feat: mecha error handling

Author: TheNuclearNexus

packages/mecha/src/mecha/api.py

@@ -54,7 +54,7 @@
 from pydantic import BaseModel, field_validator
 from tokenstream import InvalidSyntax, Preprocessor, TokenStream, set_location
 
-from .ast import AstLiteral, AstNode, AstRoot
+from .ast import AstError, AstLiteral, AstNode, AstRoot
 from .config import CommandTree
 from .database import (
     CompilationDatabase,
@@ -396,7 +396,7 @@ def parse(
                 filename=str(filename) if filename else None,
                 file=source,
             )
-            raise DiagnosticError(DiagnosticCollection([set_location(d, exc)])) from exc
+            diagnostics.append(set_location(d, exc))
         else:
             if self.cache and filename and cache_miss:
                 try:
@@ -407,6 +407,24 @@ def parse(
                     pass
             return ast
 
+        if len(diagnostics) > 0:
+            if self.cache and filename and cache_miss:
+                self.cache.invalidate_changes(self.directory / filename)
+
+            raise DiagnosticError(DiagnosticCollection(diagnostics))
+
+    def parse_stream(
+        self,
+        multiline: bool | None,
+        provide: JsonDict | None,
+        parser: str,
+        stream: TokenStream,
+    ):
+        with self.prepare_token_stream(stream, multiline=multiline):
+            with stream.provide(**provide or {}):
+                ast = delegate(parser, stream)
+        return ast
+
     @overload
     def compile(
         self,

packages/mecha/src/mecha/ast.py

@@ -4,6 +4,7 @@
     "AstNode",
     "AstChildren",
     "AstRoot",
+    "AstError",
     "AstCommand",
     "AstCommandSentinel",
     "AstString",
@@ -119,7 +120,9 @@
 from nbtlib import Byte, ByteArray, Compound, CompoundMatch, Double, Int, IntArray
 from nbtlib import List as ListTag
 from nbtlib import ListIndex, LongArray, NamedKey, Numeric, Path, String
-from tokenstream import UNKNOWN_LOCATION, SourceLocation, set_location
+from tokenstream import UNKNOWN_LOCATION, InvalidSyntax, SourceLocation, set_location
+
+from mecha.error import MechaError
 
 from .utils import string_to_number
 
@@ -266,11 +269,18 @@ class AstChildren(AbstractChildren[AstNodeType]):
 class AstRoot(AstNode):
     """Ast root node."""
 
-    commands: AstChildren["AstCommand"] = required_field()
+    commands: AstChildren["AstCommand|AstError"] = required_field()
 
     parser = "root"
 
 
+@dataclass(frozen=True, slots=True)
+class AstError(AstNode):
+    """Ast error node."""
+
+    error: InvalidSyntax = required_field()
+
+
 @dataclass(frozen=True, slots=True)
 class AstCommand(AstNode):
     """Ast command node."""

packages/mecha/src/mecha/contrib/bake_macros.py

@@ -31,6 +31,7 @@
 from mecha import (
     AstChildren,
     AstCommand,
+    AstError,
     AstMacroLine,
     AstMacroLineText,
     AstMacroLineVariable,
@@ -135,10 +136,14 @@ class MacroBaker(Visitor):
     def bake_macros(self, node: AstRoot):
         invocation = self.invocations.get(self.database.current)
 
-        result: List[AstCommand] = []
+        result: List[AstCommand|AstError] = []
         modified = False
 
         for command in node.commands:
+            if isinstance(command, AstError):
+                result.append(command)
+                continue
+            
             if isinstance(command, AstMacroLine) and invocation:
                 baked_macro_line = self.bake_macro_line(command, invocation)
                 modified |= baked_macro_line is not command

packages/mecha/src/mecha/contrib/inline_function_tag.py

@@ -17,6 +17,7 @@
 from mecha import (
     AstChildren,
     AstCommand,
+    AstError,
     AstResourceLocation,
     AstRoot,
     CommandTree,
@@ -66,9 +67,13 @@ class InlineFunctionTagHandler(Visitor):
     @rule(AstRoot)
     def inline_function_tag(self, node: AstRoot):
         changed = False
-        commands: List[AstCommand] = []
+        commands: List[AstCommand|AstError] = []
 
         for command in node.commands:
+            if isinstance(command, AstError):
+                commands.append(command)
+                continue
+
             if command.identifier == "function:tag:name" and isinstance(
                 tag := command.arguments[0], AstResourceLocation
             ):

packages/mecha/src/mecha/contrib/nested_resources.py

@@ -29,6 +29,7 @@
 from mecha import (
     AstChildren,
     AstCommand,
+    AstError,
     AstJson,
     AstNode,
     AstResourceLocation,
@@ -217,9 +218,13 @@ class NestedResourcesTransformer(MutatingReducer):
     @rule(AstRoot)
     def nested_resources(self, node: AstRoot):
         changed = False
-        commands: List[AstCommand] = []
+        commands: List[AstCommand|AstError] = []
 
         for command in node.commands:
+            if isinstance(command, AstError):
+                commands.append(command)
+                continue
+
             if file_type := self.nested_resource_identifiers.get(command.identifier):
                 name, content = command.arguments
 

packages/mecha/src/mecha/contrib/nesting.py

@@ -22,6 +22,7 @@
 from mecha import (
     AstChildren,
     AstCommand,
+    AstError,
     AstResourceLocation,
     AstRoot,
     CommandTree,
@@ -34,6 +35,7 @@
     rule,
 )
 from mecha.contrib.nested_location import NestedLocationResolver
+from mecha.parse import parse_root_item
 
 
 class NestingOptions(BaseModel):
@@ -81,8 +83,8 @@ def parse_nested_root(stream: TokenStream) -> AstRoot:
 
     level, command_level = stream.indentation[-2:]
 
-    commands: List[AstCommand] = []
-
+    errors: list[InvalidSyntax] = []
+    commands: List[AstCommand|AstError] = []
     with (
         stream.intercept("newline"),
         stream.provide(
@@ -91,7 +93,11 @@ def parse_nested_root(stream: TokenStream) -> AstRoot:
         ),
     ):
         while True:
-            commands.append(delegate("root_item", stream))
+            
+            result = parse_root_item(stream, errors, colon=True)
+            
+            if result is not None:
+                commands.append(result)
 
             # The command parser consumes the trailing newline so we need to rewind
             # to be able to use "consume_line_continuation()".
@@ -101,7 +107,6 @@ def parse_nested_root(stream: TokenStream) -> AstRoot:
             with stream.provide(multiline=True, line_indentation=level):
                 if not consume_line_continuation(stream):
                     break
-
     node = AstRoot(commands=AstChildren(commands))
     return set_location(node, commands[0], commands[-1])
 
@@ -179,6 +184,9 @@ def nesting_execute_commands(self, node: AstCommand):
 
         single_command = None
         for command in root.commands:
+            if isinstance(command, AstError):
+                continue
+
             if command.compile_hints.get("skip_execute_inline_single_command"):
                 continue
             if single_command is None:
@@ -270,12 +278,20 @@ def nesting_schedule_function(self, node: AstCommand):
     @rule(AstRoot)
     def nesting(self, node: AstRoot):
         changed = False
-        commands: List[AstCommand] = []
+        commands: List[AstCommand|AstError] = []
 
         for command in node.commands:
+            if isinstance(command, AstError):
+                commands.append(command)
+                continue
+
             if command.identifier in self.identifier_map:
                 result = yield from self.handle_function(command, top_level=True)
                 commands.extend(result)
+
+                if len(result) == 1 and result[0] is command:
+                    continue
+
                 changed = True
                 continue
 
@@ -294,6 +310,10 @@ def nesting(self, node: AstRoot):
             if expand:
                 changed = True
                 for nested_command in cast(AstRoot, expand.arguments[0]).commands:
+                    if isinstance(nested_command, AstError):
+                        commands.append(nested_command)
+                        continue
+                    
                     if nested_command.identifier == "execute:subcommand":
                         expansion = cast(AstCommand, nested_command.arguments[0])
                     else:
@@ -321,7 +341,7 @@ def handle_function(
         self,
         node: AstCommand,
         top_level: bool = False,
-    ) -> Generator[Diagnostic, None, List[AstCommand]]:
+    ) -> Generator[Diagnostic, None, List[AstCommand|AstError]]:
         name, *args, root = node.arguments
 
         if isinstance(name, AstResourceLocation) and isinstance(root, AstRoot):

packages/mecha/src/mecha/contrib/statistics.py

@@ -19,6 +19,7 @@
 from mecha import (
     AstCommand,
     AstCommandSentinel,
+    AstError,
     AstMacroLine,
     AstMacroLineText,
     AstMacroLineVariable,
@@ -110,9 +111,9 @@ def root(self, node: AstRoot):
         self.stats.function_count += 1
 
         for command in node.commands:
-            if isinstance(command, AstCommandSentinel):
+            if isinstance(command, (AstCommandSentinel, AstError)):
                 continue
-
+            
             behind_execute = False
 
             while command.identifier.startswith("execute"):

packages/mecha/src/mecha/parse.py

@@ -91,7 +91,15 @@
 from beet import LATEST_MINECRAFT_VERSION
 from beet.core.utils import VersionNumber, split_version
 from nbtlib import Byte, Double, Float, Int, Long, OutOfRange, Short, String
-from tokenstream import InvalidSyntax, SourceLocation, TokenStream, set_location
+from tokenstream import (
+    InvalidSyntax,
+    SourceLocation,
+    SyntaxRules,
+    TokenStream,
+    UnexpectedToken,
+    set_location,
+    Token,
+)
 
 from .ast import (
     AstAdvancementPredicate,
@@ -108,6 +116,7 @@
     AstDustColorTransitionParticleParameters,
     AstDustParticleParameters,
     AstEntityAnchor,
+    AstError,
     AstFallingDustParticleParameters,
     AstGamemode,
     AstGreedy,
@@ -638,6 +647,13 @@ def __init__(self, parser: str):
         self.parser = parser
 
 
+class InvalidSyntaxCollection(MechaError):
+    errors: list[InvalidSyntax]
+
+    def __init__(self, errors: list[InvalidSyntax]):
+        self.errors = errors
+
+
 @overload
 def delegate(parser: str) -> Parser: ...
 
@@ -681,6 +697,7 @@ def consume_line_continuation(stream: TokenStream) -> bool:
     return False
 
 
+# TODO: Attempt to move error recovery into AstCommand's parser
 def parse_root(stream: TokenStream) -> AstRoot:
     """Parse root."""
     start = stream.peek()
@@ -689,20 +706,59 @@ def parse_root(stream: TokenStream) -> AstRoot:
         node = AstRoot(commands=AstChildren[AstCommand]())
         return set_location(node, SourceLocation(0, 1, 1))
 
-    commands: List[AstCommand] = []
+    errors: List[InvalidSyntax] = []
+    commands: List[AstCommand | AstError] = []
 
     with stream.ignore("comment"):
         for _ in stream.peek_until():
             if stream.get("newline"):
                 continue
             if stream.get("eof"):
                 break
-            commands.append(delegate("root_item", stream))
 
-    node = AstRoot(commands=AstChildren(commands))
-    return set_location(node, start, stream.current)
+            result = parse_root_item(stream, errors)
+            if result is not None:
+                commands.append(result)
+
+    children = AstChildren(commands)
+
+    node = AstRoot(commands=children)
+
+    if stream.index < 0:
+        end_location = SourceLocation(1, 0, 0)
+    else:
+        end_location = stream.current.end_location
 
+    return set_location(node, start, end_location)
 
+def consume_error(stream: TokenStream, errors: list[InvalidSyntax]):
+    next = stream.peek()
+    with stream.syntax(unknown=r"."):
+        while next := stream.peek():
+            stream.expect()
+            if next.value == "\n":
+                break
+    node = AstError(errors[-1].location, errors[-1].end_location, errors[-1])
+    return node
+
+
+
+def parse_root_item(stream: TokenStream, errors: list[InvalidSyntax], colon: bool = False):
+
+    with stream.checkpoint() as commit:
+        try:
+            command: AstCommand = delegate("root_item", stream)
+            commit()
+            return command
+        except InvalidSyntax as exc:
+            errors.append(exc)
+
+    if commit.rollback:
+        if colon:
+            with stream.syntax(colon=r":"):
+                return consume_error(stream, errors)
+        return consume_error(stream, errors)
+        
 def parse_command(stream: TokenStream) -> AstCommand:
     """Parse command."""
     spec = get_stream_spec(stream)