/home/uke/oil/mycpp/gc_builtins.cc

Source (jump to first uncovered line)
#include <errno.h>  // errno
#include <float.h>  // DBL_MIN, DBL_MAX
#include <math.h>   // INFINITY
#include <stdio.h>  // required for readline/readline.h (man readline)

#include "_build/detected-cpp-config.h"
#include "mycpp/gc_list.h"
#include "mycpp/gc_str.h"
#ifdef HAVE_READLINE
  #include "cpp/frontend_pyreadline.h"
#endif

// Translation of Python's print().
void print(BigStr* s) {
  fputs(s->data_, stdout);  // print until first NUL
  fputc('\n', stdout);
}

BigStr* str(int i) {
  BigStr* s = OverAllocatedStr(kIntBufSize);
  int length = snprintf(s->data(), kIntBufSize, "%d", i);
  s->MaybeShrink(length);
  return s;
}

BigStr* str(double d) {
  char buf[64];  // overestimate, but we use snprintf() to be safe

  int n = sizeof(buf) - 2;  // in case we add '.0'

  // The round tripping test in mycpp/float_test.cc tells us:
  // %.9g - FLOAT round trip
  // %.17g - DOUBLE round trip
  // But this causes problems in practice, e.g. for 3.14, or 1/3
  // int length = snprintf(buf, n, "%.17g", d);

  // So use 1 less digit, which happens to match Python 3 and node.js (but not
  // Python 2)
  int length = snprintf(buf, n, "%.16g", d);

  // TODO: This may depend on LC_NUMERIC locale!

  // We may return the strings:
  //    inf  -inf   nan
  // But this shouldn't come up much, because Python code changes it to:
  //    INFINITY   -INFINITY   NAN
  if (strchr(buf, 'i') || strchr(buf, 'n')) {
    return StrFromC(buf);  // don't add .0
  }

  // Problem:
  // %f prints 3.0000000 and 3.500000
  // %g prints 3 and 3.5
  //
  // We want 3.0 and 3.5, so add '.0' in some cases
  if (!strchr(buf, '.')) {  // 12345 -> 12345.0
    buf[length] = '.';
    buf[length + 1] = '0';
    buf[length + 2] = '\0';
  }

  return StrFromC(buf);
}
// %a is a hexfloat form, probably don't need that
// int length = snprintf(buf, n, "%a", d);

// Do we need this API?  Or is mylib.InternedStr(BigStr* s, int start, int end)
// better for getting values out of Token.line without allocating?
//
// e.g. mylib.InternedStr(tok.line, tok.start, tok.start+1)
//
// Also for SmallStr, we don't care about interning.  Only for HeapStr.

BigStr* intern(BigStr* s) {
  // TODO: put in table gHeap.interned_
  return s;
}

// Print quoted string.  Called by StrFormat('%r').
// TODO: consider using J8 notation instead, since error messages show that
// string.
BigStr* repr(BigStr* s) {
  // Worst case: \0 becomes 4 bytes as '\\x00', and then two quote bytes.
  int n = len(s);
  int upper_bound = n * 4 + 2;

  BigStr* result = OverAllocatedStr(upper_bound);

  // Single quote by default.
  char quote = '\'';
  if (memchr(s->data_, '\'', n) && !memchr(s->data_, '"', n)) {
    quote = '"';
  }
  char* p = result->data_;

  // From PyString_Repr()
  *p++ = quote;
  for (int i = 0; i < n; ++i) {
    unsigned char c = static_cast<unsigned char>(s->data_[i]);
    if (c == quote || c == '\\') {
      *p++ = '\\';
      *p++ = c;
    } else if (c == '\t') {
      *p++ = '\\';
      *p++ = 't';
    } else if (c == '\n') {
      *p++ = '\\';
      *p++ = 'n';
    } else if (c == '\r') {
      *p++ = '\\';
      *p++ = 'r';
    } else if (0x20 <= c && c < 0x80) {
      *p++ = c;
    } else {
      // Unprintable becomes \xff.
      // TODO: Consider \yff.  This is similar to J8 strings, but we don't
      // decode UTF-8.
      sprintf(p, "\\x%02x", c & 0xff);
      p += 4;
    }
  }
  *p++ = quote;
  *p = '\0';

  int length = p - result->data_;
  result->MaybeShrink(length);
  return result;
}

// Helper functions that don't use exceptions.

bool StringToInt(const char* s, int length, int base, int* result) {
  if (length == 0) {
    return false;  // empty string isn't a valid integer
  }

  // Note: sizeof(int) is often 4 bytes on both 32-bit and 64-bit
  //       sizeof(long) is often 4 bytes on both 32-bit but 8 bytes on 64-bit
  // static_assert(sizeof(long) == 8);

  char* pos;  // mutated by strtol

  errno = 0;
  long v = strtol(s, &pos, base);

  if (errno == ERANGE) {
    switch (v) {
    case LONG_MIN:
      return false;  // underflow of long, which may be 64 bits
    case LONG_MAX:
      return false;  // overflow of long
    }
  }

  // It should ALSO fit in an int, not just a long
  if (v > INT_MAX) {
    return false;
  }
  if (v < INT_MIN) {
    return false;
  }

  const char* end = s + length;
  if (pos == end) {
    *result = v;
    return true;  // strtol() consumed ALL characters.
  }

  while (pos < end) {
    if (!IsAsciiWhitespace(*pos)) {
      return false;  // Trailing non-space
    }
    pos++;
  }

  *result = v;
  return true;  // Trailing space is OK
}

bool StringToInt64(const char* s, int length, int base, int64_t* result) {
  if (length == 0) {
    return false;  // empty string isn't a valid integer
  }

  // These should be the same type
  static_assert(sizeof(long long) == sizeof(int64_t), "");

  char* pos;  // mutated by strtol

  errno = 0;
  long long v = strtoll(s, &pos, base);

  if (errno == ERANGE) {
    switch (v) {
    case LLONG_MIN:
      return false;  // underflow
    case LLONG_MAX:
      return false;  // overflow
    }
  }

  const char* end = s + length;
  if (pos == end) {
    *result = v;
    return true;  // strtol() consumed ALL characters.
  }

  while (pos < end) {
    if (!IsAsciiWhitespace(*pos)) {
      return false;  // Trailing non-space
    }
    pos++;
  }

  *result = v;
  return true;  // Trailing space is OK
}

int to_int(BigStr* s, int base) {
  int i;
  if (StringToInt(s->data_, len(s), base, &i)) {
    return i;  // truncated to int
  } else {
    throw Alloc<ValueError>();
  }
}

BigStr* chr(int i) {
  // NOTE: i should be less than 256, in which we could return an object from
  // GLOBAL_STR() pool, like StrIter
  auto result = NewStr(1);
  result->data_[0] = i;
  return result;
}

int ord(BigStr* s) {
  assert(len(s) == 1);
  // signed to unsigned conversion, so we don't get values like -127
  uint8_t c = static_cast<uint8_t>(s->data_[0]);
  return c;
}

bool to_bool(BigStr* s) {
  return len(s) != 0;
}

double to_float(int i) {
  return static_cast<double>(i);
}

double to_float(BigStr* s) {
  char* begin = s->data_;
  char* end_pos = begin + len(s);
  char* orig_end = end_pos;

  errno = 0;
  double result = strtod(begin, &end_pos);

  if (errno == ERANGE) {  // error: overflow or underflow
    if (result >= HUGE_VAL) {
      return INFINITY;
    } else if (result <= -HUGE_VAL) {
      return -INFINITY;
    } else if (-DBL_MIN <= result && result <= DBL_MIN) {
      return 0.0;
    } else {
      FAIL("Invalid value after ERANGE");
    }
  }
  if (end_pos == begin) {  // error: not a floating point number
    throw Alloc<ValueError>();
  }
  if (end_pos != orig_end) {  // trailing data like '5s' not alowed
    while (end_pos < orig_end) {
      if (!IsAsciiWhitespace(*end_pos)) {
        throw Alloc<ValueError>();  // Trailing non-space
      }
      end_pos++;
    }
  }

  return result;
}

// e.g. ('a' in 'abc')
bool str_contains(BigStr* haystack, BigStr* needle) {
  // Common case
  if (len(needle) == 1) {
    return memchr(haystack->data_, needle->data_[0], len(haystack));
  }

  if (len(needle) > len(haystack)) {
    return false;
  }

  // General case. TODO: We could use a smarter substring algorithm.

  const char* end = haystack->data_ + len(haystack);
  const char* last_possible = end - len(needle);
  const char* p = haystack->data_;

  while (p <= last_possible) {
    if (memcmp(p, needle->data_, len(needle)) == 0) {
      return true;
    }
    p++;
  }
  return false;
}

BigStr* str_repeat(BigStr* s, int times) {
  // Python allows -1 too, and Oil used that
  if (times <= 0) {
    return kEmptyString;
  }
  int len_ = len(s);
  int new_len = len_ * times;
  BigStr* result = NewStr(new_len);

  char* dest = result->data_;
  for (int i = 0; i < times; i++) {
    memcpy(dest, s->data_, len_);
    dest += len_;
  }
  return result;
}

// for os_path.join()
// NOTE(Jesse): Perfect candidate for BoundedBuffer
BigStr* str_concat3(BigStr* a, BigStr* b, BigStr* c) {
  int a_len = len(a);
  int b_len = len(b);
  int c_len = len(c);

  int new_len = a_len + b_len + c_len;
  BigStr* result = NewStr(new_len);
  char* pos = result->data_;

  memcpy(pos, a->data_, a_len);
  pos += a_len;

  memcpy(pos, b->data_, b_len);
  pos += b_len;

  memcpy(pos, c->data_, c_len);

  assert(pos + c_len == result->data_ + new_len);

  return result;
}

BigStr* str_concat(BigStr* a, BigStr* b) {
  int a_len = len(a);
  int b_len = len(b);
  int new_len = a_len + b_len;
  BigStr* result = NewStr(new_len);
  char* buf = result->data_;

  memcpy(buf, a->data_, a_len);
  memcpy(buf + a_len, b->data_, b_len);

  return result;
}

//
// Comparators
//

bool str_equals(BigStr* left, BigStr* right) {
  // Fast path for identical strings.  String deduplication during GC could
  // make this more likely.  String interning could guarantee it, allowing us
  // to remove memcmp().
  if (left == right) {
    return true;
  }

  // TODO: It would be nice to remove this condition, but I think we need MyPy
  // strict None checking for it
  if (left == nullptr || right == nullptr) {
    return false;
  }

  if (left->len_ != right->len_) {
    return false;
  }

  return memcmp(left->data_, right->data_, left->len_) == 0;
}

bool maybe_str_equals(BigStr* left, BigStr* right) {
  if (left && right) {
    return str_equals(left, right);
  }

  if (!left && !right) {
    return true;  // None == None
  }

  return false;  // one is None and one is a BigStr*
}

bool items_equal(BigStr* left, BigStr* right) {
  return str_equals(left, right);
}

bool keys_equal(BigStr* left, BigStr* right) {
  return items_equal(left, right);
}

bool items_equal(Tuple2<int, int>* t1, Tuple2<int, int>* t2) {
  return (t1->at0() == t2->at0()) && (t1->at1() == t2->at1());
}

bool keys_equal(Tuple2<int, int>* t1, Tuple2<int, int>* t2) {
  return items_equal(t1, t2);
}

bool items_equal(Tuple2<BigStr*, int>* t1, Tuple2<BigStr*, int>* t2) {
  return items_equal(t1->at0(), t2->at0()) && (t1->at1() == t2->at1());
}

bool keys_equal(Tuple2<BigStr*, int>* t1, Tuple2<BigStr*, int>* t2) {
  return items_equal(t1, t2);
}

bool str_equals_c(BigStr* s, const char* c_string, int c_len) {
  // Needs SmallStr change
  if (len(s) == c_len) {
    return memcmp(s->data_, c_string, c_len) == 0;
  } else {
    return false;
  }
}

bool str_equals0(const char* c_string, BigStr* s) {
  int n = strlen(c_string);
  if (len(s) == n) {
    return memcmp(s->data_, c_string, n) == 0;
  } else {
    return false;
  }
}

int hash(BigStr* s) {
  return s->hash(fnv1);
}

int max(int a, int b) {
  return std::max(a, b);
}

int min(int a, int b) {
  return std::min(a, b);
}

int max(List<int>* elems) {
  int n = len(elems);
  if (n < 1) {
    throw Alloc<ValueError>();
  }

  int ret = elems->at(0);
  for (int i = 0; i < n; ++i) {
    int cand = elems->at(i);
    if (cand > ret) {
      ret = cand;
    }
  }

  return ret;
}

cpp

Coverage Report

Created: 2025-06-15 03:57

Line	Count	Source (jump to first uncovered line)
1		#include <errno.h> // errno
2		#include <float.h> // DBL_MIN, DBL_MAX
3		#include <math.h> // INFINITY
4		#include <stdio.h> // required for readline/readline.h (man readline)
5
6		#include "_build/detected-cpp-config.h"
7		#include "mycpp/gc_list.h"
8		#include "mycpp/gc_str.h"
9		#ifdef HAVE_READLINE
10		#include "cpp/frontend_pyreadline.h"
11		#endif
12
13		// Translation of Python's print().
14	5	void print(BigStr* s) {
15	5	fputs(s->data_, stdout); // print until first NUL
16	5	fputc('\n', stdout);
17	5	}
18
19	0	BigStr* str(int i) {
20	0	BigStr* s = OverAllocatedStr(kIntBufSize);
21	0	int length = snprintf(s->data(), kIntBufSize, "%d", i);
22	0	s->MaybeShrink(length);
23	0	return s;
24	0	}
25
26	0	BigStr* str(double d) {
27	0	char buf[64]; // overestimate, but we use snprintf() to be safe
28
29	0	int n = sizeof(buf) - 2; // in case we add '.0'
30
31		// The round tripping test in mycpp/float_test.cc tells us:
32		// %.9g - FLOAT round trip
33		// %.17g - DOUBLE round trip
34		// But this causes problems in practice, e.g. for 3.14, or 1/3
35		// int length = snprintf(buf, n, "%.17g", d);
36
37		// So use 1 less digit, which happens to match Python 3 and node.js (but not
38		// Python 2)
39	0	int length = snprintf(buf, n, "%.16g", d);
40
41		// TODO: This may depend on LC_NUMERIC locale!
42
43		// We may return the strings:
44		// inf -inf nan
45		// But this shouldn't come up much, because Python code changes it to:
46		// INFINITY -INFINITY NAN
47	0	if (strchr(buf, 'i') \|\| strchr(buf, 'n')) {
48	0	return StrFromC(buf); // don't add .0
49	0	}
50
51		// Problem:
52		// %f prints 3.0000000 and 3.500000
53		// %g prints 3 and 3.5
54		//
55		// We want 3.0 and 3.5, so add '.0' in some cases
56	0	if (!strchr(buf, '.')) { // 12345 -> 12345.0
57	0	buf[length] = '.';
58	0	buf[length + 1] = '0';
59	0	buf[length + 2] = '\0';
60	0	}
61
62	0	return StrFromC(buf);
63	0	}
64		// %a is a hexfloat form, probably don't need that
65		// int length = snprintf(buf, n, "%a", d);
66
67		// Do we need this API? Or is mylib.InternedStr(BigStr* s, int start, int end)
68		// better for getting values out of Token.line without allocating?
69		//
70		// e.g. mylib.InternedStr(tok.line, tok.start, tok.start+1)
71		//
72		// Also for SmallStr, we don't care about interning. Only for HeapStr.
73
74	0	BigStr* intern(BigStr* s) {
75		// TODO: put in table gHeap.interned_
76	0	return s;
77	0	}
78
79		// Print quoted string. Called by StrFormat('%r').
80		// TODO: consider using J8 notation instead, since error messages show that
81		// string.
82	1	BigStr* repr(BigStr* s) {
83		// Worst case: \0 becomes 4 bytes as '\\x00', and then two quote bytes.
84	1	int n = len(s);
85	1	int upper_bound = n * 4 + 2;
86
87	1	BigStr* result = OverAllocatedStr(upper_bound);
88
89		// Single quote by default.
90	1	char quote = '\'';
91	1	if (memchr(s->data_, '\'', n) && !memchr(s->data_, '"', n)) {
92	0	quote = '"';
93	0	}
94	1	char* p = result->data_;
95
96		// From PyString_Repr()
97	1	*p++ = quote;
98	8	for (int i = 0; i < n; ++i) {
99	7	unsigned char c = static_cast<unsigned char>(s->data_[i]);
100	7	if (c == quote \|\| c == '\\') {
101	0	*p++ = '\\';
102	0	*p++ = c;
103	7	} else if (c == '\t') {
104	0	*p++ = '\\';
105	0	*p++ = 't';
106	7	} else if (c == '\n') {
107	0	*p++ = '\\';
108	0	*p++ = 'n';
109	7	} else if (c == '\r') {
110	0	*p++ = '\\';
111	0	*p++ = 'r';
112	7	} else if (0x20 <= c && c < 0x80) {
113	7	*p++ = c;
114	7	} else {
115		// Unprintable becomes \xff.
116		// TODO: Consider \yff. This is similar to J8 strings, but we don't
117		// decode UTF-8.
118	0	sprintf(p, "\\x%02x", c & 0xff);
119	0	p += 4;
120	0	}
121	7	}
122	1	*p++ = quote;
123	1	*p = '\0';
124
125	1	int length = p - result->data_;
126	1	result->MaybeShrink(length);
127	1	return result;
128	1	}
129
130		// Helper functions that don't use exceptions.
131
132	2	bool StringToInt(const char* s, int length, int base, int* result) {
133	2	if (length == 0) {
134	0	return false; // empty string isn't a valid integer
135	0	}
136
137		// Note: sizeof(int) is often 4 bytes on both 32-bit and 64-bit
138		// sizeof(long) is often 4 bytes on both 32-bit but 8 bytes on 64-bit
139		// static_assert(sizeof(long) == 8);
140
141	2	char* pos; // mutated by strtol
142
143	2	errno = 0;
144	2	long v = strtol(s, &pos, base);
145
146	2	if (errno == ERANGE) {
147	0	switch (v) {
148	0	case LONG_MIN:
149	0	return false; // underflow of long, which may be 64 bits
150	0	case LONG_MAX:
151	0	return false; // overflow of long
152	0	}
153	0	}
154
155		// It should ALSO fit in an int, not just a long
156	2	if (v > INT_MAX) {
157	0	return false;
158	0	}
159	2	if (v < INT_MIN) {
160	0	return false;
161	0	}
162
163	2	const char* end = s + length;
164	2	if (pos == end) {
165	1	*result = v;
166	1	return true; // strtol() consumed ALL characters.
167	1	}
168
169	1	while (pos < end) {
170	1	if (!IsAsciiWhitespace(*pos)) {
171	1	return false; // Trailing non-space
172	1	}
173	0	pos++;
174	0	}
175
176	0	*result = v;
177	0	return true; // Trailing space is OK
178	1	}
179
180	0	bool StringToInt64(const char* s, int length, int base, int64_t* result) {
181	0	if (length == 0) {
182	0	return false; // empty string isn't a valid integer
183	0	}
184
185		// These should be the same type
186	0	static_assert(sizeof(long long) == sizeof(int64_t), "");
187
188	0	char* pos; // mutated by strtol
189
190	0	errno = 0;
191	0	long long v = strtoll(s, &pos, base);
192
193	0	if (errno == ERANGE) {
194	0	switch (v) {
195	0	case LLONG_MIN:
196	0	return false; // underflow
197	0	case LLONG_MAX:
198	0	return false; // overflow
199	0	}
200	0	}
201
202	0	const char* end = s + length;
203	0	if (pos == end) {
204	0	*result = v;
205	0	return true; // strtol() consumed ALL characters.
206	0	}
207
208	0	while (pos < end) {
209	0	if (!IsAsciiWhitespace(*pos)) {
210	0	return false; // Trailing non-space
211	0	}
212	0	pos++;
213	0	}
214
215	0	*result = v;
216	0	return true; // Trailing space is OK
217	0	}
218
219	2	int to_int(BigStr* s, int base) {
220	2	int i;
221	2	if (StringToInt(s->data_, len(s), base, &i)) {
222	1	return i; // truncated to int
223	1	} else {
224	1	throw Alloc<ValueError>();
225	1	}
226	2	}
227
228	0	BigStr* chr(int i) {
229		// NOTE: i should be less than 256, in which we could return an object from
230		// GLOBAL_STR() pool, like StrIter
231	0	auto result = NewStr(1);
232	0	result->data_[0] = i;
233	0	return result;
234	0	}
235
236	0	int ord(BigStr* s) {
237	0	assert(len(s) == 1);
238		// signed to unsigned conversion, so we don't get values like -127
239	0	uint8_t c = static_cast<uint8_t>(s->data_[0]);
240	0	return c;
241	0	}
242
243	0	bool to_bool(BigStr* s) {
244	0	return len(s) != 0;
245	0	}
246
247	0	double to_float(int i) {
248	0	return static_cast<double>(i);
249	0	}
250
251	0	double to_float(BigStr* s) {
252	0	char* begin = s->data_;
253	0	char* end_pos = begin + len(s);
254	0	char* orig_end = end_pos;
255
256	0	errno = 0;
257	0	double result = strtod(begin, &end_pos);
258
259	0	if (errno == ERANGE) { // error: overflow or underflow
260	0	if (result >= HUGE_VAL) {
261	0	return INFINITY;
262	0	} else if (result <= -HUGE_VAL) {
263	0	return -INFINITY;
264	0	} else if (-DBL_MIN <= result && result <= DBL_MIN) {
265	0	return 0.0;
266	0	} else {
267	0	FAIL("Invalid value after ERANGE");
268	0	}
269	0	}
270	0	if (end_pos == begin) { // error: not a floating point number
271	0	throw Alloc<ValueError>();
272	0	}
273	0	if (end_pos != orig_end) { // trailing data like '5s' not alowed
274	0	while (end_pos < orig_end) {
275	0	if (!IsAsciiWhitespace(*end_pos)) {
276	0	throw Alloc<ValueError>(); // Trailing non-space
277	0	}
278	0	end_pos++;
279	0	}
280	0	}
281
282	0	return result;
283	0	}
284
285		// e.g. ('a' in 'abc')
286	0	bool str_contains(BigStr* haystack, BigStr* needle) {
287		// Common case
288	0	if (len(needle) == 1) {
289	0	return memchr(haystack->data_, needle->data_[0], len(haystack));
290	0	}
291
292	0	if (len(needle) > len(haystack)) {
293	0	return false;
294	0	}
295
296		// General case. TODO: We could use a smarter substring algorithm.
297
298	0	const char* end = haystack->data_ + len(haystack);
299	0	const char* last_possible = end - len(needle);
300	0	const char* p = haystack->data_;
301
302	0	while (p <= last_possible) {
303	0	if (memcmp(p, needle->data_, len(needle)) == 0) {
304	0	return true;
305	0	}
306	0	p++;
307	0	}
308	0	return false;
309	0	}
310
311	0	BigStr* str_repeat(BigStr* s, int times) {
312		// Python allows -1 too, and Oil used that
313	0	if (times <= 0) {
314	0	return kEmptyString;
315	0	}
316	0	int len_ = len(s);
317	0	int new_len = len_ * times;
318	0	BigStr* result = NewStr(new_len);
319
320	0	char* dest = result->data_;
321	0	for (int i = 0; i < times; i++) {
322	0	memcpy(dest, s->data_, len_);
323	0	dest += len_;
324	0	}
325	0	return result;
326	0	}
327
328		// for os_path.join()
329		// NOTE(Jesse): Perfect candidate for BoundedBuffer
330	0	BigStr* str_concat3(BigStr* a, BigStr* b, BigStr* c) {
331	0	int a_len = len(a);
332	0	int b_len = len(b);
333	0	int c_len = len(c);
334
335	0	int new_len = a_len + b_len + c_len;
336	0	BigStr* result = NewStr(new_len);
337	0	char* pos = result->data_;
338
339	0	memcpy(pos, a->data_, a_len);
340	0	pos += a_len;
341
342	0	memcpy(pos, b->data_, b_len);
343	0	pos += b_len;
344
345	0	memcpy(pos, c->data_, c_len);
346
347	0	assert(pos + c_len == result->data_ + new_len);
348
349	0	return result;
350	0	}
351
352	0	BigStr* str_concat(BigStr* a, BigStr* b) {
353	0	int a_len = len(a);
354	0	int b_len = len(b);
355	0	int new_len = a_len + b_len;
356	0	BigStr* result = NewStr(new_len);
357	0	char* buf = result->data_;
358
359	0	memcpy(buf, a->data_, a_len);
360	0	memcpy(buf + a_len, b->data_, b_len);
361
362	0	return result;
363	0	}
364
365		//
366		// Comparators
367		//
368
369	32	bool str_equals(BigStr* left, BigStr* right) {
370		// Fast path for identical strings. String deduplication during GC could
371		// make this more likely. String interning could guarantee it, allowing us
372		// to remove memcmp().
373	32	if (left == right) {
374	1	return true;
375	1	}
376
377		// TODO: It would be nice to remove this condition, but I think we need MyPy
378		// strict None checking for it
379	31	if (left == nullptr \|\| right == nullptr) {
380	0	return false;
381	0	}
382
383	31	if (left->len_ != right->len_) {
384	1	return false;
385	1	}
386
387	30	return memcmp(left->data_, right->data_, left->len_) == 0;
388	31	}
389
390	0	bool maybe_str_equals(BigStr* left, BigStr* right) {
391	0	if (left && right) {
392	0	return str_equals(left, right);
393	0	}
394
395	0	if (!left && !right) {
396	0	return true; // None == None
397	0	}
398
399	0	return false; // one is None and one is a BigStr*
400	0	}
401
402	7	bool items_equal(BigStr* left, BigStr* right) {
403	7	return str_equals(left, right);
404	7	}
405
406	7	bool keys_equal(BigStr* left, BigStr* right) {
407	7	return items_equal(left, right);
408	7	}
409
410	0	bool items_equal(Tuple2<int, int>* t1, Tuple2<int, int>* t2) {
411	0	return (t1->at0() == t2->at0()) && (t1->at1() == t2->at1());
412	0	}
413
414	0	bool keys_equal(Tuple2<int, int>* t1, Tuple2<int, int>* t2) {
415	0	return items_equal(t1, t2);
416	0	}
417
418	0	bool items_equal(Tuple2<BigStr, int> t1, Tuple2<BigStr, int> t2) {
419	0	return items_equal(t1->at0(), t2->at0()) && (t1->at1() == t2->at1());
420	0	}
421
422	0	bool keys_equal(Tuple2<BigStr, int> t1, Tuple2<BigStr, int> t2) {
423	0	return items_equal(t1, t2);
424	0	}
425
426	0	bool str_equals_c(BigStr* s, const char* c_string, int c_len) {
427		// Needs SmallStr change
428	0	if (len(s) == c_len) {
429	0	return memcmp(s->data_, c_string, c_len) == 0;
430	0	} else {
431	0	return false;
432	0	}
433	0	}
434
435	129	bool str_equals0(const char* c_string, BigStr* s) {
436	129	int n = strlen(c_string);
437	129	if (len(s) == n) {
438	13	return memcmp(s->data_, c_string, n) == 0;
439	116	} else {
440	116	return false;
441	116	}
442	129	}
443
444	0	int hash(BigStr* s) {
445	0	return s->hash(fnv1);
446	0	}
447
448	0	int max(int a, int b) {
449	0	return std::max(a, b);
450	0	}
451
452	0	int min(int a, int b) {
453	0	return std::min(a, b);
454	0	}
455
456	0	int max(List<int>* elems) {
457	0	int n = len(elems);
458	0	if (n < 1) {
459	0	throw Alloc<ValueError>();
460	0	}
461
462	0	int ret = elems->at(0);
463	0	for (int i = 0; i < n; ++i) {
464	0	int cand = elems->at(i);
465	0	if (cand > ret) {
466	0	ret = cand;
467	0	}
468	0	}
469
470	0	return ret;
471	0	}