idx
+#     while i < len(src):
+#         if check_src(src, i, tag):
+#             break
+#         i += 1
+#     return i
+
+# def get_scope_start_offset(src: List[str], i: int) -> int:
+#     if check_src(src, i, "{"):
+#         return 1
+#     elif check_src(src, i+1, "{"):
+#         return 2
+#     raise ScopeNotFoundException("Unable to find scope start. Is the program ill-formed?")
+
+
+# def get_function_scopes(src: str) -> List[Function_scope]:
+#     scopes = []
+#     lines = src.splitlines(True)
+#     for match in function_pattern.finditer(src):
+
+#         span = match.span()
+#         matched_str = src[span[0]:span[1]]
+
+#         rtype, name, args = get_function_info(match, matched_str)
+
+#         child_instructions = get_function_instructions(lines, matched_str)
+
+#         scopes.append(Function_scope(child_instructions, name, rtype, args))
+#     return scopes

res/input/input_parsed.java

@@ -1,8 +1,11 @@
-//except for this line, this is what interpret_source.load_src returns
 importgreenfoot.*;//(World,Actor,GreenfootImage,GreenfootandMouseInfo)
 classRoverextendsActor
 {
 Displayanzeige;
+/**
+*thisfunctionistobeimplementedbytheuser
+*dependingontheneededactions
+*/
 voidact()
 {
 S66Nr3(7);
@@ -81,6 +84,11 @@ fahreZeileDreheRunter(true);
 }
 fahreZeileDreheRunter(false);
 }
+/**
+*DerRoverbewegtsicheinFeldinFahrtrichtungweiter.
+*SolltesichinFahrtrichtungeinObjektderKlasseHuegelbefindenoderersichanderGrenzederWeltbefinden,
+*dannerscheinteineentsprechendeMeldungaufdemDisplay.
+*/
 voidfahre()
 {
 intposX=getX();
@@ -103,6 +111,10 @@ if(posX==getX()&&posY==getY()&&!huegelVorhanden("vorne"))
 nachricht("Ichkannmichnichtbewegen");
 }
 }
+/**
+*DerRoverdrehtsichum90GradindieRichtung,diemitrichtung(ￜlinksￜoderￜrechtsￜ)ￜbergebenwurde.
+*SollteeinandererText(String)als"rechts"oder"links"ï¿œbergebenwerden,dannerscheinteineentsprechendeMeldungaufdemDisplay.
+*/
 voiddrehe(Stringrichtung)
 {
 if(richtung.equals("rechts")){
@@ -115,6 +127,10 @@ setRotation(getRotation()+180);
 nachricht("KeinenKorrekteRichtunggegeben!");
 }
 }
+/**
+*DerRovergibtdurcheinenWahrheitswert(trueoderfalse)zurï¿œck,obsichaufseinerPositioneinObjektderKlasseGesteinbefindet.
+*EineentsprechendeMeldungerscheintauchaufdemDisplay.
+*/
 booleangesteinVorhanden()
 {
 if(getOneIntersectingObject(Gestein.class)!=null)
@@ -124,6 +140,11 @@ returntrue;
 }
 returnfalse;
 }
+/**
+*DerRoverï¿œberprï¿œft,obsichinrichtung("rechts","links",oder"vorne")einObjektderKlasseHuegelbefindet.
+*DasErgebniswirdaufdemDisplayangezeigt.
+*SollteeinandererText(String)als"rechts","links"oder"vorne"ï¿œbergebenwerden,dannerscheinteineentsprechendeMeldungaufdemDisplay.
+*/
 booleanhuegelVorhanden(Stringrichtung)
 {
 introt=getRotation();
@@ -161,6 +182,10 @@ nachricht("Befehlnichtkorrekt!");
 }
 returnfalse;
 }
+/**
+*DerRoverermitteltdenWassergehaltdesGesteinsaufseinerPositionundgibtdiesenaufdemDisplayaus.
+*SolltekeinObjektderKlasseGesteinvorhandensein,dannerscheinteineentsprechendeMeldungaufdemDisplay.
+*/
 voidanalysiereGestein()
 {
 if(gesteinVorhanden())
@@ -174,10 +199,17 @@ else
 nachricht("HieristkeinGestein");
 }
 }
+/**
+*DerRovererzeugteinObjektderKlasseï¿œMarkierungï¿œaufseinerPosition.
+*/
 voidsetzeMarke()
 {
 getWorld().addObject(newMarke(),getX(),getY());
 }
+/**
+**DerRovergibtdurcheinenWahrheitswert(trueoderfalse)zurï¿œck,obsichaufseinerPositioneinObjektderMarkebefindet.
+*EineentsprechendeMeldungerscheintauchaufdemDisplay.
+*/
 booleanmarkeVorhanden()
 {
 if(getOneIntersectingObject(Marke.class)!=null)
@@ -206,7 +238,7 @@ voiddisplayAusschalten()
 {
 getWorld().removeObject(anzeige);
 }
-protectedvoidaddedToWorld(Worldworld)
+voidaddedToWorld(Worldworld)
 {
 setImage("images/rover.png");
 world=getWorld();