Files
MetaCrate/tools/generate_api_shims.py

281 lines
12 KiB
Python

#!/usr/bin/env python3
"""Generate callable, failure-only Rust shims from the authoritative catalog."""
from __future__ import annotations
import argparse
import csv
import json
import re
import subprocess
from collections import Counter
from pathlib import Path
import generate_rust_mapping as mapping
ROOT = Path(__file__).resolve().parents[1]
CATALOG = ROOT / "api/public-api.json"
LEDGER = ROOT / "api/RUST-MAPPING.tsv"
COVERAGE = ROOT / "api/SHIM-COVERAGE.md"
TARGETS = {
"LibreMetaverse.Types": ROOT / "crates/libremetaverse-types/src/generated.rs",
"LibreMetaverse.StructuredData": ROOT / "crates/libremetaverse-structured-data/src/generated.rs",
}
UNDERLYING = {
"System.SByte": "i8", "System.Byte": "u8", "System.Int16": "i16", "System.UInt16": "u16",
"System.Int32": "i32", "System.UInt32": "u32", "System.Int64": "i64", "System.UInt64": "u64",
}
PRIVATE_LAYOUTS = {
"T:LibreMetaverse.CacheDictionary`2": [("entries", "std::collections::HashMap<TKey, TValue>")],
"T:LibreMetaverse.DoubleDictionary`3": [
("by_first", "std::collections::HashMap<TKey1, TValue>"),
("by_second", "std::collections::HashMap<TKey2, TValue>"),
],
"T:LibreMetaverse.EmptyRemovalStrategy`1": [("keys", "Vec<TKey>")],
"T:LibreMetaverse.ExpiringCache`2": [("entries", "std::collections::HashMap<TKey, TValue>")],
"T:LibreMetaverse.LruRemovalStrategy`1": [("keys", "Vec<TKey>")],
"T:LibreMetaverse.MruRemovalStrategy`1": [("keys", "Vec<TKey>")],
"T:LibreMetaverse.MultiValueDictionary`2": [("entries", "std::collections::HashMap<TKey, Vec<TValue>>")],
"T:LibreMetaverse.TokenBucket": [("content", "i32"), ("max_burst", "i32"), ("drip_rate", "i32")],
"T:LibreMetaverse.UUID": [("bytes", "[u8; 16]")],
"T:LibreMetaverse.StructuredData.OSDArray": [("values", "Vec<OSD>")],
"T:LibreMetaverse.StructuredData.OSDBinary": [("value", "Vec<u8>")],
"T:LibreMetaverse.StructuredData.OSDBoolean": [("value", "bool")],
"T:LibreMetaverse.StructuredData.OSDDate": [("value", "std::time::SystemTime")],
"T:LibreMetaverse.StructuredData.OSDException": [("message", "String")],
"T:LibreMetaverse.StructuredData.OSDInteger": [("value", "i32")],
"T:LibreMetaverse.StructuredData.OSDMap": [("values", "std::collections::HashMap<String, OSD>")],
"T:LibreMetaverse.StructuredData.OSDReal": [("value", "f64")],
"T:LibreMetaverse.StructuredData.OSDString": [("value", "String")],
"T:LibreMetaverse.StructuredData.OSDUUID": [("value", "libremetaverse_types::UUID")],
"T:LibreMetaverse.StructuredData.OSDUri": [("value", "libremetaverse_types::compat::Uri")],
}
def generic_names(item: dict) -> list[str]:
return [parameter["name"] for parameter in item.get("generic_parameters", [])]
def format_rust(source: str) -> str:
result = subprocess.run(
["rustfmt", "--edition", "2024", "--emit", "stdout"],
input=source,
text=True,
capture_output=True,
check=True,
)
return result.stdout
def type_name(type_row: dict) -> str:
return type_row["rust_path"].rsplit("::", 1)[-1]
def generic_suffix(names: list[str]) -> str:
return "<" + ", ".join(names) + ">" if names else ""
def flags_enum(item: dict) -> bool:
return any(attribute["type"] == "System.FlagsAttribute" for attribute in item.get("attributes", []))
def duplicate_enum_values(item: dict) -> bool:
values = [member["value"]["value"] for member in item["members"] if member["kind"] == "enum_value"]
return len(values) != len(set(values))
def render_enum(item: dict, rust_name: str) -> list[str]:
underlying = UNDERLYING[item["enum_underlying_type"]]
values = [member for member in item["members"] if member["kind"] == "enum_value"]
lines = [f"/// C# type: `{item['doc_id']}`."]
if flags_enum(item) or duplicate_enum_values(item):
lines += [
"#[derive(Clone, Copy, Debug, Eq, Hash, PartialEq)]",
"#[repr(transparent)]",
f"pub struct {rust_name}(pub {underlying});",
f"impl {rust_name} {{",
]
for member in values:
name = mapping.snake(member["name"]).upper()
value = member["value"]["value"]
lines += [f" /// C# member: `{member['doc_id']}`.", f" pub const {name}: Self = Self({value});"]
lines.append("}")
return lines
lines += [
"#[derive(Clone, Copy, Debug, Eq, Hash, PartialEq)]",
f"#[repr({underlying})]",
f"pub enum {rust_name} {{",
]
for member in values:
lines += [f" /// C# member: `{member['doc_id']}`.", f" {mapping.pascal(member['name'])} = {member['value']['value']},"]
lines.append("}")
return lines
def render_body(signature: str, member_id: str, error_model: str, trait: bool) -> str:
if trait:
signature = signature.removeprefix("pub ")
failure = (
f"libremetaverse_types::not_implemented({json.dumps(member_id)})"
if error_model.startswith("Result")
else f"libremetaverse_types::unimplemented_api!({json.dumps(member_id)})"
)
return f" {signature} {{ {failure} }}"
def render_type(item: dict, type_row: dict, member_rows: dict[str, dict[str, str]], mapper: mapping.Mapper) -> str:
rust_name = type_name(type_row)
if item["kind"] == "enum":
return "\n".join(render_enum(item, rust_name))
names = generic_names(item)
suffix = generic_suffix(names)
trait = item["kind"] == "interface"
lines = [f"/// C# type: `{item['doc_id']}`."]
if trait:
lines.append(f"pub trait {rust_name}{suffix} {{")
else:
fields = [member for member in item["members"] if member["kind"] == "field" and not member.get("static")]
private_fields = PRIVATE_LAYOUTS.get(item["doc_id"], [])
if item["doc_id"] == "T:LibreMetaverse.StructuredData.OSD":
lines += [
"#[non_exhaustive]",
"pub enum OSD {",
" Undefined, Boolean(bool), Integer(i32), Real(f64), String(String),",
" UUID(libremetaverse_types::UUID), Date(std::time::SystemTime),",
" Uri(libremetaverse_types::compat::Uri), Binary(Vec<u8>),",
" Array(Vec<OSD>), Map(std::collections::HashMap<String, OSD>), LlsdXml(String),",
"}",
]
elif not fields and not names and not private_fields:
lines.append(f"pub struct {rust_name};")
else:
lines.append(f"pub struct {rust_name}{suffix} {{")
generics = set(names)
for field in fields:
field_type = mapper.type(field["type"], field.get("nullability"), generics)
lines += [f" /// C# member: `{field['doc_id']}`.", f" pub {mapping.snake(field['name'])}: {field_type},"]
for field_name, field_type in private_fields:
lines.append(f" {field_name}: {field_type},")
if names:
tuple_type = "(" + ", ".join(names) + ("," if len(names) == 1 else "") + ")"
lines.append(f" _marker: std::marker::PhantomData<{tuple_type}>,")
lines.append("}")
lines += [f"impl{suffix} {rust_name}{suffix} {{"]
for member in item["members"]:
if member["kind"] in {"enum_value"} or (member["kind"] == "field" and not member.get("static")):
continue
row = member_rows[member["doc_id"]]
lines.append(f" /// C# member: `{member['doc_id']}`.")
if member["kind"] == "constant":
value = member["value"]["value"]
lines.append(f" {row['rust_signature']} = {value};")
continue
signatures = row["rust_signature"].split(" ; ")
for index, signature in enumerate(signatures):
if index:
lines.append(f" /// Setter for C# member: `{member['doc_id']}`.")
lines.append(render_body(signature, member["doc_id"], row["error_model"], trait))
lines.append("}")
if not trait:
for interface in item.get("interfaces", []):
base, arguments, _ = mapping.type_parts(interface)
if base not in mapper.interfaces:
continue
mapped_arguments = [mapper.type(argument, None, set(names)) for argument in arguments]
trait_path = mapper.resolved[base] + generic_suffix(mapped_arguments)
lines.append(f"impl{suffix} {trait_path} for {rust_name}{suffix} {{}}")
return "\n".join(lines)
def generate_sources(catalog: dict) -> tuple[dict[Path, str], dict[str, tuple[int, int]]]:
type_rows, resolved = mapping.build_type_rows(catalog)
type_by_id = {row["csharp_type_id"]: row for row in type_rows}
interfaces = {mapping.type_parts(item["signature"])[0] for assembly in catalog["assemblies"] for item in assembly["types"] if item["kind"] == "interface"}
mapper = mapping.Mapper(resolved, interfaces)
member_rows = {row["csharp_id"]: row for row in csv.DictReader(LEDGER.open(), delimiter="\t")}
outputs = {}
coverage = {}
for assembly in catalog["assemblies"]:
name = assembly["identity"]["name"]
if name not in TARGETS:
continue
chunks = [
"// @generated by tools/generate_api_shims.py; do not edit by hand.",
"#![allow(clippy::missing_errors_doc)]",
"#![allow(clippy::must_use_candidate)]",
"#![allow(clippy::all, clippy::pedantic, dead_code, unused_variables)]",
"#![allow(non_camel_case_types)]",
"#![allow(non_snake_case)]",
"",
]
for item in assembly["types"]:
chunks += [render_type(item, type_by_id[item["doc_id"]], member_rows, mapper), ""]
source = format_rust("\n".join(chunks).rstrip() + "\n")
expected_types = {item["doc_id"] for item in assembly["types"]}
expected_members = {member["doc_id"] for item in assembly["types"] for member in item["members"]}
emitted_types = set(re.findall(r"C# type: `(.*)`\.", source))
emitted_members = set(re.findall(r"C# member: `(.*)`\.", source))
if emitted_types != expected_types or emitted_members != expected_members:
raise ValueError(
f"generated catalog ID mismatch for {name}: "
f"types missing/stale={len(expected_types - emitted_types)}/{len(emitted_types - expected_types)}, "
f"members missing/stale={len(expected_members - emitted_members)}/{len(emitted_members - expected_members)}"
)
outputs[TARGETS[name]] = source
coverage[name] = (len(assembly["types"]), sum(len(item["members"]) for item in assembly["types"]))
return outputs, coverage
def coverage_report(catalog: dict, coverage: dict[str, tuple[int, int]]) -> str:
lines = [
"# Callable shim coverage", "", "Generated by `python3 tools/generate_api_shims.py`; do not edit by hand.", "",
"| Assembly | Types | Members | Status |", "|---|---:|---:|---|",
]
total_types = total_members = 0
for assembly in catalog["assemblies"]:
name = assembly["identity"]["name"]
types = len(assembly["types"])
members = sum(len(item["members"]) for item in assembly["types"])
if name in coverage:
got_types, got_members = coverage[name]
if (types, members) != (got_types, got_members):
raise ValueError(f"incomplete generated slice for {name}")
status = "callable failure-only shim"
total_types += types
total_members += members
else:
status = "pending milestone issue"
lines.append(f"| `{name}` | {types:,} | {members:,} | {status} |")
lines += ["", f"Current callable coverage: **{total_types:,} types / {total_members:,} members**.", ""]
return "\n".join(lines)
def generate() -> dict[Path, str]:
catalog = json.loads(CATALOG.read_text())
sources, coverage = generate_sources(catalog)
sources[COVERAGE] = coverage_report(catalog, coverage)
return sources
def main() -> None:
parser = argparse.ArgumentParser()
parser.add_argument("--check", action="store_true")
args = parser.parse_args()
outputs = generate()
if args.check:
stale = [str(path.relative_to(ROOT)) for path, text in outputs.items() if not path.exists() or path.read_text() != text]
if stale:
raise SystemExit("stale generated API shims: " + ", ".join(stale))
else:
for path, text in outputs.items():
path.write_text(text)
print("generated " + ", ".join(str(path.relative_to(ROOT)) for path in outputs))
if __name__ == "__main__":
main()