TermiNexus
e71daeea0e
按规范「注释只讲解作用，不掺杂历史原因，尽量英文」：把 includes 下 75 个 .vp 的中文 # 注释与三引号 docstring 改写为英文，每个注释块压到 2 行以内。 保留要点：作用、约束、危险警告（不得/必须/注意）、ABI 与类型约定（如 BYTEPTR | t.CPtr = i8** 的跨模块坍缩坑）、位宽/长度约定、fail-fast 要求、内存与句柄所有权。 未动：文件头注释（规范豁免，仍是中文）、运行时中文字符串常量（printf / fatal_error / MlirFail 等用户可见消息，属代码字面量）。 验证：独立脚本比对 HEAD 与工作区的代码骨架（剥离 # 注释与三引号 docstring 后的非空行）——75 个文件逐字节一致；行尾 CRLF/LF 未被改动；全文剩余中文仅 547 行头注释、0 条行尾注释。bootstrap 三重重编 SUCCESS；全量 77/77 顺序回归通过。
609 lines
19 KiB
Plaintext
609 lines
19 KiB
Plaintext
from stdint import *
|
||
import t, c
|
||
import stdio
|
||
|
||
|
||
# String copy function
|
||
def strcpy(dest: str, src: str) -> str:
|
||
if dest == None or src == None: return None
|
||
original_dest: str = dest
|
||
for s in src:
|
||
dest[0] = s
|
||
dest += 1
|
||
dest[0] = 0
|
||
return original_dest
|
||
|
||
|
||
# String concatenate function
|
||
def strcat(dest: str, src: str) -> str:
|
||
if dest == None or src == None: return dest
|
||
original_dest: str = dest
|
||
# Find end of dest
|
||
while dest[0] != '\0':
|
||
dest += 1
|
||
# Copy src
|
||
for s in src:
|
||
dest[0] = s
|
||
dest += 1
|
||
dest[0] = 0
|
||
return original_dest
|
||
|
||
|
||
# String copy function with limited length
|
||
def strncpy(dest: str, src: str, n: sizet) -> str:
|
||
if dest == None or src == None or n == 0: return dest
|
||
original_dest: str = dest
|
||
i: sizet = 0
|
||
for s in src:
|
||
if i >= n: break
|
||
dest[0] = s
|
||
dest += 1
|
||
i += 1
|
||
# Fill remaining space with '\0'
|
||
while i < n:
|
||
dest[0] = 0
|
||
dest += 1
|
||
i += 1
|
||
return original_dest
|
||
|
||
|
||
# String concatenate function with limited length
|
||
def strncat(dest: str, src: str, n: sizet) -> str:
|
||
if dest == None or src == None or n == 0: return dest
|
||
original_dest: str = dest
|
||
# Find end of dest
|
||
while dest[0] != '\0':
|
||
dest += 1
|
||
# Copy up to n chars from src
|
||
i: sizet = 0
|
||
for s in src:
|
||
if i >= n: break
|
||
dest[0] = s
|
||
dest += 1
|
||
i += 1
|
||
dest[0] = 0
|
||
return original_dest
|
||
|
||
|
||
# String length function
|
||
export def strlen(src: str) -> sizet:
|
||
length: sizet = 0
|
||
for _ in src:
|
||
length += 1
|
||
return length
|
||
|
||
# String comparison function
|
||
# Note: str1[0] - str2[0] is an i8 subtraction that wraps (0xFF means -1), so both operands
|
||
# must be promoted to i32 before subtracting (as in memcmp below); otherwise strcmp("abc","abd") returns 255 instead of -1.
|
||
def strcmp(str1: str, str2: str) -> int:
|
||
# Pointer-equality short-circuit: identical pointers guarantee identical contents, so return 0 directly and skip the full byte loop.
|
||
# Semantics strictly unchanged (same pointer => same content => strcmp result is 0); different pointers fall through to the loop below.
|
||
if str1 == str2:
|
||
return 0
|
||
while str1[0] != '\0' and str2[0] != '\0':
|
||
if str1[0] != str2[0]:
|
||
return int(str1[0]) - int(str2[0])
|
||
str1 += 1
|
||
str2 += 1
|
||
return int(str1[0]) - int(str2[0])
|
||
|
||
def samestr(str1: str, str2: str) -> bool:
|
||
return strcmp(str1, str2) == 0
|
||
|
||
|
||
# String compare function with limited length
|
||
def strncmp(str1: str, str2: str, n: sizet) -> int:
|
||
# Pointer-equality short-circuit (same as strcmp): identical pointers compare equal for any n, so return 0.
|
||
# Semantics strictly unchanged.
|
||
if str1 == str2:
|
||
return 0
|
||
i: sizet = 0
|
||
while str1[0] != '\0' and str2[0] != '\0' and i < n:
|
||
if str1[0] != str2[0]:
|
||
return int(str1[0]) - int(str2[0])
|
||
str1 += 1
|
||
str2 += 1
|
||
i += 1
|
||
if i < n:
|
||
return int(str1[0]) - int(str2[0])
|
||
return 0
|
||
|
||
|
||
# Memory compare function
|
||
def memcmp(ptr1: t.CPtr, ptr2: t.CPtr, n: sizet) -> int:
|
||
p1: BYTEPTR = BYTEPTR(ptr1)
|
||
p2: BYTEPTR = BYTEPTR(ptr2)
|
||
for i in range(n):
|
||
if p1[i] != p2[i]:
|
||
return int(p1[i]) - int(p2[i])
|
||
return 0
|
||
|
||
|
||
# String find function
|
||
def strchr(s: str, cr: int) -> str:
|
||
while s[0] != '\0':
|
||
if s[0] == char(cr):
|
||
return str(s)
|
||
s += 1
|
||
if cr == '\0':
|
||
return str(s)
|
||
return None
|
||
|
||
# String reverse find function (find last occurrence of character)
|
||
def strrchr(s: str, cr: int) -> str:
|
||
last: str = None
|
||
while s[0] != '\0':
|
||
if s[0] == char(cr):
|
||
last = s
|
||
s += 1
|
||
if cr == '\0': return s
|
||
return last
|
||
|
||
|
||
# Substring search function (find first occurrence of needle in s, return pointer to that position, or None if not found)
|
||
# Hot-path optimization: use strchr to jump to first-char candidates instead of calling strncmp at every byte, and drop the upfront strlen(s) (strncmp stops at '\0').
|
||
# Semantics unchanged: an empty needle returns s; not found returns None.
|
||
def strstr(s: str, needle: str) -> str:
|
||
if needle[0] == 0:
|
||
return s
|
||
needle_len: sizet = strlen(needle)
|
||
if needle_len == 0:
|
||
return s
|
||
first_ch: int = int(needle[0])
|
||
cur: str = s
|
||
while True:
|
||
hit: str = strchr(cur, first_ch)
|
||
if hit is None:
|
||
return None
|
||
if strncmp(hit, needle, needle_len) == 0:
|
||
return hit
|
||
cur = hit + 1
|
||
|
||
# String skip function
|
||
def strspn(s: str, skip: str = " ") -> int:
|
||
si: int = 0
|
||
while s[si] != 0:
|
||
found: int = 0
|
||
sj: int = 0
|
||
while skip[sj] != 0:
|
||
if s[si] == skip[sj]:
|
||
found = 1
|
||
break
|
||
sj += 1
|
||
if found == 0:
|
||
break
|
||
si += 1
|
||
return si
|
||
|
||
# Memory set function
|
||
export def memset(ptr: t.CPtr, value: int, num: sizet) -> t.CPtr:
|
||
# Original implementation:
|
||
# p: BYTEPTR = BYTEPTR(ptr)
|
||
# i: sizet
|
||
# for i in range(num):
|
||
# p[i] = t.CUnsignedChar(value)
|
||
if num == 0: return ptr
|
||
c.Asm(f"""mov rdi, {c.AsmInp(ptr, t.ASM_DESCR.REG_ANY)}
|
||
mov eax, {c.AsmInp(value, t.ASM_DESCR.REG_ANY)}
|
||
mov rcx, {c.AsmInp(num, t.ASM_DESCR.REG_ANY)}
|
||
rep stosb""",
|
||
op=[ASM_DESCR.CLOBBER_MEMORY, ASM_DESCR.CLOBBER_RAX,
|
||
ASM_DESCR.CLOBBER_RCX, ASM_DESCR.CLOBBER_RDI])
|
||
return ptr
|
||
|
||
|
||
def memset32(ptr: t.CPtr, value: uint32, count: sizet) -> t.CPtr:
|
||
# Original implementation:
|
||
# p: UINT32PTR = UINT32PTR(ptr)
|
||
# i: sizet
|
||
# for i in range(count):
|
||
# p[i] = value
|
||
if count == 0: return ptr
|
||
c.Asm(f"""mov rdi, {c.AsmInp(ptr, t.ASM_DESCR.REG_ANY)}
|
||
mov eax, {c.AsmInp(value, t.ASM_DESCR.REG_ANY)}
|
||
mov rcx, {c.AsmInp(count, t.ASM_DESCR.REG_ANY)}
|
||
rep stosd""",
|
||
op=[ASM_DESCR.CLOBBER_MEMORY, ASM_DESCR.CLOBBER_RAX,
|
||
ASM_DESCR.CLOBBER_RCX, ASM_DESCR.CLOBBER_RDI])
|
||
return ptr
|
||
|
||
|
||
# Memory copy function
|
||
export def memcpy(dest: t.CPtr, src: t.CPtr, num: sizet) -> t.CPtr:
|
||
# Original implementation:
|
||
# d: BYTEPTR = BYTEPTR(dest)
|
||
# s: BYTEPTR = BYTEPTR(src)
|
||
# i: sizet
|
||
# for i in range(num):
|
||
# d[i] = s[i]
|
||
if num == 0: return dest
|
||
c.Asm(f"""mov rdi, {c.AsmInp(dest, t.ASM_DESCR.REG_ANY)}
|
||
mov rsi, {c.AsmInp(src, t.ASM_DESCR.REG_ANY)}
|
||
mov rcx, {c.AsmInp(num, t.ASM_DESCR.REG_ANY)}
|
||
rep movsb""",
|
||
op=[ASM_DESCR.CLOBBER_MEMORY, ASM_DESCR.CLOBBER_RCX,
|
||
ASM_DESCR.CLOBBER_RDI, ASM_DESCR.CLOBBER_RSI])
|
||
return dest
|
||
|
||
|
||
# Memory move function
|
||
def memmove(dest: t.CPtr, src: t.CPtr, num: sizet) -> t.CPtr:
|
||
# Original implementation:
|
||
# d: BYTEPTR = BYTEPTR(dest)
|
||
# s: BYTEPTR = BYTEPTR(src)
|
||
# i: sizet
|
||
# if d < s:
|
||
# for i in range(num):
|
||
# d[i] = s[i]
|
||
# elif d > s:
|
||
# for i in range(num, 0, -1):
|
||
# d[i - 1] = s[i - 1]
|
||
if num == 0: return dest
|
||
c.Asm(f"""mov rdi, {c.AsmInp(dest, t.ASM_DESCR.REG_ANY)}
|
||
mov rsi, {c.AsmInp(src, t.ASM_DESCR.REG_ANY)}
|
||
mov rcx, {c.AsmInp(num, t.ASM_DESCR.REG_ANY)}
|
||
cmp rdi, rsi
|
||
jb 2f
|
||
std
|
||
add rdi, rcx
|
||
dec rdi
|
||
add rsi, rcx
|
||
dec rsi
|
||
rep movsb
|
||
cld
|
||
jmp 3f
|
||
2:
|
||
rep movsb
|
||
3:""",
|
||
op=[ASM_DESCR.CLOBBER_MEMORY, ASM_DESCR.CLOBBER_RCX,
|
||
ASM_DESCR.CLOBBER_RDI, ASM_DESCR.CLOBBER_RSI])
|
||
return dest
|
||
|
||
|
||
def atoi(src: str) -> int:
|
||
num: int = 0
|
||
sign: int = 1
|
||
s: str = src
|
||
if s == None: return 0
|
||
# Skip whitespace
|
||
while s[0] == ' ' or s[0] == '\t' or s[0] == '\n' or s[0] == '\r':
|
||
s += 1
|
||
# Sign
|
||
if s[0] == '-':
|
||
sign = -1
|
||
s += 1
|
||
elif s[0] == '+':
|
||
s += 1
|
||
for ch in s:
|
||
if ch < '0' or ch > '9': break
|
||
digit: int = ch - '0'
|
||
num = num * 10 + digit
|
||
return num * sign
|
||
|
||
|
||
def atoll(src: str) -> int64:
|
||
num: int64 = 0
|
||
sign: int64 = 1
|
||
s: str = src
|
||
if s == None: return 0
|
||
while s[0] == ' ' or s[0] == '\t' or s[0] == '\n' or s[0] == '\r':
|
||
s += 1
|
||
if s[0] == '-':
|
||
sign = -1
|
||
s += 1
|
||
elif s[0] == '+':
|
||
s += 1
|
||
# Hex prefix 0x / 0X
|
||
is_hex: int = 0
|
||
if s[0] == '0' and (s[1] == 'x' or s[1] == 'X'):
|
||
is_hex = 1
|
||
s += 2
|
||
if is_hex:
|
||
for ch in s:
|
||
d: int = 0
|
||
if ch >= '0' and ch <= '9':
|
||
d = ch - '0'
|
||
elif ch >= 'a' and ch <= 'f':
|
||
d = ch - 'a' + 10
|
||
elif ch >= 'A' and ch <= 'F':
|
||
d = ch - 'A' + 10
|
||
else:
|
||
break
|
||
digit: int64 = int64(d)
|
||
num = num * 16 + digit
|
||
else:
|
||
for ch in s:
|
||
if ch < '0' or ch > '9': break
|
||
d: int = ch - '0'
|
||
digit: int64 = int64(d)
|
||
num = num * 10 + digit
|
||
return num * sign
|
||
|
||
|
||
# ============================================================
|
||
# Correctly-rounded decimal literal -> double conversion (N4): mantissa in u64 (<=19 digits, 20th digit rounds once),
|
||
# scaled once by an exact power of ten in double-double (hi+lo, ~106 bits) then rounded a single final time.
|
||
# Convention: no global table (10^0..10^22 built by repeated multiplication of 10, each step exact).
|
||
# For p<0 compute 10^p*2^600 in the normal range and scale back exactly, avoiding subnormal loss in the double-double low part below ~1e-292.
|
||
# ============================================================
|
||
|
||
# _dd_split_hi - upper half of the Dekker split (hi = top 26 bits).
|
||
# Magnitude rescaling: multiply by 2^-28 when |a| > 6.7e292 and by 2^28 when |a| < 1e-292 (powers of two, lossless for normals); caller gets the lower half via lo = a - hi.
|
||
def _dd_split_hi(a: double) -> double:
|
||
aa: double = a
|
||
big: int = 0
|
||
small: int = 0
|
||
if a > 6.7e292 or a < -6.7e292:
|
||
aa = a * 3.725290298461914e-09
|
||
big = 1
|
||
elif (a > 0.0 and a < 1.0e-292) or (a < 0.0 and a > -1.0e-292):
|
||
aa = a * 268435456.0
|
||
small = 1
|
||
tt: double = aa * 134217729.0
|
||
hi: double = tt - (tt - aa)
|
||
if big != 0:
|
||
return hi * 268435456.0
|
||
if small != 0:
|
||
return hi * 3.725290298461914e-09
|
||
return hi
|
||
|
||
|
||
# _dd_mul - double-double multiply: out = a × b (out[0]=hi, out[1]=lo)
|
||
def _dd_mul(ah: double, al: double, bh: double, bl: double,
|
||
out: t.CArray[double, 2]):
|
||
p: double = ah * bh
|
||
a_hi: double = _dd_split_hi(ah)
|
||
a_lo: double = ah - a_hi
|
||
b_hi: double = _dd_split_hi(bh)
|
||
b_lo: double = bh - b_hi
|
||
e: double = ((a_hi * b_hi - p) + a_hi * b_lo + a_lo * b_hi) + a_lo * b_lo
|
||
e = e + (ah * bl + al * bh)
|
||
s: double = p + e
|
||
out[0] = s
|
||
out[1] = e - (s - p)
|
||
|
||
|
||
# _dd_add - double-double add: out = a + b
|
||
def _dd_add(ah: double, al: double, bh: double, bl: double,
|
||
out: t.CArray[double, 2]):
|
||
s: double = ah + bh
|
||
bb: double = s - ah
|
||
e: double = (ah - (s - bb)) + (bh - bb)
|
||
e = e + (al + bl)
|
||
r0: double = s + e
|
||
out[0] = r0
|
||
out[1] = e - (r0 - s)
|
||
|
||
|
||
# _dd_div - double-double divide: out = a / b (q0 plus one residual correction step)
|
||
def _dd_div(ah: double, al: double, bh: double, bl: double,
|
||
out: t.CArray[double, 2], tmp: t.CArray[double, 2]):
|
||
q0: double = ah / bh
|
||
_dd_mul(bh, bl, q0, 0.0, tmp)
|
||
_dd_add(ah, al, -tmp[0], -tmp[1], tmp)
|
||
q1: double = tmp[0] / bh
|
||
_dd_add(q0, 0.0, q1, 0.0, out)
|
||
|
||
|
||
# _dd_from_u64 - u64 -> double-double (exact: hi is the rounded value, lo the exact residual)
|
||
def _dd_from_u64(m: uint64, out: t.CArray[double, 2]):
|
||
hi: double = double(m)
|
||
lo: double = 0.0
|
||
if m >= uint64(9007199254740992):
|
||
lo = double(m - uint64(hi))
|
||
s: double = hi + lo
|
||
bb: double = s - hi
|
||
out[0] = s
|
||
out[1] = (hi - (s - bb)) + (lo - bb)
|
||
|
||
|
||
# _dd_pow10 - double-double 10^e (e >= 0; chunked by <=22 so each chunk is exact)
|
||
def _dd_pow10(e: int, out: t.CArray[double, 2], tmp: t.CArray[double, 2]):
|
||
out[0] = 1.0
|
||
out[1] = 0.0
|
||
k: int = e
|
||
while k > 0:
|
||
step: int = 22
|
||
if k < 22:
|
||
step = k
|
||
p10: double = 1.0
|
||
for i in range(step):
|
||
p10 = p10 * 10.0
|
||
_dd_mul(out[0], out[1], p10, 0.0, tmp)
|
||
out[0] = tmp[0]
|
||
out[1] = tmp[1]
|
||
k -= step
|
||
|
||
|
||
# _pow2 - 2^n (|n| <= 1000): repeated multiply by 2/0.5, exact at every step
|
||
def _pow2(n: int) -> double:
|
||
r: double = 1.0
|
||
i: int = n
|
||
if i >= 0:
|
||
while i > 0:
|
||
r = r * 2.0
|
||
i -= 1
|
||
else:
|
||
while i < 0:
|
||
r = r * 0.5
|
||
i += 1
|
||
return r
|
||
|
||
|
||
# _atof_scale_stepwise - stepwise ×/÷10 scaling (fallback for subnormal/overflow cases)
|
||
def _atof_scale_stepwise(x: double, p: int) -> double:
|
||
r: double = x
|
||
e: int = p
|
||
while e > 0:
|
||
r = r * 10.0
|
||
e -= 1
|
||
while e < 0:
|
||
r = r / 10.0
|
||
e += 1
|
||
return r
|
||
|
||
|
||
def atof(src: str) -> double:
|
||
s: str = src
|
||
if s == None: return 0.0
|
||
# Skip whitespace
|
||
while s[0] == ' ' or s[0] == '\t' or s[0] == '\n' or s[0] == '\r':
|
||
s += 1
|
||
# Sign
|
||
sign: double = 1.0
|
||
if s[0] == '-':
|
||
sign = -1.0
|
||
s += 1
|
||
elif s[0] == '+':
|
||
s += 1
|
||
# Mantissa: significant digits packed into u64 (<=19 digits), tracking the decimal exponent
|
||
mant: uint64 = 0
|
||
nd: int = 0
|
||
dexp: int = 0
|
||
seen_dot: int = 0
|
||
started: int = 0
|
||
rounded: int = 0
|
||
while s[0] != '\0':
|
||
ch: char = s[0]
|
||
if ch >= '0' and ch <= '9':
|
||
d: int = ch - '0'
|
||
if started == 0:
|
||
# Leading zeros are not stored in the mantissa; leading zeros in the fraction still advance the decimal exponent
|
||
if d != 0:
|
||
started = 1
|
||
mant = uint64(d)
|
||
nd = 1
|
||
if seen_dot != 0:
|
||
dexp -= 1
|
||
elif seen_dot != 0:
|
||
dexp -= 1
|
||
elif nd < 19:
|
||
mant = mant * uint64(10) + uint64(d)
|
||
nd += 1
|
||
if seen_dot != 0:
|
||
dexp -= 1
|
||
else:
|
||
# From the 20th digit on: dropped integer digits bump the exponent; dropped fraction digits do not.
|
||
# Only the 20th digit is used for one rounding (dropped amount <= 1e-19 relative error).
|
||
if seen_dot == 0:
|
||
dexp += 1
|
||
elif rounded == 0:
|
||
if d >= 5:
|
||
mant += uint64(1)
|
||
rounded = 1
|
||
elif ch == '.':
|
||
seen_dot = 1
|
||
else:
|
||
break
|
||
s += 1
|
||
if started == 0:
|
||
return sign * 0.0
|
||
# Carry: rounding may turn 19 digits into 20
|
||
if mant >= uint64(10000000000000000000):
|
||
mant = mant // uint64(10)
|
||
dexp += 1
|
||
# Exponent part
|
||
exp: int = 0
|
||
if s[0] == 'e' or s[0] == 'E':
|
||
s += 1
|
||
exp_sign: int = 1
|
||
if s[0] == '-':
|
||
exp_sign = -1
|
||
s += 1
|
||
elif s[0] == '+':
|
||
s += 1
|
||
while s[0] >= '0' and s[0] <= '9':
|
||
exp = exp * 10 + (s[0] - '0')
|
||
s += 1
|
||
exp = exp * exp_sign
|
||
p: int = dexp + exp
|
||
if p == 0:
|
||
return sign * double(mant)
|
||
# Double-double scaling: mant × 10^p, single final rounding
|
||
ta: t.CArray[double, 2]
|
||
tb: t.CArray[double, 2]
|
||
tc: t.CArray[double, 2]
|
||
tr: t.CArray[double, 2]
|
||
_dd_from_u64(mant, ta)
|
||
v: double = 0.0
|
||
if p > 0:
|
||
_dd_pow10(p, tb, tr)
|
||
_dd_mul(ta[0], ta[1], tb[0], tb[1], tr)
|
||
v = tr[0] + tr[1]
|
||
else:
|
||
# p < 0: compute 2^600 / 10^(-p) (= 10^p * 2^600) in the normal range, then scale back exactly.
|
||
# Direct 10^p loses effective bits to subnormals for large |p|; divide the denominator in chunks of <=308 (10^313 overflows to inf and yields NaN).
|
||
tb[0] = _pow2(600)
|
||
tb[1] = 0.0
|
||
q: int = -p
|
||
while q > 0:
|
||
step2: int = 308
|
||
if q < 308:
|
||
step2 = q
|
||
_dd_pow10(step2, tc, tr)
|
||
_dd_div(tb[0], tb[1], tc[0], tc[1], tr, tc)
|
||
tb[0] = tr[0]
|
||
tb[1] = tr[1]
|
||
q -= step2
|
||
_dd_mul(ta[0], ta[1], tb[0], tb[1], tr)
|
||
v = (tr[0] + tr[1]) * _pow2(-600)
|
||
# Normal range (not NaN / not inf / not subnormal) -> the correctly rounded result
|
||
if v == v and v <= 1.7976931348623157e308 and v >= -1.7976931348623157e308 \
|
||
and (v == 0.0 or v >= 2.2250738585072014e-308 or v <= -2.2250738585072014e-308):
|
||
return sign * v
|
||
# Overflow / subnormal / NaN -> fall back to stepwise scaling
|
||
return sign * _atof_scale_stepwise(double(mant), p)
|
||
|
||
|
||
def atof_out(src: str, out: double*) -> int:
|
||
"""Write the atof result to out[0] (out-parameter form, avoiding the cross-module double return bug).
|
||
The return value only signals success (1) or failure (0); the value is passed via the out pointer."""
|
||
if out is None:
|
||
return 0
|
||
v: double = atof(src)
|
||
out[0] = v
|
||
return 1
|
||
|
||
def split(s: str, delim: str, result: t.CArray[str], cap: int) -> int:
|
||
"""Split s by a single-character delimiter, writing token pointers to result[0..] and a None sentinel at the end.
|
||
cap is the declared capacity (element count); token count is limited to <= cap-1, leaving one slot for the sentinel."""
|
||
count: int = 0
|
||
start: str = s
|
||
if not s or not delim or not result:
|
||
return 0
|
||
if cap <= 0:
|
||
return 0
|
||
d: char = delim[0] # single delimiter
|
||
|
||
while start[0]:
|
||
# Skip delimiters
|
||
while start[0] == d: start += 1
|
||
if not start[0]: break
|
||
# Capacity guard: keep the last slot for the end sentinel
|
||
if count >= cap - 1:
|
||
break
|
||
|
||
# Record token start
|
||
result[count] = start
|
||
count += 1
|
||
|
||
# Find the next delimiter and truncate
|
||
while start[0] and start[0] != d: start += 1
|
||
if start[0]:
|
||
start[0] = '\0'
|
||
start += 1
|
||
result[count] = None # end marker
|
||
return count
|
||
|
||
|
||
ascii_lowercase: str = "abcdefghijklmnopqrstuvwxyz"
|
||
ascii_uppercase: str = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
|
||
ascii_letters: str = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"
|
||
|
||
digits: str = "0123456789"
|
||
hexdigits: str = "0123456789abcdefABCDEF"
|
||
octdigits: str = "01234567"
|
||
punctuation: str = "!\"#$%&'()*+,-./:;<=>?@[]^_`{|}~"
|
||
whitespace: str = " \t\n\r\f\v"
|
||
printable: str = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~ \t\n\r\f\v"
|