1
0
mirror of https://github.com/ytdl-org/youtube-dl synced 2024-12-16 14:36:47 +00:00

[jsinterp] Fix and improve split/join

* improve split/join edge cases
* correctly implement regex split (not like re.split)
This commit is contained in:
dirkf 2024-12-12 04:13:06 +00:00
parent 9993e8c5a9
commit 57b0e045d4
2 changed files with 60 additions and 5 deletions

View File

@ -483,6 +483,13 @@ class TestJSInterpreter(unittest.TestCase):
self._test(jsi, 't-e-s-t', args=[test_input, '-']) self._test(jsi, 't-e-s-t', args=[test_input, '-'])
self._test(jsi, '', args=[[], '-']) self._test(jsi, '', args=[[], '-'])
self._test('function f(){return '
'[1, 1.0, "abc", {a: 1}, null, undefined, Infinity, NaN].join()}',
'1,1,abc,[object Object],,,Infinity,NaN')
self._test('function f(){return '
'[1, 1.0, "abc", {a: 1}, null, undefined, Infinity, NaN].join("~")}',
'1~1~abc~[object Object]~~~Infinity~NaN')
def test_split(self): def test_split(self):
test_result = list('test') test_result = list('test')
tests = [ tests = [
@ -496,6 +503,16 @@ class TestJSInterpreter(unittest.TestCase):
self._test(jsi, test_result, args=['t-e-s-t', '-']) self._test(jsi, test_result, args=['t-e-s-t', '-'])
self._test(jsi, [''], args=['', '-']) self._test(jsi, [''], args=['', '-'])
self._test(jsi, [], args=['', '']) self._test(jsi, [], args=['', ''])
# RegExp split
self._test('function f(){return "test".split(/(?:)/)}',
['t', 'e', 's', 't'])
self._test('function f(){return "t-e-s-t".split(/[es-]+/)}',
['t', 't'])
# from MDN: surrogate pairs aren't handled: case 1 fails
# self._test('function f(){return "😄😄".split(/(?:)/)}',
# ['\ud83d', '\ude04', '\ud83d', '\ude04'])
self._test('function f(){return "😄😄".split(/(?:)/u)}',
['😄', '😄'])
def test_slice(self): def test_slice(self):
self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice()}', [0, 1, 2, 3, 4, 5, 6, 7, 8]) self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice()}', [0, 1, 2, 3, 4, 5, 6, 7, 8])

View File

@ -397,6 +397,9 @@ class JSInterpreter(object):
RE_FLAGS = { RE_FLAGS = {
# special knowledge: Python's re flags are bitmask values, current max 128 # special knowledge: Python's re flags are bitmask values, current max 128
# invent new bitmask values well above that for literal parsing # invent new bitmask values well above that for literal parsing
# JS 'u' flag is effectively always set (surrogate pairs aren't seen),
# but \u{...} and \p{...} escapes aren't handled); no additional JS 'v'
# features are supported
# TODO: execute matches with these flags (remaining: d, y) # TODO: execute matches with these flags (remaining: d, y)
'd': 1024, # Generate indices for substring matches 'd': 1024, # Generate indices for substring matches
'g': 2048, # Global search 'g': 2048, # Global search
@ -404,6 +407,7 @@ class JSInterpreter(object):
'm': re.M, # Multi-line search 'm': re.M, # Multi-line search
's': re.S, # Allows . to match newline characters 's': re.S, # Allows . to match newline characters
'u': re.U, # Treat a pattern as a sequence of unicode code points 'u': re.U, # Treat a pattern as a sequence of unicode code points
'v': re.U, # Like 'u' with extended character class and \p{} syntax
'y': 4096, # Perform a "sticky" search that matches starting at the current position in the target string 'y': 4096, # Perform a "sticky" search that matches starting at the current position in the target string
} }
@ -1047,13 +1051,47 @@ class JSInterpreter(object):
raise self.Exception('Unsupported Math method ' + member, expr=expr) raise self.Exception('Unsupported Math method ' + member, expr=expr)
if member == 'split': if member == 'split':
assertion(argvals, 'takes one or more arguments') assertion(len(argvals) <= 2, 'takes at most two arguments')
assertion(len(argvals) == 1, 'with limit argument is not implemented') if len(argvals) > 1:
return obj.split(argvals[0]) if argvals[0] else list(obj) limit = argvals[1]
assertion(isinstance(limit, int) and limit >= 0, 'integer limit >= 0')
if limit == 0:
return []
else:
limit = 0
if len(argvals) == 0:
argvals = [JS_Undefined]
elif isinstance(argvals[0], self.JS_RegExp):
# avoid re.split(), similar but not enough
def where():
for m in argvals[0].finditer(obj):
yield m.span(0)
yield (None, None)
def splits(limit=limit):
i = 0
for j, jj in where():
if j == jj == 0:
continue
if j is None and i >= len(obj):
break
yield obj[i:j]
if jj is None or limit == 1:
break
limit -= 1
i = jj
return list(splits())
return (
obj.split(argvals[0], limit - 1) if argvals[0] and argvals[0] != JS_Undefined
else list(obj)[:limit or None])
elif member == 'join': elif member == 'join':
assertion(isinstance(obj, list), 'must be applied on a list') assertion(isinstance(obj, list), 'must be applied on a list')
assertion(len(argvals) == 1, 'takes exactly one argument') assertion(len(argvals) <= 1, 'takes at most one argument')
return argvals[0].join(obj) return (',' if len(argvals) == 0 else argvals[0]).join(
('' if x in (None, JS_Undefined) else _js_toString(x))
for x in obj)
elif member == 'reverse': elif member == 'reverse':
assertion(not argvals, 'does not take any arguments') assertion(not argvals, 'does not take any arguments')
obj.reverse() obj.reverse()