Cleanup enum variants in tuple form with multiple fields

[tslil-topos]

Low priority

I put together a quick python script (source at end) and pointed this at the repo, here's the output. Some of these will require human judgement and we could imagine increasing the threshold from 2 to 3 fields etc.

TupleVariant(file='catlog/src/dbl/model.rs',
             enum='InvalidDblModel',
             variant='Eqn',
             fields=['Option<usize>', 'NonEmpty<InvalidModelEqn>'])


TupleVariant(file='catlog/src/dbl/theory.rs',
             enum='InvalidDblTheory',
             variant='MorTypeEq',
             fields=['usize', 'NonEmpty<InvalidPathEq>'])


TupleVariant(file='catlog/src/dbl/theory.rs',
             enum='InvalidDblTheory',
             variant='ObOpEq',
             fields=['usize', 'NonEmpty<InvalidPathEq>'])


TupleVariant(file='catlog/src/one/fp_category.rs',
             enum='InvalidFpCategory',
             variant='Eqn',
             fields=['usize', 'NonEmpty<InvalidPathEq>'])


TupleVariant(file='catlog/src/tt/stx.rs',
             enum='TyS_',
             variant='Morphism',
             fields=['MorType', 'TmS', 'TmS'])


TupleVariant(file='catlog/src/tt/stx.rs',
             enum='TyS_',
             variant='Sing',
             fields=['TyS', 'TmS'])


TupleVariant(file='catlog/src/tt/stx.rs',
             enum='TyS_',
             variant='Id',
             fields=['TyS', 'TmS', 'TmS'])


TupleVariant(file='catlog/src/tt/stx.rs',
             enum='TyS_',
             variant='Specialize',
             fields=['TyS', 'Vec<(Vec<(FieldName, LabelSegment)>, TyS)>'])


TupleVariant(file='catlog/src/tt/stx.rs',
             enum='TmS_',
             variant='TopApp',
             fields=['TopVarName', 'Vec<TmS>'])


TupleVariant(file='catlog/src/tt/stx.rs',
             enum='TmS_',
             variant='Var',
             fields=['BwdIdx', 'VarName', 'LabelSegment'])


TupleVariant(file='catlog/src/tt/stx.rs',
             enum='TmS_',
             variant='Proj',
             fields=['TmS', 'FieldName', 'LabelSegment'])


TupleVariant(file='catlog/src/tt/stx.rs',
             enum='TmS_',
             variant='Compose',
             fields=['TmS', 'TmS'])


TupleVariant(file='catlog/src/tt/stx.rs',
             enum='TmS_',
             variant='ObApp',
             fields=['VarName', 'TmS'])


TupleVariant(file='catlog/src/tt/val.rs',
             enum='TyV_',
             variant='Morphism',
             fields=['MorType', 'TmV', 'TmV'])


TupleVariant(file='catlog/src/tt/val.rs',
             enum='TyV_',
             variant='Sing',
             fields=['TyV', 'TmV'])


TupleVariant(file='catlog/src/tt/val.rs',
             enum='TyV_',
             variant='Id',
             fields=['TyV', 'TmV', 'TmV'])


TupleVariant(file='catlog/src/tt/val.rs',
             enum='TmN_',
             variant='Var',
             fields=['FwdIdx', 'VarName', 'LabelSegment'])


TupleVariant(file='catlog/src/tt/val.rs',
             enum='TmN_',
             variant='Proj',
             fields=['TmN', 'FieldName', 'LabelSegment'])


TupleVariant(file='catlog/src/tt/val.rs',
             enum='TmV_',
             variant='Neu',
             fields=['TmN', 'TyV'])


TupleVariant(file='catlog/src/tt/val.rs',
             enum='TmV_',
             variant='App',
             fields=['VarName', 'TmV'])


TupleVariant(file='catlog/src/tt/val.rs',
             enum='TmV_',
             variant='Compose',
             fields=['TmV', 'TmV'])


TupleVariant(file='catlog/src/tt/text_elab.rs',
             enum='TopElabResult',
             variant='Declaration',
             fields=['TopVarName', 'TopDecl'])


TupleVariant(file='catlog/src/dbl/modal/model.rs',
             enum='ModalOb',
             variant='App',
             fields=['Box<Self>', 'QualifiedName'])


TupleVariant(file='catlog/src/dbl/modal/model.rs',
             enum='ModalOb',
             variant='List',
             fields=['List', 'Vec<Self>'])


TupleVariant(file='catlog/src/dbl/modal/model.rs',
             enum='ModalMor',
             variant='App',
             fields=['Box<Path<ModalOb, Self>>', 'QualifiedName'])


TupleVariant(file='catlog/src/dbl/modal/model.rs',
             enum='ModalMor',
             variant='HomApp',
             fields=['Box<Path<ModalOb, Self>>', 'QualifiedName'])


TupleVariant(file='catlog/src/dbl/modal/model.rs',
             enum='ModalMor',
             variant='List',
             fields=['MorListData', 'Vec<Self>'])


TupleVariant(file='catlog/src/dbl/modal/theory.rs',
             enum='ModalOp',
             variant='Concat',
             fields=['List', 'usize', 'ModalObType'])

Script

#!/usr/bin/env -S uv run --script
#
# /// script
# requires-python = ">=3.13"
# dependencies = ["tree-sitter", "tree-sitter-rust"]
# ///

import sys
from pathlib import Path
from dataclasses import dataclass
from pprint import pp

import tree_sitter_rust
from tree_sitter import Language, Parser

RUST_LANGUAGE = Language(tree_sitter_rust.language())
parser = Parser(RUST_LANGUAGE)


@dataclass
class TupleVariant:
    file: str
    enum: str
    variant: str
    fields: list[str]


def find_enum_variants_with_fields(
    repo_root: Path, file_path: Path
) -> list[TupleVariant]:
    with file_path.open("rb") as fh:
        tree = parser.parse(fh.read())

    file = file_path.relative_to(repo_root).as_posix()

    enums = []
    for node in tree.root_node.children:
        if node.type == "enum_item":
            enum_name = None
            for child in node.children:
                if child.type == "type_identifier":
                    enum_name = child.text.decode("utf-8")
                elif enum_name and child.type == "enum_variant_list":
                    for variant in child.children:
                        if variant.type == "enum_variant":
                            variant_name = None
                            fields = []
                            for v_child in variant.children:
                                if v_child.type == "identifier":
                                    variant_name = v_child.text.decode("utf-8")
                                elif v_child.type == "ordered_field_declaration_list":
                                    for f in v_child.children:
                                        if "type" in f.type:
                                            fields.append(f.text.decode("utf-8"))
                            if len(fields) > 1 and variant_name:
                                enums.append(
                                    TupleVariant(file, enum_name, variant_name, fields)
                                )

    return enums


def find_all(repo_root: Path) -> list[TupleVariant]:
    rust_files = repo_root.glob("**/*.rs")
    return [
        tv
        for rust_file in rust_files
        for tv in find_enum_variants_with_fields(repo_root, rust_file)
    ]


def go():
    if len(sys.argv) != 2:
        print(f"Usage: {sys.argv[0]} /path/to/repo/")
        sys.exit(1)
    print("```")
    for tv in find_all(Path(sys.argv[1])):
        pp(tv)
        print("\n")
    print("```")


if __name__ == "__main__":
    go()