---

part1/mutator_values.py

@@ -0,0 +1,128 @@
+import random
+import re
+
+import sqlite_static_helper
+
+
+def _quote_str(text: str) -> str:
+  """O'Reilly -> 'O''Reilly'"""
+  return "'" + text.replace("'", "''") + "'"
+
+
+_TOKEN_RE = re.compile(
+  r"""
+    (?P<line_comment>--[^\n]*)                            |
+    (?P<block_comment>/\*.*?\*/)                          |
+    (?P<string>'(?:[^']|'')*')                            |
+    (?P<dquoted>"(?:[^"]|"")*")                           |
+    (?P<bracket>\[[^\]]*\])                               |
+    (?P<backtick>`(?:[^`]|``)*`)                          |
+    (?P<blob>[xX]'[0-9a-fA-F]*')                          |
+    (?P<number>\b\d+(?:\.\d+)?(?:[eE][+-]?\d+)?\b)        |
+    (?P<ident>[A-Za-z_][A-Za-z0-9_]*)                     |
+    (?P<op>[<>!=]=|<>|\|\||::|->>?|[+\-*/%<>=&|^~,.;()@]) |
+    (?P<ws>\s+)                                           |
+    (?P<other>.)
+  """,
+  re.VERBOSE | re.DOTALL,
+)
+
+
+def _tokenize(s: str) -> list[tuple[str | None, str]]:
+  """[("line_comment", "-- This is a comment!"), ("number", 0), ...]"""
+  return [(m.lastgroup, m.group()) for m in _TOKEN_RE.finditer(s)]
+
+
+_UNSAFE_KINDS = {
+  "line_comment", "block_comment", "string", "dquoted", "bracket", "backtick",
+  "blob"
+}
+
+
+def _sub_in_safe(s: str, pattern: re.Pattern, repl, max_subs: int = 1) -> str:
+  """Run `pattern.sub(repl, …)` only outside strings/comments/blob literals."""
+  if max_subs <= 0:
+    return s
+  parts: list[str] = []
+  done = 0
+  for kind, text in _tokenize(s):
+    if done >= max_subs or kind in _UNSAFE_KINDS:
+      parts.append(text)
+      continue
+    new_text, n = pattern.subn(repl, text, count=max_subs - done)
+    parts.append(new_text)
+    done += n
+  return ''.join(parts)
+
+
+_NUM_RE = re.compile(r'\b\d+(?:\.\d+)?(?:[eE][+-]?\d+)?\b')
+
+
+def mut_replace_num_with_edge_int(s: str) -> str:
+  """Replace the left-most number with a random edge-case integer"""
+  return _sub_in_safe(
+    s, _NUM_RE, lambda m: str(random.choice(sqlite_static_helper.EDGE_INTS)), 1)
+
+
+def mut_replace_num_with_random_int(s: str) -> str:
+  """Replace the left-most number with a random integer"""
+  return _sub_in_safe(s, _NUM_RE,
+                      lambda m: str(random.randint(-(2**63), 2**63 - 1)), 1)
+
+
+def mut_replace_num_with_edge_float(s: str) -> str:
+  """Replace the left-most number with a random edge-case floating-point number"""
+  return _sub_in_safe(
+    s, _NUM_RE, lambda m: repr(random.choice(sqlite_static_helper.EDGE_FLOATS)),
+    1)
+
+
+def mut_replace_num_with_edge_string(s: str) -> str:
+  """Replace the left-most number with a random edge-case string"""
+  return _sub_in_safe(
+    s, _NUM_RE,
+    lambda m: _quote_str(random.choice(sqlite_static_helper.EDGE_STRINGS)), 1)
+
+
+def mut_replace_num_with_random_hex(s: str) -> str:
+  """Replace the left-most number with a random hex blob"""
+  n = random.randint(0, 16)
+  val = "x'" + ''.join(random.choices('0123456789abcdef', k=n * 2)) + "'"
+  return _sub_in_safe(s, _NUM_RE, lambda m: val, 1)
+
+
+def mut_replace_num_with_null(s: str) -> str:
+  """Replace the left-most number with NULL"""
+  return _sub_in_safe(s, _NUM_RE, lambda m: 'NULL', 1)
+
+
+def mut_wrap_num_in_cast(s: str) -> str:
+  """Cast the left-most number to a different data type"""
+  target = random.choice(sqlite_static_helper.TYPES)
+  return _sub_in_safe(s, _NUM_RE, lambda m: f'CAST({m.group(0)} AS {target})',
+                      1)
+
+
+def _wrap_value(s: str, templates: list[str]) -> str:
+  tpl = random.choice(templates)
+  return _sub_in_safe(s, _NUM_RE, lambda m: tpl.replace('{v}', m.group(0)), 1)
+
+
+def mut_wrap_num_in_datetime_func(s: str) -> str:
+  """Wrap the left-most number in a date/time function"""
+  return _wrap_value(s, sqlite_static_helper.DATETIME_FUNCS)
+
+
+def mut_wrap_num_in_core_func(s: str) -> str:
+  """Wrap the left-most number in core function"""
+  return _wrap_value(s, sqlite_static_helper.CORE_FUNCS)
+
+
+def mut_wrap_num_in_math_func(s: str) -> str:
+  """Wrap the left-most number in a math function"""
+  return _wrap_value(s, sqlite_static_helper.MATH_FUNCS)
+
+
+def mut_wrap_num_in_json_func(s: str) -> str:
+  """Wrap the left-most number in a JSON function"""
+  return _wrap_value(s, sqlite_static_helper.JSON_VALUE_FUNCS)