Add label extractor to preprocessor

Author: object-Object

.gitignore

@@ -12,6 +12,7 @@ debug.json
 /src/debugger.mlog
 /src/config/*.mlog
 /src/peripherals/scrolling_display.mlog
+*.labels.mlog
 
 node_modules/
 

Makefile

@@ -54,6 +54,12 @@ build/%.o: asm/%.s | build
 src/%.mlog: src/%.mlog.jinja
 	python -m mlogv32.preprocessor file -o src/$*.mlog src/$*.mlog.jinja
 
+src/init.mlog: src/main.labels.mlog
+
+# only extract uppercase labels
+src/main.labels.mlog: src/main.mlog
+	python -m mlogv32.preprocessor labels --filter '[^a-z]+' -o src/main.labels.mlog src/main.mlog
+
 .PHONY: mlog-configs
 mlog-configs: src/config/configs.yaml
 	python -m mlogv32.preprocessor configs src/config/configs.yaml

pyproject.toml

@@ -8,6 +8,7 @@ version = "0.1.0"
 requires-python = ">=3.12"
 dependencies = [
     "jinja2>=3.1.6",
+    "lark[interegular]>=1.2.2",
     "pydantic>=2.11.5",
     "pymsch>=0.0.11",
     "pyperclip>=1.9.0",

python/src/mlogv32/preprocessor/app.py

@@ -1,21 +1,22 @@
+import re
 from pathlib import Path
 from typing import Annotated, Any, TypedDict
 
 import yaml
 from jinja2 import Environment, FileSystemLoader, StrictUndefined
 from typer import Option, Typer
 
-from mlogv32.preprocessor.extensions import CommentStatement
-
+from .extensions import CommentStatement
 from .filters import FILTERS
+from .parser.mlog import Label, Statement, parse_mlog
 
 app = Typer(
     pretty_exceptions_show_locals=False,
 )
 
 
 @app.command(name="file")
-def command_file(
+def file_command(
     path: Path,
     output: Annotated[Path | None, Option("-o", "--output")] = None,
 ):
@@ -32,6 +33,37 @@ def command_file(
         print(result)
 
 
+@app.command()
+def labels(
+    path: Path,
+    output: Annotated[Path | None, Option("-o", "--output")] = None,
+    filter_str: Annotated[str | None, Option("--filter")] = None,
+):
+    if filter_str:
+        filter_re = re.compile(filter_str)
+    else:
+        filter_re = None
+
+    with path.open("r", encoding="utf-8") as f:
+        lines = parse_mlog(f.read())
+
+    result_lines = list[str]()
+    line_num = 0
+    for line in lines:
+        match line:
+            case Label(name=name):
+                if not filter_re or filter_re.fullmatch(name):
+                    result_lines.append(f"set {name} {line_num}")
+            case Statement():
+                line_num += 1
+    result = "\n".join(result_lines)
+
+    if output:
+        output.write_text(result, encoding="utf-8")
+    else:
+        print(result)
+
+
 @app.command()
 def configs(yaml_path: Path):
     with yaml_path.open("rb") as f:

python/src/mlogv32/preprocessor/parser/__init__.py

@@ -0,0 +1,7 @@
+__all__ = [
+    "Label",
+    "Statement",
+    "parse_mlog",
+]
+
+from .mlog import Label, Statement, parse_mlog

python/src/mlogv32/preprocessor/parser/mlog.lark

@@ -0,0 +1,19 @@
+start: (line? COMMENT? _NEWLINE)*
+
+?line: LABEL -> label
+     | statement
+
+statement: TOKEN (TOKEN | LABEL | STRING)*
+
+STRING.2: /"[^\n]*"/
+
+LABEL.1: /[^\n #\t;]+:/
+
+TOKEN: /[^\n #\t;]+/
+
+COMMENT: /#[^\n]*/
+
+_NEWLINE: /[\n;]+/
+
+%ignore /[ \t]+/
+%ignore COMMENT

python/src/mlogv32/preprocessor/parser/mlog.py

@@ -0,0 +1,56 @@
+from dataclasses import dataclass
+from importlib import resources
+from typing import TYPE_CHECKING, Callable
+
+from lark import Lark, Token, Transformer
+
+if TYPE_CHECKING:
+
+    def v_args[T](
+        inline: bool = False,
+        meta: bool = False,
+        tree: bool = False,
+    ) -> Callable[[T], T]: ...
+else:
+    from lark import v_args
+
+
+@dataclass
+class Label:
+    name: str
+
+
+@dataclass
+class Statement:
+    name: str
+    args: list[str]
+
+
+@dataclass
+@v_args(inline=True)
+class MlogTransformer(Transformer[Token, list[Label | Statement]]):
+    def start(self, *children: Label | Statement):
+        return list(children)
+
+    def label(self, name: str):
+        return Label(name[:-1])
+
+    def statement(self, name: str, *args: str):
+        return Statement(name, list(args))
+
+    def TOKEN(self, token: Token):
+        return str(token)
+
+
+GRAMMAR = (resources.files() / "mlog.lark").read_text("utf-8")
+
+PARSER = Lark(
+    GRAMMAR,
+    parser="lalr",
+    strict=True,
+)
+
+
+def parse_mlog(text: str):
+    tree = PARSER.parse(text)
+    return MlogTransformer().transform(tree)

src/main.mlog.jinja

@@ -273,7 +273,7 @@ done_pause_step:
     jump SLLI always
     jump SRLI always
     jump SRAI always
-    jump illegal_instruction always # null + 64 would jump here
+    jump ILLEGAL_INSTRUCTION always # null + 64 would jump here
     jump MLOGSYS always
     jump MLOGDRAW always
 
@@ -332,7 +332,7 @@ default_mtvec_handler:
     printflush {{ERROR_OUTPUT}}
     jump halt always
 
-illegal_instruction:
+ILLEGAL_INSTRUCTION:
     set mcause 2
     # TODO: return faulting instruction bits?
     # continue into trap_without_mtval
@@ -1734,7 +1734,7 @@ PRIV:
     jump EBREAK equal imm 1
     jump MRET equal imm 0b001100000010
     jump end_instruction equal imm 0b000100000101 # WFI (no-op, ignore TW bit)
-    jump illegal_instruction always
+    jump ILLEGAL_INSTRUCTION always
 
 ECALL:
     # 0b1000 (8) = ecall from U-mode
@@ -1748,7 +1748,7 @@ EBREAK:
     jump trap always
 
 MRET:
-    jump illegal_instruction lessThan privilege_mode 0b11
+    jump ILLEGAL_INSTRUCTION lessThan privilege_mode 0b11
 
     # set mstatus.MIE to mstatus.MPIE, set mstatus.MPIE to 1, set privilege_mode to mstatus.MPP, and set mstatus.MPP to U
     op and mstatus.mie csr_mstatus 0b10000000
@@ -1839,15 +1839,15 @@ modify_csr:
     # csr[9:8] encodes the lowest privilege level that can access the CSR
     op shr csr_11_8 imm 8
     op and required_privilege csr_11_8 0b1100000000
-    jump illegal_instruction lessThan privilege_mode required_privilege
+    jump ILLEGAL_INSTRUCTION lessThan privilege_mode required_privilege
 
     # disable all supervisor and hypervisor CSRs
-    jump illegal_instruction equal required_privilege 0b01
-    jump illegal_instruction equal required_privilege 0b10
+    jump ILLEGAL_INSTRUCTION equal required_privilege 0b01
+    jump ILLEGAL_INSTRUCTION equal required_privilege 0b10
 
     # 0x7b0-0x7bf are only visible to debug mode
     op shr csr_11_4 imm 4
-    jump illegal_instruction equal csr_11_4 0x7b
+    jump ILLEGAL_INSTRUCTION equal csr_11_4 0x7b
 
     # read
 
@@ -1878,7 +1878,7 @@ modify_csr__conditional_write:
     jump end_instruction_with_rd equal rs1_id 0
 
 modify_csr__always_write:
-    jump illegal_instruction equal readonly true
+    jump ILLEGAL_INSTRUCTION equal readonly true
 
     set @counter modify_csr_write
     # -----------------
@@ -1922,7 +1922,7 @@ modify_csr__read__not_machine:
     jump modify_csr__read__not_timer notEqual tmp 0
 
     # mcounteren is hardwired zero, so now that we know it's a timer, trap if we're not in M-mode
-    jump illegal_instruction lessThan privilege_mode 0b11
+    jump ILLEGAL_INSTRUCTION lessThan privilege_mode 0b11
 
     jump modify_csr__mcycle equal csr_7_0 0x00
     jump modify_csr__time equal imm 0xc01 # mtime doesn't exist
@@ -2066,7 +2066,7 @@ modify_csr__minstreth:
 
 MLOGSYS:
     # I-type: rs1, imm=funct12, rd_id
-    jump illegal_instruction greaterThan imm 3
+    jump ILLEGAL_INSTRUCTION greaterThan imm 3
 
     # TODO: we could save an instruction if we used B-type instead of I-type
     op mul jump imm 2
@@ -2108,7 +2108,7 @@ MLOGSYS__init_icache__loop:
 
 MLOGDRAW:
     # I-type: rs1, imm=funct12, rd_id
-    jump illegal_instruction greaterThan imm 15
+    jump ILLEGAL_INSTRUCTION greaterThan imm 15
 
     read a1 {{REGISTERS}} 11
     read a2 {{REGISTERS}} 12