Files
castle/packages/web/src/http/Parser.zig
2026-03-07 21:08:22 +01:00

480 lines
15 KiB
Zig

const std = @import("std");
const Parser = @This();
const Header = @import("Header.zig");
const Method = @import("Method.zig").Method;
const Response = @import("../Response.zig");
const RequestHandler = @import("../RequestHandler.zig");
const RequestRouter = @import("../RequestRouter.zig");
const Error = error{
MethodNotSupported,
HttpVersionNotSupported,
MissingLineFeed,
InvalidContentLength,
RouterError,
HandlerError,
};
const Vec: type = @Vector(std.simd.suggestVectorLength(u8).?, u8);
const vec_len = @typeInfo(Vec).vector.len;
const Pattern = struct {
value: Vec,
mask: Vec,
len: u6,
pub fn init(comptime prefix: []const u8) Pattern {
if (prefix.len > vec_len) {
@compileError("Prefix length is too high");
}
var value: [vec_len]u8 = undefined;
var mask: [vec_len]u8 = undefined;
for (0..vec_len) |i| {
if (i < prefix.len) {
value[i] = prefix[i];
mask[i] = 0xFF;
} else {
value[i] = 0x00;
mask[i] = 0x00;
}
}
}
inline fn check(self: Pattern, vec: Vec) bool {
return @reduce(.And, vec & self.mask == self.value);
}
};
const patterns = struct {
pub const methods = struct {
// NOTE These patterns are arranged in a specific order, such that the
// first ones are the most common (based on vibes only).
pub const GET = Pattern.init("GET ");
pub const POST = Pattern.init("POST ");
pub const HEAD = Pattern.init("HEAD ");
pub const PUT = Pattern.init("PUT ");
pub const DELETE = Pattern.init("DELETE ");
pub const PATCH = Pattern.init("PATCH ");
pub const OPTIONS = Pattern.init("OPTIONS ");
pub const CONNECT = Pattern.init("CONNECT ");
pub const TRACE = Pattern.init("TRACE ");
};
pub const @"version_http/1.1" = Pattern.init("HTTP/1.1\r\n");
};
inline fn hasSpace(vec: Vec) bool {
const has_space = vec == @as(Vec, @splat(' '));
return @reduce(.Or, has_space);
}
inline fn hasCRLF(vec: Vec) bool {
const has_cr = vec == @as(Vec, @splat('\r'));
const has_lf = vec == @as(Vec, @splat('\n'));
return @reduce(.Or, has_cr | has_lf);
}
const State = union(enum) {
pub fn methodComplete(method: Method) State {
return .{
.method_complete = .{
.method = method,
},
};
}
pub fn pathname(method: Method, p: []const u8) State {
return .{
.pathname_state = .{
.method = method,
.pathname = p,
},
};
}
pub fn headerValue(name: []const u8, value: []const u8) State {
return .{
.header_value = .{
.name = name,
.value = value,
},
};
}
init: void,
// TODO Add all methods here and in `consumeChar` (they are covered by
// `consumeVec`, though)
method_d: void,
method_g: void,
method_h: void,
method_p: void,
method_de: void,
method_ge: void,
method_he: void,
method_pa: void,
method_po: void,
method_pu: void,
method_del: void,
method_hea: void,
method_pat: void,
method_pos: void,
method_dele: void,
method_patc: void,
method_delet: void,
method_complete: struct { method: Method },
pathname_state: struct { method: Method, pathname: []const u8 },
pathname_complete: void,
version_h: void,
version_ht: void,
version_htt: void,
version_http: void,
@"version_http/": void,
@"version_http/1": void,
@"version_http/1.": void,
version_complete: void,
start_line_end: void,
header_name_start: void,
header_name: []const u8,
header_value: struct { name: []const u8, value: []const u8 },
header_line_end: void,
headers_end: void,
body: []const u8,
};
const ConsumeResult = struct {
consumed: usize,
done: bool,
};
const ConsumeCharResult = enum {
not_done,
done,
};
request_router: RequestRouter,
response: *Response,
state: State,
content_length: usize,
request_handler: ?RequestHandler = null,
last_router_error: anyerror = undefined,
last_handler_error: anyerror = undefined,
pub fn init(request_router: RequestRouter, response: *Response) Parser {
return .{
.request_router = request_router,
.response = response,
.state = .init,
.content_length = 0,
};
}
pub fn consume(self: *Parser, chars: []const u8) Error!ConsumeResult {
var i: usize = 0;
while (i < chars.len) {
switch (self.state) {
.body => |body| {
const to_consume = @min(chars.len - i, self.content_length - body.len);
const new_body = body.ptr[0 .. body.len + to_consume];
self.state = .{ .body = new_body };
i += to_consume;
const done = new_body.len >= self.content_length;
if (done) {
self.request_handler.?.rawBody(self.response, new_body) catch |err| {
self.last_handler_error = err;
return error.HandlerError;
};
}
return .{
.consumed = i,
.done = done,
};
},
else => {
if (chars.len - i >= vec_len) {
const vec_res = try self.consumeVec(chars[i..][0..vec_len]);
i += vec_res.consumed;
if (vec_res.done) {
return .{
.consumed = i,
.done = true,
};
}
if (vec_res.consumed > 0) {
continue;
}
}
const char_res = try self.consumeChar(&chars[i]);
i += 1;
if (char_res == .done) return .{
.consumed = i,
.done = true,
};
},
}
}
return .{
.consumed = chars.len,
.done = false,
};
}
pub fn consumeVec(self: *Parser, vec: *const [vec_len]u8) Error!ConsumeResult {
switch (self.state) {
.init => {
inline for (@typeInfo(patterns.methods).@"struct".decls) |decl| {
const pattern: Pattern = @field(patterns.methods, decl.name);
if (pattern.check(vec)) {
self.state = .methodComplete(@field(Method, decl.name));
return .{
.consumed = pattern.len,
.done = false,
};
}
}
return error.MethodNotSupported;
},
.pathname_state => |s| {
if (hasSpace(vec)) {
// Delegate to `consumeChar`.
return .{
.consumed = 0,
.done = false,
};
}
self.state = .pathname(s.method, s.pathname.ptr[0 .. s.pathname.len + vec_len]);
},
.pathname_complete => {
if (patterns.@"version_http/1.1".check(vec)) {
self.state = .header_name_start;
return .{
.consumed = patterns.@"version_http/1.1".len,
.done = false,
};
} else {
return error.HttpVersionNotSupported;
}
},
.header_value => |s| {
if (hasCRLF(vec)) {
// Delegate to `consumeChar`.
return .{
.consumed = 0,
.done = false,
};
}
self.state = .headerValue(s.name, s.value.ptr[0 .. s.value.len + vec_len]);
},
else => {
// Delegate to `consumeChar`.
return .{
.consumed = 0,
.done = false,
};
},
}
}
pub fn consumeChar(self: *Parser, c_ptr: *const u8) Error!ConsumeCharResult {
const c = c_ptr.*;
const c_slice = @as([*]const u8, @ptrCast(c_ptr))[0..1];
switch (self.state) {
.init => switch (c) {
'D' => self.state = .method_d,
'G' => self.state = .method_g,
'H' => self.state = .method_h,
'P' => self.state = .method_p,
else => return error.MethodNotSupported,
},
.method_d => switch (c) {
'E' => self.state = .method_de,
else => return error.MethodNotSupported,
},
.method_g => switch (c) {
'E' => self.state = .method_ge,
else => return error.MethodNotSupported,
},
.method_h => switch (c) {
'E' => self.state = .method_he,
else => return error.MethodNotSupported,
},
.method_p => switch (c) {
'A' => self.state = .method_pa,
'O' => self.state = .method_po,
'U' => self.state = .method_pu,
else => return error.MethodNotSupported,
},
.method_de => switch (c) {
'L' => self.state = .method_del,
else => return error.MethodNotSupported,
},
.method_ge => switch (c) {
'T' => self.state = .methodComplete(.GET),
else => return error.MethodNotSupported,
},
.method_he => switch (c) {
'A' => self.state = .method_hea,
else => return error.MethodNotSupported,
},
.method_pa => switch (c) {
'T' => self.state = .method_pat,
else => return error.MethodNotSupported,
},
.method_po => switch (c) {
'S' => self.state = .method_pos,
else => return error.MethodNotSupported,
},
.method_pu => switch (c) {
'T' => self.state = .methodComplete(.PUT),
else => return error.MethodNotSupported,
},
.method_del => switch (c) {
'E' => self.state = .method_dele,
else => return error.MethodNotSupported,
},
.method_hea => switch (c) {
'D' => self.state = .methodComplete(.HEAD),
else => return error.MethodNotSupported,
},
.method_pat => switch (c) {
'C' => self.state = .method_patc,
else => return error.MethodNotSupported,
},
.method_pos => switch (c) {
'T' => self.state = .methodComplete(.POST),
else => return error.MethodNotSupported,
},
.method_dele => switch (c) {
'T' => self.state = .method_delet,
else => return error.MethodNotSupported,
},
.method_patc => switch (c) {
'H' => self.state = .methodComplete(.PATCH),
else => return error.MethodNotSupported,
},
.method_delet => switch (c) {
'E' => self.state = .methodComplete(.DELETE),
else => return error.MethodNotSupported,
},
.method_complete => |s| switch (c) {
' ' => self.state = .pathname(s.method, @as([*]const u8, @ptrCast(c_ptr))[1..1]),
else => return error.MethodNotSupported,
},
.pathname_state => |s| switch (c) {
' ' => {
self.state = .pathname_complete;
self.request_handler = self.request_router.rawRoute(.init(s.method, s.pathname)) catch |err| {
self.last_router_error = err;
return error.RouterError;
};
},
else => self.state = .pathname(s.method, s.pathname.ptr[0 .. s.pathname.len + 1]),
},
.pathname_complete => switch (c) {
'H' => self.state = .version_h,
else => return error.HttpVersionNotSupported,
},
.version_h => switch (c) {
'T' => self.state = .version_ht,
else => return error.HttpVersionNotSupported,
},
.version_ht => switch (c) {
'T' => self.state = .version_htt,
else => return error.HttpVersionNotSupported,
},
.version_htt => switch (c) {
'P' => self.state = .version_http,
else => return error.HttpVersionNotSupported,
},
.version_http => switch (c) {
'/' => self.state = .@"version_http/",
else => return error.HttpVersionNotSupported,
},
.@"version_http/" => switch (c) {
'1' => self.state = .@"version_http/1",
else => return error.HttpVersionNotSupported,
},
.@"version_http/1" => switch (c) {
'.' => self.state = .@"version_http/1.",
else => return error.HttpVersionNotSupported,
},
.@"version_http/1." => switch (c) {
'1' => self.state = .version_complete,
else => return error.HttpVersionNotSupported,
},
.version_complete => switch (c) {
'\r' => self.state = .start_line_end,
else => return error.HttpVersionNotSupported,
},
.start_line_end => switch (c) {
'\n' => self.state = .header_name_start,
else => return error.MissingLineFeed,
},
.header_name_start => switch (c) {
'\r' => self.state = .headers_end,
else => self.state = .{ .header_name = c_slice },
},
.header_name => |name| switch (c) {
':' => {
self.state = .headerValue(name, @as([*]const u8, @ptrCast(c_ptr))[1..1]);
},
else => self.state = .{ .header_name = name.ptr[0 .. name.len + 1] },
},
.header_value => |s| switch (c) {
'\r' => {
self.state = .header_line_end;
const header: Header = .init(s.name, std.mem.trim(u8, s.value, " \t"));
if (header.isKnown(.@"Content-Length")) {
self.content_length = std.fmt.parseInt(usize, header.value, 10) catch return error.InvalidContentLength;
}
self.request_handler.rawHeader(header) catch |err| {
self.last_handler_error = err;
return error.HandlerError;
};
},
else => self.state = .headerValue(s.name, s.value.ptr[0 .. s.value.len + 1]),
},
.header_line_end => switch (c) {
'\n' => self.state = .header_name_start,
else => return error.MissingLineFeed,
},
.headers_end => switch (c) {
'\n' => {
if (self.content_length == 0) {
self.handler.rawBody(self.request, &.{});
return .done;
}
self.state = .{ .body = @as([*]const u8, @ptrCast(c_ptr))[1..1] };
},
else => return error.MissingLineFeed,
},
.body => |body| {
const new_body = body.ptr[0 .. body.len + 1];
self.state = .{ .body = new_body };
if (new_body.len >= self.content_length) {
self.handler.rawBody(self.request, new_body);
return .done;
}
},
}
return .not_done;
}