@shd101wyy/yo 0.1.29 → 0.1.30
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.github/skills/yo-async-effects/SKILL.md +3 -3
- package/.github/skills/yo-async-effects/async-effects-recipes.md +19 -11
- package/.github/skills/yo-core-patterns/core-patterns-cheatsheet.md +33 -13
- package/.github/skills/yo-project-workflow/workflow-cheatsheet.md +1 -1
- package/.github/skills/yo-syntax/syntax-cheatsheet.md +59 -21
- package/README.md +4 -3
- package/out/cjs/index.cjs +771 -676
- package/out/cjs/yo-cli.cjs +1003 -898
- package/out/cjs/yo-lsp.cjs +834 -739
- package/out/esm/index.mjs +716 -621
- package/out/types/src/codegen/exprs/async.d.ts +2 -0
- package/out/types/src/codegen/exprs/await.d.ts +1 -0
- package/out/types/src/codegen/exprs/closures.d.ts +4 -0
- package/out/types/src/codegen/functions/context.d.ts +6 -0
- package/out/types/src/env.d.ts +2 -0
- package/out/types/src/evaluator/builtins/pragma.d.ts +9 -0
- package/out/types/src/evaluator/builtins/unsafe.d.ts +8 -0
- package/out/types/src/evaluator/context.d.ts +2 -0
- package/out/types/src/evaluator/index.d.ts +1 -1
- package/out/types/src/evaluator/memory-safety.d.ts +14 -0
- package/out/types/src/evaluator/types/flowability.d.ts +6 -0
- package/out/types/src/expr-traversal.d.ts +1 -0
- package/out/types/src/expr.d.ts +4 -1
- package/out/types/src/public-safe-report.d.ts +19 -0
- package/out/types/src/tests/comptime-ref-gate.test.d.ts +1 -0
- package/out/types/src/tests/pragma-validation.test.d.ts +1 -0
- package/out/types/src/tests/public-safe-report.test.d.ts +1 -0
- package/out/types/src/tests/type-representation-pointer.test.d.ts +1 -0
- package/out/types/src/tests/unsafe-gate.test.d.ts +1 -0
- package/out/types/src/tests/unsafe-report-classify.test.d.ts +1 -0
- package/out/types/src/types/definitions.d.ts +2 -0
- package/out/types/src/types/utils.d.ts +4 -0
- package/out/types/src/unsafe-report.d.ts +29 -0
- package/out/types/src/value.d.ts +1 -0
- package/out/types/tsconfig.tsbuildinfo +1 -1
- package/package.json +1 -1
- package/scripts/add-pragma-for-pointer-decls.ts +134 -0
- package/scripts/add-pragma.ts +58 -0
- package/scripts/migrate-amp-method-calls.ts +186 -0
- package/scripts/migrate-clone-calls.ts +93 -0
- package/scripts/migrate-get-unwrap.ts +166 -0
- package/scripts/migrate-index-patterns.ts +210 -0
- package/scripts/migrate-index-trait.ts +142 -0
- package/scripts/migrate-iterator.ts +150 -0
- package/scripts/migrate-self-ptr.ts +220 -0
- package/scripts/migrate-skip-pragmas.ts +109 -0
- package/scripts/migrate-tostring.ts +134 -0
- package/scripts/trim-pragma.ts +130 -0
- package/scripts/wrap-extern-calls.ts +161 -0
- package/std/alg/hash.yo +3 -2
- package/std/allocator.yo +6 -5
- package/std/async.yo +2 -2
- package/std/collections/array_list.yo +59 -40
- package/std/collections/btree_map.yo +19 -18
- package/std/collections/deque.yo +9 -8
- package/std/collections/hash_map.yo +101 -13
- package/std/collections/hash_set.yo +5 -4
- package/std/collections/linked_list.yo +39 -4
- package/std/collections/ordered_map.yo +3 -3
- package/std/collections/priority_queue.yo +14 -13
- package/std/crypto/md5.yo +2 -1
- package/std/crypto/random.yo +16 -15
- package/std/crypto/sha256.yo +2 -1
- package/std/encoding/base64.yo +14 -14
- package/std/encoding/hex.yo +3 -3
- package/std/encoding/json.yo +59 -10
- package/std/encoding/punycode.yo +24 -23
- package/std/encoding/toml.yo +4 -3
- package/std/encoding/utf16.yo +2 -2
- package/std/env.yo +43 -28
- package/std/error.yo +6 -6
- package/std/fmt/display.yo +2 -2
- package/std/fmt/index.yo +6 -5
- package/std/fmt/to_string.yo +39 -38
- package/std/fmt/writer.yo +9 -8
- package/std/fs/dir.yo +34 -33
- package/std/fs/file.yo +52 -51
- package/std/fs/metadata.yo +10 -9
- package/std/fs/temp.yo +24 -13
- package/std/fs/walker.yo +10 -9
- package/std/gc.yo +1 -0
- package/std/glob.yo +7 -7
- package/std/http/client.yo +15 -14
- package/std/http/http.yo +6 -6
- package/std/http/index.yo +1 -1
- package/std/imm/list.yo +33 -0
- package/std/imm/map.yo +2 -1
- package/std/imm/set.yo +1 -0
- package/std/imm/sorted_map.yo +1 -0
- package/std/imm/sorted_set.yo +1 -0
- package/std/imm/string.yo +27 -23
- package/std/imm/vec.yo +18 -2
- package/std/io/reader.yo +2 -1
- package/std/io/writer.yo +3 -2
- package/std/libc/assert.yo +1 -0
- package/std/libc/ctype.yo +1 -0
- package/std/libc/dirent.yo +1 -0
- package/std/libc/errno.yo +1 -0
- package/std/libc/fcntl.yo +1 -0
- package/std/libc/float.yo +1 -0
- package/std/libc/limits.yo +1 -0
- package/std/libc/math.yo +1 -0
- package/std/libc/signal.yo +1 -0
- package/std/libc/stdatomic.yo +1 -0
- package/std/libc/stdint.yo +1 -0
- package/std/libc/stdio.yo +1 -0
- package/std/libc/stdlib.yo +1 -0
- package/std/libc/string.yo +1 -0
- package/std/libc/sys/stat.yo +1 -0
- package/std/libc/time.yo +1 -0
- package/std/libc/unistd.yo +1 -0
- package/std/libc/wctype.yo +1 -0
- package/std/libc/windows.yo +2 -0
- package/std/log.yo +7 -6
- package/std/net/addr.yo +5 -4
- package/std/net/dns.yo +7 -6
- package/std/net/errors.yo +8 -8
- package/std/net/tcp.yo +19 -18
- package/std/net/udp.yo +13 -12
- package/std/os/signal.yo +3 -3
- package/std/path.yo +1 -0
- package/std/prelude.yo +353 -182
- package/std/process/command.yo +40 -23
- package/std/process/index.yo +2 -1
- package/std/regex/compiler.yo +10 -9
- package/std/regex/index.yo +41 -41
- package/std/regex/match.yo +2 -2
- package/std/regex/parser.yo +21 -21
- package/std/regex/vm.yo +42 -41
- package/std/string/string.yo +95 -40
- package/std/string/string_builder.yo +9 -9
- package/std/string/unicode.yo +50 -49
- package/std/sync/channel.yo +2 -1
- package/std/sync/cond.yo +5 -4
- package/std/sync/mutex.yo +4 -3
- package/std/sys/advise.yo +1 -0
- package/std/sys/bufio/buf_reader.yo +17 -16
- package/std/sys/bufio/buf_writer.yo +10 -9
- package/std/sys/clock.yo +1 -0
- package/std/sys/copy.yo +1 -0
- package/std/sys/dir.yo +10 -9
- package/std/sys/dns.yo +6 -5
- package/std/sys/errors.yo +11 -11
- package/std/sys/events.yo +1 -0
- package/std/sys/externs.yo +38 -37
- package/std/sys/file.yo +17 -16
- package/std/sys/future.yo +4 -3
- package/std/sys/iov.yo +1 -0
- package/std/sys/mmap.yo +1 -0
- package/std/sys/path.yo +1 -0
- package/std/sys/perm.yo +2 -1
- package/std/sys/pipe.yo +1 -0
- package/std/sys/process.yo +5 -4
- package/std/sys/signal.yo +1 -0
- package/std/sys/socketpair.yo +1 -0
- package/std/sys/sockinfo.yo +1 -0
- package/std/sys/statfs.yo +2 -1
- package/std/sys/statx.yo +1 -0
- package/std/sys/sysinfo.yo +1 -0
- package/std/sys/tcp.yo +15 -14
- package/std/sys/temp.yo +1 -0
- package/std/sys/time.yo +2 -1
- package/std/sys/timer.yo +6 -6
- package/std/sys/tty.yo +2 -1
- package/std/sys/udp.yo +13 -12
- package/std/sys/unix.yo +12 -11
- package/std/testing/bench.yo +4 -3
- package/std/thread.yo +7 -6
- package/std/time/datetime.yo +18 -15
- package/std/time/duration.yo +11 -10
- package/std/time/instant.yo +4 -4
- package/std/time/sleep.yo +1 -0
- package/std/url/index.yo +3 -3
- package/std/worker.yo +4 -3
package/std/process/command.yo
CHANGED
|
@@ -9,7 +9,7 @@
|
|
|
9
9
|
//! { Command } :: import "std/process/command";
|
|
10
10
|
//! { Exception, AnyError } :: import "std/error";
|
|
11
11
|
//!
|
|
12
|
-
//! main :: (fn(io :
|
|
12
|
+
//! main :: (fn(io : Io, exn : Exception) -> unit)({
|
|
13
13
|
//! cmd := Command.new(`echo`);
|
|
14
14
|
//! cmd.arg(`hello`);
|
|
15
15
|
//! cmd.arg(`world`);
|
|
@@ -17,10 +17,11 @@
|
|
|
17
17
|
//! assert(out.status.success(), "echo should succeed");
|
|
18
18
|
//! });
|
|
19
19
|
//! ```
|
|
20
|
+
pragma(Pragma.AllowUnsafe);
|
|
20
21
|
open(import("../collections/array_list"));
|
|
21
22
|
open(import("../string"));
|
|
22
|
-
{ Exception,
|
|
23
|
-
{
|
|
23
|
+
{ Exception, IoExn } :: import("../error");
|
|
24
|
+
{ IoError } :: import("../sys/errors");
|
|
24
25
|
IO_process :: import("../sys/process");
|
|
25
26
|
IO_file :: import("../sys/file");
|
|
26
27
|
IO_pipe :: import("../sys/pipe");
|
|
@@ -96,12 +97,12 @@ impl(
|
|
|
96
97
|
)
|
|
97
98
|
),
|
|
98
99
|
/// Append a single argument to the argv list.
|
|
99
|
-
arg : (fn(self :
|
|
100
|
+
arg : (fn(ref(self) : Self, a : String) -> unit)({
|
|
100
101
|
self._args.push(a);
|
|
101
102
|
()
|
|
102
103
|
}),
|
|
103
104
|
/// Append multiple arguments to the argv list.
|
|
104
|
-
args : (fn(self :
|
|
105
|
+
args : (fn(ref(self) : Self, more : ArrayList(String)) -> unit)({
|
|
105
106
|
n := more.len();
|
|
106
107
|
i := usize(0);
|
|
107
108
|
while(runtime(i < n), {
|
|
@@ -122,7 +123,7 @@ impl(
|
|
|
122
123
|
// Convert each entry of `(program + args)` into an owned NUL-terminated
|
|
123
124
|
// `ArrayList(u8)`. The returned list owns the C-string storage; callers
|
|
124
125
|
// extract the raw `*(u8)` pointers separately.
|
|
125
|
-
_build_cstr_storage :: (fn(self :
|
|
126
|
+
_build_cstr_storage :: (fn(self : Command) -> ArrayList(ArrayList(u8)))({
|
|
126
127
|
storage := ArrayList(ArrayList(u8)).new();
|
|
127
128
|
storage.push(self._program.to_cstr());
|
|
128
129
|
n := self._args.len();
|
|
@@ -145,15 +146,20 @@ _build_cstr_storage :: (fn(self : *(Command)) -> ArrayList(ArrayList(u8)))({
|
|
|
145
146
|
// returns the pid (since the kernel reads argv synchronously inside the syscall).
|
|
146
147
|
_spawn_with_fds :: (
|
|
147
148
|
fn(
|
|
148
|
-
self :
|
|
149
|
+
self : Command,
|
|
149
150
|
stdin_fd : i32,
|
|
150
151
|
stdout_fd : i32,
|
|
151
152
|
stderr_fd : i32,
|
|
152
|
-
io :
|
|
153
|
-
) -> Impl(Future(i32,
|
|
154
|
-
)(
|
|
153
|
+
io : Io
|
|
154
|
+
) -> Impl(Future(i32, IoExn))
|
|
155
|
+
)({
|
|
156
|
+
// Build the C-string storage BEFORE entering the async closure so the
|
|
157
|
+
// closure doesn't have to capture `self` directly. `self : Command` is an
|
|
158
|
+
// object/ref binding that can't be safely captured across an async
|
|
159
|
+
// boundary; the resulting `storage` ArrayList is Rc/heap-allocated and
|
|
160
|
+
// can be moved into the closure's capture struct.
|
|
161
|
+
storage := _build_cstr_storage(self);
|
|
155
162
|
io.async((e) => {
|
|
156
|
-
storage := _build_cstr_storage(self);
|
|
157
163
|
argc := storage.len();
|
|
158
164
|
// Allocate argv buffer: argc entries + 1 null terminator.
|
|
159
165
|
argv_bytes := (sizeof(?*(u8)) * (argc + usize(1)));
|
|
@@ -185,12 +191,12 @@ _spawn_with_fds :: (
|
|
|
185
191
|
pid := e.io.await(IO_process.spawn(program_cstr, argv_buf,.None, stdin_fd, stdout_fd, stderr_fd), e.io);
|
|
186
192
|
// Free argv buffer (storage is dropped naturally at scope end).
|
|
187
193
|
free(.Some(argv_raw));
|
|
188
|
-
|
|
194
|
+
IoError.check(pid, e.exn)
|
|
189
195
|
})
|
|
190
|
-
);
|
|
196
|
+
});
|
|
191
197
|
// Read all available bytes from `fd` into a freshly returned ArrayList(u8)
|
|
192
198
|
// until EOF (read returns 0).
|
|
193
|
-
_drain_fd :: (fn(fd : i32, io :
|
|
199
|
+
_drain_fd :: (fn(fd : i32, io : Io) -> Impl(Future(ArrayList(u8), IoExn)))(
|
|
194
200
|
io.async((e) => {
|
|
195
201
|
buf_size := usize(4096);
|
|
196
202
|
buf := *(u8)(malloc(buf_size).unwrap());
|
|
@@ -200,7 +206,7 @@ _drain_fd :: (fn(fd : i32, io : IO) -> Impl(Future(ArrayList(u8), IOErr)))(
|
|
|
200
206
|
cond(
|
|
201
207
|
(n < i32(0)) => {
|
|
202
208
|
free(.Some(*(void)(buf)));
|
|
203
|
-
e.exn.throw(dyn(
|
|
209
|
+
e.exn.throw(dyn(IoError.from_errno(i32(0) - n)));
|
|
204
210
|
},
|
|
205
211
|
(n == i32(0)) => break,
|
|
206
212
|
true => {
|
|
@@ -223,24 +229,32 @@ impl(
|
|
|
223
229
|
Command,
|
|
224
230
|
/// Spawn the child with stdio inherited from the parent, wait for it to
|
|
225
231
|
/// exit, and return its `ExitStatus`.
|
|
226
|
-
status : (fn(self :
|
|
232
|
+
status : (fn(self : Self, io : Io) -> Impl(Future(ExitStatus, IoExn)))(
|
|
233
|
+
// `self : Self` is a value-copy of the Rc-counted Command object, so the
|
|
234
|
+
// closure can freely capture it. (Was `ref(self)` previously, which is a
|
|
235
|
+
// second-class reference and can't be captured across an async boundary.)
|
|
227
236
|
io.async((e) => {
|
|
228
|
-
pid := e.io.await(
|
|
237
|
+
pid := e.io.await(
|
|
238
|
+
_spawn_with_fds(self, i32(-(1)), i32(-(1)), i32(-(1)), io),
|
|
239
|
+
e
|
|
240
|
+
);
|
|
229
241
|
raw := e.io.await(IO_process.waitpid(pid, i32(0)), e.io);
|
|
230
|
-
|
|
242
|
+
IoError.check(raw, e.exn);
|
|
231
243
|
ExitStatus(raw : raw)
|
|
232
244
|
})
|
|
233
245
|
),
|
|
234
246
|
/// Spawn the child with stdout and stderr captured through pipes. Waits for
|
|
235
247
|
/// the child to exit and returns the exit status plus the captured bytes.
|
|
236
|
-
output : (fn(self :
|
|
248
|
+
output : (fn(self : Self, io : Io) -> Impl(Future(Output, IoExn)))(
|
|
249
|
+
// `self : Self` is a value-copy of the Rc-counted Command object; the
|
|
250
|
+
// closure can freely capture it instead of needing `ref(self)`.
|
|
237
251
|
io.async((e) => {
|
|
238
252
|
// Create stdout and stderr pipes.
|
|
239
253
|
out_fd_buf := MaybeUninit(Array(i32, usize(2))).new();
|
|
240
254
|
out_fd := *(i32)(out_fd_buf.as_ptr());
|
|
241
255
|
pr1 := IO_pipe.pipe(out_fd);
|
|
242
256
|
cond(
|
|
243
|
-
(pr1 < i32(0)) => e.exn.throw(dyn(
|
|
257
|
+
(pr1 < i32(0)) => e.exn.throw(dyn(IoError.from_errno(i32(0) - pr1))),
|
|
244
258
|
true => ()
|
|
245
259
|
);
|
|
246
260
|
out_read := (out_fd &+ usize(0)).*;
|
|
@@ -252,14 +266,17 @@ impl(
|
|
|
252
266
|
(pr2 < i32(0)) => {
|
|
253
267
|
e.io.await(IO_file.close(out_read), e.io);
|
|
254
268
|
e.io.await(IO_file.close(out_write), e.io);
|
|
255
|
-
e.exn.throw(dyn(
|
|
269
|
+
e.exn.throw(dyn(IoError.from_errno(i32(0) - pr2)));
|
|
256
270
|
},
|
|
257
271
|
true => ()
|
|
258
272
|
);
|
|
259
273
|
err_read := (err_fd &+ usize(0)).*;
|
|
260
274
|
err_write := (err_fd &+ usize(1)).*;
|
|
261
275
|
// Spawn child with the write ends of both pipes.
|
|
262
|
-
pid := e.io.await(
|
|
276
|
+
pid := e.io.await(
|
|
277
|
+
_spawn_with_fds(self, i32(-(1)), out_write, err_write, io),
|
|
278
|
+
e
|
|
279
|
+
);
|
|
263
280
|
// Close write ends in parent so reads see EOF when child exits.
|
|
264
281
|
e.io.await(IO_file.close(out_write), e.io);
|
|
265
282
|
e.io.await(IO_file.close(err_write), e.io);
|
|
@@ -269,7 +286,7 @@ impl(
|
|
|
269
286
|
e.io.await(IO_file.close(out_read), e.io);
|
|
270
287
|
e.io.await(IO_file.close(err_read), e.io);
|
|
271
288
|
raw := e.io.await(IO_process.waitpid(pid, i32(0)), e.io);
|
|
272
|
-
|
|
289
|
+
IoError.check(raw, e.exn);
|
|
273
290
|
Output(
|
|
274
291
|
status : ExitStatus(raw : raw),
|
|
275
292
|
stdout : stdout_buf,
|
package/std/process/index.yo
CHANGED
|
@@ -2,6 +2,7 @@
|
|
|
2
2
|
//!
|
|
3
3
|
//! Environment variables, command-line arguments, and current working
|
|
4
4
|
//! directory live in `std/env`.
|
|
5
|
+
pragma(Pragma.AllowUnsafe);
|
|
5
6
|
/// Current target platform as a compile-time string.
|
|
6
7
|
/// One of: "linux", "macos", "windows", "freebsd", "emscripten", "wasi".
|
|
7
8
|
platform :: __yo_process_platform();
|
|
@@ -19,7 +20,7 @@ export(Arch);
|
|
|
19
20
|
/// Exit the process with the given status code.
|
|
20
21
|
exit :: (fn(code : usize) -> unit)({
|
|
21
22
|
{ exit : _exit } :: import("../libc/stdlib");
|
|
22
|
-
_exit(int(code));
|
|
23
|
+
unsafe(_exit(int(code)));
|
|
23
24
|
});
|
|
24
25
|
export(exit);
|
|
25
26
|
_command :: import("./command.yo");
|
package/std/regex/compiler.yo
CHANGED
|
@@ -1,5 +1,6 @@
|
|
|
1
1
|
//! NFA compiler — compiles a `RegexNode` AST into a flat list of NFA instructions
|
|
2
2
|
//! using Thompson's construction algorithm.
|
|
3
|
+
pragma(Pragma.AllowUnsafe);
|
|
3
4
|
open(import("std/collections/array_list"));
|
|
4
5
|
open(import("std/string"));
|
|
5
6
|
{ RegexNode, NodeKind, CharRange, AnchorKind, GroupNameEntry } :: import("./node.yo");
|
|
@@ -269,13 +270,13 @@ impl(
|
|
|
269
270
|
.Sequence => {
|
|
270
271
|
i := usize(0);
|
|
271
272
|
while(i < node.children.len(), i = (i + usize(1)), {
|
|
272
|
-
child := node.children
|
|
273
|
+
child := node.children(i);
|
|
273
274
|
recur(self, child);
|
|
274
275
|
});
|
|
275
276
|
},
|
|
276
277
|
.Alternation => {
|
|
277
|
-
left := node.children
|
|
278
|
-
right := node.children
|
|
278
|
+
left := node.children(usize(0));
|
|
279
|
+
right := node.children(usize(1));
|
|
279
280
|
split_pc := self._emit(Instr.split_instr(usize(0), usize(0)));
|
|
280
281
|
left_start := self._current_pc();
|
|
281
282
|
&(self._program.instructions(split_pc)).* = Instr.split_instr(left_start, usize(0));
|
|
@@ -288,7 +289,7 @@ impl(
|
|
|
288
289
|
&(self._program.instructions(jump_pc)).* = Instr.jump_instr(end_pc);
|
|
289
290
|
},
|
|
290
291
|
.Quantifier => {
|
|
291
|
-
child := node.children
|
|
292
|
+
child := node.children(usize(0));
|
|
292
293
|
min_val := node.q_min;
|
|
293
294
|
max_val := node.q_max;
|
|
294
295
|
greedy := node.q_greedy;
|
|
@@ -352,7 +353,7 @@ impl(
|
|
|
352
353
|
);
|
|
353
354
|
},
|
|
354
355
|
.Group => {
|
|
355
|
-
child := node.children
|
|
356
|
+
child := node.children(usize(0));
|
|
356
357
|
start_slot := (node.group_index * usize(2));
|
|
357
358
|
end_slot := ((node.group_index * usize(2)) + usize(1));
|
|
358
359
|
self._emit(Instr.save_instr(start_slot));
|
|
@@ -360,14 +361,14 @@ impl(
|
|
|
360
361
|
self._emit(Instr.save_instr(end_slot));
|
|
361
362
|
},
|
|
362
363
|
.NonCapturingGroup => {
|
|
363
|
-
child := node.children
|
|
364
|
+
child := node.children(usize(0));
|
|
364
365
|
recur(self, child);
|
|
365
366
|
},
|
|
366
367
|
.Backreference => {
|
|
367
368
|
self._emit(Instr.backref_instr(node.group_index));
|
|
368
369
|
},
|
|
369
370
|
.Lookahead => {
|
|
370
|
-
child := node.children
|
|
371
|
+
child := node.children(usize(0));
|
|
371
372
|
positive := (!(node.negated));
|
|
372
373
|
// Emit lookahead instruction with placeholder sub_end
|
|
373
374
|
la_pc := self._emit(Instr.lookahead_instr(usize(0), usize(0), positive));
|
|
@@ -378,7 +379,7 @@ impl(
|
|
|
378
379
|
&(self._program.instructions(la_pc)).* = Instr.lookahead_instr(sub_start, sub_end, positive);
|
|
379
380
|
},
|
|
380
381
|
.Lookbehind => {
|
|
381
|
-
child := node.children
|
|
382
|
+
child := node.children(usize(0));
|
|
382
383
|
positive := (!(node.negated));
|
|
383
384
|
lb_pc := self._emit(Instr.lookbehind_instr(usize(0), usize(0), positive));
|
|
384
385
|
sub_start := self._current_pc();
|
|
@@ -401,7 +402,7 @@ impl(
|
|
|
401
402
|
instrs := self._program.instructions;
|
|
402
403
|
(done : bool) = false;
|
|
403
404
|
while((pc < instrs.len()) && (!(done)), {
|
|
404
|
-
instr := instrs
|
|
405
|
+
instr := instrs(pc);
|
|
405
406
|
match(
|
|
406
407
|
instr.kind,
|
|
407
408
|
.Save => {
|
package/std/regex/index.yo
CHANGED
|
@@ -61,7 +61,7 @@ impl(
|
|
|
61
61
|
result_bytes := ArrayList(u8).with_capacity(end_pos - start);
|
|
62
62
|
i := start;
|
|
63
63
|
while(i < end_pos, i = (i + usize(1)), {
|
|
64
|
-
result_bytes.push(bytes
|
|
64
|
+
result_bytes.push(bytes(i));
|
|
65
65
|
});
|
|
66
66
|
String.from_bytes(result_bytes)
|
|
67
67
|
}),
|
|
@@ -69,7 +69,7 @@ impl(
|
|
|
69
69
|
char_idx := usize(0);
|
|
70
70
|
i := usize(0);
|
|
71
71
|
while(i < byte_pos, {
|
|
72
|
-
b := bytes
|
|
72
|
+
b := bytes(i);
|
|
73
73
|
char_len := cond(
|
|
74
74
|
(b < u8(0x80)) => usize(1),
|
|
75
75
|
((b >= u8(0xC0)) && (b < u8(0xE0))) => usize(2),
|
|
@@ -91,18 +91,18 @@ impl(
|
|
|
91
91
|
(prefix_len == usize(0)) => from_byte,
|
|
92
92
|
(input_len < prefix_len) => (input_len + usize(1)),
|
|
93
93
|
true => {
|
|
94
|
-
first_byte := prefix
|
|
94
|
+
first_byte := prefix(usize(0));
|
|
95
95
|
(pos : usize) = from_byte;
|
|
96
96
|
(found : bool) = false;
|
|
97
97
|
while((pos <= (input_len - prefix_len)) && (!(found)), {
|
|
98
98
|
cond(
|
|
99
|
-
(input_bytes
|
|
99
|
+
(input_bytes(pos) == first_byte) => {
|
|
100
100
|
// Check remaining prefix bytes
|
|
101
101
|
(match_ok : bool) = true;
|
|
102
102
|
pi := usize(1);
|
|
103
103
|
while((pi < prefix_len) && match_ok, pi = (pi + usize(1)), {
|
|
104
104
|
cond(
|
|
105
|
-
(input_bytes
|
|
105
|
+
(input_bytes(pos + pi) != prefix(pi)) => {
|
|
106
106
|
match_ok = false;
|
|
107
107
|
},
|
|
108
108
|
true => ()
|
|
@@ -136,8 +136,8 @@ impl(
|
|
|
136
136
|
_build_match : (fn(self : Self, slots : ArrayList(usize), input : String) -> RegexMatch)({
|
|
137
137
|
bytes := input.as_bytes();
|
|
138
138
|
unset := usize.MAX;
|
|
139
|
-
match_start_byte := slots
|
|
140
|
-
match_end_byte := slots
|
|
139
|
+
match_start_byte := slots(usize(0));
|
|
140
|
+
match_end_byte := slots(usize(1));
|
|
141
141
|
match_text := self._extract_substring(bytes, match_start_byte, match_end_byte);
|
|
142
142
|
match_char_index := self._byte_to_char_index(bytes, match_start_byte);
|
|
143
143
|
groups := ArrayList(Option(String)).new();
|
|
@@ -147,8 +147,8 @@ impl(
|
|
|
147
147
|
end_slot := ((g * usize(2)) + usize(1));
|
|
148
148
|
cond(
|
|
149
149
|
((start_slot < slots.len()) && (end_slot < slots.len())) => {
|
|
150
|
-
gs := slots
|
|
151
|
-
ge := slots
|
|
150
|
+
gs := slots(start_slot);
|
|
151
|
+
ge := slots(end_slot);
|
|
152
152
|
cond(
|
|
153
153
|
((gs != unset) && (ge != unset)) => {
|
|
154
154
|
group_text := self._extract_substring(bytes, gs, ge);
|
|
@@ -203,7 +203,7 @@ impl(
|
|
|
203
203
|
break;
|
|
204
204
|
},
|
|
205
205
|
true => {
|
|
206
|
-
b := bytes
|
|
206
|
+
b := bytes(byte_pos);
|
|
207
207
|
char_len := cond(
|
|
208
208
|
(b < u8(0x80)) => usize(1),
|
|
209
209
|
((b >= u8(0xC0)) && (b < u8(0xE0))) => usize(2),
|
|
@@ -237,8 +237,8 @@ impl(
|
|
|
237
237
|
result.matched => {
|
|
238
238
|
m := self._build_match(result.slots, input);
|
|
239
239
|
matches.push(m);
|
|
240
|
-
match_start := result.slots
|
|
241
|
-
match_end := result.slots
|
|
240
|
+
match_start := result.slots(usize(0));
|
|
241
|
+
match_end := result.slots(usize(1));
|
|
242
242
|
cond(
|
|
243
243
|
(match_end == match_start) => {
|
|
244
244
|
// Empty match: advance one char to avoid infinite loop
|
|
@@ -247,7 +247,7 @@ impl(
|
|
|
247
247
|
break;
|
|
248
248
|
},
|
|
249
249
|
true => {
|
|
250
|
-
b := bytes
|
|
250
|
+
b := bytes(byte_pos);
|
|
251
251
|
char_len := cond(
|
|
252
252
|
(b < u8(0x80)) => usize(1),
|
|
253
253
|
((b >= u8(0xC0)) && (b < u8(0xE0))) => usize(2),
|
|
@@ -281,8 +281,8 @@ impl(
|
|
|
281
281
|
result.matched => {
|
|
282
282
|
m := self._build_match(result.slots, input);
|
|
283
283
|
matches.push(m);
|
|
284
|
-
match_start := result.slots
|
|
285
|
-
match_end := result.slots
|
|
284
|
+
match_start := result.slots(usize(0));
|
|
285
|
+
match_end := result.slots(usize(1));
|
|
286
286
|
cond(
|
|
287
287
|
(match_end == match_start) => {
|
|
288
288
|
cond(
|
|
@@ -290,7 +290,7 @@ impl(
|
|
|
290
290
|
break;
|
|
291
291
|
},
|
|
292
292
|
true => {
|
|
293
|
-
b := bytes
|
|
293
|
+
b := bytes(byte_pos);
|
|
294
294
|
char_len := cond(
|
|
295
295
|
(b < u8(0x80)) => usize(1),
|
|
296
296
|
((b >= u8(0xC0)) && (b < u8(0xE0))) => usize(2),
|
|
@@ -319,7 +319,7 @@ impl(
|
|
|
319
319
|
break;
|
|
320
320
|
},
|
|
321
321
|
true => {
|
|
322
|
-
b := bytes
|
|
322
|
+
b := bytes(byte_pos);
|
|
323
323
|
char_len := cond(
|
|
324
324
|
(b < u8(0x80)) => usize(1),
|
|
325
325
|
((b >= u8(0xC0)) && (b < u8(0xE0))) => usize(2),
|
|
@@ -368,11 +368,11 @@ impl(
|
|
|
368
368
|
result := ArrayList(u8).new();
|
|
369
369
|
i := usize(0);
|
|
370
370
|
while(i < rep_len, {
|
|
371
|
-
b := rep_bytes
|
|
371
|
+
b := rep_bytes(i);
|
|
372
372
|
cond(
|
|
373
373
|
((b == u8(36)) && ((i + usize(1)) < rep_len)) => {
|
|
374
374
|
// '$' character — check next char
|
|
375
|
-
next_b := rep_bytes
|
|
375
|
+
next_b := rep_bytes(i + usize(1));
|
|
376
376
|
cond(
|
|
377
377
|
(next_b == u8(36)) => {
|
|
378
378
|
// $$ → literal $
|
|
@@ -384,7 +384,7 @@ impl(
|
|
|
384
384
|
match_bytes := m.value().as_bytes();
|
|
385
385
|
mi := usize(0);
|
|
386
386
|
while(mi < match_bytes.len(), mi = (mi + usize(1)), {
|
|
387
|
-
result.push(match_bytes
|
|
387
|
+
result.push(match_bytes(mi));
|
|
388
388
|
});
|
|
389
389
|
i = (i + usize(2));
|
|
390
390
|
},
|
|
@@ -396,7 +396,7 @@ impl(
|
|
|
396
396
|
byte_idx := usize(0);
|
|
397
397
|
ci := usize(0);
|
|
398
398
|
while(ci < char_idx, {
|
|
399
|
-
cb := input_bytes
|
|
399
|
+
cb := input_bytes(byte_idx);
|
|
400
400
|
cbl := cond(
|
|
401
401
|
(cb < u8(0x80)) => usize(1),
|
|
402
402
|
((cb >= u8(0xC0)) && (cb < u8(0xE0))) => usize(2),
|
|
@@ -408,7 +408,7 @@ impl(
|
|
|
408
408
|
});
|
|
409
409
|
pi := usize(0);
|
|
410
410
|
while(pi < byte_idx, pi = (pi + usize(1)), {
|
|
411
|
-
result.push(input_bytes
|
|
411
|
+
result.push(input_bytes(pi));
|
|
412
412
|
});
|
|
413
413
|
i = (i + usize(2));
|
|
414
414
|
},
|
|
@@ -420,7 +420,7 @@ impl(
|
|
|
420
420
|
byte_idx := usize(0);
|
|
421
421
|
ci := usize(0);
|
|
422
422
|
while(ci < char_idx, {
|
|
423
|
-
cb := input_bytes
|
|
423
|
+
cb := input_bytes(byte_idx);
|
|
424
424
|
cbl := cond(
|
|
425
425
|
(cb < u8(0x80)) => usize(1),
|
|
426
426
|
((cb >= u8(0xC0)) && (cb < u8(0xE0))) => usize(2),
|
|
@@ -435,7 +435,7 @@ impl(
|
|
|
435
435
|
byte_idx = (byte_idx + match_bytes.len());
|
|
436
436
|
pi := byte_idx;
|
|
437
437
|
while(pi < input_bytes.len(), pi = (pi + usize(1)), {
|
|
438
|
-
result.push(input_bytes
|
|
438
|
+
result.push(input_bytes(pi));
|
|
439
439
|
});
|
|
440
440
|
i = (i + usize(2));
|
|
441
441
|
},
|
|
@@ -449,7 +449,7 @@ impl(
|
|
|
449
449
|
g_bytes := g.as_bytes();
|
|
450
450
|
gi := usize(0);
|
|
451
451
|
while(gi < g_bytes.len(), gi = (gi + usize(1)), {
|
|
452
|
-
result.push(g_bytes
|
|
452
|
+
result.push(g_bytes(gi));
|
|
453
453
|
});
|
|
454
454
|
},
|
|
455
455
|
.None => ()
|
|
@@ -462,7 +462,7 @@ impl(
|
|
|
462
462
|
(name_end : usize) = name_start;
|
|
463
463
|
(found_close : bool) = false;
|
|
464
464
|
while(name_end < rep_len, {
|
|
465
|
-
nb := rep_bytes
|
|
465
|
+
nb := rep_bytes(name_end);
|
|
466
466
|
cond(
|
|
467
467
|
(nb == u8(125)) => {
|
|
468
468
|
found_close = true;
|
|
@@ -478,7 +478,7 @@ impl(
|
|
|
478
478
|
name_bytes := ArrayList(u8).with_capacity(name_end - name_start);
|
|
479
479
|
ni := name_start;
|
|
480
480
|
while(ni < name_end, ni = (ni + usize(1)), {
|
|
481
|
-
name_bytes.push(rep_bytes
|
|
481
|
+
name_bytes.push(rep_bytes(ni));
|
|
482
482
|
});
|
|
483
483
|
name := String.from_bytes(name_bytes);
|
|
484
484
|
grp := m.named_group(name);
|
|
@@ -488,7 +488,7 @@ impl(
|
|
|
488
488
|
g_bytes := g.as_bytes();
|
|
489
489
|
gi := usize(0);
|
|
490
490
|
while(gi < g_bytes.len(), gi = (gi + usize(1)), {
|
|
491
|
-
result.push(g_bytes
|
|
491
|
+
result.push(g_bytes(gi));
|
|
492
492
|
});
|
|
493
493
|
},
|
|
494
494
|
.None => ()
|
|
@@ -533,7 +533,7 @@ impl(
|
|
|
533
533
|
(match_start_byte : usize) = usize(0);
|
|
534
534
|
ci := usize(0);
|
|
535
535
|
while(ci < char_idx, {
|
|
536
|
-
cb := input_bytes
|
|
536
|
+
cb := input_bytes(match_start_byte);
|
|
537
537
|
cbl := cond(
|
|
538
538
|
(cb < u8(0x80)) => usize(1),
|
|
539
539
|
((cb >= u8(0xC0)) && (cb < u8(0xE0))) => usize(2),
|
|
@@ -548,17 +548,17 @@ impl(
|
|
|
548
548
|
out := ArrayList(u8).new();
|
|
549
549
|
pi := usize(0);
|
|
550
550
|
while(pi < match_start_byte, pi = (pi + usize(1)), {
|
|
551
|
-
out.push(input_bytes
|
|
551
|
+
out.push(input_bytes(pi));
|
|
552
552
|
});
|
|
553
553
|
rep := self._apply_replacement(replacement, m);
|
|
554
554
|
rep_bytes := rep.as_bytes();
|
|
555
555
|
ri := usize(0);
|
|
556
556
|
while(ri < rep_bytes.len(), ri = (ri + usize(1)), {
|
|
557
|
-
out.push(rep_bytes
|
|
557
|
+
out.push(rep_bytes(ri));
|
|
558
558
|
});
|
|
559
559
|
pi = match_end_byte;
|
|
560
560
|
while(pi < input_bytes.len(), pi = (pi + usize(1)), {
|
|
561
|
-
out.push(input_bytes
|
|
561
|
+
out.push(input_bytes(pi));
|
|
562
562
|
});
|
|
563
563
|
String.from_bytes(out)
|
|
564
564
|
}
|
|
@@ -574,13 +574,13 @@ impl(
|
|
|
574
574
|
(last_end_byte : usize) = usize(0);
|
|
575
575
|
mi := usize(0);
|
|
576
576
|
while(mi < all_matches.len(), mi = (mi + usize(1)), {
|
|
577
|
-
m := all_matches
|
|
577
|
+
m := all_matches(mi);
|
|
578
578
|
// Convert char index to byte index for match start
|
|
579
579
|
char_idx := m.index();
|
|
580
580
|
(match_start_byte : usize) = usize(0);
|
|
581
581
|
ci := usize(0);
|
|
582
582
|
while(ci < char_idx, {
|
|
583
|
-
cb := input_bytes
|
|
583
|
+
cb := input_bytes(match_start_byte);
|
|
584
584
|
cbl := cond(
|
|
585
585
|
(cb < u8(0x80)) => usize(1),
|
|
586
586
|
((cb >= u8(0xC0)) && (cb < u8(0xE0))) => usize(2),
|
|
@@ -594,21 +594,21 @@ impl(
|
|
|
594
594
|
// Copy text between last match end and this match start
|
|
595
595
|
pi := last_end_byte;
|
|
596
596
|
while(pi < match_start_byte, pi = (pi + usize(1)), {
|
|
597
|
-
out.push(input_bytes
|
|
597
|
+
out.push(input_bytes(pi));
|
|
598
598
|
});
|
|
599
599
|
// Apply replacement
|
|
600
600
|
rep := self._apply_replacement(replacement, m);
|
|
601
601
|
rep_bytes := rep.as_bytes();
|
|
602
602
|
ri := usize(0);
|
|
603
603
|
while(ri < rep_bytes.len(), ri = (ri + usize(1)), {
|
|
604
|
-
out.push(rep_bytes
|
|
604
|
+
out.push(rep_bytes(ri));
|
|
605
605
|
});
|
|
606
606
|
last_end_byte = match_end_byte;
|
|
607
607
|
});
|
|
608
608
|
// Copy remaining text after last match
|
|
609
609
|
pi := last_end_byte;
|
|
610
610
|
while(pi < input_bytes.len(), pi = (pi + usize(1)), {
|
|
611
|
-
out.push(input_bytes
|
|
611
|
+
out.push(input_bytes(pi));
|
|
612
612
|
});
|
|
613
613
|
String.from_bytes(out)
|
|
614
614
|
}
|
|
@@ -626,13 +626,13 @@ impl(
|
|
|
626
626
|
(last_end_byte : usize) = usize(0);
|
|
627
627
|
mi := usize(0);
|
|
628
628
|
while(mi < all_matches.len(), mi = (mi + usize(1)), {
|
|
629
|
-
m := all_matches
|
|
629
|
+
m := all_matches(mi);
|
|
630
630
|
// Convert char index to byte index for match start
|
|
631
631
|
char_idx := m.index();
|
|
632
632
|
(match_start_byte : usize) = usize(0);
|
|
633
633
|
ci := usize(0);
|
|
634
634
|
while(ci < char_idx, {
|
|
635
|
-
cb := input_bytes
|
|
635
|
+
cb := input_bytes(match_start_byte);
|
|
636
636
|
cbl := cond(
|
|
637
637
|
(cb < u8(0x80)) => usize(1),
|
|
638
638
|
((cb >= u8(0xC0)) && (cb < u8(0xE0))) => usize(2),
|
|
@@ -647,7 +647,7 @@ impl(
|
|
|
647
647
|
part_bytes := ArrayList(u8).new();
|
|
648
648
|
pi := last_end_byte;
|
|
649
649
|
while(pi < match_start_byte, pi = (pi + usize(1)), {
|
|
650
|
-
part_bytes.push(input_bytes
|
|
650
|
+
part_bytes.push(input_bytes(pi));
|
|
651
651
|
});
|
|
652
652
|
parts.push(String.from_bytes(part_bytes));
|
|
653
653
|
// Include capture groups in split result (like JS)
|
|
@@ -670,7 +670,7 @@ impl(
|
|
|
670
670
|
remaining := ArrayList(u8).new();
|
|
671
671
|
pi := last_end_byte;
|
|
672
672
|
while(pi < input_bytes.len(), pi = (pi + usize(1)), {
|
|
673
|
-
remaining.push(input_bytes
|
|
673
|
+
remaining.push(input_bytes(pi));
|
|
674
674
|
});
|
|
675
675
|
parts.push(String.from_bytes(remaining));
|
|
676
676
|
}
|
package/std/regex/match.yo
CHANGED
|
@@ -42,7 +42,7 @@ impl(
|
|
|
42
42
|
true => {
|
|
43
43
|
actual_idx := (idx - usize(1));
|
|
44
44
|
cond(
|
|
45
|
-
(actual_idx < self._groups.len()) => self._groups
|
|
45
|
+
(actual_idx < self._groups.len()) => self._groups(actual_idx),
|
|
46
46
|
true =>.None
|
|
47
47
|
)
|
|
48
48
|
}
|
|
@@ -52,7 +52,7 @@ impl(
|
|
|
52
52
|
named_group : (fn(self : Self, name : String) -> Option(String))({
|
|
53
53
|
i := usize(0);
|
|
54
54
|
while(i < self._group_names.len(), i = (i + usize(1)), {
|
|
55
|
-
entry := self._group_names
|
|
55
|
+
entry := self._group_names(i);
|
|
56
56
|
cond(
|
|
57
57
|
(entry.name == name) => {
|
|
58
58
|
return(self.group(entry.index));
|