From eee450516e3ef03749f6f1565830ba726899413e Mon Sep 17 00:00:00 2001 From: gingerBill Date: Wed, 26 Mar 2025 17:45:04 +0000 Subject: Improve bodge for android execute build --- src/string.cpp | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'src/string.cpp') diff --git a/src/string.cpp b/src/string.cpp index b001adf0e..88b679540 100644 --- a/src/string.cpp +++ b/src/string.cpp @@ -273,6 +273,15 @@ gb_internal String path_extension(String const &str, bool include_dot = true) { return substring(str, include_dot ? pos : pos + 1, str.len); } + +gb_internal String path_remove_extension(String const &str) { + isize pos = string_extension_position(str); + if (pos < 0) { + return str; + } + return substring(str, 0, pos); +} + gb_internal String string_trim_whitespace(String str) { while (str.len > 0 && rune_is_whitespace(str[str.len-1])) { str.len--; -- cgit v1.2.3 From 1f814c33dcfccec9073d15214268457c0d6ea273 Mon Sep 17 00:00:00 2001 From: gingerBill Date: Thu, 10 Apr 2025 11:36:03 +0100 Subject: Support subtargets in build tags: `#build darwin:generic` and `#build linux:android, darwin:ios` --- base/runtime/core.odin | 4 +++ core/odin/parser/file_tags.odin | 24 ++++++++++--- src/build_settings.cpp | 34 +++++++++++++++--- src/parser.cpp | 12 ++++--- src/string.cpp | 77 +++++++++++++++++++++++++++++++++++++++++ 5 files changed, 138 insertions(+), 13 deletions(-) (limited to 'src/string.cpp') diff --git a/base/runtime/core.odin b/base/runtime/core.odin index cd7d35540..5ab8a7008 100644 --- a/base/runtime/core.odin +++ b/base/runtime/core.odin @@ -559,10 +559,14 @@ ALL_ODIN_OS_TYPES :: Odin_OS_Types{ Odin_Platform_Subtarget_Type :: enum int { Default, iOS, + Android, } */ Odin_Platform_Subtarget_Type :: type_of(ODIN_PLATFORM_SUBTARGET) +Odin_Platform_Subtarget_Types :: bit_set[Odin_Platform_Subtarget_Type] + + /* // Defined internally by the compiler Odin_Sanitizer_Flag :: enum u32 { diff --git a/core/odin/parser/file_tags.odin b/core/odin/parser/file_tags.odin index c5c6637c3..fff59ee1e 100644 --- a/core/odin/parser/file_tags.odin +++ b/core/odin/parser/file_tags.odin @@ -30,14 +30,27 @@ File_Tags :: struct { } @require_results -get_build_os_from_string :: proc(str: string) -> runtime.Odin_OS_Type { +get_build_os_from_string :: proc(str: string) -> (found_os: runtime.Odin_OS_Type, found_subtarget: runtime.Odin_Platform_Subtarget_Type) { + str_os, _, str_subtarget := strings.partition(str, ":") + fields := reflect.enum_fields_zipped(runtime.Odin_OS_Type) for os in fields { - if strings.equal_fold(os.name, str) { - return runtime.Odin_OS_Type(os.value) + if strings.equal_fold(os.name, str_os) { + found_os = runtime.Odin_OS_Type(os.value) + break } } - return .Unknown + if str_subtarget != "" { + fields := reflect.enum_fields_zipped(runtime.Odin_Platform_Subtarget_Type) + for subtarget in fields { + if strings.equal_fold(subtarget.name, str_subtarget) { + found_subtarget = runtime.Odin_Platform_Subtarget_Type(subtarget.value) + break + } + } + } + + return } @require_results get_build_arch_from_string :: proc(str: string) -> runtime.Odin_Arch_Type { @@ -187,7 +200,8 @@ parse_file_tags :: proc(file: ast.File, allocator := context.allocator) -> (tags if value == "ignore" { tags.ignore = true - } else if os := get_build_os_from_string(value); os != .Unknown { + } else if os, subtarget := get_build_os_from_string(value); os != .Unknown { + _ = subtarget // TODO(bill): figure out how to handle the subtarget logic if is_notted { os_negative += {os} } else { diff --git a/src/build_settings.cpp b/src/build_settings.cpp index c2575e2d4..c941e0f68 100644 --- a/src/build_settings.cpp +++ b/src/build_settings.cpp @@ -847,13 +847,39 @@ gb_global NamedTargetMetrics *selected_target_metrics; gb_global Subtarget selected_subtarget; -gb_internal TargetOsKind get_target_os_from_string(String str) { +gb_internal TargetOsKind get_target_os_from_string(String str, Subtarget *subtarget_ = nullptr) { + String os_name = str; + String subtarget = {}; + auto part = string_partition(str, str_lit(":")); + if (part.match.len == 1) { + os_name = part.head; + subtarget = part.tail; + } + + TargetOsKind kind = TargetOs_Invalid; + for (isize i = 0; i < TargetOs_COUNT; i++) { - if (str_eq_ignore_case(target_os_names[i], str)) { - return cast(TargetOsKind)i; + if (str_eq_ignore_case(target_os_names[i], os_name)) { + kind = cast(TargetOsKind)i; + break; } } - return TargetOs_Invalid; + if (subtarget_) *subtarget_ = Subtarget_Default; + + if (subtarget.len != 0) { + if (str_eq_ignore_case(subtarget, "generic") || str_eq_ignore_case(subtarget, "default")) { + if (subtarget_) *subtarget_ = Subtarget_Default; + } else { + for (isize i = 1; i < Subtarget_COUNT; i++) { + if (str_eq_ignore_case(subtarget_strings[i], subtarget)) { + if (subtarget_) *subtarget_ = cast(Subtarget)i; + break; + } + } + } + } + + return kind; } gb_internal TargetArchKind get_target_arch_from_string(String str) { diff --git a/src/parser.cpp b/src/parser.cpp index f38f79607..a397585e8 100644 --- a/src/parser.cpp +++ b/src/parser.cpp @@ -6157,7 +6157,7 @@ gb_internal String build_tag_get_token(String s, String *out) { isize width = utf8_decode(&s[n], s.len-n, &rune); if (n == 0 && rune == '!') { - } else if (!rune_is_letter(rune) && !rune_is_digit(rune)) { + } else if (!rune_is_letter(rune) && !rune_is_digit(rune) && rune != ':') { isize k = gb_max(gb_max(n, width), 1); *out = substring(s, k, s.len); return substring(s, 0, k); @@ -6209,7 +6209,9 @@ gb_internal bool parse_build_tag(Token token_for_pos, String s) { continue; } - TargetOsKind os = get_target_os_from_string(p); + Subtarget subtarget = Subtarget_Default; + + TargetOsKind os = get_target_os_from_string(p, &subtarget); TargetArchKind arch = get_target_arch_from_string(p); num_tokens += 1; @@ -6223,11 +6225,13 @@ gb_internal bool parse_build_tag(Token token_for_pos, String s) { if (os != TargetOs_Invalid) { this_kind_os_seen = true; + bool same_subtarget = (subtarget == Subtarget_Default) || (subtarget == selected_subtarget); + GB_ASSERT(arch == TargetArch_Invalid); if (is_notted) { - this_kind_correct = this_kind_correct && (os != build_context.metrics.os); + this_kind_correct = this_kind_correct && (os != build_context.metrics.os || !same_subtarget); } else { - this_kind_correct = this_kind_correct && (os == build_context.metrics.os); + this_kind_correct = this_kind_correct && (os == build_context.metrics.os && same_subtarget); } } else if (arch != TargetArch_Invalid) { this_kind_arch_seen = true; diff --git a/src/string.cpp b/src/string.cpp index 88b679540..ae8d066b1 100644 --- a/src/string.cpp +++ b/src/string.cpp @@ -336,6 +336,83 @@ gb_internal Array split_lines_from_array(Array const &array, gbAlloc return lines; } +enum : u32 { PRIME_RABIN_KARP = 16777619u }; + +gb_internal u32 hash_str_rabin_karp(String const &s, u32 *pow_) { + u32 hash = 0; + u32 pow = 1; + for (isize i = 0; i < s.len; i++) { + hash = hash*PRIME_RABIN_KARP + cast(u32)s.text[i]; + } + u32 sq = PRIME_RABIN_KARP; + for (isize i = s.len; i > 0; i >>= 1) { + if ((i & 1) != 0) { + pow *= sq; + } + sq *= sq; + } + if (pow_) *pow_ = pow; + return hash; + +} + + +gb_internal isize string_index(String const &s, String const &substr) { + isize n = substr.len; + if (n == 0) { + return 0; + } else if (n == 1) { + return string_index_byte(s, substr[0]); + } else if (n == s.len) { + if (s == substr) { + return 0; + } + return -1; + } else if (n > s.len) { + return -1; + } + u32 pow = 1; + u32 hash = hash_str_rabin_karp(s, &pow); + u32 h = 0; + for (isize i = 0; i < n; i++) { + h = h*PRIME_RABIN_KARP + cast(u32)s.text[i]; + } + if (h == hash && substring(s, 0, n) == substr) { + return 0; + } + for (isize i = n; i < s.len; /**/) { + h *= PRIME_RABIN_KARP; + h += cast(u32)s.text[i]; + h -= pow * u32(s.text[i-n]); + i += 1; + if (h == hash && substring(s, i-n, i) == substr) { + return i - n; + } + } + return -1; +} + + +struct StringPartition { + String head; + String match; + String tail; +}; + +gb_internal StringPartition string_partition(String const &str, String const &sep) { + StringPartition res = {}; + isize i = string_index(str, sep); + if (i < 0) { + res.head = str; + return res; + } + + res.head = substring(str, 0, i); + res.match = substring(str, i, i+sep.len); + res.tail = substring(str, i+sep.len, str.len); + return res; +} + gb_internal bool string_contains_char(String const &s, u8 c) { isize i; for (i = 0; i < s.len; i++) { -- cgit v1.2.3 From 2561427dd396a69cd49eb02c0814c4e8e8b3a08f Mon Sep 17 00:00:00 2001 From: gingerBill Date: Sat, 2 Aug 2025 11:00:15 +0100 Subject: Add `string16` and `cstring16` (UTF-16 based strings) --- base/runtime/core.odin | 13 +++- base/runtime/core_builtin.odin | 28 ++++++- base/runtime/internal.odin | 87 +++++++++++++++++++++ core/fmt/fmt.odin | 76 ++++++++++++++++++ core/io/io.odin | 24 ++++++ core/io/util.odin | 27 +++++++ core/unicode/utf16/utf16.odin | 20 +++++ src/build_settings.cpp | 8 +- src/cached.cpp | 2 +- src/check_builtin.cpp | 9 ++- src/check_decl.cpp | 6 ++ src/check_expr.cpp | 88 +++++++++++++++++++++ src/checker.cpp | 16 ++-- src/common.cpp | 2 +- src/exact_value.cpp | 52 ++++++++++++- src/llvm_backend_expr.cpp | 72 +++++++++++++++++ src/llvm_backend_general.cpp | 31 ++++++++ src/llvm_backend_proc.cpp | 9 +++ src/llvm_backend_type.cpp | 28 ++++++- src/llvm_backend_utility.cpp | 12 +++ src/main.cpp | 8 +- src/microsoft_craziness.h | 4 +- src/path.cpp | 8 +- src/string.cpp | 172 +++++++++++++++++++++++++++++++++++++---- src/types.cpp | 133 +++++++++++++++++++++++++++---- 25 files changed, 873 insertions(+), 62 deletions(-) (limited to 'src/string.cpp') diff --git a/base/runtime/core.odin b/base/runtime/core.odin index baecb4146..fe40427ff 100644 --- a/base/runtime/core.odin +++ b/base/runtime/core.odin @@ -73,7 +73,7 @@ Type_Info_Rune :: struct {} Type_Info_Float :: struct {endianness: Platform_Endianness} Type_Info_Complex :: struct {} Type_Info_Quaternion :: struct {} -Type_Info_String :: struct {is_cstring: bool} +Type_Info_String :: struct {is_cstring: bool, is_utf16: bool} Type_Info_Boolean :: struct {} Type_Info_Any :: struct {} Type_Info_Type_Id :: struct {} @@ -397,6 +397,11 @@ Raw_String :: struct { len: int, } +Raw_String16 :: struct { + data: [^]u16, + len: int, +} + Raw_Slice :: struct { data: rawptr, len: int, @@ -450,6 +455,12 @@ Raw_Cstring :: struct { } #assert(size_of(Raw_Cstring) == size_of(cstring)) +Raw_Cstring16 :: struct { + data: [^]u16, +} +#assert(size_of(Raw_Cstring16) == size_of(cstring16)) + + Raw_Soa_Pointer :: struct { data: rawptr, index: int, diff --git a/base/runtime/core_builtin.odin b/base/runtime/core_builtin.odin index e2ba14f3a..09118998c 100644 --- a/base/runtime/core_builtin.odin +++ b/base/runtime/core_builtin.odin @@ -86,11 +86,26 @@ copy_from_string :: proc "contextless" (dst: $T/[]$E/u8, src: $S/string) -> int } return n } + +// `copy_from_string16` is a built-in procedure that copies elements from a source string `src` to a destination slice `dst`. +// The source and destination may overlap. Copy returns the number of elements copied, which will be the minimum +// of len(src) and len(dst). +// +// Prefer the procedure group `copy`. +@builtin +copy_from_string16 :: proc "contextless" (dst: $T/[]$E/u16, src: $S/string16) -> int { + n := min(len(dst), len(src)) + if n > 0 { + intrinsics.mem_copy(raw_data(dst), raw_data(src), n*size_of(u16)) + } + return n +} + // `copy` is a built-in procedure that copies elements from a source slice/string `src` to a destination slice `dst`. // The source and destination may overlap. Copy returns the number of elements copied, which will be the minimum // of len(src) and len(dst). @builtin -copy :: proc{copy_slice, copy_from_string} +copy :: proc{copy_slice, copy_from_string, copy_from_string16} @@ -285,6 +300,15 @@ delete_map :: proc(m: $T/map[$K]$V, loc := #caller_location) -> Allocator_Error } +@builtin +delete_string16 :: proc(str: string16, allocator := context.allocator, loc := #caller_location) -> Allocator_Error { + return mem_free_with_size(raw_data(str), len(str)*size_of(u16), allocator, loc) +} +@builtin +delete_cstring16 :: proc(str: cstring16, allocator := context.allocator, loc := #caller_location) -> Allocator_Error { + return mem_free((^u16)(str), allocator, loc) +} + // `delete` will try to free the underlying data of the passed built-in data structure (string, cstring, dynamic array, slice, or map), with the given `allocator` if the allocator supports this operation. // // Note: Prefer `delete` over the specific `delete_*` procedures where possible. @@ -297,6 +321,8 @@ delete :: proc{ delete_map, delete_soa_slice, delete_soa_dynamic_array, + delete_string16, + delete_cstring16, } diff --git a/base/runtime/internal.odin b/base/runtime/internal.odin index 907b187f1..660af58ab 100644 --- a/base/runtime/internal.odin +++ b/base/runtime/internal.odin @@ -493,12 +493,40 @@ string_cmp :: proc "contextless" (a, b: string) -> int { return ret } + +string16_eq :: proc "contextless" (lhs, rhs: string16) -> bool { + x := transmute(Raw_String16)lhs + y := transmute(Raw_String16)rhs + if x.len != y.len { + return false + } + return #force_inline memory_equal(x.data, y.data, x.len*size_of(u16)) +} + +string16_cmp :: proc "contextless" (a, b: string16) -> int { + x := transmute(Raw_String16)a + y := transmute(Raw_String16)b + + ret := memory_compare(x.data, y.data, min(x.len, y.len)*size_of(u16)) + if ret == 0 && x.len != y.len { + return -1 if x.len < y.len else +1 + } + return ret +} + string_ne :: #force_inline proc "contextless" (a, b: string) -> bool { return !string_eq(a, b) } string_lt :: #force_inline proc "contextless" (a, b: string) -> bool { return string_cmp(a, b) < 0 } string_gt :: #force_inline proc "contextless" (a, b: string) -> bool { return string_cmp(a, b) > 0 } string_le :: #force_inline proc "contextless" (a, b: string) -> bool { return string_cmp(a, b) <= 0 } string_ge :: #force_inline proc "contextless" (a, b: string) -> bool { return string_cmp(a, b) >= 0 } +string16_ne :: #force_inline proc "contextless" (a, b: string16) -> bool { return !string16_eq(a, b) } +string16_lt :: #force_inline proc "contextless" (a, b: string16) -> bool { return string16_cmp(a, b) < 0 } +string16_gt :: #force_inline proc "contextless" (a, b: string16) -> bool { return string16_cmp(a, b) > 0 } +string16_le :: #force_inline proc "contextless" (a, b: string16) -> bool { return string16_cmp(a, b) <= 0 } +string16_ge :: #force_inline proc "contextless" (a, b: string16) -> bool { return string16_cmp(a, b) >= 0 } + + cstring_len :: proc "contextless" (s: cstring) -> int { p0 := uintptr((^byte)(s)) p := p0 @@ -508,6 +536,16 @@ cstring_len :: proc "contextless" (s: cstring) -> int { return int(p - p0) } +cstring16_len :: proc "contextless" (s: cstring16) -> int { + p := ([^]u16)(s) + n := 0 + for p != nil && p[0] != 0 { + p = p[1:] + n += 1 + } + return n +} + cstring_to_string :: proc "contextless" (s: cstring) -> string { if s == nil { return "" @@ -517,6 +555,15 @@ cstring_to_string :: proc "contextless" (s: cstring) -> string { return transmute(string)Raw_String{ptr, n} } +cstring16_to_string16 :: proc "contextless" (s: cstring16) -> string16 { + if s == nil { + return "" + } + ptr := (^u16)(s) + n := cstring16_len(s) + return transmute(string16)Raw_String16{ptr, n} +} + cstring_eq :: proc "contextless" (lhs, rhs: cstring) -> bool { x := ([^]byte)(lhs) @@ -559,6 +606,46 @@ cstring_gt :: #force_inline proc "contextless" (a, b: cstring) -> bool { return cstring_le :: #force_inline proc "contextless" (a, b: cstring) -> bool { return cstring_cmp(a, b) <= 0 } cstring_ge :: #force_inline proc "contextless" (a, b: cstring) -> bool { return cstring_cmp(a, b) >= 0 } +cstring16_eq :: proc "contextless" (lhs, rhs: cstring16) -> bool { + x := ([^]u16)(lhs) + y := ([^]u16)(rhs) + if x == y { + return true + } + if (x == nil) ~ (y == nil) { + return false + } + xn := cstring16_len(lhs) + yn := cstring16_len(rhs) + if xn != yn { + return false + } + return #force_inline memory_equal(x, y, xn*size_of(u16)) +} + +cstring16_cmp :: proc "contextless" (lhs, rhs: cstring16) -> int { + x := ([^]u16)(lhs) + y := ([^]u16)(rhs) + if x == y { + return 0 + } + if (x == nil) ~ (y == nil) { + return -1 if x == nil else +1 + } + xn := cstring16_len(lhs) + yn := cstring16_len(rhs) + ret := memory_compare(x, y, min(xn, yn)*size_of(u16)) + if ret == 0 && xn != yn { + return -1 if xn < yn else +1 + } + return ret +} + +cstring16_ne :: #force_inline proc "contextless" (a, b: cstring16) -> bool { return !cstring16_eq(a, b) } +cstring16_lt :: #force_inline proc "contextless" (a, b: cstring16) -> bool { return cstring16_cmp(a, b) < 0 } +cstring16_gt :: #force_inline proc "contextless" (a, b: cstring16) -> bool { return cstring16_cmp(a, b) > 0 } +cstring16_le :: #force_inline proc "contextless" (a, b: cstring16) -> bool { return cstring16_cmp(a, b) <= 0 } +cstring16_ge :: #force_inline proc "contextless" (a, b: cstring16) -> bool { return cstring16_cmp(a, b) >= 0 } complex32_eq :: #force_inline proc "contextless" (a, b: complex32) -> bool { return real(a) == real(b) && imag(a) == imag(b) } complex32_ne :: #force_inline proc "contextless" (a, b: complex32) -> bool { return real(a) != real(b) || imag(a) != imag(b) } diff --git a/core/fmt/fmt.odin b/core/fmt/fmt.odin index 0f6470cca..7fe6287d4 100644 --- a/core/fmt/fmt.odin +++ b/core/fmt/fmt.odin @@ -1551,6 +1551,79 @@ fmt_string :: proc(fi: ^Info, s: string, verb: rune) { fmt_cstring :: proc(fi: ^Info, s: cstring, verb: rune) { fmt_string(fi, string(s), verb) } + +// Formats a string UTF-16 with a specific format. +// +// Inputs: +// - fi: Pointer to the Info struct containing format settings. +// - s: The string to format. +// - verb: The format specifier character (e.g. 's', 'v', 'q', 'x', 'X'). +// +fmt_string16 :: proc(fi: ^Info, s: string16, verb: rune) { + s, verb := s, verb + if ol, ok := fi.optional_len.?; ok { + s = s[:clamp(ol, 0, len(s))] + } + if !fi.in_bad && fi.record_level > 0 && verb == 'v' { + verb = 'q' + } + + switch verb { + case 's', 'v': + if fi.width_set { + if fi.width > len(s) { + if fi.minus { + io.write_string16(fi.writer, s, &fi.n) + } + + for _ in 0.. 0 && space { + io.write_byte(fi.writer, ' ', &fi.n) + } + char_set := __DIGITS_UPPER + if verb == 'x' { + char_set = __DIGITS_LOWER + } + _fmt_int(fi, u64(s[i]), 16, false, bit_size=16, digits=char_set) + } + + case: + fmt_bad_verb(fi, verb) + } +} +// Formats a C-style UTF-16 string with a specific format. +// +// Inputs: +// - fi: Pointer to the Info struct containing format settings. +// - s: The C-style string to format. +// - verb: The format specifier character (Ref fmt_string). +// +fmt_cstring16 :: proc(fi: ^Info, s: cstring16, verb: rune) { + fmt_string16(fi, string16(s), verb) +} + // Formats a raw pointer with a specific format. // // Inputs: @@ -3210,6 +3283,9 @@ fmt_arg :: proc(fi: ^Info, arg: any, verb: rune) { case string: fmt_string(fi, a, verb) case cstring: fmt_cstring(fi, a, verb) + case string16: fmt_string16(fi, a, verb) + case cstring16: fmt_cstring16(fi, a, verb) + case typeid: reflect.write_typeid(fi.writer, a, &fi.n) case i16le: fmt_int(fi, u64(a), true, 16, verb) diff --git a/core/io/io.odin b/core/io/io.odin index c2b44cbdb..5431519bf 100644 --- a/core/io/io.odin +++ b/core/io/io.odin @@ -5,6 +5,7 @@ package io import "base:intrinsics" import "core:unicode/utf8" +import "core:unicode/utf16" // Seek whence values Seek_From :: enum { @@ -314,6 +315,29 @@ write_string :: proc(s: Writer, str: string, n_written: ^int = nil) -> (n: int, return write(s, transmute([]byte)str, n_written) } +// write_string16 writes the contents of the string16 s to w reencoded as utf-8 +write_string16 :: proc(s: Writer, str: string16, n_written: ^int = nil) -> (n: int, err: Error) { + for i := 0; i < len(str); i += 1 { + r := rune(utf16.REPLACEMENT_CHAR) + + switch c := str[i]; { + case c < utf16._surr1, utf16._surr3 <= c: + r = rune(c) + case utf16._surr1 <= c && c < utf16._surr2 && i+1 < len(str) && + utf16._surr2 <= str[i+1] && str[i+1] < utf16._surr3: + r = utf16.decode_surrogate_pair(rune(c), rune(str[i+1])) + i += 1 + } + + w, err := write_rune(s, r, n_written) + n += w + if err != nil { + return + } + } + return +} + // write_rune writes a UTF-8 encoded rune to w. write_rune :: proc(s: Writer, r: rune, n_written: ^int = nil) -> (size: int, err: Error) { defer if err == nil && n_written != nil { diff --git a/core/io/util.odin b/core/io/util.odin index fa98e007b..72983523a 100644 --- a/core/io/util.odin +++ b/core/io/util.odin @@ -264,6 +264,33 @@ write_quoted_string :: proc(w: Writer, str: string, quote: byte = '"', n_written return } +write_quoted_string16 :: proc(w: Writer, str: string16, quote: byte = '"', n_written: ^int = nil, for_json := false) -> (n: int, err: Error) { + defer if n_written != nil { + n_written^ += n + } + write_byte(w, quote, &n) or_return + for width, s := 0, str; len(s) > 0; s = s[width:] { + r := rune(s[0]) + width = 1 + if r >= utf8.RUNE_SELF { + r, width = utf16.decode_rune_in_string(s) + } + if width == 1 && r == utf8.RUNE_ERROR { + write_byte(w, '\\', &n) or_return + write_byte(w, 'x', &n) or_return + write_byte(w, DIGITS_LOWER[s[0]>>4], &n) or_return + write_byte(w, DIGITS_LOWER[s[0]&0xf], &n) or_return + continue + } + + n_wrapper(write_escaped_rune(w, r, quote, false, nil, for_json), &n) or_return + + } + write_byte(w, quote, &n) or_return + return +} + + // writer append a quoted rune into the byte buffer, return the written size write_quoted_rune :: proc(w: Writer, r: rune) -> (n: int) { _write_byte :: #force_inline proc(w: Writer, c: byte) -> int { diff --git a/core/unicode/utf16/utf16.odin b/core/unicode/utf16/utf16.odin index e2bcf7f68..9a8cfe438 100644 --- a/core/unicode/utf16/utf16.odin +++ b/core/unicode/utf16/utf16.odin @@ -106,6 +106,26 @@ decode :: proc(d: []rune, s: []u16) -> (n: int) { return } +decode_rune_in_string :: proc(s: string16) -> (r: rune, width: int) { + r = rune(REPLACEMENT_CHAR) + n := len(s) + if n < 1 { + return + } + width = 1 + + + switch c := s[0]; { + case c < _surr1, _surr3 <= c: + r = rune(c) + case _surr1 <= c && c < _surr2 && 1 < len(s) && + _surr2 <= s[1] && s[1] < _surr3: + r = decode_surrogate_pair(rune(c), rune(s[1])) + width += 1 + } + return +} + rune_count :: proc(s: []u16) -> (n: int) { for i := 0; i < len(s); i += 1 { c := s[i] diff --git a/src/build_settings.cpp b/src/build_settings.cpp index 46a4f9ae5..40bbe41e5 100644 --- a/src/build_settings.cpp +++ b/src/build_settings.cpp @@ -1089,7 +1089,7 @@ gb_internal String internal_odin_root_dir(void) { text = gb_alloc_array(permanent_allocator(), wchar_t, len+1); GetModuleFileNameW(nullptr, text, cast(int)len); - path = string16_to_string(heap_allocator(), make_string16(text, len)); + path = string16_to_string(heap_allocator(), make_string16(cast(u16 *)text, len)); for (i = path.len-1; i >= 0; i--) { u8 c = path[i]; @@ -1387,14 +1387,14 @@ gb_internal String path_to_fullpath(gbAllocator a, String s, bool *ok_) { mutex_lock(&fullpath_mutex); - len = GetFullPathNameW(&string16[0], 0, nullptr, nullptr); + len = GetFullPathNameW(cast(wchar_t *)&string16[0], 0, nullptr, nullptr); if (len != 0) { wchar_t *text = gb_alloc_array(permanent_allocator(), wchar_t, len+1); - GetFullPathNameW(&string16[0], len, text, nullptr); + GetFullPathNameW(cast(wchar_t *)&string16[0], len, text, nullptr); mutex_unlock(&fullpath_mutex); text[len] = 0; - result = string16_to_string(a, make_string16(text, len)); + result = string16_to_string(a, make_string16(cast(u16 *)text, len)); result = string_trim_whitespace(result); // Replace Windows style separators diff --git a/src/cached.cpp b/src/cached.cpp index efdadce7b..61b5d01b4 100644 --- a/src/cached.cpp +++ b/src/cached.cpp @@ -231,7 +231,7 @@ Array cache_gather_envs() { wchar_t *curr_string = strings; while (curr_string && *curr_string) { - String16 wstr = make_string16_c(curr_string); + String16 wstr = make_string16_c(cast(u16 *)curr_string); curr_string += wstr.len+1; String str = string16_to_string(temporary_allocator(), wstr); if (string_starts_with(str, str_lit("CURR_DATE_TIME="))) { diff --git a/src/check_builtin.cpp b/src/check_builtin.cpp index 974224ed2..d36cf4520 100644 --- a/src/check_builtin.cpp +++ b/src/check_builtin.cpp @@ -2327,6 +2327,9 @@ gb_internal bool check_builtin_procedure(CheckerContext *c, Operand *operand, As if (is_type_string(op_type) && id == BuiltinProc_len) { if (operand->mode == Addressing_Constant) { mode = Addressing_Constant; + + GB_ASSERT_MSG(!is_type_string16(op_type), "TODO(bill): constant utf-16 string len"); + String str = operand->value.value_string; value = exact_value_i64(str.len); type = t_untyped_integer; @@ -2334,6 +2337,8 @@ gb_internal bool check_builtin_procedure(CheckerContext *c, Operand *operand, As mode = Addressing_Value; if (is_type_cstring(op_type)) { add_package_dependency(c, "runtime", "cstring_len"); + } else if (is_type_cstring16(op_type)) { + add_package_dependency(c, "runtime", "cstring16_len"); } } } else if (is_type_array(op_type)) { @@ -4683,7 +4688,9 @@ gb_internal bool check_builtin_procedure(CheckerContext *c, Operand *operand, As break; case Type_Basic: if (t->Basic.kind == Basic_string) { - operand->type = alloc_type_multi_pointer(t_u8); + operand->type = t_u8_multi_ptr; + } else if (t->Basic.kind == Basic_string16) { + operand->type = t_u16_multi_ptr; } break; case Type_Pointer: diff --git a/src/check_decl.cpp b/src/check_decl.cpp index dd4c09e85..af46ee40e 100644 --- a/src/check_decl.cpp +++ b/src/check_decl.cpp @@ -815,6 +815,12 @@ gb_internal bool signature_parameter_similar_enough(Type *x, Type *y) { if (sig_compare(is_type_cstring, is_type_u8_multi_ptr, x, y)) { return true; } + if (sig_compare(is_type_cstring16, is_type_u16_ptr, x, y)) { + return true; + } + if (sig_compare(is_type_cstring16, is_type_u16_multi_ptr, x, y)) { + return true; + } if (sig_compare(is_type_uintptr, is_type_rawptr, x, y)) { return true; diff --git a/src/check_expr.cpp b/src/check_expr.cpp index 6723a7580..57073e22f 100644 --- a/src/check_expr.cpp +++ b/src/check_expr.cpp @@ -2862,6 +2862,14 @@ gb_internal void add_comparison_procedures_for_fields(CheckerContext *c, Type *t add_package_dependency(c, "runtime", "string_eq"); add_package_dependency(c, "runtime", "string_ne"); break; + case Basic_cstring16: + add_package_dependency(c, "runtime", "cstring16_eq"); + add_package_dependency(c, "runtime", "cstring16_ne"); + break; + case Basic_string16: + add_package_dependency(c, "runtime", "string16_eq"); + add_package_dependency(c, "runtime", "string16_ne"); + break; } break; case Type_Struct: @@ -3035,6 +3043,24 @@ gb_internal void check_comparison(CheckerContext *c, Ast *node, Operand *x, Oper case Token_LtEq: add_package_dependency(c, "runtime", "cstring_le"); break; case Token_GtEq: add_package_dependency(c, "runtime", "cstring_gt"); break; } + } else if (is_type_cstring16(x->type) && is_type_cstring16(y->type)) { + switch (op) { + case Token_CmpEq: add_package_dependency(c, "runtime", "cstring16_eq"); break; + case Token_NotEq: add_package_dependency(c, "runtime", "cstring16_ne"); break; + case Token_Lt: add_package_dependency(c, "runtime", "cstring16_lt"); break; + case Token_Gt: add_package_dependency(c, "runtime", "cstring16_gt"); break; + case Token_LtEq: add_package_dependency(c, "runtime", "cstring16_le"); break; + case Token_GtEq: add_package_dependency(c, "runtime", "cstring16_gt"); break; + } + } else if (is_type_string16(x->type) || is_type_string16(y->type)) { + switch (op) { + case Token_CmpEq: add_package_dependency(c, "runtime", "string16_eq"); break; + case Token_NotEq: add_package_dependency(c, "runtime", "string16_ne"); break; + case Token_Lt: add_package_dependency(c, "runtime", "string16_lt"); break; + case Token_Gt: add_package_dependency(c, "runtime", "string16_gt"); break; + case Token_LtEq: add_package_dependency(c, "runtime", "string16_le"); break; + case Token_GtEq: add_package_dependency(c, "runtime", "string16_gt"); break; + } } else if (is_type_string(x->type) || is_type_string(y->type)) { switch (op) { case Token_CmpEq: add_package_dependency(c, "runtime", "string_eq"); break; @@ -3340,6 +3366,11 @@ gb_internal bool check_is_castable_to(CheckerContext *c, Operand *operand, Type return true; } + // []u16 <-> string16 (not cstring16) + if (is_type_u16_slice(src) && (is_type_string16(dst) && !is_type_cstring16(dst))) { + return true; + } + // cstring -> string if (are_types_identical(src, t_cstring) && are_types_identical(dst, t_string)) { if (operand->mode != Addressing_Constant) { @@ -3347,6 +3378,14 @@ gb_internal bool check_is_castable_to(CheckerContext *c, Operand *operand, Type } return true; } + // cstring16 -> string16 + if (are_types_identical(src, t_cstring16) && are_types_identical(dst, t_string16)) { + if (operand->mode != Addressing_Constant) { + add_package_dependency(c, "runtime", "cstring16_to_string16"); + } + return true; + } + // cstring -> ^u8 if (are_types_identical(src, t_cstring) && is_type_u8_ptr(dst)) { return !is_constant; @@ -3372,6 +3411,34 @@ gb_internal bool check_is_castable_to(CheckerContext *c, Operand *operand, Type if (is_type_rawptr(src) && are_types_identical(dst, t_cstring)) { return !is_constant; } + + // cstring -> ^u16 + if (are_types_identical(src, t_cstring16) && is_type_u16_ptr(dst)) { + return !is_constant; + } + // cstring -> [^]u16 + if (are_types_identical(src, t_cstring16) && is_type_u16_multi_ptr(dst)) { + return !is_constant; + } + // cstring -> rawptr + if (are_types_identical(src, t_cstring16) && is_type_rawptr(dst)) { + return !is_constant; + } + + + // ^u16 -> cstring16 + if (is_type_u16_ptr(src) && are_types_identical(dst, t_cstring16)) { + return !is_constant; + } + // [^]u16 -> cstring + if (is_type_u16_multi_ptr(src) && are_types_identical(dst, t_cstring16)) { + return !is_constant; + } + // rawptr -> cstring16 + if (is_type_rawptr(src) && are_types_identical(dst, t_cstring16)) { + return !is_constant; + } + // proc <-> proc if (is_type_proc(src) && is_type_proc(dst)) { if (is_type_polymorphic(dst)) { @@ -4558,6 +4625,8 @@ gb_internal void convert_to_typed(CheckerContext *c, Operand *operand, Type *tar // target_type = t_untyped_nil; } else if (is_type_cstring(target_type)) { // target_type = t_untyped_nil; + } else if (is_type_cstring16(target_type)) { + // target_type = t_untyped_nil; } else if (!type_has_nil(target_type)) { operand->mode = Addressing_Invalid; convert_untyped_error(c, operand, target_type); @@ -8226,6 +8295,7 @@ gb_internal bool check_set_index_data(Operand *o, Type *t, bool indirection, i64 case Type_Basic: if (t->Basic.kind == Basic_string) { if (o->mode == Addressing_Constant) { + GB_ASSERT(o->value.kind == ExactValue_String); *max_count = o->value.value_string.len; } if (o->mode != Addressing_Constant) { @@ -8233,6 +8303,16 @@ gb_internal bool check_set_index_data(Operand *o, Type *t, bool indirection, i64 } o->type = t_u8; return true; + } else if (t->Basic.kind == Basic_string16) { + if (o->mode == Addressing_Constant) { + GB_ASSERT(o->value.kind == ExactValue_String16); + *max_count = o->value.value_string16.len; + } + if (o->mode != Addressing_Constant) { + o->mode = Addressing_Value; + } + o->type = t_u16; + return true; } else if (t->Basic.kind == Basic_UntypedString) { if (o->mode == Addressing_Constant) { *max_count = o->value.value_string.len; @@ -10879,9 +10959,17 @@ gb_internal ExprKind check_slice_expr(CheckerContext *c, Operand *o, Ast *node, if (t->Basic.kind == Basic_string || t->Basic.kind == Basic_UntypedString) { valid = true; if (o->mode == Addressing_Constant) { + GB_ASSERT(o->value.kind == ExactValue_String); max_count = o->value.value_string.len; } o->type = type_deref(o->type); + } else if (t->Basic.kind == Basic_string16) { + valid = true; + if (o->mode == Addressing_Constant) { + GB_ASSERT(o->value.kind == ExactValue_String16); + max_count = o->value.value_string16.len; + } + o->type = type_deref(o->type); } break; diff --git a/src/checker.cpp b/src/checker.cpp index a1d8f98d7..20da5b19b 100644 --- a/src/checker.cpp +++ b/src/checker.cpp @@ -1363,13 +1363,15 @@ gb_internal void init_universal(void) { } - t_u8_ptr = alloc_type_pointer(t_u8); - t_u8_multi_ptr = alloc_type_multi_pointer(t_u8); - t_int_ptr = alloc_type_pointer(t_int); - t_i64_ptr = alloc_type_pointer(t_i64); - t_f64_ptr = alloc_type_pointer(t_f64); - t_u8_slice = alloc_type_slice(t_u8); - t_string_slice = alloc_type_slice(t_string); + t_u8_ptr = alloc_type_pointer(t_u8); + t_u8_multi_ptr = alloc_type_multi_pointer(t_u8); + t_u16_ptr = alloc_type_pointer(t_u16); + t_u16_multi_ptr = alloc_type_multi_pointer(t_u16); + t_int_ptr = alloc_type_pointer(t_int); + t_i64_ptr = alloc_type_pointer(t_i64); + t_f64_ptr = alloc_type_pointer(t_f64); + t_u8_slice = alloc_type_slice(t_u8); + t_string_slice = alloc_type_slice(t_string); // intrinsics types for objective-c stuff { diff --git a/src/common.cpp b/src/common.cpp index ad1e5a851..b3761fc36 100644 --- a/src/common.cpp +++ b/src/common.cpp @@ -669,7 +669,7 @@ gb_internal gb_inline f64 gb_sqrt(f64 x) { gb_internal wchar_t **command_line_to_wargv(wchar_t *cmd_line, int *_argc) { u32 i, j; - u32 len = cast(u32)string16_len(cmd_line); + u32 len = cast(u32)string16_len(cast(u16 *)cmd_line); i = ((len+2)/2)*gb_size_of(void *) + gb_size_of(void *); wchar_t **argv = cast(wchar_t **)GlobalAlloc(GMEM_FIXED, i + (len+2)*gb_size_of(wchar_t)); diff --git a/src/exact_value.cpp b/src/exact_value.cpp index 37751c8f1..f2aed84c2 100644 --- a/src/exact_value.cpp +++ b/src/exact_value.cpp @@ -29,6 +29,7 @@ enum ExactValueKind { ExactValue_Compound = 8, ExactValue_Procedure = 9, ExactValue_Typeid = 10, + ExactValue_String16 = 11, ExactValue_Count, }; @@ -46,6 +47,7 @@ struct ExactValue { Ast * value_compound; Ast * value_procedure; Type * value_typeid; + String16 value_string16; }; }; @@ -66,6 +68,9 @@ gb_internal uintptr hash_exact_value(ExactValue v) { case ExactValue_String: res = gb_fnv32a(v.value_string.text, v.value_string.len); break; + case ExactValue_String16: + res = gb_fnv32a(v.value_string.text, v.value_string.len*gb_size_of(u16)); + break; case ExactValue_Integer: { u32 key = gb_fnv32a(v.value_integer.dp, gb_size_of(*v.value_integer.dp) * v.value_integer.used); @@ -118,6 +123,11 @@ gb_internal ExactValue exact_value_string(String string) { result.value_string = string; return result; } +gb_internal ExactValue exact_value_string16(String16 string) { + ExactValue result = {ExactValue_String16}; + result.value_string16 = string; + return result; +} gb_internal ExactValue exact_value_i64(i64 i) { ExactValue result = {ExactValue_Integer}; @@ -656,6 +666,7 @@ gb_internal i32 exact_value_order(ExactValue const &v) { return 0; case ExactValue_Bool: case ExactValue_String: + case ExactValue_String16: return 1; case ExactValue_Integer: return 2; @@ -689,6 +700,7 @@ gb_internal void match_exact_values(ExactValue *x, ExactValue *y) { case ExactValue_Bool: case ExactValue_String: + case ExactValue_String16: case ExactValue_Quaternion: case ExactValue_Pointer: case ExactValue_Compound: @@ -891,7 +903,18 @@ gb_internal ExactValue exact_binary_operator_value(TokenKind op, ExactValue x, E gb_memmove(data, sx.text, sx.len); gb_memmove(data+sx.len, sy.text, sy.len); return exact_value_string(make_string(data, len)); - break; + } + case ExactValue_String16: { + if (op != Token_Add) goto error; + + // NOTE(bill): How do you minimize this over allocation? + String sx = x.value_string; + String sy = y.value_string; + isize len = sx.len+sy.len; + u16 *data = gb_alloc_array(permanent_allocator(), u16, len); + gb_memmove(data, sx.text, sx.len*gb_size_of(u16)); + gb_memmove(data+sx.len, sy.text, sy.len*gb_size_of(u16)); + return exact_value_string16(make_string16(data, len)); } } @@ -994,6 +1017,19 @@ gb_internal bool compare_exact_values(TokenKind op, ExactValue x, ExactValue y) } break; } + case ExactValue_String16: { + String16 a = x.value_string16; + String16 b = y.value_string16; + switch (op) { + case Token_CmpEq: return a == b; + case Token_NotEq: return a != b; + case Token_Lt: return a < b; + case Token_LtEq: return a <= b; + case Token_Gt: return a > b; + case Token_GtEq: return a >= b; + } + break; + } case ExactValue_Pointer: { switch (op) { @@ -1050,6 +1086,20 @@ gb_internal gbString write_exact_value_to_string(gbString str, ExactValue const gb_free(heap_allocator(), s.text); return str; } + case ExactValue_String16: { + String s = quote_to_ascii(heap_allocator(), v.value_string16); + string_limit = gb_max(string_limit, 36); + if (s.len <= string_limit) { + str = gb_string_append_length(str, s.text, s.len); + } else { + isize n = string_limit/5; + str = gb_string_append_length(str, s.text, n); + str = gb_string_append_fmt(str, "\"..%lld chars..\"", s.len-(2*n)); + str = gb_string_append_length(str, s.text+s.len-n, n); + } + gb_free(heap_allocator(), s.text); + return str; + } case ExactValue_Integer: { String s = big_int_to_string(heap_allocator(), &v.value_integer); str = gb_string_append_length(str, s.text, s.len); diff --git a/src/llvm_backend_expr.cpp b/src/llvm_backend_expr.cpp index 74aea82f1..fbf0dea11 100644 --- a/src/llvm_backend_expr.cpp +++ b/src/llvm_backend_expr.cpp @@ -1656,6 +1656,8 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) { res.type = t; res.value = llvm_cstring(m, str); return res; + } else if (src->kind == Type_Basic && src->Basic.kind == Basic_string16 && dst->Basic.kind == Basic_cstring16) { + GB_PANIC("TODO(bill): UTF-16 string"); } // if (is_type_float(dst)) { // return value; @@ -1795,6 +1797,38 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) { } + + if (is_type_cstring16(src) && is_type_u16_ptr(dst)) { + return lb_emit_transmute(p, value, dst); + } + if (is_type_u16_ptr(src) && is_type_cstring16(dst)) { + return lb_emit_transmute(p, value, dst); + } + if (is_type_cstring16(src) && is_type_u16_multi_ptr(dst)) { + return lb_emit_transmute(p, value, dst); + } + if (is_type_u8_multi_ptr(src) && is_type_cstring16(dst)) { + return lb_emit_transmute(p, value, dst); + } + if (is_type_cstring16(src) && is_type_rawptr(dst)) { + return lb_emit_transmute(p, value, dst); + } + if (is_type_rawptr(src) && is_type_cstring16(dst)) { + return lb_emit_transmute(p, value, dst); + } + + if (are_types_identical(src, t_cstring16) && are_types_identical(dst, t_string16)) { + TEMPORARY_ALLOCATOR_GUARD(); + + lbValue c = lb_emit_conv(p, value, t_cstring16); + auto args = array_make(temporary_allocator(), 1); + args[0] = c; + lbValue s = lb_emit_runtime_call(p, "cstring16_to_string16", args); + return lb_emit_conv(p, s, dst); + } + + + // integer -> boolean if (is_type_integer(src) && is_type_boolean(dst)) { lbValue res = {}; @@ -2296,6 +2330,14 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) { return res; } + // []u16 <-> string16 + if (is_type_u16_slice(src) && is_type_string16(dst)) { + return lb_emit_transmute(p, value, t); + } + if (is_type_string16(src) && is_type_u16_slice(dst)) { + return lb_emit_transmute(p, value, t); + } + // []byte/[]u8 <-> string if (is_type_u8_slice(src) && is_type_string(dst)) { return lb_emit_transmute(p, value, t); @@ -2304,6 +2346,7 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) { return lb_emit_transmute(p, value, t); } + if (is_type_array_like(dst)) { Type *elem = base_array_type(dst); isize index_count = cast(isize)get_array_type_count(dst); @@ -2483,6 +2526,12 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) { if (is_type_untyped(src)) { + if (is_type_string(src) && is_type_string16(dst)) { + GB_PANIC("TODO(bill): UTF-16 string"); + lbAddr result = lb_add_local_generated(p, t, false); + lb_addr_store(p, result, value); + return lb_addr_load(p, result); + } if (is_type_string(src) && is_type_string(dst)) { lbAddr result = lb_add_local_generated(p, t, false); lb_addr_store(p, result, value); @@ -3056,6 +3105,13 @@ gb_internal lbValue lb_emit_comp_against_nil(lbProcedure *p, TokenKind op_kind, res.value = LLVMBuildIsNotNull(p->builder, x.value, ""); } return res; + case Basic_cstring16: + if (op_kind == Token_CmpEq) { + res.value = LLVMBuildIsNull(p->builder, x.value, ""); + } else if (op_kind == Token_NotEq) { + res.value = LLVMBuildIsNotNull(p->builder, x.value, ""); + } + return res; case Basic_any: { // TODO(bill): is this correct behaviour for nil comparison for any? @@ -4432,6 +4488,22 @@ gb_internal lbAddr lb_build_addr_slice_expr(lbProcedure *p, Ast *expr) { } case Type_Basic: { + if (is_type_string16(type)) { + GB_ASSERT_MSG(are_types_identical(type, t_string16), "got %s", type_to_string(type)); + lbValue len = lb_string_len(p, base); + if (high.value == nullptr) high = len; + + if (!no_indices) { + lb_emit_slice_bounds_check(p, se->open, low, high, len, se->low != nullptr); + } + + lbValue elem = lb_emit_ptr_offset(p, lb_string_elem(p, base), low); + lbValue new_len = lb_emit_arith(p, Token_Sub, high, low, t_int); + + lbAddr str = lb_add_local_generated(p, t_string16, false); + lb_fill_string(p, str, elem, new_len); + return str; + } GB_ASSERT_MSG(are_types_identical(type, t_string), "got %s", type_to_string(type)); lbValue len = lb_string_len(p, base); if (high.value == nullptr) high = len; diff --git a/src/llvm_backend_general.cpp b/src/llvm_backend_general.cpp index 3ce0c725f..d9771a75b 100644 --- a/src/llvm_backend_general.cpp +++ b/src/llvm_backend_general.cpp @@ -1812,6 +1812,37 @@ gb_internal LLVMTypeRef lb_type_internal(lbModule *m, Type *type) { return type; } case Basic_cstring: return LLVMPointerType(LLVMInt8TypeInContext(ctx), 0); + + + case Basic_string16: + { + char const *name = "..string16"; + LLVMTypeRef type = LLVMGetTypeByName(m->mod, name); + if (type != nullptr) { + return type; + } + type = LLVMStructCreateNamed(ctx, name); + + if (build_context.metrics.ptr_size < build_context.metrics.int_size) { + GB_ASSERT(build_context.metrics.ptr_size == 4); + GB_ASSERT(build_context.metrics.int_size == 8); + LLVMTypeRef fields[3] = { + LLVMPointerType(lb_type(m, t_u16), 0), + lb_type(m, t_i32), + lb_type(m, t_int), + }; + LLVMStructSetBody(type, fields, 3, false); + } else { + LLVMTypeRef fields[2] = { + LLVMPointerType(lb_type(m, t_u16), 0), + lb_type(m, t_int), + }; + LLVMStructSetBody(type, fields, 2, false); + } + return type; + } + case Basic_cstring16: return LLVMPointerType(LLVMInt16TypeInContext(ctx), 0); + case Basic_any: { char const *name = "..any"; diff --git a/src/llvm_backend_proc.cpp b/src/llvm_backend_proc.cpp index e63c92f6f..8f306b771 100644 --- a/src/llvm_backend_proc.cpp +++ b/src/llvm_backend_proc.cpp @@ -2289,6 +2289,10 @@ gb_internal lbValue lb_build_builtin_proc(lbProcedure *p, Ast *expr, TypeAndValu } if (is_type_cstring(t)) { return lb_cstring_len(p, v); + } else if (is_type_cstring16(t)) { + return lb_cstring16_len(p, v); + } else if (is_type_string16(t)) { + return lb_string_len(p, v); } else if (is_type_string(t)) { return lb_string_len(p, v); } else if (is_type_array(t)) { @@ -2728,6 +2732,11 @@ gb_internal lbValue lb_build_builtin_proc(lbProcedure *p, Ast *expr, TypeAndValu res = lb_emit_conv(p, res, tv.type); } else if (t->Basic.kind == Basic_cstring) { res = lb_emit_conv(p, x, tv.type); + } else if (t->Basic.kind == Basic_string16) { + res = lb_string_elem(p, x); + res = lb_emit_conv(p, res, tv.type); + } else if (t->Basic.kind == Basic_cstring16) { + res = lb_emit_conv(p, x, tv.type); } break; case Type_Pointer: diff --git a/src/llvm_backend_type.cpp b/src/llvm_backend_type.cpp index 43c5f0b40..a91d77fe5 100644 --- a/src/llvm_backend_type.cpp +++ b/src/llvm_backend_type.cpp @@ -531,7 +531,33 @@ gb_internal void lb_setup_type_info_data_giant_array(lbModule *m, i64 global_typ case Basic_cstring: { tag_type = t_type_info_string; - LLVMValueRef vals[1] = { + LLVMValueRef vals[2] = { + lb_const_bool(m, t_bool, true).value, + lb_const_bool(m, t_bool, false).value, + }; + + variant_value = llvm_const_named_struct(m, tag_type, vals, gb_count_of(vals)); + } + break; + + case Basic_string16: + { + tag_type = t_type_info_string; + LLVMValueRef vals[2] = { + lb_const_bool(m, t_bool, false).value, + lb_const_bool(m, t_bool, true).value, + }; + + variant_value = llvm_const_named_struct(m, tag_type, vals, gb_count_of(vals)); + } + break; + + + case Basic_cstring16: + { + tag_type = t_type_info_string; + LLVMValueRef vals[2] = { + lb_const_bool(m, t_bool, true).value, lb_const_bool(m, t_bool, true).value, }; diff --git a/src/llvm_backend_utility.cpp b/src/llvm_backend_utility.cpp index 521553147..d4117b7ff 100644 --- a/src/llvm_backend_utility.cpp +++ b/src/llvm_backend_utility.cpp @@ -1626,11 +1626,17 @@ gb_internal void lb_fill_string(lbProcedure *p, lbAddr const &string, lbValue ba gb_internal lbValue lb_string_elem(lbProcedure *p, lbValue string) { Type *t = base_type(string.type); + if (t->kind == Type_Basic && t->Basic.kind == Basic_string16) { + return lb_emit_struct_ev(p, string, 0); + } GB_ASSERT(t->kind == Type_Basic && t->Basic.kind == Basic_string); return lb_emit_struct_ev(p, string, 0); } gb_internal lbValue lb_string_len(lbProcedure *p, lbValue string) { Type *t = base_type(string.type); + if (t->kind == Type_Basic && t->Basic.kind == Basic_string16) { + return lb_emit_struct_ev(p, string, 1); + } GB_ASSERT_MSG(t->kind == Type_Basic && t->Basic.kind == Basic_string, "%s", type_to_string(t)); return lb_emit_struct_ev(p, string, 1); } @@ -1641,6 +1647,12 @@ gb_internal lbValue lb_cstring_len(lbProcedure *p, lbValue value) { args[0] = lb_emit_conv(p, value, t_cstring); return lb_emit_runtime_call(p, "cstring_len", args); } +gb_internal lbValue lb_cstring16_len(lbProcedure *p, lbValue value) { + GB_ASSERT(is_type_cstring16(value.type)); + auto args = array_make(permanent_allocator(), 1); + args[0] = lb_emit_conv(p, value, t_cstring16); + return lb_emit_runtime_call(p, "cstring16_len", args); +} gb_internal lbValue lb_array_elem(lbProcedure *p, lbValue array_ptr) { diff --git a/src/main.cpp b/src/main.cpp index 112d1208a..5a43e3c02 100644 --- a/src/main.cpp +++ b/src/main.cpp @@ -142,9 +142,9 @@ gb_internal i32 system_exec_command_line_app_internal(bool exit_on_err, char con } wcmd = string_to_string16(permanent_allocator(), make_string(cast(u8 *)cmd_line, cmd_len-1)); - if (CreateProcessW(nullptr, wcmd.text, - nullptr, nullptr, true, 0, nullptr, nullptr, - &start_info, &pi)) { + if (CreateProcessW(nullptr, cast(wchar_t *)wcmd.text, + nullptr, nullptr, true, 0, nullptr, nullptr, + &start_info, &pi)) { WaitForSingleObject(pi.hProcess, INFINITE); GetExitCodeProcess(pi.hProcess, cast(DWORD *)&exit_code); @@ -232,7 +232,7 @@ gb_internal Array setup_args(int argc, char const **argv) { wchar_t **wargv = command_line_to_wargv(GetCommandLineW(), &wargc); auto args = array_make(a, 0, wargc); for (isize i = 0; i < wargc; i++) { - wchar_t *warg = wargv[i]; + u16 *warg = cast(u16 *)wargv[i]; isize wlen = string16_len(warg); String16 wstr = make_string16(warg, wlen); String arg = string16_to_string(a, wstr); diff --git a/src/microsoft_craziness.h b/src/microsoft_craziness.h index b0fd22a23..933607a2a 100644 --- a/src/microsoft_craziness.h +++ b/src/microsoft_craziness.h @@ -59,7 +59,7 @@ struct Find_Result { }; gb_internal String mc_wstring_to_string(wchar_t const *str) { - return string16_to_string(mc_allocator, make_string16_c(str)); + return string16_to_string(mc_allocator, make_string16_c(cast(u16 *)str)); } gb_internal String16 mc_string_to_wstring(String str) { @@ -103,7 +103,7 @@ gb_internal HANDLE mc_find_first(String wildcard, MC_Find_Data *find_data) { String16 wildcard_wide = mc_string_to_wstring(wildcard); defer (mc_free(wildcard_wide)); - HANDLE handle = FindFirstFileW(wildcard_wide.text, &_find_data); + HANDLE handle = FindFirstFileW(cast(wchar_t *)wildcard_wide.text, &_find_data); if (handle == INVALID_HANDLE_VALUE) return INVALID_HANDLE_VALUE; find_data->file_attributes = _find_data.dwFileAttributes; diff --git a/src/path.cpp b/src/path.cpp index d5e982088..2b97a04df 100644 --- a/src/path.cpp +++ b/src/path.cpp @@ -130,7 +130,7 @@ gb_internal String directory_from_path(String const &s) { String16 wstr = string_to_string16(a, path); defer (gb_free(a, wstr.text)); - i32 attribs = GetFileAttributesW(wstr.text); + i32 attribs = GetFileAttributesW(cast(wchar_t *)wstr.text); if (attribs < 0) return false; return (attribs & FILE_ATTRIBUTE_DIRECTORY) != 0; @@ -360,7 +360,7 @@ gb_internal ReadDirectoryError read_directory(String path, Array *fi) defer (gb_free(a, wstr.text)); WIN32_FIND_DATAW file_data = {}; - HANDLE find_file = FindFirstFileW(wstr.text, &file_data); + HANDLE find_file = FindFirstFileW(cast(wchar_t *)wstr.text, &file_data); if (find_file == INVALID_HANDLE_VALUE) { return ReadDirectory_Unknown; } @@ -372,7 +372,7 @@ gb_internal ReadDirectoryError read_directory(String path, Array *fi) wchar_t *filename_w = file_data.cFileName; u64 size = cast(u64)file_data.nFileSizeLow; size |= (cast(u64)file_data.nFileSizeHigh) << 32; - String name = string16_to_string(a, make_string16_c(filename_w)); + String name = string16_to_string(a, make_string16_c(cast(u16 *)filename_w)); if (name == "." || name == "..") { gb_free(a, name.text); continue; @@ -494,7 +494,7 @@ gb_internal bool write_directory(String path) { #else gb_internal bool write_directory(String path) { String16 wstr = string_to_string16(heap_allocator(), path); - LPCWSTR wdirectory_name = wstr.text; + LPCWSTR wdirectory_name = cast(wchar_t *)wstr.text; HANDLE directory = CreateFileW(wdirectory_name, GENERIC_WRITE, diff --git a/src/string.cpp b/src/string.cpp index ae8d066b1..8405938f4 100644 --- a/src/string.cpp +++ b/src/string.cpp @@ -26,15 +26,14 @@ struct String_Iterator { // NOTE(bill): String16 is only used for Windows due to its file directories struct String16 { - wchar_t *text; - isize len; - wchar_t const &operator[](isize i) const { + u16 * text; + isize len; + u16 const &operator[](isize i) const { GB_ASSERT_MSG(0 <= i && i < len, "[%td]", i); return text[i]; } }; - gb_internal gb_inline String make_string(u8 const *text, isize len) { String s; s.text = cast(u8 *)text; @@ -45,19 +44,19 @@ gb_internal gb_inline String make_string(u8 const *text, isize len) { return s; } - -gb_internal gb_inline String16 make_string16(wchar_t const *text, isize len) { +gb_internal gb_inline String16 make_string16(u16 const *text, isize len) { String16 s; - s.text = cast(wchar_t *)text; + s.text = cast(u16 *)text; s.len = len; return s; } -gb_internal isize string16_len(wchar_t const *s) { + +gb_internal isize string16_len(u16 const *s) { if (s == nullptr) { return 0; } - wchar_t const *p = s; + u16 const *p = s; while (*p) { p++; } @@ -69,7 +68,7 @@ gb_internal gb_inline String make_string_c(char const *text) { return make_string(cast(u8 *)cast(void *)text, gb_strlen(text)); } -gb_internal gb_inline String16 make_string16_c(wchar_t const *text) { +gb_internal gb_inline String16 make_string16_c(u16 const *text) { return make_string16(text, string16_len(text)); } @@ -145,6 +144,27 @@ gb_internal int string_compare(String const &a, String const &b) { return res; } + +gb_internal int string16_compare(String16 const &a, String16 const &b) { + if (a.text == b.text) { + return cast(int)(a.len - b.len); + } + if (a.text == nullptr) { + return -1; + } + if (b.text == nullptr) { + return +1; + } + + uintptr n = gb_min(a.len, b.len); + int res = memcmp(a.text, b.text, n*gb_size_of(u16)); + if (res == 0) { + res = cast(int)(a.len - b.len); + } + return res; +} + + gb_internal isize string_index_byte(String const &s, u8 x) { for (isize i = 0; i < s.len; i++) { if (s.text[i] == x) { @@ -182,6 +202,26 @@ template gb_internal bool operator >= (String const &a, char const (&b template <> bool operator == (String const &a, char const (&b)[1]) { return a.len == 0; } template <> bool operator != (String const &a, char const (&b)[1]) { return a.len != 0; } + +gb_internal gb_inline bool str_eq(String16 const &a, String16 const &b) { + if (a.len != b.len) return false; + if (a.len == 0) return true; + return memcmp(a.text, b.text, a.len) == 0; +} +gb_internal gb_inline bool str_ne(String16 const &a, String16 const &b) { return !str_eq(a, b); } +gb_internal gb_inline bool str_lt(String16 const &a, String16 const &b) { return string16_compare(a, b) < 0; } +gb_internal gb_inline bool str_gt(String16 const &a, String16 const &b) { return string16_compare(a, b) > 0; } +gb_internal gb_inline bool str_le(String16 const &a, String16 const &b) { return string16_compare(a, b) <= 0; } +gb_internal gb_inline bool str_ge(String16 const &a, String16 const &b) { return string16_compare(a, b) >= 0; } + +gb_internal gb_inline bool operator == (String16 const &a, String16 const &b) { return str_eq(a, b); } +gb_internal gb_inline bool operator != (String16 const &a, String16 const &b) { return str_ne(a, b); } +gb_internal gb_inline bool operator < (String16 const &a, String16 const &b) { return str_lt(a, b); } +gb_internal gb_inline bool operator > (String16 const &a, String16 const &b) { return str_gt(a, b); } +gb_internal gb_inline bool operator <= (String16 const &a, String16 const &b) { return str_le(a, b); } +gb_internal gb_inline bool operator >= (String16 const &a, String16 const &b) { return str_ge(a, b); } + + gb_internal gb_inline bool string_starts_with(String const &s, String const &prefix) { if (prefix.len > s.len) { return false; @@ -614,7 +654,7 @@ gb_internal String normalize_path(gbAllocator a, String const &path, String cons // TODO(bill): Make this non-windows specific gb_internal String16 string_to_string16(gbAllocator a, String s) { int len, len1; - wchar_t *text; + u16 *text; if (s.len < 1) { return make_string16(nullptr, 0); @@ -625,9 +665,9 @@ gb_internal String16 string_to_string16(gbAllocator a, String s) { return make_string16(nullptr, 0); } - text = gb_alloc_array(a, wchar_t, len+1); + text = gb_alloc_array(a, u16, len+1); - len1 = convert_multibyte_to_widechar(cast(char *)s.text, cast(int)s.len, text, cast(int)len); + len1 = convert_multibyte_to_widechar(cast(char *)s.text, cast(int)s.len, cast(wchar_t *)text, cast(int)len); if (len1 == 0) { gb_free(a, text); return make_string16(nullptr, 0); @@ -646,7 +686,7 @@ gb_internal String string16_to_string(gbAllocator a, String16 s) { return make_string(nullptr, 0); } - len = convert_widechar_to_multibyte(s.text, cast(int)s.len, nullptr, 0); + len = convert_widechar_to_multibyte(cast(wchar_t *)s.text, cast(int)s.len, nullptr, 0); if (len == 0) { return make_string(nullptr, 0); } @@ -654,7 +694,7 @@ gb_internal String string16_to_string(gbAllocator a, String16 s) { text = gb_alloc_array(a, u8, len+1); - len1 = convert_widechar_to_multibyte(s.text, cast(int)s.len, cast(char *)text, cast(int)len); + len1 = convert_widechar_to_multibyte(cast(wchar_t *)s.text, cast(int)s.len, cast(char *)text, cast(int)len); if (len1 == 0) { gb_free(a, text); return make_string(nullptr, 0); @@ -674,9 +714,9 @@ gb_internal String temporary_directory(gbAllocator allocator) { return String{0}; } DWORD len = gb_max(MAX_PATH, n); - wchar_t *b = gb_alloc_array(heap_allocator(), wchar_t, len+1); + u16 *b = gb_alloc_array(heap_allocator(), u16, len+1); defer (gb_free(heap_allocator(), b)); - n = GetTempPathW(len, b); + n = GetTempPathW(len, cast(wchar_t *)b); if (n == 3 && b[1] == ':' && b[2] == '\\') { } else if (n > 0 && b[n-1] == '\\') { @@ -791,6 +831,104 @@ gb_internal String quote_to_ascii(gbAllocator a, String str, u8 quote='"') { return res; } +gb_internal Rune decode_surrogate_pair(u16 r1, u16 r2) { + static Rune const _surr1 = 0xd800; + static Rune const _surr2 = 0xdc00; + static Rune const _surr3 = 0xe000; + static Rune const _surr_self = 0x10000; + + if (_surr1 <= r1 && r1 < _surr2 && _surr2 <= r2 && r2 < _surr3) { + return (((r1-_surr1)<<10) | (r2 - _surr2)) + _surr_self; + } + return GB_RUNE_INVALID; +} + +gb_internal String quote_to_ascii(gbAllocator a, String16 str, u8 quote='"') { + static Rune const _surr1 = 0xd800; + static Rune const _surr2 = 0xdc00; + static Rune const _surr3 = 0xe000; + static Rune const _surr_self = 0x10000; + + u16 *s = cast(u16 *)str.text; + isize n = str.len; + auto buf = array_make(a, 0, n*2); + array_add(&buf, quote); + for (isize width = 0; n > 0; s += width, n -= width) { + Rune r = cast(Rune)s[0]; + width = 1; + if (r < _surr1 || _surr3 <= r) { + r = cast(Rune)r; + } else if (_surr1 <= r && r < _surr2) { + if (n>1) { + r = decode_surrogate_pair(s[0], s[1]); + if (r != GB_RUNE_INVALID) { + width = 2; + } + } else { + r = GB_RUNE_INVALID; + } + } + if (width == 1 && r == GB_RUNE_INVALID) { + array_add(&buf, cast(u8)'\\'); + array_add(&buf, cast(u8)'x'); + array_add(&buf, cast(u8)lower_hex[s[0]>>4]); + array_add(&buf, cast(u8)lower_hex[s[0]&0xf]); + continue; + } + + if (r == quote || r == '\\') { + array_add(&buf, cast(u8)'\\'); + array_add(&buf, u8(r)); + continue; + } + if (r < 0x80 && is_printable(r)) { + array_add(&buf, u8(r)); + continue; + } + switch (r) { + case '\a': + case '\b': + case '\f': + case '\n': + case '\r': + case '\t': + case '\v': + default: + if (r < ' ') { + u8 b = cast(u8)r; + array_add(&buf, cast(u8)'\\'); + array_add(&buf, cast(u8)'x'); + array_add(&buf, cast(u8)lower_hex[b>>4]); + array_add(&buf, cast(u8)lower_hex[b&0xf]); + } + if (r > GB_RUNE_MAX) { + r = 0XFFFD; + } + if (r < 0x10000) { + array_add(&buf, cast(u8)'\\'); + array_add(&buf, cast(u8)'u'); + for (isize i = 12; i >= 0; i -= 4) { + array_add(&buf, cast(u8)lower_hex[(r>>i)&0xf]); + } + } else { + array_add(&buf, cast(u8)'\\'); + array_add(&buf, cast(u8)'U'); + for (isize i = 28; i >= 0; i -= 4) { + array_add(&buf, cast(u8)lower_hex[(r>>i)&0xf]); + } + } + } + } + + + + array_add(&buf, quote); + String res = {}; + res.text = buf.data; + res.len = buf.count; + return res; +} + diff --git a/src/types.cpp b/src/types.cpp index 2e696810d..ad576c8af 100644 --- a/src/types.cpp +++ b/src/types.cpp @@ -41,8 +41,13 @@ enum BasicKind { Basic_uint, Basic_uintptr, Basic_rawptr, - Basic_string, // ^u8 + int - Basic_cstring, // ^u8 + + Basic_string, // [^]u8 + int + Basic_cstring, // [^]u8 + + Basic_string16, // [^]u16 + int + Basic_cstring16, // [^]u16 + int + Basic_any, // rawptr + ^Type_Info Basic_typeid, @@ -500,8 +505,14 @@ gb_global Type basic_types[] = { {Type_Basic, {Basic_uintptr, BasicFlag_Integer | BasicFlag_Unsigned, -1, STR_LIT("uintptr")}}, {Type_Basic, {Basic_rawptr, BasicFlag_Pointer, -1, STR_LIT("rawptr")}}, + {Type_Basic, {Basic_string, BasicFlag_String, -1, STR_LIT("string")}}, {Type_Basic, {Basic_cstring, BasicFlag_String, -1, STR_LIT("cstring")}}, + + {Type_Basic, {Basic_string16, BasicFlag_String, -1, STR_LIT("string16")}}, + {Type_Basic, {Basic_cstring16, BasicFlag_String, -1, STR_LIT("cstring16")}}, + + {Type_Basic, {Basic_any, 0, 16, STR_LIT("any")}}, {Type_Basic, {Basic_typeid, 0, 8, STR_LIT("typeid")}}, @@ -591,8 +602,12 @@ gb_global Type *t_uint = &basic_types[Basic_uint]; gb_global Type *t_uintptr = &basic_types[Basic_uintptr]; gb_global Type *t_rawptr = &basic_types[Basic_rawptr]; + gb_global Type *t_string = &basic_types[Basic_string]; gb_global Type *t_cstring = &basic_types[Basic_cstring]; +gb_global Type *t_string16 = &basic_types[Basic_string16]; +gb_global Type *t_cstring16 = &basic_types[Basic_cstring16]; + gb_global Type *t_any = &basic_types[Basic_any]; gb_global Type *t_typeid = &basic_types[Basic_typeid]; @@ -630,6 +645,8 @@ gb_global Type *t_untyped_uninit = &basic_types[Basic_UntypedUninit]; gb_global Type *t_u8_ptr = nullptr; gb_global Type *t_u8_multi_ptr = nullptr; +gb_global Type *t_u16_ptr = nullptr; +gb_global Type *t_u16_multi_ptr = nullptr; gb_global Type *t_int_ptr = nullptr; gb_global Type *t_i64_ptr = nullptr; gb_global Type *t_f64_ptr = nullptr; @@ -1292,6 +1309,14 @@ gb_internal bool is_type_string(Type *t) { } return false; } +gb_internal bool is_type_string16(Type *t) { + t = base_type(t); + if (t == nullptr) { return false; } + if (t->kind == Type_Basic) { + return t->Basic.kind == Basic_string16; + } + return false; +} gb_internal bool is_type_cstring(Type *t) { t = base_type(t); if (t == nullptr) { return false; } @@ -1300,6 +1325,14 @@ gb_internal bool is_type_cstring(Type *t) { } return false; } +gb_internal bool is_type_cstring16(Type *t) { + t = base_type(t); + if (t == nullptr) { return false; } + if (t->kind == Type_Basic) { + return t->Basic.kind == Basic_cstring16; + } + return false; +} gb_internal bool is_type_typed(Type *t) { t = base_type(t); if (t == nullptr) { return false; } @@ -1429,6 +1462,12 @@ gb_internal bool is_type_u8(Type *t) { } return false; } +gb_internal bool is_type_u16(Type *t) { + if (t->kind == Type_Basic) { + return t->Basic.kind == Basic_u16; + } + return false; +} gb_internal bool is_type_array(Type *t) { t = base_type(t); if (t == nullptr) { return false; } @@ -1690,6 +1729,39 @@ gb_internal bool is_type_rune_array(Type *t) { return false; } +gb_internal bool is_type_u16_slice(Type *t) { + t = base_type(t); + if (t == nullptr) { return false; } + if (t->kind == Type_Slice) { + return is_type_u16(t->Slice.elem); + } + return false; +} +gb_internal bool is_type_u16_array(Type *t) { + t = base_type(t); + if (t == nullptr) { return false; } + if (t->kind == Type_Array) { + return is_type_u16(t->Array.elem); + } + return false; +} +gb_internal bool is_type_u16_ptr(Type *t) { + t = base_type(t); + if (t == nullptr) { return false; } + if (t->kind == Type_Pointer) { + return is_type_u16(t->Slice.elem); + } + return false; +} +gb_internal bool is_type_u16_multi_ptr(Type *t) { + t = base_type(t); + if (t == nullptr) { return false; } + if (t->kind == Type_MultiPointer) { + return is_type_u16(t->Slice.elem); + } + return false; +} + gb_internal bool is_type_array_like(Type *t) { return is_type_array(t) || is_type_enumerated_array(t); @@ -2109,7 +2181,7 @@ gb_internal bool is_type_indexable(Type *t) { Type *bt = base_type(t); switch (bt->kind) { case Type_Basic: - return bt->Basic.kind == Basic_string; + return bt->Basic.kind == Basic_string || bt->Basic.kind == Basic_string16; case Type_Array: case Type_Slice: case Type_DynamicArray: @@ -2129,7 +2201,7 @@ gb_internal bool is_type_sliceable(Type *t) { Type *bt = base_type(t); switch (bt->kind) { case Type_Basic: - return bt->Basic.kind == Basic_string; + return bt->Basic.kind == Basic_string || bt->Basic.kind == Basic_string16; case Type_Array: case Type_Slice: case Type_DynamicArray: @@ -2376,6 +2448,7 @@ gb_internal bool type_has_nil(Type *t) { case Basic_any: return true; case Basic_cstring: + case Basic_cstring16: return true; case Basic_typeid: return true; @@ -2443,8 +2516,9 @@ gb_internal bool is_type_comparable(Type *t) { case Basic_rune: return true; case Basic_string: - return true; case Basic_cstring: + case Basic_string16: + case Basic_cstring16: return true; case Basic_typeid: return true; @@ -3774,10 +3848,12 @@ gb_internal i64 type_size_of(Type *t) { if (t->kind == Type_Basic) { GB_ASSERT_MSG(is_type_typed(t), "%s", type_to_string(t)); switch (t->Basic.kind) { - case Basic_string: size = 2*build_context.int_size; break; - case Basic_cstring: size = build_context.ptr_size; break; - case Basic_any: size = 16; break; - case Basic_typeid: size = 8; break; + case Basic_string: size = 2*build_context.int_size; break; + case Basic_cstring: size = build_context.ptr_size; break; + case Basic_string16: size = 2*build_context.int_size; break; + case Basic_cstring16: size = build_context.ptr_size; break; + case Basic_any: size = 16; break; + case Basic_typeid: size = 8; break; case Basic_int: case Basic_uint: size = build_context.int_size; @@ -3837,10 +3913,12 @@ gb_internal i64 type_align_of_internal(Type *t, TypePath *path) { case Type_Basic: { GB_ASSERT(is_type_typed(t)); switch (t->Basic.kind) { - case Basic_string: return build_context.int_size; - case Basic_cstring: return build_context.ptr_size; - case Basic_any: return 8; - case Basic_typeid: return 8; + case Basic_string: return build_context.int_size; + case Basic_cstring: return build_context.ptr_size; + case Basic_string16: return build_context.int_size; + case Basic_cstring16: return build_context.ptr_size; + case Basic_any: return 8; + case Basic_typeid: return 8; case Basic_int: case Basic_uint: return build_context.int_size; @@ -4088,10 +4166,12 @@ gb_internal i64 type_size_of_internal(Type *t, TypePath *path) { return size; } switch (kind) { - case Basic_string: return 2*build_context.int_size; - case Basic_cstring: return build_context.ptr_size; - case Basic_any: return 16; - case Basic_typeid: return 8; + case Basic_string: return 2*build_context.int_size; + case Basic_cstring: return build_context.ptr_size; + case Basic_string16: return 2*build_context.int_size; + case Basic_cstring16: return build_context.ptr_size; + case Basic_any: return 16; + case Basic_typeid: return 8; case Basic_int: case Basic_uint: return build_context.int_size; @@ -4320,6 +4400,15 @@ gb_internal i64 type_offset_of(Type *t, i64 index, Type **field_type_) { if (field_type_) *field_type_ = t_int; return build_context.int_size; // len } + } else if (t->Basic.kind == Basic_string16) { + switch (index) { + case 0: + if (field_type_) *field_type_ = t_u16_ptr; + return 0; // data + case 1: + if (field_type_) *field_type_ = t_int; + return build_context.int_size; // len + } } else if (t->Basic.kind == Basic_any) { switch (index) { case 0: @@ -4396,6 +4485,11 @@ gb_internal i64 type_offset_of_from_selection(Type *type, Selection sel) { case 0: t = t_rawptr; break; case 1: t = t_int; break; } + } else if (t->Basic.kind == Basic_string16) { + switch (index) { + case 0: t = t_rawptr; break; + case 1: t = t_int; break; + } } else if (t->Basic.kind == Basic_any) { switch (index) { case 0: t = t_rawptr; break; @@ -4637,6 +4731,11 @@ gb_internal Type *type_internal_index(Type *t, isize index) { GB_ASSERT(index == 0 || index == 1); return index == 0 ? t_u8_ptr : t_int; } + case Basic_string16: + { + GB_ASSERT(index == 0 || index == 1); + return index == 0 ? t_u16_ptr : t_int; + } case Basic_any: { GB_ASSERT(index == 0 || index == 1); -- cgit v1.2.3 From ae02d3d02d2eb5132fa7c6573ed7db20d7e18f3e Mon Sep 17 00:00:00 2001 From: gingerBill Date: Sat, 2 Aug 2025 11:55:16 +0100 Subject: Begin supporting `string16` across the core library --- base/intrinsics/intrinsics.odin | 1 + base/runtime/print.odin | 6 +++ core/encoding/cbor/tags.odin | 2 +- core/encoding/cbor/unmarshal.odin | 2 + core/encoding/json/marshal.odin | 8 ++-- core/encoding/json/unmarshal.odin | 4 +- core/flags/internal_rtti.odin | 2 + core/fmt/fmt.odin | 16 ++++---- core/io/io.odin | 4 +- core/reflect/types.odin | 8 ++-- src/check_builtin.cpp | 2 + src/check_expr.cpp | 42 +++++++++++++++++---- src/checker_builtin_procs.hpp | 2 + src/llvm_backend.cpp | 6 +++ src/llvm_backend_const.cpp | 77 +++++++++++++++++++++++++++++++++++++-- src/llvm_backend_debug.cpp | 14 +++++++ src/llvm_backend_expr.cpp | 3 +- src/llvm_backend_general.cpp | 37 +++++++++++++++++++ src/llvm_backend_utility.cpp | 19 +++++++++- src/string.cpp | 7 ++++ 20 files changed, 230 insertions(+), 32 deletions(-) (limited to 'src/string.cpp') diff --git a/base/intrinsics/intrinsics.odin b/base/intrinsics/intrinsics.odin index be75739fe..d45d24f48 100644 --- a/base/intrinsics/intrinsics.odin +++ b/base/intrinsics/intrinsics.odin @@ -141,6 +141,7 @@ type_is_quaternion :: proc($T: typeid) -> bool --- type_is_string :: proc($T: typeid) -> bool --- type_is_typeid :: proc($T: typeid) -> bool --- type_is_any :: proc($T: typeid) -> bool --- +type_is_string16 :: proc($T: typeid) -> bool --- type_is_endian_platform :: proc($T: typeid) -> bool --- type_is_endian_little :: proc($T: typeid) -> bool --- diff --git a/base/runtime/print.odin b/base/runtime/print.odin index 145f002d1..85ed49445 100644 --- a/base/runtime/print.odin +++ b/base/runtime/print.odin @@ -293,7 +293,13 @@ print_type :: #force_no_inline proc "contextless" (ti: ^Type_Info) { print_string("quaternion") print_u64(u64(8*ti.size)) case Type_Info_String: + if info.is_cstring { + print_byte('c') + } print_string("string") + if info.is_utf16 { + print_string("16") + } case Type_Info_Boolean: switch ti.id { case bool: print_string("bool") diff --git a/core/encoding/cbor/tags.odin b/core/encoding/cbor/tags.odin index 17420af46..e0e69cbf5 100644 --- a/core/encoding/cbor/tags.odin +++ b/core/encoding/cbor/tags.odin @@ -298,7 +298,7 @@ tag_base64_unmarshal :: proc(_: ^Tag_Implementation, d: Decoder, _: Tag_Number, #partial switch t in ti.variant { case reflect.Type_Info_String: - + assert(!t.is_utf16) if t.is_cstring { length := base64.decoded_len(bytes) builder := strings.builder_make(0, length+1) diff --git a/core/encoding/cbor/unmarshal.odin b/core/encoding/cbor/unmarshal.odin index 365ac5d6f..2840429f5 100644 --- a/core/encoding/cbor/unmarshal.odin +++ b/core/encoding/cbor/unmarshal.odin @@ -335,6 +335,8 @@ _unmarshal_value :: proc(d: Decoder, v: any, hdr: Header, allocator := context.a _unmarshal_bytes :: proc(d: Decoder, v: any, ti: ^reflect.Type_Info, hdr: Header, add: Add, allocator := context.allocator, loc := #caller_location) -> (err: Unmarshal_Error) { #partial switch t in ti.variant { case reflect.Type_Info_String: + assert(!t.is_utf16) + bytes := err_conv(_decode_bytes(d, add, allocator=allocator, loc=loc)) or_return if t.is_cstring { diff --git a/core/encoding/json/marshal.odin b/core/encoding/json/marshal.odin index ebb9a639c..cdb00a354 100644 --- a/core/encoding/json/marshal.odin +++ b/core/encoding/json/marshal.odin @@ -353,10 +353,10 @@ marshal_to_writer :: proc(w: io.Writer, v: any, opt: ^Marshal_Options) -> (err: #partial switch info in ti.variant { case runtime.Type_Info_String: switch x in v { - case string: - return x == "" - case cstring: - return x == nil || x == "" + case string: return x == "" + case cstring: return x == nil || x == "" + case string16: return x == "" + case cstring16: return x == nil || x == "" } case runtime.Type_Info_Any: return v.(any) == nil diff --git a/core/encoding/json/unmarshal.odin b/core/encoding/json/unmarshal.odin index b9ed1476f..51e7e3b81 100644 --- a/core/encoding/json/unmarshal.odin +++ b/core/encoding/json/unmarshal.odin @@ -570,7 +570,9 @@ unmarshal_object :: proc(p: ^Parser, v: any, end_token: Token_Kind) -> (err: Unm key_ptr: rawptr #partial switch tk in t.key.variant { - case runtime.Type_Info_String: + case runtime.Type_Info_String: + assert(!tk.is_utf16) + key_ptr = rawptr(&key) key_cstr: cstring if reflect.is_cstring(t.key) { diff --git a/core/flags/internal_rtti.odin b/core/flags/internal_rtti.odin index 1c559ca55..58224cc87 100644 --- a/core/flags/internal_rtti.odin +++ b/core/flags/internal_rtti.odin @@ -127,6 +127,8 @@ parse_and_set_pointer_by_base_type :: proc(ptr: rawptr, str: string, type_info: } case runtime.Type_Info_String: + assert(!specific_type_info.is_utf16) + if specific_type_info.is_cstring { cstr_ptr := (^cstring)(ptr) if cstr_ptr != nil { diff --git a/core/fmt/fmt.odin b/core/fmt/fmt.odin index 7fe6287d4..9c245de94 100644 --- a/core/fmt/fmt.odin +++ b/core/fmt/fmt.odin @@ -2346,14 +2346,14 @@ fmt_array :: proc(fi: ^Info, data: rawptr, n: int, elem_size: int, elem: ^reflec } switch reflect.type_info_base(elem).id { - case byte: fmt_string(fi, string(([^]byte)(data)[:n]), verb); return - case u16: print_utf16(fi, ([^]u16)(data)[:n]); return - case u16le: print_utf16(fi, ([^]u16le)(data)[:n]); return - case u16be: print_utf16(fi, ([^]u16be)(data)[:n]); return - case u32: print_utf32(fi, ([^]u32)(data)[:n]); return - case u32le: print_utf32(fi, ([^]u32le)(data)[:n]); return - case u32be: print_utf32(fi, ([^]u32be)(data)[:n]); return - case rune: print_utf32(fi, ([^]rune)(data)[:n]); return + case byte: fmt_string(fi, string (([^]byte)(data)[:n]), verb); return + case u16: fmt_string16(fi, string16(([^]u16) (data)[:n]), verb); return + case u16le: print_utf16(fi, ([^]u16le)(data)[:n]); return + case u16be: print_utf16(fi, ([^]u16be)(data)[:n]); return + case u32: print_utf32(fi, ([^]u32)(data)[:n]); return + case u32le: print_utf32(fi, ([^]u32le)(data)[:n]); return + case u32be: print_utf32(fi, ([^]u32be)(data)[:n]); return + case rune: print_utf32(fi, ([^]rune)(data)[:n]); return } } if verb == 'p' { diff --git a/core/io/io.odin b/core/io/io.odin index 5431519bf..c4eb6a073 100644 --- a/core/io/io.odin +++ b/core/io/io.odin @@ -319,7 +319,6 @@ write_string :: proc(s: Writer, str: string, n_written: ^int = nil) -> (n: int, write_string16 :: proc(s: Writer, str: string16, n_written: ^int = nil) -> (n: int, err: Error) { for i := 0; i < len(str); i += 1 { r := rune(utf16.REPLACEMENT_CHAR) - switch c := str[i]; { case c < utf16._surr1, utf16._surr3 <= c: r = rune(c) @@ -329,7 +328,8 @@ write_string16 :: proc(s: Writer, str: string16, n_written: ^int = nil) -> (n: i i += 1 } - w, err := write_rune(s, r, n_written) + w: int + w, err = write_rune(s, r, n_written) n += w if err != nil { return diff --git a/core/reflect/types.odin b/core/reflect/types.odin index 511c5c9bd..2351408cc 100644 --- a/core/reflect/types.odin +++ b/core/reflect/types.odin @@ -511,9 +511,11 @@ write_type_writer :: #force_no_inline proc(w: io.Writer, ti: ^Type_Info, n_writt io.write_i64(w, i64(8*ti.size), 10, &n) or_return case Type_Info_String: if info.is_cstring { - io.write_string(w, "cstring", &n) or_return - } else { - io.write_string(w, "string", &n) or_return + io.write_byte(w, 'c', &n) or_return + } + io.write_string(w, "string", &n) or_return + if info.is_utf16 { + io.write_string(w, "16", &n) or_return } case Type_Info_Boolean: switch ti.id { diff --git a/src/check_builtin.cpp b/src/check_builtin.cpp index d36cf4520..4abace637 100644 --- a/src/check_builtin.cpp +++ b/src/check_builtin.cpp @@ -19,6 +19,7 @@ gb_global BuiltinTypeIsProc *builtin_type_is_procs[BuiltinProc__type_simple_bool is_type_complex, is_type_quaternion, is_type_string, + is_type_string16, is_type_typeid, is_type_any, is_type_endian_platform, @@ -6139,6 +6140,7 @@ gb_internal bool check_builtin_procedure(CheckerContext *c, Operand *operand, As case BuiltinProc_type_is_complex: case BuiltinProc_type_is_quaternion: case BuiltinProc_type_is_string: + case BuiltinProc_type_is_string16: case BuiltinProc_type_is_typeid: case BuiltinProc_type_is_any: case BuiltinProc_type_is_endian_platform: diff --git a/src/check_expr.cpp b/src/check_expr.cpp index 57073e22f..8d2e4d637 100644 --- a/src/check_expr.cpp +++ b/src/check_expr.cpp @@ -2106,6 +2106,9 @@ gb_internal bool check_representable_as_constant(CheckerContext *c, ExactValue i } else if (is_type_boolean(type)) { return in_value.kind == ExactValue_Bool; } else if (is_type_string(type)) { + if (in_value.kind == ExactValue_String16) { + return is_type_string16(type) || is_type_cstring16(type); + } return in_value.kind == ExactValue_String; } else if (is_type_integer(type) || is_type_rune(type)) { if (in_value.kind == ExactValue_Bool) { @@ -2320,6 +2323,9 @@ gb_internal bool check_representable_as_constant(CheckerContext *c, ExactValue i if (in_value.kind == ExactValue_String) { return false; } + if (in_value.kind == ExactValue_String16) { + return false; + } if (out_value) *out_value = in_value; } else if (is_type_bit_set(type)) { if (in_value.kind == ExactValue_Integer) { @@ -4654,6 +4660,13 @@ gb_internal void convert_to_typed(CheckerContext *c, Operand *operand, Type *tar break; } } + } else if (operand->value.kind == ExactValue_String16) { + String16 s = operand->value.value_string16; + if (is_type_u16_array(t)) { + if (s.len == t->Array.count) { + break; + } + } } operand->mode = Addressing_Invalid; convert_untyped_error(c, operand, target_type); @@ -4983,6 +4996,12 @@ gb_internal ExactValue get_constant_field_single(CheckerContext *c, ExactValue v if (success_) *success_ = true; if (finish_) *finish_ = true; return exact_value_u64(val); + } else if (value.kind == ExactValue_String16) { + GB_ASSERT(0 <= index && index < value.value_string.len); + u16 val = value.value_string16[index]; + if (success_) *success_ = true; + if (finish_) *finish_ = true; + return exact_value_u64(val); } if (value.kind != ExactValue_Compound) { if (success_) *success_ = true; @@ -11124,15 +11143,21 @@ gb_internal ExprKind check_slice_expr(CheckerContext *c, Operand *o, Ast *node, o->expr = node; return kind; } - - String s = {}; - if (o->value.kind == ExactValue_String) { - s = o->value.value_string; - } - o->mode = Addressing_Constant; o->type = t; - o->value = exact_value_string(substring(s, cast(isize)indices[0], cast(isize)indices[1])); + + if (o->value.kind == ExactValue_String16) { + String16 s = o->value.value_string16; + + o->value = exact_value_string16(substring(s, cast(isize)indices[0], cast(isize)indices[1])); + } else { + String s = {}; + if (o->value.kind == ExactValue_String) { + s = o->value.value_string; + } + + o->value = exact_value_string(substring(s, cast(isize)indices[0], cast(isize)indices[1])); + } } return kind; } @@ -11221,6 +11246,7 @@ gb_internal ExprKind check_expr_base_internal(CheckerContext *c, Operand *o, Ast Type *t = t_invalid; switch (node->tav.value.kind) { case ExactValue_String: t = t_untyped_string; break; + case ExactValue_String16: t = t_string16; break; // TODO(bill): determine this correctly case ExactValue_Float: t = t_untyped_float; break; case ExactValue_Complex: t = t_untyped_complex; break; case ExactValue_Quaternion: t = t_untyped_quaternion; break; @@ -11657,6 +11683,8 @@ gb_internal bool is_exact_value_zero(ExactValue const &v) { return !v.value_bool; case ExactValue_String: return v.value_string.len == 0; + case ExactValue_String16: + return v.value_string16.len == 0; case ExactValue_Integer: return big_int_is_zero(&v.value_integer); case ExactValue_Float: diff --git a/src/checker_builtin_procs.hpp b/src/checker_builtin_procs.hpp index 8e135ab10..bff887d9e 100644 --- a/src/checker_builtin_procs.hpp +++ b/src/checker_builtin_procs.hpp @@ -250,6 +250,7 @@ BuiltinProc__type_simple_boolean_begin, BuiltinProc_type_is_complex, BuiltinProc_type_is_quaternion, BuiltinProc_type_is_string, + BuiltinProc_type_is_string16, BuiltinProc_type_is_typeid, BuiltinProc_type_is_any, @@ -607,6 +608,7 @@ gb_global BuiltinProc builtin_procs[BuiltinProc_COUNT] = { {STR_LIT("type_is_complex"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics}, {STR_LIT("type_is_quaternion"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics}, {STR_LIT("type_is_string"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics}, + {STR_LIT("type_is_string16"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics}, {STR_LIT("type_is_typeid"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics}, {STR_LIT("type_is_any"), 1, false, Expr_Expr, BuiltinProcPkg_intrinsics}, diff --git a/src/llvm_backend.cpp b/src/llvm_backend.cpp index 13a1d8cf3..f37415cc1 100644 --- a/src/llvm_backend.cpp +++ b/src/llvm_backend.cpp @@ -1264,7 +1264,13 @@ String lb_get_objc_type_encoding(Type *t, isize pointer_depth = 0) { case Basic_string: return build_context.metrics.int_size == 4 ? str_lit("{string=*i}") : str_lit("{string=*q}"); + case Basic_string16: + return build_context.metrics.int_size == 4 ? str_lit("{string16=*i}") : str_lit("{string16=*q}"); + case Basic_cstring: return str_lit("*"); + case Basic_cstring16: return str_lit("*"); + + case Basic_any: return str_lit("{any=^v^v}"); // rawptr + ^Type_Info case Basic_typeid: diff --git a/src/llvm_backend_const.cpp b/src/llvm_backend_const.cpp index c3112934e..8c05ed4a2 100644 --- a/src/llvm_backend_const.cpp +++ b/src/llvm_backend_const.cpp @@ -122,6 +122,25 @@ gb_internal lbValue lb_const_ptr_cast(lbModule *m, lbValue value, Type *t) { gb_internal LLVMValueRef llvm_const_string_internal(lbModule *m, Type *t, LLVMValueRef data, LLVMValueRef len) { + GB_ASSERT(!is_type_string16(t)); + if (build_context.metrics.ptr_size < build_context.metrics.int_size) { + LLVMValueRef values[3] = { + data, + LLVMConstNull(lb_type(m, t_i32)), + len, + }; + return llvm_const_named_struct_internal(lb_type(m, t), values, 3); + } else { + LLVMValueRef values[2] = { + data, + len, + }; + return llvm_const_named_struct_internal(lb_type(m, t), values, 2); + } +} + +gb_internal LLVMValueRef llvm_const_string16_internal(lbModule *m, Type *t, LLVMValueRef data, LLVMValueRef len) { + GB_ASSERT(is_type_string16(t)); if (build_context.metrics.ptr_size < build_context.metrics.int_size) { LLVMValueRef values[3] = { data, @@ -238,6 +257,10 @@ gb_internal lbValue lb_const_string(lbModule *m, String const &value) { return lb_const_value(m, t_string, exact_value_string(value)); } +gb_internal lbValue lb_const_string(lbModule *m, String16 const &value) { + return lb_const_value(m, t_string16, exact_value_string16(value)); +} + gb_internal lbValue lb_const_bool(lbModule *m, Type *type, bool value) { lbValue res = {}; @@ -569,7 +592,11 @@ gb_internal lbValue lb_const_value(lbModule *m, Type *type, ExactValue value, lb GB_ASSERT(is_type_slice(type)); res.value = lb_find_or_add_entity_string_byte_slice_with_type(m, value.value_string, original_type).value; return res; - } else { + } else if (value.kind == ExactValue_String16) { + GB_ASSERT(is_type_slice(type)); + GB_PANIC("TODO(bill): UTF-16 String"); + return res; + }else { ast_node(cl, CompoundLit, value.value_compound); isize count = cl->elems.count; @@ -751,15 +778,23 @@ gb_internal lbValue lb_const_value(lbModule *m, Type *type, ExactValue value, lb { bool custom_link_section = cc.link_section.len > 0; - LLVMValueRef ptr = lb_find_or_add_entity_string_ptr(m, value.value_string, custom_link_section); + LLVMValueRef ptr = nullptr; lbValue res = {}; res.type = default_type(original_type); + if (is_type_string16(res.type) || is_type_cstring16(res.type)) { + TEMPORARY_ALLOCATOR_GUARD(); + String16 s16 = string_to_string16(temporary_allocator(), value.value_string); + ptr = lb_find_or_add_entity_string16_ptr(m, s16, custom_link_section); + } else { + ptr = lb_find_or_add_entity_string_ptr(m, value.value_string, custom_link_section); + } + if (custom_link_section) { LLVMSetSection(ptr, alloc_cstring(permanent_allocator(), cc.link_section)); } - if (is_type_cstring(res.type)) { + if (is_type_cstring(res.type) || is_type_cstring16(res.type)) { res.value = ptr; } else { if (value.value_string.len == 0) { @@ -768,12 +803,46 @@ gb_internal lbValue lb_const_value(lbModule *m, Type *type, ExactValue value, lb LLVMValueRef str_len = LLVMConstInt(lb_type(m, t_int), value.value_string.len, true); GB_ASSERT(is_type_string(original_type)); - res.value = llvm_const_string_internal(m, original_type, ptr, str_len); + if (is_type_string16(res.type)) { + res.value = llvm_const_string16_internal(m, original_type, ptr, str_len); + } else { + res.value = llvm_const_string_internal(m, original_type, ptr, str_len); + } + } + + return res; + } + + case ExactValue_String16: + { + GB_ASSERT(is_type_string16(res.type) || is_type_cstring16(res.type)); + + bool custom_link_section = cc.link_section.len > 0; + + LLVMValueRef ptr = lb_find_or_add_entity_string16_ptr(m, value.value_string16, custom_link_section); + lbValue res = {}; + res.type = default_type(original_type); + + if (custom_link_section) { + LLVMSetSection(ptr, alloc_cstring(permanent_allocator(), cc.link_section)); + } + + if (is_type_cstring16(res.type)) { + res.value = ptr; + } else { + if (value.value_string16.len == 0) { + ptr = LLVMConstNull(lb_type(m, t_u8_ptr)); + } + LLVMValueRef str_len = LLVMConstInt(lb_type(m, t_int), value.value_string16.len, true); + GB_ASSERT(is_type_string(original_type)); + + res.value = llvm_const_string16_internal(m, original_type, ptr, str_len); } return res; } + case ExactValue_Integer: if (is_type_pointer(type) || is_type_multi_pointer(type) || is_type_proc(type)) { LLVMTypeRef t = lb_type(m, original_type); diff --git a/src/llvm_backend_debug.cpp b/src/llvm_backend_debug.cpp index 024c5564e..182920fc7 100644 --- a/src/llvm_backend_debug.cpp +++ b/src/llvm_backend_debug.cpp @@ -802,6 +802,20 @@ gb_internal LLVMMetadataRef lb_debug_type_internal(lbModule *m, Type *type) { LLVMMetadataRef char_type = lb_debug_type_basic_type(m, str_lit("char"), 8, LLVMDWARFTypeEncoding_Unsigned); return LLVMDIBuilderCreatePointerType(m->debug_builder, char_type, ptr_bits, ptr_bits, 0, "cstring", 7); } + + case Basic_string16: + { + LLVMMetadataRef elements[2] = {}; + elements[0] = lb_debug_struct_field(m, str_lit("data"), t_u16_ptr, 0); + elements[1] = lb_debug_struct_field(m, str_lit("len"), t_int, int_bits); + return lb_debug_basic_struct(m, str_lit("string16"), 2*int_bits, int_bits, elements, gb_count_of(elements)); + } + case Basic_cstring16: + { + LLVMMetadataRef char_type = lb_debug_type_basic_type(m, str_lit("wchar_t"), 16, LLVMDWARFTypeEncoding_Unsigned); + return LLVMDIBuilderCreatePointerType(m->debug_builder, char_type, ptr_bits, ptr_bits, 0, "cstring16", 7); + } + case Basic_any: { LLVMMetadataRef elements[2] = {}; diff --git a/src/llvm_backend_expr.cpp b/src/llvm_backend_expr.cpp index fbf0dea11..3463b6083 100644 --- a/src/llvm_backend_expr.cpp +++ b/src/llvm_backend_expr.cpp @@ -4354,12 +4354,13 @@ gb_internal lbAddr lb_build_addr_index_expr(lbProcedure *p, Ast *expr) { } - case Type_Basic: { // Basic_string + case Type_Basic: { // Basic_string/Basic_string16 lbValue str; lbValue elem; lbValue len; lbValue index; + str = lb_build_expr(p, ie->expr); if (deref) { str = lb_emit_load(p, str); diff --git a/src/llvm_backend_general.cpp b/src/llvm_backend_general.cpp index d9771a75b..9ef1c23c0 100644 --- a/src/llvm_backend_general.cpp +++ b/src/llvm_backend_general.cpp @@ -2715,6 +2715,43 @@ gb_internal LLVMValueRef lb_find_or_add_entity_string_ptr(lbModule *m, String co } } +gb_internal LLVMValueRef lb_find_or_add_entity_string16_ptr(lbModule *m, String16 const &str, bool custom_link_section) { + // TODO(bill): caching for UTF-16 strings + + LLVMValueRef indices[2] = {llvm_zero(m), llvm_zero(m)}; + + LLVMValueRef data = nullptr; + { + LLVMTypeRef llvm_u16 = LLVMInt16TypeInContext(m->ctx); + + TEMPORARY_ALLOCATOR_GUARD(); + + LLVMValueRef *values = gb_alloc_array(temporary_allocator(), LLVMValueRef, str.len+1); + + for (isize i = 0; i < str.len; i++) { + values[i] = LLVMConstInt(llvm_u16, str.text[i], false); + } + values[str.len] = LLVMConstInt(llvm_u16, 0, false); + + data = LLVMConstArray(llvm_u16, values, cast(unsigned)(str.len+1)); + } + + + u32 id = m->global_array_index.fetch_add(1); + gbString name = gb_string_make(temporary_allocator(), "csbs$"); + name = gb_string_appendc(name, m->module_name); + name = gb_string_append_fmt(name, "$%x", id); + + LLVMTypeRef type = LLVMTypeOf(data); + LLVMValueRef global_data = LLVMAddGlobal(m->mod, type, name); + LLVMSetInitializer(global_data, data); + lb_make_global_private_const(global_data); + LLVMSetAlignment(global_data, 1); + + LLVMValueRef ptr = LLVMConstInBoundsGEP2(type, global_data, indices, 2); + return ptr; +} + gb_internal lbValue lb_find_or_add_entity_string(lbModule *m, String const &str, bool custom_link_section) { LLVMValueRef ptr = nullptr; if (str.len != 0) { diff --git a/src/llvm_backend_utility.cpp b/src/llvm_backend_utility.cpp index d4117b7ff..ea1bae4e9 100644 --- a/src/llvm_backend_utility.cpp +++ b/src/llvm_backend_utility.cpp @@ -6,6 +6,7 @@ gb_internal bool lb_is_type_aggregate(Type *t) { case Type_Basic: switch (t->Basic.kind) { case Basic_string: + case Basic_string16: case Basic_any: return true; @@ -981,7 +982,8 @@ gb_internal i32 lb_convert_struct_index(lbModule *m, Type *t, i32 index) { } else if (build_context.ptr_size != build_context.int_size) { switch (t->kind) { case Type_Basic: - if (t->Basic.kind != Basic_string) { + if (t->Basic.kind != Basic_string && + t->Basic.kind != Basic_string16) { break; } /*fallthrough*/ @@ -1160,6 +1162,11 @@ gb_internal lbValue lb_emit_struct_ep(lbProcedure *p, lbValue s, i32 index) { case 0: result_type = alloc_type_pointer(t->Slice.elem); break; case 1: result_type = t_int; break; } + } else if (is_type_string16(t)) { + switch (index) { + case 0: result_type = t_u16_ptr; break; + case 1: result_type = t_int; break; + } } else if (is_type_string(t)) { switch (index) { case 0: result_type = t_u8_ptr; break; @@ -1273,6 +1280,12 @@ gb_internal lbValue lb_emit_struct_ev(lbProcedure *p, lbValue s, i32 index) { switch (t->kind) { case Type_Basic: switch (t->Basic.kind) { + case Basic_string16: + switch (index) { + case 0: result_type = t_u16_ptr; break; + case 1: result_type = t_int; break; + } + break; case Basic_string: switch (index) { case 0: result_type = t_u8_ptr; break; @@ -1440,6 +1453,10 @@ gb_internal lbValue lb_emit_deep_field_gep(lbProcedure *p, lbValue e, Selection e = lb_emit_struct_ep(p, e, index); break; + case Basic_string16: + e = lb_emit_struct_ep(p, e, index); + break; + default: GB_PANIC("un-gep-able type %s", type_to_string(type)); break; diff --git a/src/string.cpp b/src/string.cpp index 8405938f4..8cc0e93f3 100644 --- a/src/string.cpp +++ b/src/string.cpp @@ -79,6 +79,13 @@ gb_internal String substring(String const &s, isize lo, isize hi) { return make_string(s.text+lo, hi-lo); } +gb_internal String16 substring(String16 const &s, isize lo, isize hi) { + isize max = s.len; + GB_ASSERT_MSG(lo <= hi && hi <= max, "%td..%td..%td", lo, hi, max); + + return make_string16(s.text+lo, hi-lo); +} + gb_internal char *alloc_cstring(gbAllocator a, String s) { char *c_str = gb_alloc_array(a, char, s.len+1); -- cgit v1.2.3 From dca9bf0b0c8a3ad2ac6854c901d99868d3a296d5 Mon Sep 17 00:00:00 2001 From: gingerBill Date: Sat, 2 Aug 2025 13:11:34 +0100 Subject: Fix string16 literal length set in LLVM --- src/llvm_backend_const.cpp | 13 ++++++++++--- src/llvm_backend_general.cpp | 4 ++-- src/string.cpp | 2 -- 3 files changed, 12 insertions(+), 7 deletions(-) (limited to 'src/string.cpp') diff --git a/src/llvm_backend_const.cpp b/src/llvm_backend_const.cpp index cba0000cd..e64be49f2 100644 --- a/src/llvm_backend_const.cpp +++ b/src/llvm_backend_const.cpp @@ -782,9 +782,12 @@ gb_internal lbValue lb_const_value(lbModule *m, Type *type, ExactValue value, lb lbValue res = {}; res.type = default_type(original_type); + isize len = value.value_string.len; + if (is_type_string16(res.type) || is_type_cstring16(res.type)) { TEMPORARY_ALLOCATOR_GUARD(); String16 s16 = string_to_string16(temporary_allocator(), value.value_string); + len = s16.len; ptr = lb_find_or_add_entity_string16_ptr(m, s16, custom_link_section); } else { ptr = lb_find_or_add_entity_string_ptr(m, value.value_string, custom_link_section); @@ -797,10 +800,14 @@ gb_internal lbValue lb_const_value(lbModule *m, Type *type, ExactValue value, lb if (is_type_cstring(res.type) || is_type_cstring16(res.type)) { res.value = ptr; } else { - if (value.value_string.len == 0) { - ptr = LLVMConstNull(lb_type(m, t_u8_ptr)); + if (len == 0) { + if (is_type_string16(res.type)) { + ptr = LLVMConstNull(lb_type(m, t_u16_ptr)); + } else { + ptr = LLVMConstNull(lb_type(m, t_u8_ptr)); + } } - LLVMValueRef str_len = LLVMConstInt(lb_type(m, t_int), value.value_string.len, true); + LLVMValueRef str_len = LLVMConstInt(lb_type(m, t_int), len, true); GB_ASSERT(is_type_string(original_type)); if (is_type_string16(res.type)) { diff --git a/src/llvm_backend_general.cpp b/src/llvm_backend_general.cpp index 064d0ef39..d84b8302b 100644 --- a/src/llvm_backend_general.cpp +++ b/src/llvm_backend_general.cpp @@ -2758,7 +2758,7 @@ gb_internal LLVMValueRef lb_find_or_add_entity_string16_ptr(lbModule *m, String1 LLVMValueRef global_data = LLVMAddGlobal(m->mod, type, name); LLVMSetInitializer(global_data, data); lb_make_global_private_const(global_data); - LLVMSetAlignment(global_data, 1); + LLVMSetAlignment(global_data, 2); LLVMValueRef ptr = LLVMConstInBoundsGEP2(type, global_data, indices, 2); if (!custom_link_section) { @@ -2855,7 +2855,7 @@ gb_internal lbValue lb_find_or_add_entity_string16_slice_with_type(lbModule *m, LLVMValueRef global_data = LLVMAddGlobal(m->mod, type, name); LLVMSetInitializer(global_data, data); lb_make_global_private_const(global_data); - LLVMSetAlignment(global_data, 1); + LLVMSetAlignment(global_data, 2); i64 data_len = str.len; LLVMValueRef ptr = nullptr; diff --git a/src/string.cpp b/src/string.cpp index 8cc0e93f3..2087a5fee 100644 --- a/src/string.cpp +++ b/src/string.cpp @@ -658,7 +658,6 @@ gb_internal String normalize_path(gbAllocator a, String const &path, String cons -// TODO(bill): Make this non-windows specific gb_internal String16 string_to_string16(gbAllocator a, String s) { int len, len1; u16 *text; @@ -680,7 +679,6 @@ gb_internal String16 string_to_string16(gbAllocator a, String s) { return make_string16(nullptr, 0); } text[len] = 0; - return make_string16(text, len); } -- cgit v1.2.3 From 3e7395eba60d147e25183961d93d00c806be244c Mon Sep 17 00:00:00 2001 From: xenobas Date: Tue, 16 Sep 2025 21:40:57 +0100 Subject: fix: fix segfault on string_to_string16 --- src/string.cpp | 25 +++++++++++++++---------- 1 file changed, 15 insertions(+), 10 deletions(-) (limited to 'src/string.cpp') diff --git a/src/string.cpp b/src/string.cpp index 2087a5fee..9c08114a7 100644 --- a/src/string.cpp +++ b/src/string.cpp @@ -633,23 +633,28 @@ gb_internal String normalize_path(gbAllocator a, String const &path, String cons return WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, widechar_input, input_length, output, output_size, nullptr, nullptr); } #elif defined(GB_SYSTEM_UNIX) || defined(GB_SYSTEM_OSX) - - #include + #include gb_internal int convert_multibyte_to_widechar(char const *multibyte_input, usize input_length, wchar_t *output, usize output_size) { - iconv_t conv = iconv_open("WCHAR_T", "UTF-8"); - size_t result = iconv(conv, cast(char **)&multibyte_input, &input_length, cast(char **)&output, &output_size); - iconv_close(conv); + String string = copy_string(heap_allocator(), make_string(cast(u8 const*)multibyte_input, input_length)); /* Guarantee NULL terminator */ + u8* input = string.text; + + mbstate_t ps = { 0 }; + size_t result = mbsrtowcs(output, cast(const char**)&input, output_size, &ps); - return cast(int)result; + gb_free(heap_allocator(), string.text); + return (result == (size_t)-1) ? -1 : (int)result; } gb_internal int convert_widechar_to_multibyte(wchar_t const *widechar_input, usize input_length, char* output, usize output_size) { - iconv_t conv = iconv_open("UTF-8", "WCHAR_T"); - size_t result = iconv(conv, cast(char**) &widechar_input, &input_length, cast(char **)&output, &output_size); - iconv_close(conv); + String string = copy_string(heap_allocator(), make_string(cast(u8 const*)widechar_input, input_length)); /* Guarantee NULL terminator */ + u8* input = string.text; + + mbstate_t ps = { 0 }; + size_t result = wcsrtombs(output, cast(const wchar_t**)&input, output_size, &ps); - return cast(int)result; + gb_free(heap_allocator(), string.text); + return (result == (size_t)-1) ? -1 : (int)result; } #else #error Implement system -- cgit v1.2.3