Page MenuHomePhorge
Files F13480122

D249.1764989544.diff
No OneTemporary
Actions

Size
9 KB
Referenced Files
None
Subscribers
None

D249.1764989544.diff
View Options

diff --git a/lilybuild/lilybuild/ci_syntax/rules.py b/lilybuild/lilybuild/ci_syntax/rules.py
new file mode 100644
--- /dev/null
+++ b/lilybuild/lilybuild/ci_syntax/rules.py
@@ -0,0 +1,122 @@
+
+import re
+
+# XXX: Since we are stuck with Python 3.9, we can't disable backtrack yet.
+# https://docs.python.org/3/library/re.html
+patterns = {
+ 'OP': re.compile(r'\s*(==|\|\||&&|=~|\!=|\!~|\!)'),
+ 'PAREN_LEFT': re.compile(r'\s*\('),
+ 'PAREN_RIGHT': re.compile(r'\s*\)'),
+ 'STR_DOUBLE': re.compile(r'''\s*"((?:[^\\"]|\\.)*)"'''),
+ 'STR_SINGLE': re.compile(r"""\s*'((?:[^\\']|\\.)*)'"""),
+ 'VAR': re.compile(r'\s*\$([A-Za-z_][A-Za-z0-9_]*)'),
+ 'NULL': re.compile(r'\s*null'),
+ 'END': re.compile(r'\s*$'),
+}
+
+ops = {
+ '==': (2, lambda a, b: a == b),
+ '!=': (2, lambda a, b: a != b),
+ '=~': (2, lambda a, b: False),
+ '!~': (2, lambda a, b: False),
+ '!': (1, lambda a: not a),
+ '&&': (2, lambda a, b: a and b),
+ '||': (2, lambda a, b: a or b),
+}
+TERM_PRECEDENCE = 9
+PAREN_PRECEDENCE = 8
+
+def is_term(token):
+ return token[0] in ['STR_DOUBLE', 'STR_SINGLE', 'VAR', 'NULL']
+
+def get_precedence(token):
+ if is_term(token):
+ return TERM_PRECEDENCE
+ elif token[0] in ['PAREN_LEFT', 'PAREN_RIGHT']:
+ return PAREN_PRECEDENCE
+ elif token[0] == 'OP' and (token[1][0] in ['==', '!=', '=~', '!~']):
+ return 6
+ elif token[0] == 'OP' and (token[1][0] in ['!']):
+ return 5
+ elif token[0] == 'OP' and (token[1][0] in ['&&']):
+ return 4
+ elif token[0] == 'OP' and (token[1][0] in ['||']):
+ return 3
+ raise SyntaxError('Unknown token')
+
+def tokenize_rule(rule_str):
+ pos = 0
+ tokenized = []
+ while not (len(tokenized) and tokenized[-1][0] == 'END'):
+ match = None
+ for t in patterns:
+ regex = patterns[t]
+ match = regex.match(rule_str, pos)
+ if match:
+ tokenized.append((t, match.groups()))
+ pos = match.end()
+ break
+ if not match:
+ raise SyntaxError(f'Bad rule "{rule_str}", at pos {pos}')
+ return tokenized[:-1]
+
+def make_tree(tokenized):
+ # https://en.wikipedia.org/wiki/Shunting_yard_algorithm
+ stack = []
+ res = []
+ for t in tokenized:
+ if is_term(t):
+ res.append(t)
+ elif t[0] == 'OP':
+ if len(stack) and stack[-1][0] == 'OP' and get_precedence(stack[-1]) >= get_precedence(t):
+ res.append(stack.pop())
+ stack.append(t)
+ elif t[0] == 'PAREN_LEFT':
+ stack.append(t)
+ elif t[0] == 'PAREN_RIGHT':
+ while len(stack) and stack[-1][0] != 'PAREN_LEFT':
+ res.append(stack.pop())
+ if not len(stack):
+ raise SyntaxError('Mismatched parentheses')
+ stack.pop()
+ while len(stack):
+ t = stack.pop()
+ if t[0] == 'PAREN_LEFT':
+ raise SyntaxError('Mismatched parentheses')
+ res.append(t)
+
+ stack = []
+ for t in res:
+ if t[0] == 'OP':
+ opname = t[1][0]
+ arity = ops[opname][0]
+ if len(stack) < arity:
+ raise SyntaxError('Missing operands')
+ operands = tuple(reversed([stack.pop() for i in range(arity)]))
+ stack.append((t[0], (opname,) + operands))
+ else:
+ stack.append(t)
+ if len(stack) != 1:
+ raise SyntaxError('Too many operands')
+ return stack[0]
+
+def parse_rule(rule_str):
+ return make_tree(tokenize_rule(rule_str))
+
+backslash_re = re.compile(r'\\(.)')
+def replace_backslash(match):
+ c = match.groups()[0]
+ return c
+
+def evaluate_rule(expr, variables):
+ if expr[0] == 'VAR':
+ return variables.get(expr[1][0])
+ elif expr[0] == 'STR_DOUBLE' or expr[0] == 'STR_SINGLE':
+ return backslash_re.sub(replace_backslash, expr[1][0])
+ elif expr[0] == 'NULL':
+ return None
+ elif expr[0] == 'OP':
+ opname = expr[1][0]
+ operands = [evaluate_rule(o, variables) for o in expr[1][1:]]
+ return ops[opname][1](*operands)
+ raise SyntaxError(f'Cannot evaluate expression {expr}')
diff --git a/lilybuild/lilybuild/tests/ci_syntax/rules_test.py b/lilybuild/lilybuild/tests/ci_syntax/rules_test.py
new file mode 100644
--- /dev/null
+++ b/lilybuild/lilybuild/tests/ci_syntax/rules_test.py
@@ -0,0 +1,118 @@
+
+import unittest
+from lilybuild.ci_syntax.rules import tokenize_rule, make_tree, parse_rule, evaluate_rule
+
+class RulesTest(unittest.TestCase):
+ def test_tokenize(self):
+ self.assertEqual(tokenize_rule("$VAR_FOO == ''"), [
+ ('VAR', ('VAR_FOO',)),
+ ('OP', ('==',)),
+ ('STR_SINGLE', ('',)),
+ ])
+ self.assertEqual(tokenize_rule(r'$VAR_FOO == "xxx\""'), [
+ ('VAR', ('VAR_FOO',)),
+ ('OP', ('==',)),
+ ('STR_DOUBLE', ('xxx\\"',)),
+ ])
+ self.assertEqual(tokenize_rule(r'!($VAR_FOO == "xxx\"")'), [
+ ('OP', ('!',)),
+ ('PAREN_LEFT', ()),
+ ('VAR', ('VAR_FOO',)),
+ ('OP', ('==',)),
+ ('STR_DOUBLE', ('xxx\\"',)),
+ ('PAREN_RIGHT', ()),
+ ])
+ with self.assertRaises(SyntaxError) as m:
+ tokenize_rule(r'$VAR_FOO =/= "foo"')
+ with self.assertRaises(SyntaxError) as m:
+ tokenize_rule(r'$VAR_FOO > "foo"')
+ # bareword
+ with self.assertRaises(SyntaxError) as m:
+ tokenize_rule(r'$VAR_FOO == foo')
+ # unterminated strings
+ with self.assertRaises(SyntaxError) as m:
+ tokenize_rule(r'$VAR_FOO == "foo\"')
+ with self.assertRaises(SyntaxError) as m:
+ tokenize_rule(r'$VAR_FOO == "foo')
+ with self.assertRaises(SyntaxError) as m:
+ tokenize_rule(r'$VAR_FOO == "foo\\\"')
+
+ def test_make_tree(self):
+ self.assertEqual(make_tree(tokenize_rule(r'($FOO)')), ('VAR', ('FOO',)))
+ self.assertEqual(make_tree(tokenize_rule(r'($FOO=="foo(") || "xxx"')), (
+ 'OP', (
+ '||',
+ ('OP', ('==', ('VAR', ('FOO',)), ('STR_DOUBLE', ('foo(',)))),
+ ('STR_DOUBLE', ('xxx',)),
+ )
+ ))
+ self.assertEqual(make_tree(tokenize_rule(r'$FOO=="foo("|| "xxx"')), (
+ 'OP', (
+ '||',
+ ('OP', ('==', ('VAR', ('FOO',)), ('STR_DOUBLE', ('foo(',)))),
+ ('STR_DOUBLE', ('xxx',)),
+ )
+ ))
+ self.assertEqual(make_tree(tokenize_rule(r'$FOO==("foo("|| "xxx")')), (
+ 'OP', (
+ '==',
+ ('VAR', ('FOO',)),
+ ('OP', ('||', ('STR_DOUBLE', ('foo(',)), ('STR_DOUBLE', ('xxx',)))),
+ )
+ ))
+ self.assertEqual(make_tree(tokenize_rule(r'!$FOO==("foo("|| "xxx")')), (
+ 'OP', (
+ '!',
+ ('OP', (
+ '==',
+ ('VAR', ('FOO',)),
+ ('OP', ('||', ('STR_DOUBLE', ('foo(',)), ('STR_DOUBLE', ('xxx',)))),
+ ))
+ )
+ ))
+ with self.assertRaises(SyntaxError) as m:
+ make_tree(tokenize_rule(r'())'))
+ self.assertEqual(str(m.exception), 'Mismatched parentheses')
+ with self.assertRaises(SyntaxError) as m:
+ make_tree(tokenize_rule(r'('))
+ self.assertEqual(str(m.exception), 'Mismatched parentheses')
+ with self.assertRaises(SyntaxError) as m:
+ make_tree(tokenize_rule(r'("("))'))
+ self.assertEqual(str(m.exception), 'Mismatched parentheses')
+
+ with self.assertRaises(SyntaxError) as m:
+ make_tree(tokenize_rule(r'$FOO== != "0"'))
+ self.assertEqual(str(m.exception), 'Missing operands')
+
+ with self.assertRaises(SyntaxError) as m:
+ make_tree(tokenize_rule(r'( == "1" && "0" || "0")'))
+ self.assertEqual(str(m.exception), 'Missing operands')
+
+ with self.assertRaises(SyntaxError) as m:
+ make_tree(tokenize_rule(r'$VAR( $FOO == "0")'))
+ self.assertEqual(str(m.exception), 'Too many operands')
+
+ def test_eval(self):
+ self.assertEqual(evaluate_rule(parse_rule("$VAR_FOO == ''"), {}), False)
+ self.assertEqual(evaluate_rule(parse_rule("$VAR_FOO == null"), {}), True)
+ self.assertEqual(evaluate_rule(parse_rule("$VAR_FOO == 'xx'"), {}), False)
+ self.assertEqual(evaluate_rule(parse_rule("$VAR_FOO == 'xx'"), { 'VAR_FOO': 'xx' }), True)
+ self.assertEqual(
+ evaluate_rule(parse_rule("$VAR_FOO == 'xx' && $BAR"), {
+ 'VAR_FOO': 'xx',
+ 'BAR': 'yyy',
+ }), 'yyy')
+ self.assertEqual(
+ evaluate_rule(parse_rule("$VAR_FOO == 'xx' && !$BAR"), {
+ 'VAR_FOO': 'xx',
+ 'BAR': 'yyy',
+ }), False)
+ self.assertEqual(
+ evaluate_rule(parse_rule("$VAR_FOO == 'xx' && $Z"), {
+ 'VAR_FOO': 'xx',
+ 'BAR': 'yyy',
+ }), None)
+ self.assertEqual(
+ evaluate_rule(parse_rule(r"$VAR_FOO == 'x\a\\\'x'"), {
+ 'VAR_FOO': r"xa\'x",
+ }), True)

File Metadata

Mime Type
text/plain
Expires
Fri, Dec 5, 6:52 PM (23 h, 20 m)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
697336
Default Alt Text
D249.1764989544.diff (9 KB)

Event Timeline