From 4eba23686e58008faa7798182ef3673b799b7e64 Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Fri, 20 Dec 2024 01:18:37 -0700
Subject: lib9p/idl: Include directory name in file headers

---
 lib9p/idl/0000-README.md       | 2 +-
 lib9p/idl/1992-9P0.9p.wip      | 2 +-
 lib9p/idl/1995-9P1.9p.wip      | 2 +-
 lib9p/idl/1996-Styx.9p.wip     | 2 +-
 lib9p/idl/2002-9P2000.9p       | 2 +-
 lib9p/idl/2005-9P2000.u.9p     | 2 +-
 lib9p/idl/2010-9P2000.L.9p.wip | 2 +-
 lib9p/idl/2012-9P2000.e.9p     | 2 +-
 8 files changed, 8 insertions(+), 8 deletions(-)

(limited to 'lib9p')

diff --git a/lib9p/idl/0000-README.md b/lib9p/idl/0000-README.md
index cec27e2..86862b7 100644
--- a/lib9p/idl/0000-README.md
+++ b/lib9p/idl/0000-README.md
@@ -1,5 +1,5 @@
 <!--
-  0000-README.md - Overview of 9P protocol definitions
+  lib9p/idl/0000-README.md - Overview of 9P protocol definitions
 
   Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
   SPDX-License-Identifier: AGPL-3.0-or-later
diff --git a/lib9p/idl/1992-9P0.9p.wip b/lib9p/idl/1992-9P0.9p.wip
index 4278fa3..27d6b33 100644
--- a/lib9p/idl/1992-9P0.9p.wip
+++ b/lib9p/idl/1992-9P0.9p.wip
@@ -1,4 +1,4 @@
-# 1992-9P0.9p - Definitions of 9P0 (Plan 9 1st ed) messages
+# lib9p/idl/1992-9P0.9p - Definitions of 9P0 (Plan 9 1st ed) messages
 #
 # Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
 # SPDX-License-Identifier: AGPL-3.0-or-later
diff --git a/lib9p/idl/1995-9P1.9p.wip b/lib9p/idl/1995-9P1.9p.wip
index 55814d4..30b9112 100644
--- a/lib9p/idl/1995-9P1.9p.wip
+++ b/lib9p/idl/1995-9P1.9p.wip
@@ -1,4 +1,4 @@
-# 1995-9P1.9p - Definitions of 9P1 (Plan 9 2nd ed and 3rd ed) messages
+# lib9p/idl/1995-9P1.9p - Definitions of 9P1 (Plan 9 2nd ed and 3rd ed) messages
 #
 # Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
 # SPDX-License-Identifier: AGPL-3.0-or-later
diff --git a/lib9p/idl/1996-Styx.9p.wip b/lib9p/idl/1996-Styx.9p.wip
index 2feb24f..d9d3399 100644
--- a/lib9p/idl/1996-Styx.9p.wip
+++ b/lib9p/idl/1996-Styx.9p.wip
@@ -1,4 +1,4 @@
-# 1996-Styx.9p - Definitions of Styx messages
+# lib9p/idl/1996-Styx.9p - Definitions of Styx messages
 #
 # Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
 # SPDX-License-Identifier: AGPL-3.0-or-later
diff --git a/lib9p/idl/2002-9P2000.9p b/lib9p/idl/2002-9P2000.9p
index 47d402a..fb7c428 100644
--- a/lib9p/idl/2002-9P2000.9p
+++ b/lib9p/idl/2002-9P2000.9p
@@ -1,4 +1,4 @@
-# 2002-9P2000.9p - Definitions of 9P2000 messages
+# lib9p/idl/2002-9P2000.9p - Definitions of 9P2000 messages
 #
 # Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
 # SPDX-License-Identifier: AGPL-3.0-or-later
diff --git a/lib9p/idl/2005-9P2000.u.9p b/lib9p/idl/2005-9P2000.u.9p
index 8b59efa..3eab4ad 100644
--- a/lib9p/idl/2005-9P2000.u.9p
+++ b/lib9p/idl/2005-9P2000.u.9p
@@ -1,4 +1,4 @@
-# 2005-9P2000.u.9p - Definitions of 9P2000.u messages
+# lib9p/idl/2005-9P2000.u.9p - Definitions of 9P2000.u messages
 #
 # Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
 # SPDX-License-Identifier: AGPL-3.0-or-later
diff --git a/lib9p/idl/2010-9P2000.L.9p.wip b/lib9p/idl/2010-9P2000.L.9p.wip
index 5261f7e..f5e79c5 100644
--- a/lib9p/idl/2010-9P2000.L.9p.wip
+++ b/lib9p/idl/2010-9P2000.L.9p.wip
@@ -1,4 +1,4 @@
-# 2010-9P2000.L.9p - Definitions of 9P2000.L messages
+# lib9p/idl/2010-9P2000.L.9p - Definitions of 9P2000.L messages
 #
 # Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
 # SPDX-License-Identifier: AGPL-3.0-or-later
diff --git a/lib9p/idl/2012-9P2000.e.9p b/lib9p/idl/2012-9P2000.e.9p
index 27db50f..c113618 100644
--- a/lib9p/idl/2012-9P2000.e.9p
+++ b/lib9p/idl/2012-9P2000.e.9p
@@ -1,4 +1,4 @@
-# 2012-9P2000.e.9p - Definitions of 9P2000.e messages
+# lib9p/idl/2012-9P2000.e.9p - Definitions of 9P2000.e messages
 #
 # Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
 # SPDX-License-Identifier: AGPL-3.0-or-later
-- 
cgit v1.2.3-2-g168b


From d6bac0f1ca6489e1d532c4e1a5fb3e6b774aafd7 Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Fri, 20 Dec 2024 01:18:37 -0700
Subject: lib9p: Pull the IDL parser into a separate file

---
 lib9p/idl.gen         | 558 +++++---------------------------------------------
 lib9p/idl/__init__.py | 491 ++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 538 insertions(+), 511 deletions(-)
 create mode 100644 lib9p/idl/__init__.py

(limited to 'lib9p')

diff --git a/lib9p/idl.gen b/lib9p/idl.gen
index ae7f1a5..47d1102 100755
--- a/lib9p/idl.gen
+++ b/lib9p/idl.gen
@@ -5,452 +5,17 @@
 # Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
 # SPDX-License-Identifier: AGPL-3.0-or-later
 
-import enum
 import os.path
-import re
-from abc import ABC, abstractmethod
-from typing import Callable, Final, Literal, TypeAlias, TypeVar, cast
+import sys
+
+sys.path.insert(0, os.path.normpath(os.path.join(__file__, "..")))
+
+import idl
 
 # This strives to be "general-purpose" in that it just acts on the
 # *.9p inputs; but (unfortunately?) there are a few special-cases in
 # this script, marked with "SPECIAL".
 
-# Types ########################################################################
-
-
-class Primitive(enum.Enum):
-    u8 = 1
-    u16 = 2
-    u32 = 4
-    u64 = 8
-
-    @property
-    def in_versions(self) -> set[str]:
-        return set()
-
-    @property
-    def name(self) -> str:
-        return str(self.value)
-
-    @property
-    def static_size(self) -> int:
-        return self.value
-
-
-class Number:
-    name: str
-    in_versions: set[str]
-
-    prim: Primitive
-
-    def __init__(self) -> None:
-        self.in_versions = set()
-
-    @property
-    def static_size(self) -> int:
-        return self.prim.static_size
-
-
-class BitfieldVal:
-    name: str
-    in_versions: set[str]
-
-    val: str
-
-    def __init__(self) -> None:
-        self.in_versions = set()
-
-
-class Bitfield:
-    name: str
-    in_versions: set[str]
-
-    prim: Primitive
-
-    bits: list[str]  # bitnames
-    names: dict[str, BitfieldVal]  # bits *and* aliases
-
-    def __init__(self) -> None:
-        self.in_versions = set()
-        self.names = {}
-
-    @property
-    def static_size(self) -> int:
-        return self.prim.static_size
-
-    def bit_is_valid(self, bit: str | int, ver: str | None = None) -> bool:
-        """Return whether the given bit is valid in the given protocol
-        version.
-
-        """
-        bitname = self.bits[bit] if isinstance(bit, int) else bit
-        assert bitname in self.bits
-        if not bitname:
-            return False
-        if bitname.startswith("_"):
-            return False
-        if ver and (ver not in self.names[bitname].in_versions):
-            return False
-        return True
-
-
-class ExprLit:
-    val: int
-
-    def __init__(self, val: int) -> None:
-        self.val = val
-
-
-class ExprSym:
-    name: str
-
-    def __init__(self, name: str) -> None:
-        self.name = name
-
-
-class ExprOp:
-    op: Literal["-", "+"]
-
-    def __init__(self, op: Literal["-", "+"]) -> None:
-        self.op = op
-
-
-class Expr:
-    tokens: list[ExprLit | ExprSym | ExprOp]
-
-    def __init__(self) -> None:
-        self.tokens = []
-
-    def __bool__(self) -> bool:
-        return len(self.tokens) > 0
-
-
-class StructMember:
-    # from left-to-right when parsing
-    cnt: str | None = None
-    name: str
-    typ: "Type"
-    max: Expr
-    val: Expr
-
-    in_versions: set[str]
-
-    @property
-    def static_size(self) -> int | None:
-        if self.cnt:
-            return None
-        return self.typ.static_size
-
-
-class Struct:
-    name: str
-    in_versions: set[str]
-
-    members: list[StructMember]
-
-    def __init__(self) -> None:
-        self.in_versions = set()
-
-    @property
-    def static_size(self) -> int | None:
-        size = 0
-        for member in self.members:
-            msize = member.static_size
-            if msize is None:
-                return None
-            size += msize
-        return size
-
-
-class Message(Struct):
-    @property
-    def msgid(self) -> int:
-        assert len(self.members) >= 3
-        assert self.members[1].name == "typ"
-        assert self.members[1].static_size == 1
-        assert self.members[1].val
-        assert len(self.members[1].val.tokens) == 1
-        assert isinstance(self.members[1].val.tokens[0], ExprLit)
-        return self.members[1].val.tokens[0].val
-
-
-Type: TypeAlias = Primitive | Number | Bitfield | Struct | Message
-# type Type = Primitive | Number | Bitfield | Struct | Message # Change to this once we have Python 3.13
-T = TypeVar("T", Number, Bitfield, Struct, Message)
-
-# Parse *.9p ###################################################################
-
-re_priname = "(?:1|2|4|8)"  # primitive names
-re_symname = "(?:[a-zA-Z_][a-zA-Z_0-9]*)"  # "symbol" names; most *.9p-defined names
-re_impname = r"(?:\*|" + re_symname + ")"  # names we can import
-re_msgname = r"(?:[TR][a-zA-Z_0-9]*)"  # names a message can be
-
-re_memtype = f"(?:{re_symname}|{re_priname})"  # typenames that a struct member can be
-
-re_expr = f"(?:(?:-|\\+|[0-9]+|&?{re_symname})+)"
-
-re_bitspec_bit = f"(?P<bit>[0-9]+)\\s*=\\s*(?P<name>{re_symname})"
-re_bitspec_alias = f"(?P<name>{re_symname})\\s*=\\s*(?P<val>\\S+)"
-
-re_memberspec = f"(?:(?P<cnt>{re_symname})\\*\\()?(?P<name>{re_symname})\\[(?P<typ>{re_memtype})(?:,max=(?P<max>{re_expr})|,val=(?P<val>{re_expr}))*\\]\\)?"
-
-
-def parse_bitspec(ver: str, bf: Bitfield, spec: str) -> None:
-    spec = spec.strip()
-
-    bit: int | None
-    val: BitfieldVal
-    if m := re.fullmatch(re_bitspec_bit, spec):
-        bit = int(m.group("bit"))
-        name = m.group("name")
-
-        val = BitfieldVal()
-        val.name = name
-        val.val = f"1<<{bit}"
-        val.in_versions.add(ver)
-
-        if bit < 0 or bit >= len(bf.bits):
-            raise ValueError(f"{bf.name}: bit {bit} is out-of-bounds")
-        if bf.bits[bit]:
-            raise ValueError(f"{bf.name}: bit {bit} already assigned")
-        bf.bits[bit] = val.name
-    elif m := re.fullmatch(re_bitspec_alias, spec):
-        name = m.group("name")
-        valstr = m.group("val")
-
-        val = BitfieldVal()
-        val.name = name
-        val.val = valstr
-        val.in_versions.add(ver)
-    else:
-        raise SyntaxError(f"invalid bitfield spec {repr(spec)}")
-
-    if val.name in bf.names:
-        raise ValueError(f"{bf.name}: name {val.name} already assigned")
-    bf.names[val.name] = val
-
-
-def parse_expr(expr: str) -> Expr:
-    assert re.fullmatch(re_expr, expr)
-    ret = Expr()
-    for tok in re.split("([-+])", expr):
-        if tok == "-" or tok == "+":
-            # I, for the life of me, do not understand why I need this
-            # cast() to keep mypy happy.
-            ret.tokens += [ExprOp(cast(Literal["-", "+"], tok))]
-        elif re.fullmatch("[0-9]+", tok):
-            ret.tokens += [ExprLit(int(tok))]
-        else:
-            ret.tokens += [ExprSym(tok)]
-    return ret
-
-
-def parse_members(ver: str, env: dict[str, Type], struct: Struct, specs: str) -> None:
-    for spec in specs.split():
-        m = re.fullmatch(re_memberspec, spec)
-        if not m:
-            raise SyntaxError(f"invalid member spec {repr(spec)}")
-
-        member = StructMember()
-        member.in_versions = {ver}
-
-        member.name = m.group("name")
-        if any(x.name == member.name for x in struct.members):
-            raise ValueError(f"duplicate member name {repr(member.name)}")
-
-        if m.group("typ") not in env:
-            raise NameError(f"Unknown type {repr(m.group(2))}")
-        member.typ = env[m.group("typ")]
-
-        if cnt := m.group("cnt"):
-            if len(struct.members) == 0 or struct.members[-1].name != cnt:
-                raise ValueError(f"list count must be previous item: {repr(cnt)}")
-            if not isinstance(struct.members[-1].typ, Primitive):
-                raise ValueError(f"list count must be an integer type: {repr(cnt)}")
-            member.cnt = cnt
-
-        if maxstr := m.group("max"):
-            if (not isinstance(member.typ, Primitive)) or member.cnt:
-                raise ValueError("',max=' may only be specified on a non-repeated atom")
-            member.max = parse_expr(maxstr)
-        else:
-            member.max = Expr()
-
-        if valstr := m.group("val"):
-            if (not isinstance(member.typ, Primitive)) or member.cnt:
-                raise ValueError("',val=' may only be specified on a non-repeated atom")
-            member.val = parse_expr(valstr)
-        else:
-            member.val = Expr()
-
-        struct.members += [member]
-
-
-def re_string(grpname: str) -> str:
-    return f'"(?P<{grpname}>[^"]*)"'
-
-
-re_line_version = f"version\\s+{re_string('version')}"
-re_line_import = f"from\\s+(?P<file>\\S+)\\s+import\\s+(?P<syms>{re_impname}(?:\\s*,\\s*{re_impname})*)"
-re_line_num = f"num\\s+(?P<name>{re_symname})\\s*=\\s*(?P<prim>{re_priname})"
-re_line_bitfield = f"bitfield\\s+(?P<name>{re_symname})\\s*=\\s*(?P<prim>{re_priname})"
-re_line_bitfield_ = (
-    f"bitfield\\s+(?P<name>{re_symname})\\s*\\+=\\s*{re_string('member')}"
-)
-re_line_struct = (
-    f"struct\\s+(?P<name>{re_symname})\\s*(?P<op>\\+?=)\\s*{re_string('members')}"
-)
-re_line_msg = (
-    f"msg\\s+(?P<name>{re_msgname})\\s*(?P<op>\\+?=)\\s*{re_string('members')}"
-)
-re_line_cont = f"\\s+{re_string('specs')}"  # could be bitfield/struct/msg
-
-
-def parse_file(
-    filename: str, get_include: Callable[[str], tuple[str, list[Type]]]
-) -> tuple[str, list[Type]]:
-    version: str | None = None
-    env: dict[str, Type] = {
-        "1": Primitive.u8,
-        "2": Primitive.u16,
-        "4": Primitive.u32,
-        "8": Primitive.u64,
-    }
-
-    def get_type(name: str, tc: type[T]) -> T:
-        nonlocal env
-        if name not in env:
-            raise NameError(f"Unknown type {repr(name)}")
-        ret = env[name]
-        if (not isinstance(ret, tc)) or (ret.__class__.__name__ != tc.__name__):
-            raise NameError(f"Type {repr(ret.name)} is not a {tc.__name__}")
-        return ret
-
-    with open(filename, "r") as fh:
-        prev: Type | None = None
-        for line in fh:
-            line = line.split("#", 1)[0].rstrip()
-            if not line:
-                continue
-            if m := re.fullmatch(re_line_version, line):
-                if version:
-                    raise SyntaxError("must have exactly 1 version line")
-                version = m.group("version")
-                continue
-            if not version:
-                raise SyntaxError("must have exactly 1 version line")
-
-            if m := re.fullmatch(re_line_import, line):
-                other_version, other_typs = get_include(m.group("file"))
-                for symname in m.group("syms").split(sep=","):
-                    symname = symname.strip()
-                    for typ in other_typs:
-                        if typ.name == symname or symname == "*":
-                            match typ:
-                                case Primitive():
-                                    pass
-                                case Number():
-                                    typ.in_versions.add(version)
-                                case Bitfield():
-                                    typ.in_versions.add(version)
-                                    for val in typ.names.values():
-                                        if other_version in val.in_versions:
-                                            val.in_versions.add(version)
-                                case Struct():  # and Message()
-                                    typ.in_versions.add(version)
-                                    for member in typ.members:
-                                        if other_version in member.in_versions:
-                                            member.in_versions.add(version)
-                            env[typ.name] = typ
-            elif m := re.fullmatch(re_line_num, line):
-                num = Number()
-                num.name = m.group("name")
-                num.in_versions.add(version)
-
-                prim = env[m.group("prim")]
-                assert isinstance(prim, Primitive)
-                num.prim = prim
-
-                env[num.name] = num
-                prev = num
-            elif m := re.fullmatch(re_line_bitfield, line):
-                bf = Bitfield()
-                bf.name = m.group("name")
-                bf.in_versions.add(version)
-
-                prim = env[m.group("prim")]
-                assert isinstance(prim, Primitive)
-                bf.prim = prim
-
-                bf.bits = (prim.static_size * 8) * [""]
-
-                env[bf.name] = bf
-                prev = bf
-            elif m := re.fullmatch(re_line_bitfield_, line):
-                bf = get_type(m.group("name"), Bitfield)
-                parse_bitspec(version, bf, m.group("member"))
-
-                prev = bf
-            elif m := re.fullmatch(re_line_struct, line):
-                match m.group("op"):
-                    case "=":
-                        struct = Struct()
-                        struct.name = m.group("name")
-                        struct.in_versions.add(version)
-                        struct.members = []
-                        parse_members(version, env, struct, m.group("members"))
-
-                        env[struct.name] = struct
-                        prev = struct
-                    case "+=":
-                        struct = get_type(m.group("name"), Struct)
-                        parse_members(version, env, struct, m.group("members"))
-
-                        prev = struct
-            elif m := re.fullmatch(re_line_msg, line):
-                match m.group("op"):
-                    case "=":
-                        msg = Message()
-                        msg.name = m.group("name")
-                        msg.in_versions.add(version)
-                        msg.members = []
-                        parse_members(version, env, msg, m.group("members"))
-
-                        env[msg.name] = msg
-                        prev = msg
-                    case "+=":
-                        msg = get_type(m.group("name"), Message)
-                        parse_members(version, env, msg, m.group("members"))
-
-                        prev = msg
-            elif m := re.fullmatch(re_line_cont, line):
-                match prev:
-                    case Bitfield():
-                        parse_bitspec(version, prev, m.group("specs"))
-                    case Struct():  # and Message()
-                        parse_members(version, env, prev, m.group("specs"))
-                    case _:
-                        raise SyntaxError(
-                            "continuation line must come after a bitfield, struct, or msg line"
-                        )
-            else:
-                raise SyntaxError(f"invalid line {repr(line)}")
-    if not version:
-        raise SyntaxError("must have exactly 1 version line")
-
-    typs: list[Type] = [x for x in env.values() if not isinstance(x, Primitive)]
-
-    for typ in [typ for typ in typs if isinstance(typ, Struct)]:
-        valid_syms = ["end", *["&" + m.name for m in typ.members]]
-        for member in typ.members:
-            for tok in [*member.max.tokens, *member.val.tokens]:
-                if isinstance(tok, ExprSym) and tok.name not in valid_syms:
-                    raise ValueError(
-                        f"{typ.name}.{member.name}: invalid sym: {tok.name}"
-                    )
-
-    return version, typs
-
 
 # Generate C ###################################################################
 
@@ -473,17 +38,17 @@ def c_ver_cond(versions: set[str]) -> str:
     return "( " + (" || ".join(c_ver_cond({v}) for v in sorted(versions))) + " )"
 
 
-def c_typename(typ: Type) -> str:
+def c_typename(typ: idl.Type) -> str:
     match typ:
-        case Primitive():
+        case idl.Primitive():
             return f"uint{typ.value*8}_t"
-        case Number():
+        case idl.Number():
             return f"{idprefix}{typ.name}_t"
-        case Bitfield():
+        case idl.Bitfield():
             return f"{idprefix}{typ.name}_t"
-        case Message():
+        case idl.Message():
             return f"struct {idprefix}msg_{typ.name}"
-        case Struct():
+        case idl.Struct():
             return f"struct {idprefix}{typ.name}"
         case _:
             raise ValueError(f"not a type: {typ.__class__.__name__}")
@@ -531,7 +96,7 @@ def ifdef_pop(n: int) -> str:
     return ret
 
 
-def gen_h(versions: set[str], typs: list[Type]) -> str:
+def gen_h(versions: set[str], typs: list[idl.Type]) -> str:
     global _ifdef_stack
     _ifdef_stack = []
 
@@ -576,13 +141,13 @@ enum {idprefix}version {{
     ret += """
 /* non-message types **********************************************************/
 """
-    for typ in [typ for typ in typs if not isinstance(typ, Message)]:
+    for typ in [typ for typ in typs if not isinstance(typ, idl.Message)]:
         ret += "\n"
         ret += ifdef_push(1, c_ver_ifdef(typ.in_versions))
         match typ:
-            case Number():
+            case idl.Number():
                 ret += f"typedef {c_typename(typ.prim)} {c_typename(typ)};\n"
-            case Bitfield():
+            case idl.Bitfield():
                 ret += f"typedef {c_typename(typ.prim)} {c_typename(typ)};\n"
                 names = [
                     *reversed(
@@ -620,7 +185,7 @@ enum {idprefix}version {{
                         sp3 = " " * (2 + namewidth - len(name))
                         ret += f"#{sp1}define{sp2}{c_name}{sp3}(({c_typename(typ)})({typ.names[name].val}))\n"
                 ret += ifdef_pop(1)
-            case Struct():
+            case idl.Struct():
                 typewidth = max(len(c_typename(m.typ)) for m in typ.members)
 
                 ret += c_typename(typ) + " {\n"
@@ -641,14 +206,14 @@ enum {idprefix}version {{
 
 """
     ret += f"enum {idprefix}msg_type {{ /* uint8_t */\n"
-    namewidth = max(len(msg.name) for msg in typs if isinstance(msg, Message))
-    for msg in [msg for msg in typs if isinstance(msg, Message)]:
+    namewidth = max(len(msg.name) for msg in typs if isinstance(msg, idl.Message))
+    for msg in [msg for msg in typs if isinstance(msg, idl.Message)]:
         ret += ifdef_push(1, c_ver_ifdef(msg.in_versions))
         ret += f"\t{idprefix.upper()}TYP_{msg.name.ljust(namewidth)} = {msg.msgid},\n"
     ret += ifdef_pop(0)
     ret += "};\n"
 
-    for msg in [msg for msg in typs if isinstance(msg, Message)]:
+    for msg in [msg for msg in typs if isinstance(msg, idl.Message)]:
         ret += "\n"
         ret += ifdef_push(1, c_ver_ifdef(msg.in_versions))
         ret += c_typename(msg) + " {"
@@ -671,22 +236,22 @@ enum {idprefix}version {{
     return ret
 
 
-def c_expr(expr: Expr) -> str:
+def c_expr(expr: idl.Expr) -> str:
     ret: list[str] = []
     for tok in expr.tokens:
         match tok:
-            case ExprOp():
+            case idl.ExprOp():
                 ret += [tok.op]
-            case ExprLit():
+            case idl.ExprLit():
                 ret += [str(tok.val)]
-            case ExprSym(name="end"):
+            case idl.ExprSym(name="end"):
                 ret += ["ctx->net_offset"]
-            case ExprSym():
+            case idl.ExprSym():
                 ret += [f"_{tok.name[1:]}_offset"]
     return " ".join(ret)
 
 
-def gen_c(versions: set[str], typs: list[Type]) -> str:
+def gen_c(versions: set[str], typs: list[idl.Type]) -> str:
     global _ifdef_stack
     _ifdef_stack = []
 
@@ -771,12 +336,12 @@ LM_ALWAYS_INLINE static bool _validate_list(struct _validate_ctx *ctx,
 #define validate_8(ctx) _validate_size_net(ctx, 8)
 """
     for typ in typs:
-        inline = "LM_FLATTEN" if isinstance(typ, Message) else "LM_ALWAYS_INLINE"
-        argfn = unused if (isinstance(typ, Struct) and not typ.members) else used
+        inline = "LM_FLATTEN" if isinstance(typ, idl.Message) else "LM_ALWAYS_INLINE"
+        argfn = unused if (isinstance(typ, idl.Struct) and not typ.members) else used
         ret += "\n"
         ret += ifdef_push(1, c_ver_ifdef(typ.in_versions))
 
-        if isinstance(typ, Bitfield):
+        if isinstance(typ, idl.Bitfield):
             ret += f"static const {c_typename(typ)} {typ.name}_masks[{c_ver_enum('NUM')}] = {{\n"
             verwidth = max(len(ver) for ver in versions)
             for ver in sorted(versions):
@@ -820,9 +385,9 @@ LM_ALWAYS_INLINE static bool _validate_list(struct _validate_ctx *ctx,
             continue
 
         match typ:
-            case Number():
+            case idl.Number():
                 ret += f"\treturn validate_{typ.prim.name}(ctx);\n"
-            case Bitfield():
+            case idl.Bitfield():
                 ret += f"\t if (validate_{typ.static_size}(ctx))\n"
                 ret += "\t\treturn true;\n"
                 ret += (
@@ -832,7 +397,7 @@ LM_ALWAYS_INLINE static bool _validate_list(struct _validate_ctx *ctx,
                 ret += f"\tif (val & ~mask)\n"
                 ret += f'\t\treturn lib9p_errorf(ctx->ctx, LINUX_EBADMSG, "unknown bits in {typ.name} bitfield: %#0{typ.static_size}"PRIx{typ.static_size*8}, val & ~mask);\n'
                 ret += "\treturn false;\n"
-            case Struct():  # and Message()
+            case idl.Struct():  # and idl.Message()
                 if len(typ.members) == 0:
                     ret += "\treturn false;\n"
                     ret += "}\n"
@@ -849,7 +414,7 @@ LM_ALWAYS_INLINE static bool _validate_list(struct _validate_ctx *ctx,
                 mark_offset: set[str] = set()
                 for member in typ.members:
                     for tok in [*member.max.tokens, *member.val.tokens]:
-                        if isinstance(tok, ExprSym) and tok.name.startswith("&"):
+                        if isinstance(tok, idl.ExprSym) and tok.name.startswith("&"):
                             if tok.name[1:] not in mark_offset:
                                 ret += f"\tuint32_t _{tok.name[1:]}_offset;\n"
                             mark_offset.add(tok.name[1:])
@@ -924,17 +489,17 @@ LM_ALWAYS_INLINE static void unmarshal_8(struct _unmarshal_ctx *ctx, uint64_t *o
 }
 """
     for typ in typs:
-        inline = "LM_FLATTEN" if isinstance(typ, Message) else "LM_ALWAYS_INLINE"
-        argfn = unused if (isinstance(typ, Struct) and not typ.members) else used
+        inline = "LM_FLATTEN" if isinstance(typ, idl.Message) else "LM_ALWAYS_INLINE"
+        argfn = unused if (isinstance(typ, idl.Struct) and not typ.members) else used
         ret += "\n"
         ret += ifdef_push(1, c_ver_ifdef(typ.in_versions))
         ret += f"{inline} static void unmarshal_{typ.name}(struct _unmarshal_ctx *{argfn('ctx')}, {c_typename(typ)} *out) {{\n"
         match typ:
-            case Number():
+            case idl.Number():
                 ret += f"\tunmarshal_{typ.prim.name}(ctx, ({c_typename(typ.prim)} *)out);\n"
-            case Bitfield():
+            case idl.Bitfield():
                 ret += f"\tunmarshal_{typ.prim.name}(ctx, ({c_typename(typ.prim)} *)out);\n"
-            case Struct():
+            case idl.Struct():
                 ret += "\tmemset(out, 0, sizeof(*out));\n"
 
                 for member in typ.members:
@@ -1017,18 +582,18 @@ LM_ALWAYS_INLINE static bool marshal_8(struct _marshal_ctx *ctx, uint64_t *val)
 }
 """
     for typ in typs:
-        inline = "LM_FLATTEN" if isinstance(typ, Message) else "LM_ALWAYS_INLINE"
-        argfn = unused if (isinstance(typ, Struct) and not typ.members) else used
+        inline = "LM_FLATTEN" if isinstance(typ, idl.Message) else "LM_ALWAYS_INLINE"
+        argfn = unused if (isinstance(typ, idl.Struct) and not typ.members) else used
         ret += "\n"
         ret += ifdef_push(1, c_ver_ifdef(typ.in_versions))
         ret += f"{inline} static bool marshal_{typ.name}(struct _marshal_ctx *{argfn('ctx')}, {c_typename(typ)} *{argfn('val')}) {{\n"
         match typ:
-            case Number():
+            case idl.Number():
                 ret += f"\treturn marshal_{typ.prim.name}(ctx, ({c_typename(typ.prim)} *)val);\n"
-            case Bitfield():
+            case idl.Bitfield():
                 ret += f"\t{c_typename(typ)} masked_val = *val & {typ.name}_masks[ctx->ctx->version];\n"
                 ret += f"\treturn marshal_{typ.prim.name}(ctx, ({c_typename(typ.prim)} *)&masked_val);\n"
-            case Struct():
+            case idl.Struct():
                 if len(typ.members) == 0:
                     ret += "\treturn false;\n"
                     ret += "}\n"
@@ -1042,7 +607,7 @@ LM_ALWAYS_INLINE static bool marshal_8(struct _marshal_ctx *ctx, uint64_t *val)
                             ret += f"\tuint32_t _{member.name}_offset;\n"
                         mark_offset.add(member.name)
                     for tok in member.val.tokens:
-                        if isinstance(tok, ExprSym) and tok.name.startswith("&"):
+                        if isinstance(tok, idl.ExprSym) and tok.name.startswith("&"):
                             if tok.name[1:] not in mark_offset:
                                 ret += f"\tuint32_t _{tok.name[1:]}_offset;\n"
                             mark_offset.add(tok.name[1:])
@@ -1109,8 +674,8 @@ LM_ALWAYS_INLINE static bool marshal_8(struct _marshal_ctx *ctx, uint64_t *val)
 
 struct _table_version _{idprefix}versions[{c_ver_enum('NUM')}] = {{
 """
-    id2typ: dict[int, Message] = {}
-    for msg in [msg for msg in typs if isinstance(msg, Message)]:
+    id2typ: dict[int, idl.Message] = {}
+    for msg in [msg for msg in typs if isinstance(msg, idl.Message)]:
         id2typ[msg.msgid] = msg
 
     for ver in ["unknown", *sorted(versions)]:
@@ -1119,7 +684,7 @@ struct _table_version _{idprefix}versions[{c_ver_enum('NUM')}] = {{
         ret += f"\t[{c_ver_enum(ver)}] = {{ .msgs = {{\n"
 
         for n in range(0, 0x100):
-            xmsg: Message | None = id2typ.get(n, None)
+            xmsg: idl.Message | None = id2typ.get(n, None)
             if xmsg:
                 if ver == "unknown":  # SPECIAL
                     if xmsg.name not in ["Tversion", "Rversion", "Rerror"]:
@@ -1154,41 +719,12 @@ LM_FLATTEN bool _{idprefix}marshal_stat(struct _marshal_ctx *ctx, struct lib9p_s
 ################################################################################
 
 
-class Parser:
-    cache: dict[str, tuple[str, list[Type]]] = {}
-
-    def parse_file(self, filename: str) -> tuple[str, list[Type]]:
-        filename = os.path.normpath(filename)
-        if filename not in self.cache:
-
-            def get_include(other_filename: str) -> tuple[str, list[Type]]:
-                return self.parse_file(os.path.join(filename, "..", other_filename))
-
-            self.cache[filename] = parse_file(filename, get_include)
-        return self.cache[filename]
-
-    def all(self) -> tuple[set[str], list[Type]]:
-        ret_versions: set[str] = set()
-        ret_typs: dict[str, Type] = {}
-        for version, typs in self.cache.values():
-            if version in ret_versions:
-                raise ValueError(f"duplicate protocol version {repr(version)}")
-            ret_versions.add(version)
-            for typ in typs:
-                if typ.name in ret_typs:
-                    if typ != ret_typs[typ.name]:
-                        raise ValueError(f"duplicate type name {repr(typ.name)}")
-                else:
-                    ret_typs[typ.name] = typ
-        return ret_versions, list(ret_typs.values())
-
-
 if __name__ == "__main__":
     import sys
 
     if len(sys.argv) < 2:
         raise ValueError("requires at least 1 .9p filename")
-    parser = Parser()
+    parser = idl.Parser()
     for txtname in sys.argv[1:]:
         parser.parse_file(txtname)
     versions, typs = parser.all()
diff --git a/lib9p/idl/__init__.py b/lib9p/idl/__init__.py
new file mode 100644
index 0000000..920d02d
--- /dev/null
+++ b/lib9p/idl/__init__.py
@@ -0,0 +1,491 @@
+# lib9p/idl/__init__.py - A parser for .9p specification files.
+#
+# Copyright (C) 2024  Luke T. Shumaker <lukeshu@lukeshu.com>
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+import enum
+import os.path
+import re
+from typing import Callable, Literal, TypeAlias, TypeVar, cast
+
+__all__ = [
+    # entrypoint
+    "Parser",
+    # types
+    "Type",
+    "Primitive",
+    "Number",
+    *["Bitfield", "BitfieldVal"],
+    *["Struct", "StructMember", "Expr", "ExprOp", "ExprSym", "ExprLit"],
+    "Message",
+]
+
+# The syntax that this parses is described in `./0000-README.md`.
+
+# Types ########################################################################
+
+
+class Primitive(enum.Enum):
+    u8 = 1
+    u16 = 2
+    u32 = 4
+    u64 = 8
+
+    @property
+    def in_versions(self) -> set[str]:
+        return set()
+
+    @property
+    def name(self) -> str:
+        return str(self.value)
+
+    @property
+    def static_size(self) -> int:
+        return self.value
+
+
+class Number:
+    name: str
+    in_versions: set[str]
+
+    prim: Primitive
+
+    def __init__(self) -> None:
+        self.in_versions = set()
+
+    @property
+    def static_size(self) -> int:
+        return self.prim.static_size
+
+
+class BitfieldVal:
+    name: str
+    in_versions: set[str]
+
+    val: str
+
+    def __init__(self) -> None:
+        self.in_versions = set()
+
+
+class Bitfield:
+    name: str
+    in_versions: set[str]
+
+    prim: Primitive
+
+    bits: list[str]  # bitnames
+    names: dict[str, BitfieldVal]  # bits *and* aliases
+
+    def __init__(self) -> None:
+        self.in_versions = set()
+        self.names = {}
+
+    @property
+    def static_size(self) -> int:
+        return self.prim.static_size
+
+    def bit_is_valid(self, bit: str | int, ver: str | None = None) -> bool:
+        """Return whether the given bit is valid in the given protocol
+        version.
+
+        """
+        bitname = self.bits[bit] if isinstance(bit, int) else bit
+        assert bitname in self.bits
+        if not bitname:
+            return False
+        if bitname.startswith("_"):
+            return False
+        if ver and (ver not in self.names[bitname].in_versions):
+            return False
+        return True
+
+
+class ExprLit:
+    val: int
+
+    def __init__(self, val: int) -> None:
+        self.val = val
+
+
+class ExprSym:
+    name: str
+
+    def __init__(self, name: str) -> None:
+        self.name = name
+
+
+class ExprOp:
+    op: Literal["-", "+"]
+
+    def __init__(self, op: Literal["-", "+"]) -> None:
+        self.op = op
+
+
+class Expr:
+    tokens: list[ExprLit | ExprSym | ExprOp]
+
+    def __init__(self) -> None:
+        self.tokens = []
+
+    def __bool__(self) -> bool:
+        return len(self.tokens) > 0
+
+
+class StructMember:
+    # from left-to-right when parsing
+    cnt: str | None = None
+    name: str
+    typ: "Type"
+    max: Expr
+    val: Expr
+
+    in_versions: set[str]
+
+    @property
+    def static_size(self) -> int | None:
+        if self.cnt:
+            return None
+        return self.typ.static_size
+
+
+class Struct:
+    name: str
+    in_versions: set[str]
+
+    members: list[StructMember]
+
+    def __init__(self) -> None:
+        self.in_versions = set()
+
+    @property
+    def static_size(self) -> int | None:
+        size = 0
+        for member in self.members:
+            msize = member.static_size
+            if msize is None:
+                return None
+            size += msize
+        return size
+
+
+class Message(Struct):
+    @property
+    def msgid(self) -> int:
+        assert len(self.members) >= 3
+        assert self.members[1].name == "typ"
+        assert self.members[1].static_size == 1
+        assert self.members[1].val
+        assert len(self.members[1].val.tokens) == 1
+        assert isinstance(self.members[1].val.tokens[0], ExprLit)
+        return self.members[1].val.tokens[0].val
+
+
+Type: TypeAlias = Primitive | Number | Bitfield | Struct | Message
+# type Type = Primitive | Number | Bitfield | Struct | Message # Change to this once we have Python 3.13
+T = TypeVar("T", Number, Bitfield, Struct, Message)
+
+# Parse ########################################################################
+
+re_priname = "(?:1|2|4|8)"  # primitive names
+re_symname = "(?:[a-zA-Z_][a-zA-Z_0-9]*)"  # "symbol" names; most *.9p-defined names
+re_impname = r"(?:\*|" + re_symname + ")"  # names we can import
+re_msgname = r"(?:[TR][a-zA-Z_0-9]*)"  # names a message can be
+
+re_memtype = f"(?:{re_symname}|{re_priname})"  # typenames that a struct member can be
+
+re_expr = f"(?:(?:-|\\+|[0-9]+|&?{re_symname})+)"
+
+re_bitspec_bit = f"(?P<bit>[0-9]+)\\s*=\\s*(?P<name>{re_symname})"
+re_bitspec_alias = f"(?P<name>{re_symname})\\s*=\\s*(?P<val>\\S+)"
+
+re_memberspec = f"(?:(?P<cnt>{re_symname})\\*\\()?(?P<name>{re_symname})\\[(?P<typ>{re_memtype})(?:,max=(?P<max>{re_expr})|,val=(?P<val>{re_expr}))*\\]\\)?"
+
+
+def parse_bitspec(ver: str, bf: Bitfield, spec: str) -> None:
+    spec = spec.strip()
+
+    bit: int | None
+    val: BitfieldVal
+    if m := re.fullmatch(re_bitspec_bit, spec):
+        bit = int(m.group("bit"))
+        name = m.group("name")
+
+        val = BitfieldVal()
+        val.name = name
+        val.val = f"1<<{bit}"
+        val.in_versions.add(ver)
+
+        if bit < 0 or bit >= len(bf.bits):
+            raise ValueError(f"{bf.name}: bit {bit} is out-of-bounds")
+        if bf.bits[bit]:
+            raise ValueError(f"{bf.name}: bit {bit} already assigned")
+        bf.bits[bit] = val.name
+    elif m := re.fullmatch(re_bitspec_alias, spec):
+        name = m.group("name")
+        valstr = m.group("val")
+
+        val = BitfieldVal()
+        val.name = name
+        val.val = valstr
+        val.in_versions.add(ver)
+    else:
+        raise SyntaxError(f"invalid bitfield spec {repr(spec)}")
+
+    if val.name in bf.names:
+        raise ValueError(f"{bf.name}: name {val.name} already assigned")
+    bf.names[val.name] = val
+
+
+def parse_expr(expr: str) -> Expr:
+    assert re.fullmatch(re_expr, expr)
+    ret = Expr()
+    for tok in re.split("([-+])", expr):
+        if tok == "-" or tok == "+":
+            # I, for the life of me, do not understand why I need this
+            # cast() to keep mypy happy.
+            ret.tokens += [ExprOp(cast(Literal["-", "+"], tok))]
+        elif re.fullmatch("[0-9]+", tok):
+            ret.tokens += [ExprLit(int(tok))]
+        else:
+            ret.tokens += [ExprSym(tok)]
+    return ret
+
+
+def parse_members(ver: str, env: dict[str, Type], struct: Struct, specs: str) -> None:
+    for spec in specs.split():
+        m = re.fullmatch(re_memberspec, spec)
+        if not m:
+            raise SyntaxError(f"invalid member spec {repr(spec)}")
+
+        member = StructMember()
+        member.in_versions = {ver}
+
+        member.name = m.group("name")
+        if any(x.name == member.name for x in struct.members):
+            raise ValueError(f"duplicate member name {repr(member.name)}")
+
+        if m.group("typ") not in env:
+            raise NameError(f"Unknown type {repr(m.group(2))}")
+        member.typ = env[m.group("typ")]
+
+        if cnt := m.group("cnt"):
+            if len(struct.members) == 0 or struct.members[-1].name != cnt:
+                raise ValueError(f"list count must be previous item: {repr(cnt)}")
+            if not isinstance(struct.members[-1].typ, Primitive):
+                raise ValueError(f"list count must be an integer type: {repr(cnt)}")
+            member.cnt = cnt
+
+        if maxstr := m.group("max"):
+            if (not isinstance(member.typ, Primitive)) or member.cnt:
+                raise ValueError("',max=' may only be specified on a non-repeated atom")
+            member.max = parse_expr(maxstr)
+        else:
+            member.max = Expr()
+
+        if valstr := m.group("val"):
+            if (not isinstance(member.typ, Primitive)) or member.cnt:
+                raise ValueError("',val=' may only be specified on a non-repeated atom")
+            member.val = parse_expr(valstr)
+        else:
+            member.val = Expr()
+
+        struct.members += [member]
+
+
+def re_string(grpname: str) -> str:
+    return f'"(?P<{grpname}>[^"]*)"'
+
+
+re_line_version = f"version\\s+{re_string('version')}"
+re_line_import = f"from\\s+(?P<file>\\S+)\\s+import\\s+(?P<syms>{re_impname}(?:\\s*,\\s*{re_impname})*)"
+re_line_num = f"num\\s+(?P<name>{re_symname})\\s*=\\s*(?P<prim>{re_priname})"
+re_line_bitfield = f"bitfield\\s+(?P<name>{re_symname})\\s*=\\s*(?P<prim>{re_priname})"
+re_line_bitfield_ = (
+    f"bitfield\\s+(?P<name>{re_symname})\\s*\\+=\\s*{re_string('member')}"
+)
+re_line_struct = (
+    f"struct\\s+(?P<name>{re_symname})\\s*(?P<op>\\+?=)\\s*{re_string('members')}"
+)
+re_line_msg = (
+    f"msg\\s+(?P<name>{re_msgname})\\s*(?P<op>\\+?=)\\s*{re_string('members')}"
+)
+re_line_cont = f"\\s+{re_string('specs')}"  # could be bitfield/struct/msg
+
+
+def parse_file(
+    filename: str, get_include: Callable[[str], tuple[str, list[Type]]]
+) -> tuple[str, list[Type]]:
+    version: str | None = None
+    env: dict[str, Type] = {
+        "1": Primitive.u8,
+        "2": Primitive.u16,
+        "4": Primitive.u32,
+        "8": Primitive.u64,
+    }
+
+    def get_type(name: str, tc: type[T]) -> T:
+        nonlocal env
+        if name not in env:
+            raise NameError(f"Unknown type {repr(name)}")
+        ret = env[name]
+        if (not isinstance(ret, tc)) or (ret.__class__.__name__ != tc.__name__):
+            raise NameError(f"Type {repr(ret.name)} is not a {tc.__name__}")
+        return ret
+
+    with open(filename, "r") as fh:
+        prev: Type | None = None
+        for line in fh:
+            line = line.split("#", 1)[0].rstrip()
+            if not line:
+                continue
+            if m := re.fullmatch(re_line_version, line):
+                if version:
+                    raise SyntaxError("must have exactly 1 version line")
+                version = m.group("version")
+                continue
+            if not version:
+                raise SyntaxError("must have exactly 1 version line")
+
+            if m := re.fullmatch(re_line_import, line):
+                other_version, other_typs = get_include(m.group("file"))
+                for symname in m.group("syms").split(sep=","):
+                    symname = symname.strip()
+                    for typ in other_typs:
+                        if typ.name == symname or symname == "*":
+                            match typ:
+                                case Primitive():
+                                    pass
+                                case Number():
+                                    typ.in_versions.add(version)
+                                case Bitfield():
+                                    typ.in_versions.add(version)
+                                    for val in typ.names.values():
+                                        if other_version in val.in_versions:
+                                            val.in_versions.add(version)
+                                case Struct():  # and Message()
+                                    typ.in_versions.add(version)
+                                    for member in typ.members:
+                                        if other_version in member.in_versions:
+                                            member.in_versions.add(version)
+                            env[typ.name] = typ
+            elif m := re.fullmatch(re_line_num, line):
+                num = Number()
+                num.name = m.group("name")
+                num.in_versions.add(version)
+
+                prim = env[m.group("prim")]
+                assert isinstance(prim, Primitive)
+                num.prim = prim
+
+                env[num.name] = num
+                prev = num
+            elif m := re.fullmatch(re_line_bitfield, line):
+                bf = Bitfield()
+                bf.name = m.group("name")
+                bf.in_versions.add(version)
+
+                prim = env[m.group("prim")]
+                assert isinstance(prim, Primitive)
+                bf.prim = prim
+
+                bf.bits = (prim.static_size * 8) * [""]
+
+                env[bf.name] = bf
+                prev = bf
+            elif m := re.fullmatch(re_line_bitfield_, line):
+                bf = get_type(m.group("name"), Bitfield)
+                parse_bitspec(version, bf, m.group("member"))
+
+                prev = bf
+            elif m := re.fullmatch(re_line_struct, line):
+                match m.group("op"):
+                    case "=":
+                        struct = Struct()
+                        struct.name = m.group("name")
+                        struct.in_versions.add(version)
+                        struct.members = []
+                        parse_members(version, env, struct, m.group("members"))
+
+                        env[struct.name] = struct
+                        prev = struct
+                    case "+=":
+                        struct = get_type(m.group("name"), Struct)
+                        parse_members(version, env, struct, m.group("members"))
+
+                        prev = struct
+            elif m := re.fullmatch(re_line_msg, line):
+                match m.group("op"):
+                    case "=":
+                        msg = Message()
+                        msg.name = m.group("name")
+                        msg.in_versions.add(version)
+                        msg.members = []
+                        parse_members(version, env, msg, m.group("members"))
+
+                        env[msg.name] = msg
+                        prev = msg
+                    case "+=":
+                        msg = get_type(m.group("name"), Message)
+                        parse_members(version, env, msg, m.group("members"))
+
+                        prev = msg
+            elif m := re.fullmatch(re_line_cont, line):
+                match prev:
+                    case Bitfield():
+                        parse_bitspec(version, prev, m.group("specs"))
+                    case Struct():  # and Message()
+                        parse_members(version, env, prev, m.group("specs"))
+                    case _:
+                        raise SyntaxError(
+                            "continuation line must come after a bitfield, struct, or msg line"
+                        )
+            else:
+                raise SyntaxError(f"invalid line {repr(line)}")
+    if not version:
+        raise SyntaxError("must have exactly 1 version line")
+
+    typs: list[Type] = [x for x in env.values() if not isinstance(x, Primitive)]
+
+    for typ in [typ for typ in typs if isinstance(typ, Struct)]:
+        valid_syms = ["end", *["&" + m.name for m in typ.members]]
+        for member in typ.members:
+            for tok in [*member.max.tokens, *member.val.tokens]:
+                if isinstance(tok, ExprSym) and tok.name not in valid_syms:
+                    raise ValueError(
+                        f"{typ.name}.{member.name}: invalid sym: {tok.name}"
+                    )
+
+    return version, typs
+
+
+# Filesystem ###################################################################
+
+
+class Parser:
+    cache: dict[str, tuple[str, list[Type]]] = {}
+
+    def parse_file(self, filename: str) -> tuple[str, list[Type]]:
+        filename = os.path.normpath(filename)
+        if filename not in self.cache:
+
+            def get_include(other_filename: str) -> tuple[str, list[Type]]:
+                return self.parse_file(os.path.join(filename, "..", other_filename))
+
+            self.cache[filename] = parse_file(filename, get_include)
+        return self.cache[filename]
+
+    def all(self) -> tuple[set[str], list[Type]]:
+        ret_versions: set[str] = set()
+        ret_typs: dict[str, Type] = {}
+        for version, typs in self.cache.values():
+            if version in ret_versions:
+                raise ValueError(f"duplicate protocol version {repr(version)}")
+            ret_versions.add(version)
+            for typ in typs:
+                if typ.name in ret_typs:
+                    if typ != ret_typs[typ.name]:
+                        raise ValueError(f"duplicate type name {repr(typ.name)}")
+                else:
+                    ret_typs[typ.name] = typ
+        return ret_versions, list(ret_typs.values())
-- 
cgit v1.2.3-2-g168b


From d018906715b395e0fb9624cbd5b7fbf1c0e3f559 Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Sat, 21 Dec 2024 21:39:02 -0700
Subject: lib9p: idl.gen: Do a better job of organizing the Python into
 sections

---
 lib9p/idl.gen | 36 +++++++++++++++++++++---------------
 1 file changed, 21 insertions(+), 15 deletions(-)

(limited to 'lib9p')

diff --git a/lib9p/idl.gen b/lib9p/idl.gen
index 47d1102..4f1e8b8 100755
--- a/lib9p/idl.gen
+++ b/lib9p/idl.gen
@@ -17,7 +17,7 @@ import idl
 # this script, marked with "SPECIAL".
 
 
-# Generate C ###################################################################
+# Utilities ####################################################################
 
 idprefix = "lib9p_"
 
@@ -54,6 +54,21 @@ def c_typename(typ: idl.Type) -> str:
             raise ValueError(f"not a type: {typ.__class__.__name__}")
 
 
+def c_expr(expr: idl.Expr) -> str:
+    ret: list[str] = []
+    for tok in expr.tokens:
+        match tok:
+            case idl.ExprOp():
+                ret += [tok.op]
+            case idl.ExprLit():
+                ret += [str(tok.val)]
+            case idl.ExprSym(name="end"):
+                ret += ["ctx->net_offset"]
+            case idl.ExprSym():
+                ret += [f"_{tok.name[1:]}_offset"]
+    return " ".join(ret)
+
+
 _ifdef_stack: list[str | None] = []
 
 
@@ -96,6 +111,9 @@ def ifdef_pop(n: int) -> str:
     return ret
 
 
+# Generate .h ##################################################################
+
+
 def gen_h(versions: set[str], typs: list[idl.Type]) -> str:
     global _ifdef_stack
     _ifdef_stack = []
@@ -236,19 +254,7 @@ enum {idprefix}version {{
     return ret
 
 
-def c_expr(expr: idl.Expr) -> str:
-    ret: list[str] = []
-    for tok in expr.tokens:
-        match tok:
-            case idl.ExprOp():
-                ret += [tok.op]
-            case idl.ExprLit():
-                ret += [str(tok.val)]
-            case idl.ExprSym(name="end"):
-                ret += ["ctx->net_offset"]
-            case idl.ExprSym():
-                ret += [f"_{tok.name[1:]}_offset"]
-    return " ".join(ret)
+# Generate .c ##################################################################
 
 
 def gen_c(versions: set[str], typs: list[idl.Type]) -> str:
@@ -716,7 +722,7 @@ LM_FLATTEN bool _{idprefix}marshal_stat(struct _marshal_ctx *ctx, struct lib9p_s
     return ret
 
 
-################################################################################
+# Main #########################################################################
 
 
 if __name__ == "__main__":
-- 
cgit v1.2.3-2-g168b


From 5fafccfcd54e0e309f084ff3891d0239250cfde3 Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Sat, 21 Dec 2024 21:39:02 -0700
Subject: lib9p: idl.gen: Add a c_macro helper

---
 lib9p/9p.generated.c | 16 ++++++++--------
 lib9p/idl.gen        | 51 +++++++++++++++++++++++++++++++++++++--------------
 2 files changed, 45 insertions(+), 22 deletions(-)

(limited to 'lib9p')

diff --git a/lib9p/9p.generated.c b/lib9p/9p.generated.c
index d3dc36a..373d95e 100644
--- a/lib9p/9p.generated.c
+++ b/lib9p/9p.generated.c
@@ -1784,15 +1784,15 @@ LM_FLATTEN static bool marshal_Rswrite(struct _marshal_ctx *ctx, struct lib9p_ms
 
 /* tables / exports ***********************************************************/
 
-#define _MSG(typ) [LIB9P_TYP_##typ] = {                          \
-		.name      = #typ,                               \
-		.basesize  = sizeof(struct lib9p_msg_##typ),     \
-		.validate  = validate_##typ,                     \
-		.unmarshal = (_unmarshal_fn_t)unmarshal_##typ,   \
-		.marshal   = (_marshal_fn_t)marshal_##typ,       \
+#define _MSG(typ) [LIB9P_TYP_##typ] = {                         \
+		.name      = #typ,                              \
+		.basesize  = sizeof(struct lib9p_msg_##typ),    \
+		.validate  = validate_##typ,                    \
+		.unmarshal = (_unmarshal_fn_t)unmarshal_##typ,  \
+		.marshal   = (_marshal_fn_t)marshal_##typ,      \
 	}
-#define _NONMSG(num) [num] = {                                   \
-		.name      = #num,                               \
+#define _NONMSG(num) [num] = {      \
+		.name      = #num,  \
 	}
 
 struct _table_version _lib9p_versions[LIB9P_VER_NUM] = {
diff --git a/lib9p/idl.gen b/lib9p/idl.gen
index 4f1e8b8..bab36af 100755
--- a/lib9p/idl.gen
+++ b/lib9p/idl.gen
@@ -22,6 +22,26 @@ import idl
 idprefix = "lib9p_"
 
 
+def tab_ljust(s: str, width: int) -> str:
+    cur = len(s.expandtabs(tabsize=8))
+    if cur >= width:
+        return s
+    return s + " " * (width - cur)
+
+
+def join_lines(*args: str) -> str:
+    return "\n".join([a.rstrip() for a in args]).rstrip() + "\n"
+
+
+def c_macro(*args: str) -> str:
+    full = join_lines(*args).rstrip()
+    assert "\n" in full
+    lines = [l.rstrip() for l in full.split("\n")]
+    width = max(len(l.expandtabs(tabsize=8)) for l in lines[:-1])
+    lines = [tab_ljust(l, width) for l in lines]
+    return "  \\\n".join(lines).rstrip() + "\n"
+
+
 def c_ver_enum(ver: str) -> str:
     return f"{idprefix.upper()}VER_{ver.replace('.', '_')}"
 
@@ -664,26 +684,29 @@ LM_ALWAYS_INLINE static bool marshal_8(struct _marshal_ctx *ctx, uint64_t *val)
     ret += ifdef_pop(0)
 
     # tables / exports #########################################################
-    ret += f"""
+    ret += """
 /* tables / exports ***********************************************************/
-
-#define _MSG(typ) [{idprefix.upper()}TYP_##typ] = {{                          \\
-		.name      = #typ,                               \\
-		.basesize  = sizeof(struct {idprefix}msg_##typ),     \\
-		.validate  = validate_##typ,                     \\
-		.unmarshal = (_unmarshal_fn_t)unmarshal_##typ,   \\
-		.marshal   = (_marshal_fn_t)marshal_##typ,       \\
-	}}
-#define _NONMSG(num) [num] = {{                                   \\
-		.name      = #num,                               \\
-	}}
-
-struct _table_version _{idprefix}versions[{c_ver_enum('NUM')}] = {{
 """
     id2typ: dict[int, idl.Message] = {}
     for msg in [msg for msg in typs if isinstance(msg, idl.Message)]:
         id2typ[msg.msgid] = msg
 
+    ret += "\n"
+    ret += c_macro(
+        f"#define _MSG(typ) [{idprefix.upper()}TYP_##typ] = {{",
+        f"\t\t.name      = #typ,",
+        f"\t\t.basesize  = sizeof(struct {idprefix}msg_##typ),",
+        f"\t\t.validate  = validate_##typ,",
+        f"\t\t.unmarshal = (_unmarshal_fn_t)unmarshal_##typ,",
+        f"\t\t.marshal   = (_marshal_fn_t)marshal_##typ,",
+        f"\t}}",
+    )
+    ret += c_macro(
+        f"#define _NONMSG(num) [num] = {{", f"\t\t.name      = #num,", f"\t}}"
+    )
+
+    ret += "\n"
+    ret += f"struct _table_version _{idprefix}versions[{c_ver_enum('NUM')}] = {{\n"
     for ver in ["unknown", *sorted(versions)]:
         if ver != "unknown":
             ret += ifdef_push(1, c_ver_ifdef({ver}))
-- 
cgit v1.2.3-2-g168b


From 820540bd706214eee1ae27bb01e6ff5496679880 Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Thu, 26 Dec 2024 18:03:24 -0700
Subject: lib9p: idl.gen: Use \t instead of literal tab, to avoid mistakes

---
 lib9p/9p.generated.c |   2 +-
 lib9p/idl.gen        | 110 +++++++++++++++++++++++++--------------------------
 2 files changed, 56 insertions(+), 56 deletions(-)

(limited to 'lib9p')

diff --git a/lib9p/9p.generated.c b/lib9p/9p.generated.c
index 373d95e..3d296de 100644
--- a/lib9p/9p.generated.c
+++ b/lib9p/9p.generated.c
@@ -2843,5 +2843,5 @@ LM_FLATTEN void _lib9p_unmarshal_stat(struct _unmarshal_ctx *ctx, struct lib9p_s
 	unmarshal_stat(ctx, out);
 }
 LM_FLATTEN bool _lib9p_marshal_stat(struct _marshal_ctx *ctx, struct lib9p_stat *val) {
-    return marshal_stat(ctx, val);
+	return marshal_stat(ctx, val);
 }
diff --git a/lib9p/idl.gen b/lib9p/idl.gen
index bab36af..f68d983 100755
--- a/lib9p/idl.gen
+++ b/lib9p/idl.gen
@@ -141,7 +141,7 @@ def gen_h(versions: set[str], typs: list[idl.Type]) -> str:
     ret = f"""/* Generated by `{' '.join(sys.argv)}`.  DO NOT EDIT!  */
 
 #ifndef _LIB9P_9P_H_
-	#error Do not include <lib9p/9p.generated.h> directly; include <lib9p/9p.h> instead
+\t#error Do not include <lib9p/9p.generated.h> directly; include <lib9p/9p.h> instead
 #endif
 
 #include <stdint.h> /* for uint{{n}}_t types */
@@ -328,32 +328,32 @@ const char *{idprefix}version_str(enum {idprefix}version ver) {{
 /* validate_* *****************************************************************/
 
 LM_ALWAYS_INLINE static bool _validate_size_net(struct _validate_ctx *ctx, uint32_t n) {
-	if (__builtin_add_overflow(ctx->net_offset, n, &ctx->net_offset))
-		/* If needed-net-size overflowed uint32_t, then
-		 * there's no way that actual-net-size will live up to
-		 * that.  */
-		return lib9p_error(ctx->ctx, LINUX_EBADMSG, "message is too short for content");
-	if (ctx->net_offset > ctx->net_size)
-		return lib9p_error(ctx->ctx, LINUX_EBADMSG, "message is too short for content");
-	return false;
+\tif (__builtin_add_overflow(ctx->net_offset, n, &ctx->net_offset))
+\t\t/* If needed-net-size overflowed uint32_t, then
+\t\t * there's no way that actual-net-size will live up to
+\t\t * that.  */
+\t\treturn lib9p_error(ctx->ctx, LINUX_EBADMSG, "message is too short for content");
+\tif (ctx->net_offset > ctx->net_size)
+\t\treturn lib9p_error(ctx->ctx, LINUX_EBADMSG, "message is too short for content");
+\treturn false;
 }
 
 LM_ALWAYS_INLINE static bool _validate_size_host(struct _validate_ctx *ctx, size_t n) {
-	if (__builtin_add_overflow(ctx->host_extra, n, &ctx->host_extra))
-		/* If needed-host-size overflowed size_t, then there's
-		 * no way that actual-net-size will live up to
-		 * that.  */
-		return lib9p_error(ctx->ctx, LINUX_EBADMSG, "message is too short for content");
-	return false;
+\tif (__builtin_add_overflow(ctx->host_extra, n, &ctx->host_extra))
+\t\t/* If needed-host-size overflowed size_t, then there's
+\t\t * no way that actual-net-size will live up to
+\t\t * that.  */
+\t\treturn lib9p_error(ctx->ctx, LINUX_EBADMSG, "message is too short for content");
+\treturn false;
 }
 
 LM_ALWAYS_INLINE static bool _validate_list(struct _validate_ctx *ctx,
                                          size_t cnt,
                                          _validate_fn_t item_fn, size_t item_host_size) {
-	for (size_t i = 0; i < cnt; i++)
-		if (_validate_size_host(ctx, item_host_size) || item_fn(ctx))
-			return true;
-	return false;
+\tfor (size_t i = 0; i < cnt; i++)
+\t\tif (_validate_size_host(ctx, item_host_size) || item_fn(ctx))
+\t\t\treturn true;
+\treturn false;
 }
 
 #define validate_1(ctx) _validate_size_net(ctx, 1)
@@ -495,23 +495,23 @@ LM_ALWAYS_INLINE static bool _validate_list(struct _validate_ctx *ctx,
 /* unmarshal_* ****************************************************************/
 
 LM_ALWAYS_INLINE static void unmarshal_1(struct _unmarshal_ctx *ctx, uint8_t *out) {
-	*out = decode_u8le(&ctx->net_bytes[ctx->net_offset]);
-	ctx->net_offset += 1;
+\t*out = decode_u8le(&ctx->net_bytes[ctx->net_offset]);
+\tctx->net_offset += 1;
 }
 
 LM_ALWAYS_INLINE static void unmarshal_2(struct _unmarshal_ctx *ctx, uint16_t *out) {
-	*out = decode_u16le(&ctx->net_bytes[ctx->net_offset]);
-	ctx->net_offset += 2;
+\t*out = decode_u16le(&ctx->net_bytes[ctx->net_offset]);
+\tctx->net_offset += 2;
 }
 
 LM_ALWAYS_INLINE static void unmarshal_4(struct _unmarshal_ctx *ctx, uint32_t *out) {
-	*out = decode_u32le(&ctx->net_bytes[ctx->net_offset]);
-	ctx->net_offset += 4;
+\t*out = decode_u32le(&ctx->net_bytes[ctx->net_offset]);
+\tctx->net_offset += 4;
 }
 
 LM_ALWAYS_INLINE static void unmarshal_8(struct _unmarshal_ctx *ctx, uint64_t *out) {
-	*out = decode_u64le(&ctx->net_bytes[ctx->net_offset]);
-	ctx->net_offset += 8;
+\t*out = decode_u64le(&ctx->net_bytes[ctx->net_offset]);
+\tctx->net_offset += 8;
 }
 """
     for typ in typs:
@@ -568,43 +568,43 @@ LM_ALWAYS_INLINE static void unmarshal_8(struct _unmarshal_ctx *ctx, uint64_t *o
 /* marshal_* ******************************************************************/
 
 LM_ALWAYS_INLINE static bool _marshal_too_large(struct _marshal_ctx *ctx) {
-	lib9p_errorf(ctx->ctx, LINUX_ERANGE, "%s too large to marshal into %s limit (limit=%"PRIu32")",
-		(ctx->net_bytes[4] % 2 == 0) ? "T-message" : "R-message",
-		ctx->ctx->version ? "negotiated" : ((ctx->net_bytes[4] % 2 == 0) ? "client" : "server"),
-		ctx->ctx->max_msg_size);
-	return true;
+\tlib9p_errorf(ctx->ctx, LINUX_ERANGE, "%s too large to marshal into %s limit (limit=%"PRIu32")",
+\t\t(ctx->net_bytes[4] % 2 == 0) ? "T-message" : "R-message",
+\t\tctx->ctx->version ? "negotiated" : ((ctx->net_bytes[4] % 2 == 0) ? "client" : "server"),
+\t\tctx->ctx->max_msg_size);
+\treturn true;
 }
 
 LM_ALWAYS_INLINE static bool marshal_1(struct _marshal_ctx *ctx, uint8_t *val) {
-	if (ctx->net_offset + 1 > ctx->ctx->max_msg_size)
-		return _marshal_too_large(ctx);
-	ctx->net_bytes[ctx->net_offset] = *val;
-	ctx->net_offset += 1;
-	return false;
+\tif (ctx->net_offset + 1 > ctx->ctx->max_msg_size)
+\t\treturn _marshal_too_large(ctx);
+\tctx->net_bytes[ctx->net_offset] = *val;
+\tctx->net_offset += 1;
+\treturn false;
 }
 
 LM_ALWAYS_INLINE static bool marshal_2(struct _marshal_ctx *ctx, uint16_t *val) {
-	if (ctx->net_offset + 2 > ctx->ctx->max_msg_size)
-		return _marshal_too_large(ctx);
-	encode_u16le(*val, &ctx->net_bytes[ctx->net_offset]);
-	ctx->net_offset += 2;
-	return false;
+\tif (ctx->net_offset + 2 > ctx->ctx->max_msg_size)
+\t\treturn _marshal_too_large(ctx);
+\tencode_u16le(*val, &ctx->net_bytes[ctx->net_offset]);
+\tctx->net_offset += 2;
+\treturn false;
 }
 
 LM_ALWAYS_INLINE static bool marshal_4(struct _marshal_ctx *ctx, uint32_t *val) {
-	if (ctx->net_offset + 4 > ctx->ctx->max_msg_size)
-		return true;
-	encode_u32le(*val, &ctx->net_bytes[ctx->net_offset]);
-	ctx->net_offset += 4;
-	return false;
+\tif (ctx->net_offset + 4 > ctx->ctx->max_msg_size)
+\t\treturn true;
+\tencode_u32le(*val, &ctx->net_bytes[ctx->net_offset]);
+\tctx->net_offset += 4;
+\treturn false;
 }
 
 LM_ALWAYS_INLINE static bool marshal_8(struct _marshal_ctx *ctx, uint64_t *val) {
-	if (ctx->net_offset + 8 > ctx->ctx->max_msg_size)
-		return true;
-	encode_u64le(*val, &ctx->net_bytes[ctx->net_offset]);
-	ctx->net_offset += 8;
-	return false;
+\tif (ctx->net_offset + 8 > ctx->ctx->max_msg_size)
+\t\treturn true;
+\tencode_u64le(*val, &ctx->net_bytes[ctx->net_offset]);
+\tctx->net_offset += 8;
+\treturn false;
 }
 """
     for typ in typs:
@@ -731,13 +731,13 @@ LM_ALWAYS_INLINE static bool marshal_8(struct _marshal_ctx *ctx, uint64_t *val)
 
     ret += f"""
 LM_FLATTEN bool _{idprefix}validate_stat(struct _validate_ctx *ctx) {{
-	return validate_stat(ctx);
+\treturn validate_stat(ctx);
 }}
 LM_FLATTEN void _{idprefix}unmarshal_stat(struct _unmarshal_ctx *ctx, struct lib9p_stat *out) {{
-	unmarshal_stat(ctx, out);
+\tunmarshal_stat(ctx, out);
 }}
 LM_FLATTEN bool _{idprefix}marshal_stat(struct _marshal_ctx *ctx, struct lib9p_stat *val) {{
-    return marshal_stat(ctx, val);
+\treturn marshal_stat(ctx, val);
 }}
 """
 
-- 
cgit v1.2.3-2-g168b


From 3d02c7b07286b76dfe3b00dee0918749f383eb84 Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Fri, 20 Dec 2024 00:36:47 -0700
Subject: lib9p: 9P2000.9p: Fix a typo in a comment

---
 lib9p/idl/2002-9P2000.9p | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'lib9p')

diff --git a/lib9p/idl/2002-9P2000.9p b/lib9p/idl/2002-9P2000.9p
index fb7c428..c83f439 100644
--- a/lib9p/idl/2002-9P2000.9p
+++ b/lib9p/idl/2002-9P2000.9p
@@ -104,7 +104,7 @@ struct stat = "stat_size[2,val=end-&kern_type]"
 # "O"pen flags (flags to pass to Topen and Tcreate)
 bitfield o = 1
 	"0=mode_0" # low bit of the 2-bit READ/WRITE/RDWR/EXEC enum
-	"1=mode_1" # high bit of the 2-bit READ/WRITE/RDWR/EXEC enum"
+	"1=mode_1" # high bit of the 2-bit READ/WRITE/RDWR/EXEC enum
 	#"2=unused"
 	#"3=unused"
 	"4=TRUNC"
-- 
cgit v1.2.3-2-g168b


From 39818a59fdc6cd1249867b7954838222b397f4cd Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Sun, 22 Dec 2024 14:43:55 -0700
Subject: lib9p: idl.gen: Fix a typo in a comment

---
 lib9p/idl.gen | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'lib9p')

diff --git a/lib9p/idl.gen b/lib9p/idl.gen
index f68d983..31bada3 100755
--- a/lib9p/idl.gen
+++ b/lib9p/idl.gen
@@ -658,7 +658,7 @@ LM_ALWAYS_INLINE static bool marshal_8(struct _marshal_ctx *ctx, uint64_t *val)
                             ret += f"marshal_{member.typ.name}(ctx, &val->{member.name}[i]);\n"
                         ret += f"\t          err; }})"
                     elif member.val:
-                        # Just increment net_offset, don't actually marsha anything (yet).
+                        # Just increment net_offset, don't actually marshal anything (yet).
                         assert member.static_size
                         ret += (
                             f"({{ ctx->net_offset += {member.static_size}; false; }})"
-- 
cgit v1.2.3-2-g168b


From 93337489aaf8bc337ef70d72c059901867f415ea Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Sat, 21 Dec 2024 22:40:21 -0700
Subject: lib9p: Reorganize 9p.generated.h

---
 lib9p/idl.gen                      | 59 ++++++++++---------------
 lib9p/include/lib9p/9p.generated.h | 90 +++++++++++++++++++-------------------
 2 files changed, 66 insertions(+), 83 deletions(-)

(limited to 'lib9p')

diff --git a/lib9p/idl.gen b/lib9p/idl.gen
index 31bada3..66161e2 100755
--- a/lib9p/idl.gen
+++ b/lib9p/idl.gen
@@ -159,7 +159,7 @@ def gen_h(versions: set[str], typs: list[idl.Type]) -> str:
         ret += "#endif\n"
 
     ret += f"""
-/* versions *******************************************************************/
+/* enum version ***************************************************************/
 
 enum {idprefix}version {{
 """
@@ -177,9 +177,21 @@ enum {idprefix}version {{
     ret += f"const char *{idprefix}version_str(enum {idprefix}version);\n"
 
     ret += """
-/* non-message types **********************************************************/
+/* enum msg_type **************************************************************/
+
 """
-    for typ in [typ for typ in typs if not isinstance(typ, idl.Message)]:
+    ret += f"enum {idprefix}msg_type {{ /* uint8_t */\n"
+    namewidth = max(len(msg.name) for msg in typs if isinstance(msg, idl.Message))
+    for msg in [msg for msg in typs if isinstance(msg, idl.Message)]:
+        ret += ifdef_push(1, c_ver_ifdef(msg.in_versions))
+        ret += f"\t{idprefix.upper()}TYP_{msg.name.ljust(namewidth)} = {msg.msgid},\n"
+    ret += ifdef_pop(0)
+    ret += "};\n"
+
+    ret += """
+/* payload types **************************************************************/
+"""
+    for typ in typs:
         ret += "\n"
         ret += ifdef_push(1, c_ver_ifdef(typ.in_versions))
         match typ:
@@ -223,10 +235,15 @@ enum {idprefix}version {{
                         sp3 = " " * (2 + namewidth - len(name))
                         ret += f"#{sp1}define{sp2}{c_name}{sp3}(({c_typename(typ)})({typ.names[name].val}))\n"
                 ret += ifdef_pop(1)
-            case idl.Struct():
+            case idl.Struct():  # and idl.Message():
+                ret += c_typename(typ) + " {"
+                if not typ.members:
+                    ret += "};\n"
+                    continue
+                ret += "\n"
+
                 typewidth = max(len(c_typename(m.typ)) for m in typ.members)
 
-                ret += c_typename(typ) + " {\n"
                 for member in typ.members:
                     if member.val:
                         continue
@@ -239,38 +256,6 @@ enum {idprefix}version {{
                 ret += "};\n"
     ret += ifdef_pop(0)
 
-    ret += """
-/* messages *******************************************************************/
-
-"""
-    ret += f"enum {idprefix}msg_type {{ /* uint8_t */\n"
-    namewidth = max(len(msg.name) for msg in typs if isinstance(msg, idl.Message))
-    for msg in [msg for msg in typs if isinstance(msg, idl.Message)]:
-        ret += ifdef_push(1, c_ver_ifdef(msg.in_versions))
-        ret += f"\t{idprefix.upper()}TYP_{msg.name.ljust(namewidth)} = {msg.msgid},\n"
-    ret += ifdef_pop(0)
-    ret += "};\n"
-
-    for msg in [msg for msg in typs if isinstance(msg, idl.Message)]:
-        ret += "\n"
-        ret += ifdef_push(1, c_ver_ifdef(msg.in_versions))
-        ret += c_typename(msg) + " {"
-        if not msg.members:
-            ret += "};\n"
-            continue
-        ret += "\n"
-
-        typewidth = max(len(c_typename(m.typ)) for m in msg.members)
-
-        for member in msg.members:
-            if member.val:
-                continue
-            ret += ifdef_push(2, c_ver_ifdef(member.in_versions))
-            ret += f"\t{c_typename(member.typ).ljust(typewidth)}  {'*' if member.cnt else ' '}{member.name};\n"
-        ret += ifdef_pop(1)
-        ret += "};\n"
-    ret += ifdef_pop(0)
-
     return ret
 
 
diff --git a/lib9p/include/lib9p/9p.generated.h b/lib9p/include/lib9p/9p.generated.h
index 25aacfe..a6dc13c 100644
--- a/lib9p/include/lib9p/9p.generated.h
+++ b/lib9p/include/lib9p/9p.generated.h
@@ -22,7 +22,7 @@
 	#error config.h must define CONFIG_9P_ENABLE_9P2000_u
 #endif
 
-/* versions *******************************************************************/
+/* enum version ***************************************************************/
 
 enum lib9p_version {
 	LIB9P_VER_unknown = 0, /* "unknown" */
@@ -40,7 +40,49 @@ enum lib9p_version {
 
 const char *lib9p_version_str(enum lib9p_version);
 
-/* non-message types **********************************************************/
+/* enum msg_type **************************************************************/
+
+enum lib9p_msg_type { /* uint8_t */
+#if CONFIG_9P_ENABLE_9P2000 || CONFIG_9P_ENABLE_9P2000_e || CONFIG_9P_ENABLE_9P2000_u
+	LIB9P_TYP_Tversion = 100,
+	LIB9P_TYP_Rversion = 101,
+	LIB9P_TYP_Tauth    = 102,
+	LIB9P_TYP_Rauth    = 103,
+	LIB9P_TYP_Tattach  = 104,
+	LIB9P_TYP_Rattach  = 105,
+	LIB9P_TYP_Rerror   = 107,
+	LIB9P_TYP_Tflush   = 108,
+	LIB9P_TYP_Rflush   = 109,
+	LIB9P_TYP_Twalk    = 110,
+	LIB9P_TYP_Rwalk    = 111,
+	LIB9P_TYP_Topen    = 112,
+	LIB9P_TYP_Ropen    = 113,
+	LIB9P_TYP_Tcreate  = 114,
+	LIB9P_TYP_Rcreate  = 115,
+	LIB9P_TYP_Tread    = 116,
+	LIB9P_TYP_Rread    = 117,
+	LIB9P_TYP_Twrite   = 118,
+	LIB9P_TYP_Rwrite   = 119,
+	LIB9P_TYP_Tclunk   = 120,
+	LIB9P_TYP_Rclunk   = 121,
+	LIB9P_TYP_Tremove  = 122,
+	LIB9P_TYP_Rremove  = 123,
+	LIB9P_TYP_Tstat    = 124,
+	LIB9P_TYP_Rstat    = 125,
+	LIB9P_TYP_Twstat   = 126,
+	LIB9P_TYP_Rwstat   = 127,
+#endif /* CONFIG_9P_ENABLE_9P2000 || CONFIG_9P_ENABLE_9P2000_e || CONFIG_9P_ENABLE_9P2000_u */
+#if CONFIG_9P_ENABLE_9P2000_e
+	LIB9P_TYP_Tsession = 150,
+	LIB9P_TYP_Rsession = 151,
+	LIB9P_TYP_Tsread   = 152,
+	LIB9P_TYP_Rsread   = 153,
+	LIB9P_TYP_Tswrite  = 154,
+	LIB9P_TYP_Rswrite  = 155,
+#endif /* CONFIG_9P_ENABLE_9P2000_e */
+};
+
+/* payload types **************************************************************/
 
 #if CONFIG_9P_ENABLE_9P2000 || CONFIG_9P_ENABLE_9P2000_e || CONFIG_9P_ENABLE_9P2000_u
 typedef uint16_t lib9p_tag_t;
@@ -156,51 +198,7 @@ typedef uint8_t lib9p_o_t;
 #define   LIB9P_O_EXEC       ((lib9p_o_t)(3))
 #define   LIB9P_O_MODE_MASK  ((lib9p_o_t)(0b00000011))
 #define   LIB9P_O_FLAG_MASK  ((lib9p_o_t)(0b11111100))
-#endif /* CONFIG_9P_ENABLE_9P2000 || CONFIG_9P_ENABLE_9P2000_e || CONFIG_9P_ENABLE_9P2000_u */
 
-/* messages *******************************************************************/
-
-enum lib9p_msg_type { /* uint8_t */
-#if CONFIG_9P_ENABLE_9P2000 || CONFIG_9P_ENABLE_9P2000_e || CONFIG_9P_ENABLE_9P2000_u
-	LIB9P_TYP_Tversion = 100,
-	LIB9P_TYP_Rversion = 101,
-	LIB9P_TYP_Tauth    = 102,
-	LIB9P_TYP_Rauth    = 103,
-	LIB9P_TYP_Tattach  = 104,
-	LIB9P_TYP_Rattach  = 105,
-	LIB9P_TYP_Rerror   = 107,
-	LIB9P_TYP_Tflush   = 108,
-	LIB9P_TYP_Rflush   = 109,
-	LIB9P_TYP_Twalk    = 110,
-	LIB9P_TYP_Rwalk    = 111,
-	LIB9P_TYP_Topen    = 112,
-	LIB9P_TYP_Ropen    = 113,
-	LIB9P_TYP_Tcreate  = 114,
-	LIB9P_TYP_Rcreate  = 115,
-	LIB9P_TYP_Tread    = 116,
-	LIB9P_TYP_Rread    = 117,
-	LIB9P_TYP_Twrite   = 118,
-	LIB9P_TYP_Rwrite   = 119,
-	LIB9P_TYP_Tclunk   = 120,
-	LIB9P_TYP_Rclunk   = 121,
-	LIB9P_TYP_Tremove  = 122,
-	LIB9P_TYP_Rremove  = 123,
-	LIB9P_TYP_Tstat    = 124,
-	LIB9P_TYP_Rstat    = 125,
-	LIB9P_TYP_Twstat   = 126,
-	LIB9P_TYP_Rwstat   = 127,
-#endif /* CONFIG_9P_ENABLE_9P2000 || CONFIG_9P_ENABLE_9P2000_e || CONFIG_9P_ENABLE_9P2000_u */
-#if CONFIG_9P_ENABLE_9P2000_e
-	LIB9P_TYP_Tsession = 150,
-	LIB9P_TYP_Rsession = 151,
-	LIB9P_TYP_Tsread   = 152,
-	LIB9P_TYP_Rsread   = 153,
-	LIB9P_TYP_Tswrite  = 154,
-	LIB9P_TYP_Rswrite  = 155,
-#endif /* CONFIG_9P_ENABLE_9P2000_e */
-};
-
-#if CONFIG_9P_ENABLE_9P2000 || CONFIG_9P_ENABLE_9P2000_e || CONFIG_9P_ENABLE_9P2000_u
 struct lib9p_msg_Tversion {
 	lib9p_tag_t      tag;
 	uint32_t         max_msg_size;
-- 
cgit v1.2.3-2-g168b


From 339e4741a6822fbd4ebd343084ade44cbdb4a86a Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Sat, 21 Dec 2024 23:12:34 -0700
Subject: lib9p: idl.gen: Clean up the bitfield code

---
 lib9p/idl.gen | 54 ++++++++++++++++++++++++++++++++----------------------
 1 file changed, 32 insertions(+), 22 deletions(-)

(limited to 'lib9p')

diff --git a/lib9p/idl.gen b/lib9p/idl.gen
index 66161e2..047ea96 100755
--- a/lib9p/idl.gen
+++ b/lib9p/idl.gen
@@ -29,6 +29,12 @@ def tab_ljust(s: str, width: int) -> str:
     return s + " " * (width - cur)
 
 
+def add_prefix(p: str, s: str) -> str:
+    if s.startswith("_"):
+        return "_" + p + s[1:]
+    return p + s
+
+
 def join_lines(*args: str) -> str:
     return "\n".join([a.rstrip() for a in args]).rstrip() + "\n"
 
@@ -206,34 +212,38 @@ enum {idprefix}version {{
                     "",
                     *[k for k in typ.names if k not in typ.bits],
                 ]
-                namewidth = max(len(name) for name in names)
+                prefix = f"{idprefix.upper()}{typ.name.upper()}_"
+                namewidth = max(len(add_prefix(prefix, name)) for name in names)
 
                 ret += "\n"
                 for name in names:
                     if name == "":
                         ret += "\n"
-                    elif name.startswith(" "):
-                        ret += ifdef_push(2, c_ver_ifdef(typ.in_versions))
-                        sp = " " * (
-                            len("#  define ")
-                            + len(idprefix)
-                            + len(typ.name)
-                            + 1
-                            + namewidth
-                            + 2
-                            - len("/* unused")
-                        )
-                        ret += f"/* unused{sp}(({c_typename(typ)})(1<<{name[1:]})) */\n"
+                        continue
+
+                    if name.startswith(" "):
+                        vers = typ.in_versions
+                        c_name = ""
+                        c_val = f"1<<{name[1:]}"
                     else:
-                        ret += ifdef_push(2, c_ver_ifdef(typ.names[name].in_versions))
-                        if name.startswith("_"):
-                            c_name = f"_{idprefix.upper()}{typ.name.upper()}_{name[1:]}"
-                        else:
-                            c_name = f"{idprefix.upper()}{typ.name.upper()}_{name}"
-                        sp1 = "  " if _ifdef_stack[-1] else ""
-                        sp2 = " " if _ifdef_stack[-1] else "   "
-                        sp3 = " " * (2 + namewidth - len(name))
-                        ret += f"#{sp1}define{sp2}{c_name}{sp3}(({c_typename(typ)})({typ.names[name].val}))\n"
+                        vers = typ.names[name].in_versions
+                        c_name = add_prefix(prefix, name)
+                        c_val = f"{typ.names[name].val}"
+
+                    ret += ifdef_push(2, c_ver_ifdef(vers))
+
+                    # It is important all of the `beg` strings have
+                    # the same length.
+                    end = ""
+                    if name.startswith(" "):
+                        beg = "/* unused"
+                        end = " */"
+                    elif _ifdef_stack[-1]:
+                        beg = "#  define"
+                    else:
+                        beg = "#define  "
+
+                    ret += f"{beg} {c_name.ljust(namewidth)}  (({c_typename(typ)})({c_val})){end}\n"
                 ret += ifdef_pop(1)
             case idl.Struct():  # and idl.Message():
                 ret += c_typename(typ) + " {"
-- 
cgit v1.2.3-2-g168b


From 1aad8aa5947960e26d8efcdabb52babd6b52005b Mon Sep 17 00:00:00 2001
From: "Luke T. Shumaker" <lukeshu@lukeshu.com>
Date: Thu, 26 Dec 2024 17:41:32 -0700
Subject: lib9p: idl.gen: Give bitmasks their own section in .c

---
 lib9p/9p.generated.c | 73 ++++++++++++++++++++++++++++------------------------
 lib9p/idl.gen        | 42 ++++++++++++++++++------------
 2 files changed, 65 insertions(+), 50 deletions(-)

(limited to 'lib9p')

diff --git a/lib9p/9p.generated.c b/lib9p/9p.generated.c
index 3d296de..c260647 100644
--- a/lib9p/9p.generated.c
+++ b/lib9p/9p.generated.c
@@ -34,6 +34,46 @@ const char *lib9p_version_str(enum lib9p_version ver) {
     return version_strs[ver];
 }
 
+/* bitmasks *******************************************************************/
+
+#if CONFIG_9P_ENABLE_9P2000 || CONFIG_9P_ENABLE_9P2000_e || CONFIG_9P_ENABLE_9P2000_u
+static const lib9p_dm_t dm_masks[LIB9P_VER_NUM] = {
+#if CONFIG_9P_ENABLE_9P2000
+	[LIB9P_VER_9P2000]   = 0b11101100000000000000000111111111,
+#endif /* CONFIG_9P_ENABLE_9P2000 */
+#if CONFIG_9P_ENABLE_9P2000_e
+	[LIB9P_VER_9P2000_e] = 0b11101100000000000000000111111111,
+#endif /* CONFIG_9P_ENABLE_9P2000_e */
+#if CONFIG_9P_ENABLE_9P2000_u
+	[LIB9P_VER_9P2000_u] = 0b11101100101111000000000111111111,
+#endif /* CONFIG_9P_ENABLE_9P2000_u */
+};
+
+static const lib9p_qt_t qt_masks[LIB9P_VER_NUM] = {
+#if CONFIG_9P_ENABLE_9P2000
+	[LIB9P_VER_9P2000]   = 0b11101100,
+#endif /* CONFIG_9P_ENABLE_9P2000 */
+#if CONFIG_9P_ENABLE_9P2000_e
+	[LIB9P_VER_9P2000_e] = 0b11101100,
+#endif /* CONFIG_9P_ENABLE_9P2000_e */
+#if CONFIG_9P_ENABLE_9P2000_u
+	[LIB9P_VER_9P2000_u] = 0b11101110,
+#endif /* CONFIG_9P_ENABLE_9P2000_u */
+};
+
+static const lib9p_o_t o_masks[LIB9P_VER_NUM] = {
+#if CONFIG_9P_ENABLE_9P2000
+	[LIB9P_VER_9P2000]   = 0b01010011,
+#endif /* CONFIG_9P_ENABLE_9P2000 */
+#if CONFIG_9P_ENABLE_9P2000_e
+	[LIB9P_VER_9P2000_e] = 0b01010011,
+#endif /* CONFIG_9P_ENABLE_9P2000_e */
+#if CONFIG_9P_ENABLE_9P2000_u
+	[LIB9P_VER_9P2000_u] = 0b01010011,
+#endif /* CONFIG_9P_ENABLE_9P2000_u */
+};
+#endif /* CONFIG_9P_ENABLE_9P2000 || CONFIG_9P_ENABLE_9P2000_e || CONFIG_9P_ENABLE_9P2000_u */
+
 /* validate_* *****************************************************************/
 
 LM_ALWAYS_INLINE static bool _validate_size_net(struct _validate_ctx *ctx, uint32_t n) {
@@ -99,17 +139,6 @@ LM_ALWAYS_INLINE static bool validate_s(struct _validate_ctx *ctx) {
 	return false;
 }
 
-static const lib9p_dm_t dm_masks[LIB9P_VER_NUM] = {
-#if CONFIG_9P_ENABLE_9P2000
-	[LIB9P_VER_9P2000]   = 0b11101100000000000000000111111111,
-#endif /* CONFIG_9P_ENABLE_9P2000 */
-#if CONFIG_9P_ENABLE_9P2000_e
-	[LIB9P_VER_9P2000_e] = 0b11101100000000000000000111111111,
-#endif /* CONFIG_9P_ENABLE_9P2000_e */
-#if CONFIG_9P_ENABLE_9P2000_u
-	[LIB9P_VER_9P2000_u] = 0b11101100101111000000000111111111,
-#endif /* CONFIG_9P_ENABLE_9P2000_u */
-};
 LM_ALWAYS_INLINE static bool validate_dm(struct _validate_ctx *ctx) {
 	 if (validate_4(ctx))
 		return true;
@@ -120,17 +149,6 @@ LM_ALWAYS_INLINE static bool validate_dm(struct _validate_ctx *ctx) {
 	return false;
 }
 
-static const lib9p_qt_t qt_masks[LIB9P_VER_NUM] = {
-#if CONFIG_9P_ENABLE_9P2000
-	[LIB9P_VER_9P2000]   = 0b11101100,
-#endif /* CONFIG_9P_ENABLE_9P2000 */
-#if CONFIG_9P_ENABLE_9P2000_e
-	[LIB9P_VER_9P2000_e] = 0b11101100,
-#endif /* CONFIG_9P_ENABLE_9P2000_e */
-#if CONFIG_9P_ENABLE_9P2000_u
-	[LIB9P_VER_9P2000_u] = 0b11101110,
-#endif /* CONFIG_9P_ENABLE_9P2000_u */
-};
 LM_ALWAYS_INLINE static bool validate_qt(struct _validate_ctx *ctx) {
 	 if (validate_1(ctx))
 		return true;
@@ -176,17 +194,6 @@ LM_ALWAYS_INLINE static bool validate_stat(struct _validate_ctx *ctx) {
 	    ;
 }
 
-static const lib9p_o_t o_masks[LIB9P_VER_NUM] = {
-#if CONFIG_9P_ENABLE_9P2000
-	[LIB9P_VER_9P2000]   = 0b01010011,
-#endif /* CONFIG_9P_ENABLE_9P2000 */
-#if CONFIG_9P_ENABLE_9P2000_e
-	[LIB9P_VER_9P2000_e] = 0b01010011,
-#endif /* CONFIG_9P_ENABLE_9P2000_e */
-#if CONFIG_9P_ENABLE_9P2000_u
-	[LIB9P_VER_9P2000_u] = 0b01010011,
-#endif /* CONFIG_9P_ENABLE_9P2000_u */
-};
 LM_ALWAYS_INLINE static bool validate_o(struct _validate_ctx *ctx) {
 	 if (validate_1(ctx))
 		return true;
diff --git a/lib9p/idl.gen b/lib9p/idl.gen
index 047ea96..72efb7b 100755
--- a/lib9p/idl.gen
+++ b/lib9p/idl.gen
@@ -318,6 +318,31 @@ const char *{idprefix}version_str(enum {idprefix}version ver) {{
 }}
 """
 
+    # bitmasks #################################################################
+    ret += f"""
+/* bitmasks *******************************************************************/
+"""
+    for typ in typs:
+        if not isinstance(typ, idl.Bitfield):
+            continue
+        ret += "\n"
+        ret += ifdef_push(1, c_ver_ifdef(typ.in_versions))
+        ret += f"static const {c_typename(typ)} {typ.name}_masks[{c_ver_enum('NUM')}] = {{\n"
+        verwidth = max(len(ver) for ver in versions)
+        for ver in sorted(versions):
+            ret += ifdef_push(2, c_ver_ifdef({ver}))
+            ret += (
+                f"\t[{c_ver_enum(ver)}]{' '*(verwidth-len(ver))} = 0b"
+                + "".join(
+                    "1" if typ.bit_is_valid(bitname, ver) else "0"
+                    for bitname in reversed(typ.bits)
+                )
+                + ",\n"
+            )
+        ret += ifdef_pop(1)
+        ret += "};\n"
+    ret += ifdef_pop(0)
+
     # validate_* ###############################################################
     ret += """
 /* validate_* *****************************************************************/
@@ -361,23 +386,6 @@ LM_ALWAYS_INLINE static bool _validate_list(struct _validate_ctx *ctx,
         argfn = unused if (isinstance(typ, idl.Struct) and not typ.members) else used
         ret += "\n"
         ret += ifdef_push(1, c_ver_ifdef(typ.in_versions))
-
-        if isinstance(typ, idl.Bitfield):
-            ret += f"static const {c_typename(typ)} {typ.name}_masks[{c_ver_enum('NUM')}] = {{\n"
-            verwidth = max(len(ver) for ver in versions)
-            for ver in sorted(versions):
-                ret += ifdef_push(2, c_ver_ifdef({ver}))
-                ret += (
-                    f"\t[{c_ver_enum(ver)}]{' '*(verwidth-len(ver))} = 0b"
-                    + "".join(
-                        "1" if typ.bit_is_valid(bitname, ver) else "0"
-                        for bitname in reversed(typ.bits)
-                    )
-                    + ",\n"
-                )
-            ret += ifdef_pop(1)
-            ret += "};\n"
-
         ret += f"{inline} static bool validate_{typ.name}(struct _validate_ctx *{argfn('ctx')}) {{\n"
 
         if typ.name == "d":  # SPECIAL
-- 
cgit v1.2.3-2-g168b