mirror of
https://github.com/norandom/OpenUltraSAST.git
synced 2026-11-14 02:08:18 +00:00
refactor(p1): one source-facts module owns function ranges, callers, declaration tables and language aliases
src/openultrasast/source.py replaces the copies in semantic/functions.py, plane/tasks/repo_facts.py, learn/features.py, learn/excerpt.py and plane/tasks/verify.py. Two engines share one sentinel (<module>), one DECLARATION/KEYWORDS/RESERVED set and one alias map: the parser engine (ranges, enclosing, bounds, call_sites, callers_of, facts_for) for the scorer, the overlay and the hunter's annotations; the line-pattern engine (declared_functions, enclosing_at, function_span, function_bounds, the brace matcher) that labels, features, roles, excerpts and plane facts were built on. learn no longer imports plane.tasks. Frozen records that store the old sentinel are mapped on read (module_function). Gate: labels rebuilt before and after are identical, 7,921 rows, after sentinel normalisation (benchmarks/measurements/2026-10-13-source-module- label-diff/record.json). 490 targeted tests pass; 3,312 collect. Net: 657 lines deleted, 212 added. benchmarks/unseen keeps <global> for its own frozen manifests. Co-Authored-By: Claude Fable 5.1 <[email protected]>
This commit is contained in:
Marius Ciepluch
co-authored by
Claude Fable 5.1
Claude Fable 5.1
parent
f63703a604
commit
062a7b780e
@@ -32,16 +32,17 @@ import re
|
||||
from collections.abc import Iterable, Sequence
|
||||
from dataclasses import dataclass
|
||||
|
||||
from ..plane.tasks.repo_facts import DECLARATION, GLOBAL, _declared_name
|
||||
from .features import function_span
|
||||
from ..source import ( # noqa: F401 (re-exported: labels, examples and harvest import them from here)
|
||||
HASH_COMMENTS,
|
||||
SLASH_COMMENTS,
|
||||
_comment_ranges,
|
||||
canonical_language,
|
||||
function_bounds,
|
||||
resolve_language,
|
||||
)
|
||||
from .labels import _ADVISORY, security_reason
|
||||
|
||||
REDACTED = "<redacted>"
|
||||
LANGUAGE_ALIASES = {"c_cpp": "cpp", "c++": "cpp", "js": "javascript", "ts": "typescript", "py": "python"}
|
||||
HASH_COMMENTS = frozenset({"python", "php", "ruby", "shell", "perl", "yaml", "r"})
|
||||
SLASH_COMMENTS = frozenset(
|
||||
{"c", "cpp", "java", "javascript", "typescript", "php", "go", "groovy", "csharp", "kotlin", "rust", "swift", "scala"}
|
||||
) # fmt: skip
|
||||
MAX_LINE_CHARS = 240
|
||||
_HEX = re.compile(r"\b[0-9a-f]{7,40}\b", re.IGNORECASE)
|
||||
|
||||
@@ -87,55 +88,6 @@ def excerpt_sha(text: str) -> str:
|
||||
# --- comments ---------------------------------------------------------------------------------------------------------
|
||||
|
||||
|
||||
def _comment_ranges(line: str, state: str | None, language: str) -> tuple[list[tuple[int, int]], str | None]:
|
||||
"""The comment spans of one line and the block state carried to the next (``block`` inside ``/* */``, a triple
|
||||
quote inside a Python docstring). Strings are skipped so a ``#`` in a literal is not a comment."""
|
||||
hashes, slashes, python = language in HASH_COMMENTS, language in SLASH_COMMENTS, language == "python"
|
||||
ranges: list[tuple[int, int]] = []
|
||||
i, n, quote = 0, len(line), None
|
||||
if state is not None:
|
||||
close = "*/" if state == "block" else state
|
||||
j = line.find(close)
|
||||
if j < 0:
|
||||
return [(0, n)], state
|
||||
ranges.append((0, j + len(close)))
|
||||
i, state = j + len(close), None
|
||||
while i < n:
|
||||
c = line[i]
|
||||
if quote:
|
||||
if c == "\\":
|
||||
i += 2
|
||||
continue
|
||||
if c == quote:
|
||||
quote = None
|
||||
i += 1
|
||||
continue
|
||||
if python and line.startswith(('"""', "'''"), i) and not line[:i].strip():
|
||||
delim = line[i : i + 3]
|
||||
j = line.find(delim, i + 3)
|
||||
if j < 0:
|
||||
ranges.append((i, n))
|
||||
return ranges, delim
|
||||
ranges.append((i, j + 3))
|
||||
i = j + 3
|
||||
continue
|
||||
if c in "\"'`":
|
||||
quote = c
|
||||
elif (slashes and line.startswith("//", i)) or (hashes and c == "#"):
|
||||
ranges.append((i, n))
|
||||
break
|
||||
elif slashes and line.startswith("/*", i):
|
||||
j = line.find("*/", i + 2)
|
||||
if j < 0:
|
||||
ranges.append((i, n))
|
||||
return ranges, "block"
|
||||
ranges.append((i, j + 2))
|
||||
i = j + 2
|
||||
continue
|
||||
i += 1
|
||||
return ranges, state
|
||||
|
||||
|
||||
_OPEN = ("/**", "/*", "//", "#", '"""', "'''", "*")
|
||||
_CLOSE = ("*/", '"""', "'''")
|
||||
|
||||
@@ -153,7 +105,7 @@ def _redact_comment(text: str) -> str:
|
||||
def redact(lines: Sequence[str], language: str, identities: Iterable[str] = ()) -> list[str]:
|
||||
"""``lines`` with security wording in comments and advisory ids anywhere replaced by ``<redacted>``, and every
|
||||
identity string (case-insensitive, at least 4 characters) replaced too."""
|
||||
language = LANGUAGE_ALIASES.get(language, language)
|
||||
language = canonical_language(language)
|
||||
names = sorted({i for i in identities if len(i) >= 4}, key=len, reverse=True)
|
||||
identity = re.compile("|".join(re.escape(n) for n in names), re.IGNORECASE) if names else None
|
||||
out: list[str] = []
|
||||
@@ -188,135 +140,6 @@ def identities_of(repo: str = "", path: str = "", commits: Iterable[str] = ()) -
|
||||
# --- spans and excerpts -----------------------------------------------------------------------------------------------
|
||||
|
||||
|
||||
_UNKNOWN_LANGUAGES = frozenset({"", "other", "unknown"})
|
||||
_PERL_LINE = re.compile(r"^\s*(?:sub\s+\w+\s*[{(;]|use\s+strict\s*;|my\s+[$@%]\w+)")
|
||||
_RUBY_DEF = re.compile(r"^\s*def\s+(?:self\.)?\w+[?!=]?(?:\s*\(|\s|$)")
|
||||
_RUBY_END = re.compile(r"^\s*end\s*$")
|
||||
_BODY_SCAN_LINES = 40 # how far past a declaration head the body's ``{`` (or a prototype's ``;``) may be
|
||||
_STATEMENT_HEAD = re.compile(r"\s*(return|if|while|for|switch|else|case|throw|await|yield)\b")
|
||||
_FUNCTION_EXPR = r"(?:async\s+)?(?:function\b\s*\*?\s*[\w$]*\s*)?(?=\(|[\w$]+\s*=>)" # function (...) / (...) => / a =>
|
||||
|
||||
|
||||
def resolve_language(lines: Sequence[str], language: str) -> str:
|
||||
"""The alias-normalised language; a file given as unknown (``other``, the feature vocabulary's bucket for Go, Ruby,
|
||||
C#, Perl, ...) is sniffed for Perl (``sub name {``, ``use strict;``) and Ruby (``def`` ... ``end``), whose
|
||||
declaration forms a brace matcher never finds and whose ``#`` comments must be redacted."""
|
||||
language = LANGUAGE_ALIASES.get(language, language)
|
||||
if language in _UNKNOWN_LANGUAGES and lines:
|
||||
if any(_PERL_LINE.match(line) for line in lines):
|
||||
return "perl"
|
||||
if any(_RUBY_DEF.match(line) for line in lines) and any(_RUBY_END.match(line) for line in lines):
|
||||
return "ruby"
|
||||
return language
|
||||
|
||||
|
||||
def _blank_comments(lines: Sequence[str], language: str) -> list[str]:
|
||||
style = language if language in SLASH_COMMENTS or language in HASH_COMMENTS else "c"
|
||||
code: list[str] = []
|
||||
state: str | None = None
|
||||
for line in lines:
|
||||
ranges, state = _comment_ranges(line, state, style)
|
||||
for start, end in reversed(ranges):
|
||||
line = line[:start] + " " * (end - start) + line[end:]
|
||||
code.append(line)
|
||||
return code
|
||||
|
||||
|
||||
def _declaration_heads(name: str) -> re.Pattern[str]:
|
||||
"""Where a brace language may declare ``name`` (the body follows): ``name(`` (a definition -- or a call, told
|
||||
apart by :func:`_body_start`); Perl ``sub name {``; an object-literal method ``name: function (`` /
|
||||
``name: (a) =>``; an assignment ``x.y.name = function (`` / ``= (a) =>`` / ``const name = m.name = async a =>``
|
||||
(any chain). ``$`` is an identifier character (``Glance$serveRequest``)."""
|
||||
quoted = re.escape(name)
|
||||
return re.compile(
|
||||
rf"(?<![\w.>$]){quoted}\s*(?:<[^<>(){{}};]*>\s*)?(?=\()"
|
||||
rf"|^\s*sub\s+{quoted}\s*(?=[{{(;])"
|
||||
rf"|(?<![\w$]){quoted}\s*:\s*{_FUNCTION_EXPR}"
|
||||
rf"|(?<![\w$]){quoted}\s*=\s*(?:[\w$.\[\]'\"]+\s*=\s*)*{_FUNCTION_EXPR}"
|
||||
)
|
||||
|
||||
|
||||
def _body_start(code: Sequence[str], line: int, column: int) -> tuple[str, int, int] | None:
|
||||
"""Scan from a declaration head for its body: ``("brace", line, column)`` of the ``{`` at parenthesis depth 0,
|
||||
``("expression", line, column)`` of the ``;`` or ``,`` ending an arrow function's expression body, ``None`` for a
|
||||
prototype (``;`` before any body), a call (a ``)`` closing an outer parenthesis, a ``}``) or no body within
|
||||
:data:`_BODY_SCAN_LINES`. Parentheses are tracked so ``f(a, function () {})`` is not a definition of ``f``."""
|
||||
depth, arrow = 0, False
|
||||
for j in range(line, min(line + _BODY_SCAN_LINES, len(code))):
|
||||
text = code[j]
|
||||
k = column if j == line else 0
|
||||
while k < len(text):
|
||||
c = text[k]
|
||||
if c == "(":
|
||||
depth += 1
|
||||
elif c == ")":
|
||||
depth -= 1
|
||||
if depth < 0:
|
||||
return None
|
||||
elif depth == 0:
|
||||
if c == "{":
|
||||
return ("brace", j, k)
|
||||
if arrow and c in ";,":
|
||||
return ("expression", j, k)
|
||||
if c in "};": # a statement ends (a call, a prototype) before any body opened
|
||||
return None
|
||||
if text.startswith("=>", k):
|
||||
arrow = True
|
||||
k += 1
|
||||
k += 1
|
||||
return None
|
||||
|
||||
|
||||
def _brace_bounds(lines: Sequence[str], language: str, function: str) -> tuple[int, int] | None:
|
||||
"""A brace language's definition of ``function`` (``Class::name`` and ``a.b`` reduced to the last name): the first
|
||||
declaration head (:func:`_declaration_heads`) whose body :func:`_body_start` finds, to the matching ``}`` (or the
|
||||
end of an expression body). Comments are blanked before anything is matched or counted."""
|
||||
name = re.split(r"::|\.|->", function)[-1]
|
||||
if not re.fullmatch(r"[A-Za-z_$][\w$]*", name):
|
||||
return None
|
||||
heads = _declaration_heads(name)
|
||||
code = _blank_comments(lines, language)
|
||||
for i, line in enumerate(code):
|
||||
if _STATEMENT_HEAD.match(line):
|
||||
continue
|
||||
for found in heads.finditer(line):
|
||||
body = _body_start(code, i, found.end())
|
||||
if body is None:
|
||||
continue
|
||||
kind, j, k = body
|
||||
if kind == "expression":
|
||||
return i, j + 1
|
||||
depth = 0
|
||||
for row in range(j, len(code)):
|
||||
for c in code[row][k:] if row == j else code[row]:
|
||||
if c == "{":
|
||||
depth += 1
|
||||
elif c == "}":
|
||||
depth -= 1
|
||||
if depth == 0:
|
||||
return i, row + 1
|
||||
return i, len(code)
|
||||
return None
|
||||
|
||||
|
||||
def function_bounds(lines: Sequence[str], language: str, function: str) -> tuple[int, int] | None:
|
||||
"""0-based ``[start, end)`` of the first declaration of ``function`` (the whole file for ``<global>``), or
|
||||
``None`` when the file does not declare it. Languages without a repo-facts declaration pattern (C, C++, C#, Perl,
|
||||
...) and brace languages whose pattern misses the name use the brace matcher; an unknown language is sniffed
|
||||
(:func:`resolve_language`)."""
|
||||
language = resolve_language(lines, language)
|
||||
if function == GLOBAL:
|
||||
return (0, len(lines)) if lines else None
|
||||
pattern = DECLARATION.get(language)
|
||||
start = next((i for i, line in enumerate(lines) if _declared_name(pattern, line) == function), None) if pattern else None
|
||||
if start is None:
|
||||
return _brace_bounds(lines, language, function) if language in SLASH_COMMENTS or pattern is None else None
|
||||
span = function_span(lines, language, function)
|
||||
if not span:
|
||||
return None
|
||||
return start, min(start + span, len(lines))
|
||||
|
||||
|
||||
def _window(start: int, end: int, focus: Sequence[int], size: int) -> tuple[int, int]:
|
||||
if end - start <= size:
|
||||
return start, end
|
||||
|
||||
Reference in New Issue
Block a user