baseline: post wave1+i2b+b2i, vfbuild13=377 errors

This commit is contained in:
loki 2026-10-03 13:53:24 +02:00
commit e11e749e4b
Signed by untrusted user who does not match committer: boba
GPG key ID: 253067914055423B
2248 changed files with 378506 additions and 0 deletions

670
tools/fixwaves.py Normal file
View file

@ -0,0 +1,670 @@
"""Error-driven фиксы по логу javac: волна boolean-кастов и private→package-private."""
import re, sys, os, collections
LOG = sys.argv[1]
WAVES = sys.argv[2] if len(sys.argv) > 2 else "ab"
HDR = re.compile(r"^(\S+\.java):(\d+): error: (.+)$")
def load_errors(path):
errs = []
lines = open(path, encoding="utf-8", errors="replace").read().splitlines()
i = 0
while i < len(lines):
m = HDR.match(lines[i])
if m:
col = None
srcline = None
if i + 2 < len(lines) and lines[i + 2].lstrip().startswith("^"):
srcline = lines[i + 1]
col = len(lines[i + 2]) - len(lines[i + 2].lstrip()) + 1
extra = []
j = i + 1
while j < len(lines) and j < i + 8 and lines[j].strip():
if (
lines[j]
.lstrip()
.startswith(("symbol:", "location:", "required:", "found:"))
):
extra.append(lines[j].strip())
j += 1
errs.append(
(
m.group(1),
int(m.group(2)),
col,
srcline,
m.group(3) + (" | " + " | ".join(extra) if extra else ""),
)
)
i += 3
else:
i += 1
return errs
OPS_STOP = set("+-*/%&|^<>!=?:,")
def scan_end(line, start):
"""Конец primary-выражения, начиная с start (0-based). None = не найден."""
depth = 0
i = start
n = len(line)
first = True
while i < n:
c = line[i]
if c in "([{":
depth += 1
elif c in ")]}":
if depth == 0:
break
depth -= 1
elif depth == 0 and not first and (c in OPS_STOP or c == ";"):
break
elif depth == 0 and c == ")":
break
first = False
i += 1
return i
def scan_multiline(lines, ln, start):
"""Возвращает (li, idx, ok): конец operand, начиная с lines[ln-1][start]."""
depth = 0
first = True
li = ln - 1
i = start
guard = 0
while li < len(lines) and guard < 400:
guard += 1
line = lines[li]
while i < len(line):
c = line[i]
if c in "([{":
depth += 1
elif c in ")]}":
if depth == 0:
return li, i, True
depth -= 1
elif depth == 0 and not first and (c in OPS_STOP or c == ";"):
return li, i, True
first = False
i += 1
li += 1
i = 0
return li, i, False
def wave_a(errs, stats, i2b=True, b2i=True):
per_file = collections.defaultdict(list)
for f, ln, col, srcline, msg in errs:
if i2b and "int cannot be converted to boolean" in msg and srcline and col:
per_file[f].append((ln, col, srcline, "i2b"))
elif b2i and "boolean cannot be converted to int" in msg and srcline and col:
per_file[f].append((ln, col, srcline, "b2i"))
for f, items in per_file.items():
lines = open(f, encoding="utf-8").read().splitlines(keepends=True)
by_line = collections.defaultdict(list)
for ln, col, srcline, kind in items:
by_line[ln].append((col, kind))
touched = False
for ln, cols in sorted(by_line.items(), reverse=True):
if ln - 1 >= len(lines):
continue
i2b_done = False
for col, kind in sorted(cols, reverse=True):
if kind != "i2b":
continue
if i2b_done:
stats["i2b_dup_line"] += 1
continue
pos = col - 1
# найти (boolean): на этой строке перед колонкой, иначе выше до 3 строк
cast_line, cast_pos = None, None
for back in range(0, 4):
bl = ln - 1 - back
if bl < 0:
break
hay = lines[bl]
limit = pos + 1 if back == 0 else len(hay)
cp = hay.rfind("(boolean)", 0, limit)
if cp != -1:
cast_line, cast_pos = bl, cp
break
if cast_line is None:
stats["i2b_no_cast"] += 1
continue
if cast_line == ln - 1:
op_start = cast_pos + len("(boolean)")
while (
op_start < len(lines[cast_line])
and lines[cast_line][op_start] == " "
):
op_start += 1
else:
# cast на предыдущей строке: operand начинается в начале строки ln
op_start = 0
cast_line = ln - 1
while (
op_start < len(lines[cast_line])
and lines[cast_line][op_start] == " "
):
op_start += 1
end_li, end_i, ok = scan_multiline(lines, ln, op_start)
if not ok:
stats["i2b_no_end"] += 1
continue
# собрать span
if end_li == ln - 1:
span = lines[ln - 1][op_start:end_i]
new_line = (
lines[ln - 1][:op_start]
+ "(("
+ span
+ ") != 0)"
+ lines[ln - 1][end_i:]
)
lines[ln - 1] = new_line
else:
span = (
lines[ln - 1][op_start:]
+ "".join(lines[ln:end_li])
+ lines[end_li][:end_i]
)
tail = lines[end_li][end_i:]
head = lines[ln - 1][:op_start]
merged = head + "((" + span + ") != 0)" + tail
if not merged.endswith("\n"):
merged += "\n"
del lines[ln - 1 : end_li + 1]
lines.insert(ln - 1, merged)
stats["i2b"] += 1
touched = True
i2b_done = True
if touched:
open(f, "w", encoding="utf-8").write("".join(lines))
# b2i оставлен на потом
if b2i:
stats["b2i_skipped"] = sum(
1 for x in errs if "boolean cannot be converted to int" in x[4]
)
def _scan_stmt_end(lines, li, start):
"""Конец statement: ';' на depth 0, начиная с lines[li][start]. (end_li, end_i, ok)."""
depth = 0
i = start
guard = 0
while li < len(lines) and guard < 500:
guard += 1
line = lines[li]
while i < len(line):
c = line[i]
if c in "([{":
depth += 1
elif c in ")]}":
if depth == 0 and c == "}":
return li, i, False
if depth > 0:
depth -= 1
elif c == ";" and depth == 0:
return li, i, True
i += 1
li += 1
i = 0
return li, i, False
def wave_b2i(errs, stats):
"""boolean-выражение в int-контексте: обернуть RHS в (EXPR) ? 1 : 0."""
per_file = collections.defaultdict(list)
for f, ln, col, srcline, msg in errs:
if "boolean cannot be converted to int" in msg and srcline and col:
per_file[f].append((ln, col))
for f, items in per_file.items():
lines = open(f, encoding="utf-8").read().splitlines(keepends=True)
for ln, col in sorted(set(items), reverse=True):
if ln - 1 >= len(lines):
continue
eq = None
for back in range(0, 8):
li = ln - 1 - back
if li < 0:
break
hay = lines[li]
limit = col if back == 0 else len(hay)
for m in re.finditer(r"(?<![=!<>+\-*/&|^])=(?!=)", hay[:limit]):
idx = m.start()
e_li, e_i, ok = _scan_stmt_end(lines, li, idx + 1)
if not ok:
continue
if (e_li, e_i) >= (ln - 1, col):
eq = (li, idx)
break
if eq:
break
if not eq:
stats["b2i_no_eq"] += 1
continue
li, idx = eq
rs = idx + 1
while rs < len(lines[li]) and lines[li][rs] in " \t":
rs += 1
end_li, end_i, ok = _scan_stmt_end(lines, li, rs)
if not ok:
stats["b2i_no_end"] += 1
continue
if end_li == li:
span = lines[li][rs:end_i]
else:
span = (
lines[li][rs:]
+ "".join(lines[li + 1 : end_li])
+ lines[end_li][:end_i]
)
if span.rstrip().endswith("? 1 : 0"):
stats["b2i_dup"] += 1
continue
if end_li == li:
lines[li] = (
lines[li][:rs] + "(" + span + ") ? 1 : 0" + lines[li][end_i:]
)
else:
head = lines[li][:rs]
tail = lines[end_li][end_i:]
merged = head + "(" + span + ") ? 1 : 0" + tail
if not merged.endswith("\n"):
merged += "\n"
del lines[li : end_li + 1]
lines.insert(li, merged)
stats["b2i"] += 1
open(f, "w", encoding="utf-8").write("".join(lines))
PRIV = re.compile(r"has private access in (\w+)$")
def wave_b(errs, stats):
members = collections.defaultdict(set)
for f, ln, col, srcline, msg in errs:
m = PRIV.search(msg)
if not m:
continue
owner = m.group(1)
sig = msg[: m.start()].strip()
name = sig.split("(")[0].strip()
# убрать квалификаторы вида SomeClass.member -> member
name = name.split(".")[-1]
if re.fullmatch(r"[A-Za-z_$][\w$]*", name):
members[owner].add(name)
for owner, names in members.items():
path = None
for cand in (
"/storage/aboba/evovis-client/src/main/java/defpackage/%s.java" % owner,
"/storage/aboba/evovis-client/src/main/java/a/%s.java" % owner,
"/storage/aboba/evovis-client/src/main/java/b/%s.java" % owner,
"/storage/aboba/evovis-client/src/main/java/evo/%s.java" % owner,
):
import os
if os.path.exists(cand):
path = cand
break
if not path:
stats["b_noowner"] += 1
continue
text = open(path, encoding="utf-8").read()
orig = text
for name in names:
rx = re.compile(
r"^(\s*)private\s+((?:static\s+|final\s+|volatile\s+|transient\s+|abstract\s+|native\s+|default\s+|synchronized\s+)*"
r"[\w.$<>\[\], ?]+\s+(?<![\w$])"
+ re.escape(name)
+ r"(?![\w$])\s*[;=(])",
re.M,
)
text, n = rx.subn(r"\1\2", text)
stats["b_priv"] += n
if text != orig:
open(path, "w", encoding="utf-8").write(text)
RECORD_RX = re.compile(
r"^(?P<mods>(?:public\s+|final\s+|abstract\s+)*)record\s+(?P<name>[A-Za-z_$][\w$]*)\s*\(",
re.M,
)
def _split_top(s, sep=","):
out, depth, cur = [], 0, ""
for ch in s:
if ch in "<([":
depth += 1
elif ch in ">)]":
depth -= 1
if ch == sep and depth == 0:
out.append(cur)
cur = ""
else:
cur += ch
if cur.strip():
out.append(cur)
return out
def wave_records(errs, stats):
owners = set()
for f, ln, col, srcline, msg in errs:
m = PRIV.search(msg.split(" | ")[0])
if m:
owners.add(m.group(1))
base = "/storage/aboba/evovis-client/src/main/java/"
for owner in sorted(owners):
path = None
for pkg in ("defpackage/", "a/", "b/", "evo/"):
cand = base + pkg + owner + ".java"
if os.path.exists(cand):
path = cand
break
if not path:
continue
text = open(path, encoding="utf-8").read()
head = text[:2000]
m = RECORD_RX.search(text)
if not m or m.group("name") != owner:
continue
# найти компоненты: от '(' до matching ')' с балансом
start = m.end() - 1
depth = 0
i = start
while i < len(text):
if text[i] == "(":
depth += 1
elif text[i] == ")":
depth -= 1
if depth == 0:
break
i += 1
comps_raw = text[start + 1 : i]
comps = []
for piece in _split_top(comps_raw):
piece = " ".join(piece.split())
if not piece:
continue
parts = piece.rsplit(None, 1)
if len(parts) != 2:
continue
typ, name = parts
typ = re.sub(r"\s*\bfinal\b\s*", " ", typ).strip()
comps.append((typ, name))
if not comps:
stats["rec_skip"] += 1
continue
# тело: от ')' до конца класса — сохранить как есть
rest = text[i + 1 :]
brace = rest.find("{")
if brace == -1:
stats["rec_skip"] += 1
continue
body = rest[brace + 1 :]
header = text[: m.start()]
mods = (m.group("mods") or "").replace("record", "").strip()
is_public = "public" in mods
lines_cls = []
vis = "public " if is_public else ""
lines_cls.append(f"{vis}final class {owner} {{")
for typ, name in comps:
lines_cls.append(f" {typ} {name};")
lines_cls.append("")
ctor_args = ", ".join(f"{typ} {name}" for typ, name in comps)
lines_cls.append(f" {vis}{owner}({ctor_args}) {{")
for typ, name in comps:
lines_cls.append(f" this.{name} = {name};")
lines_cls.append(" }")
lines_cls.append("")
for typ, name in comps:
lines_cls.append(f" public {typ} {name}() {{")
lines_cls.append(f" return this.{name};")
lines_cls.append(" }")
lines_cls.append("")
# equals
lines_cls.append(" @Override")
lines_cls.append(" public boolean equals(Object o) {")
lines_cls.append(f" if (!(o instanceof {owner} that)) {{")
lines_cls.append(" return false;")
lines_cls.append(" }")
for typ, name in comps:
if "[]" in typ:
if typ.endswith("[]") and typ[:-2] in (
"int",
"long",
"double",
"float",
"boolean",
"byte",
"short",
"char",
):
lines_cls.append(
f" if (!java.util.Arrays.equals(this.{name}, that.{name})) {{ return false; }}"
)
else:
lines_cls.append(
f" if (!java.util.Objects.deepEquals(this.{name}, that.{name})) {{ return false; }}"
)
elif typ in (
"int",
"long",
"double",
"float",
"boolean",
"byte",
"short",
"char",
):
if typ == "float":
lines_cls.append(
f" if (Float.compare(this.{name}, that.{name}) != 0) {{ return false; }}"
)
elif typ == "double":
lines_cls.append(
f" if (Double.compare(this.{name}, that.{name}) != 0) {{ return false; }}"
)
else:
lines_cls.append(
f" if (this.{name} != that.{name}) {{ return false; }}"
)
else:
lines_cls.append(
f" if (!java.util.Objects.equals(this.{name}, that.{name})) {{ return false; }}"
)
lines_cls.append(" return true;")
lines_cls.append(" }")
lines_cls.append("")
lines_cls.append(" @Override")
lines_cls.append(" public int hashCode() {")
args = ", ".join(
(
"java.util.Arrays.deepHashCode(this.%s)" % n
if "[]" in ty
else "this.%s" % n
)
for ty, n in comps
)
lines_cls.append(f" return java.util.Objects.hash({args});")
lines_cls.append(" }")
lines_cls.append("")
lines_cls.append(" @Override")
lines_cls.append(" public String toString() {")
sb = 'return "%s[" + ' % owner
parts = []
for ty, n in comps:
parts.append('"%s=" + this.%s' % (n, n))
lines_cls.append(" " + sb + ' + ", " + '.join(parts) + ' + "]";')
lines_cls.append(" }")
lines_cls.append(" ")
new_text = header + "\n".join(lines_cls) + body
open(path, "w", encoding="utf-8").write(new_text)
stats["records"] += 1
def wave_ctors(errs, stats):
pairs = set()
for f, ln, col, srcline, msg in errs:
m = PRIV.search(msg.split(" | ")[0])
if not m:
continue
owner = m.group(1)
sig = msg.split(" | ")[0][: m.start()].strip()
if sig.startswith(owner + "("):
pairs.add(owner)
base = "/storage/aboba/evovis-client/src/main/java/"
for owner in pairs:
path = None
for pkg in ("defpackage/", "a/", "b/", "evo/"):
cand = base + pkg + owner + ".java"
if os.path.exists(cand):
path = cand
break
if not path:
continue
text = open(path, encoding="utf-8").read()
new = re.sub(
r"(?m)^(\s*)private(\s+)" + re.escape(owner) + r"\s*\(",
r"\1\2" + owner + "(",
text,
)
if new != text:
open(path, "w", encoding="utf-8").write(new)
stats["ctors"] += 1
def wave_pkgqual(errs, stats):
base = "/storage/aboba/evovis-client/src/main/java/"
per_file = collections.defaultdict(list)
for f, ln, col, srcline, msg in errs:
if not f.startswith(base + "b/"):
continue
if "cannot find symbol" not in msg:
continue
parts = msg.split(" | ")
symbol = location = None
for p in parts:
if p.startswith("symbol:"):
symbol = p[len("symbol:") :].strip()
elif p.startswith("location:"):
location = p[len("location:") :].strip()
if not location:
continue
lm = re.match(r"(?:class|interface|enum|record)\s+([\w$]+)", location)
if not lm:
continue
loc = lm.group(1)
if symbol and symbol.startswith("method "):
member = symbol[len("method ") :].split("(")[0]
elif symbol and symbol.startswith("class "):
member = None
else:
member = None
per_file[f].append((ln, col, loc, member))
for f, items in per_file.items():
lines = open(f, encoding="utf-8").read().splitlines(keepends=True)
for ln, col, loc, member in sorted(set(items), reverse=True):
if ln - 1 >= len(lines):
continue
line = lines[ln - 1]
if member:
needle = loc + "." + member
idx = line.find(needle)
if idx == -1:
stats["qual_miss"] += 1
continue
lines[ln - 1] = (
line[:idx] + "defpackage." + needle + line[idx + len(needle) :]
)
stats["qual"] += 1
else:
# класс: заменить идентификатор на позиции колонки
pos = col - 1
m = re.match(r"[A-Za-z_$][\w$]*", line[pos:])
if not m or m.group() != loc:
# найти loc как целый идентификатор
m2 = re.search(r"(?<![\w.$])" + re.escape(loc) + r"(?![\w$])", line)
if not m2:
stats["qual_miss"] += 1
continue
idx = m2.start()
else:
idx = pos
lines[ln - 1] = (
line[:idx] + "defpackage." + loc + line[idx + len(loc) :]
)
stats["qual"] += 1
open(f, "w", encoding="utf-8").write("".join(lines))
def wave_vec(errs, stats):
per_file = collections.defaultdict(list)
for f, ln, col, srcline, msg in errs:
if "cannot find symbol" not in msg or not col:
continue
parts = msg.split(" | ")
symbol = location = None
for p in parts:
if p.startswith("symbol:"):
symbol = p[len("symbol:") :].strip()
elif p.startswith("location:"):
location = p[len("location:") :].strip()
if not symbol or not location or "Vector3fc" not in location:
continue
vm = re.match(r"variable\s+([\w$]+)", location)
sm = re.match(r"variable\s+([\w$]+)", symbol)
if not vm or not sm:
continue
per_file[f].append((ln, col, vm.group(1), sm.group(1)))
for f, items in per_file.items():
lines = open(f, encoding="utf-8").read().splitlines(keepends=True)
for ln, col, var, field in sorted(items, reverse=True):
if ln - 1 >= len(lines):
continue
line = lines[ln - 1]
needle = var + "." + field
idx = line.rfind(needle, 0, col + 10)
if idx == -1:
idx = line.find(needle)
if idx == -1:
stats["vec_miss"] += 1
continue
lines[ln - 1] = (
line[:idx] + var + "." + field + "()" + line[idx + len(needle) :]
)
stats["vec"] += 1
open(f, "w", encoding="utf-8").write("".join(lines))
def main():
errs = load_errors(LOG)
stats = collections.Counter()
if "a" in WAVES:
wave_a(errs, stats, i2b=("1" in WAVES or WAVES == "a"), b2i=False)
elif "1" in WAVES or "2" in WAVES:
wave_a(errs, stats, i2b="1" in WAVES, b2i=False)
if "2" in WAVES:
wave_b2i(errs, stats)
if "b" in WAVES:
wave_b(errs, stats)
if "c" in WAVES:
wave_records(errs, stats)
wave_ctors(errs, stats)
if "e" in WAVES:
wave_pkgqual(errs, stats)
if "f" in WAVES:
wave_vec(errs, stats)
print(dict(stats))
if __name__ == "__main__":
main()