Update part of the js interpreter
This commit is contained in:
parent
3a7010f478
commit
a5eb1d789c
1 changed files with 95 additions and 26 deletions
|
|
@ -187,7 +187,7 @@ class VideoPage(BaseYoutubePage):
|
||||||
self._player_cache = {}
|
self._player_cache = {}
|
||||||
|
|
||||||
def _extract_signature_function(self, video_id, player_url, slen):
|
def _extract_signature_function(self, video_id, player_url, slen):
|
||||||
id_m = re.match(r'.*-(?P<id>[a-zA-Z0-9_-]+)\.(?P<ext>[a-z]+)$',
|
id_m = re.match(r'.*-(?P<id>[a-zA-Z0-9_-]+)(?:/watch_as3|/html5player)?\.(?P<ext>[a-z]+)$',
|
||||||
player_url)
|
player_url)
|
||||||
player_type = id_m.group('ext')
|
player_type = id_m.group('ext')
|
||||||
player_id = id_m.group('id')
|
player_id = id_m.group('id')
|
||||||
|
|
@ -214,6 +214,8 @@ class VideoPage(BaseYoutubePage):
|
||||||
u'Initial JS player signature function name')
|
u'Initial JS player signature function name')
|
||||||
|
|
||||||
functions = {}
|
functions = {}
|
||||||
|
objects = {}
|
||||||
|
code = jscode
|
||||||
|
|
||||||
def argidx(varname):
|
def argidx(varname):
|
||||||
return string.lowercase.index(varname)
|
return string.lowercase.index(varname)
|
||||||
|
|
@ -245,12 +247,40 @@ class VideoPage(BaseYoutubePage):
|
||||||
assign = lambda v: v
|
assign = lambda v: v
|
||||||
expr = stmt[len(u'return '):]
|
expr = stmt[len(u'return '):]
|
||||||
else:
|
else:
|
||||||
raise BrokenPageError(
|
# Try interpreting it as an expression
|
||||||
u'Cannot determine left side of statement in %r' % stmt)
|
expr = stmt
|
||||||
|
assign = lambda v: v
|
||||||
|
|
||||||
v = interpret_expression(expr, local_vars, allow_recursion)
|
v = interpret_expression(expr, local_vars, allow_recursion)
|
||||||
return assign(v)
|
return assign(v)
|
||||||
|
|
||||||
|
def extract_object(objname):
|
||||||
|
obj = {}
|
||||||
|
obj_m = re.search(
|
||||||
|
(r'(?:var\s+)?%s\s*=\s*\{' % re.escape(objname)) +
|
||||||
|
r'\s*(?P<fields>([a-zA-Z$0-9]+\s*:\s*function\(.*?\)\s*\{.*?\})*)' +
|
||||||
|
r'\}\s*;',
|
||||||
|
code)
|
||||||
|
fields = obj_m.group('fields')
|
||||||
|
# Currently, it only supports function definitions
|
||||||
|
fields_m = re.finditer(
|
||||||
|
r'(?P<key>[a-zA-Z$0-9]+)\s*:\s*function'
|
||||||
|
r'\((?P<args>[a-z,]+)\){(?P<code>[^}]+)}',
|
||||||
|
fields)
|
||||||
|
for f in fields_m:
|
||||||
|
argnames = f.group('args').split(',')
|
||||||
|
obj[f.group('key')] = build_function(argnames, f.group('code'))
|
||||||
|
|
||||||
|
return obj
|
||||||
|
|
||||||
|
def build_function(argnames, code):
|
||||||
|
def resf(args):
|
||||||
|
local_vars = dict(zip(argnames, args))
|
||||||
|
for stmt in code.split(';'):
|
||||||
|
res = interpret_statement(stmt, local_vars)
|
||||||
|
return res
|
||||||
|
return resf
|
||||||
|
|
||||||
def interpret_expression(expr, local_vars, allow_recursion):
|
def interpret_expression(expr, local_vars, allow_recursion):
|
||||||
if expr.isdigit():
|
if expr.isdigit():
|
||||||
return int(expr)
|
return int(expr)
|
||||||
|
|
@ -258,48 +288,87 @@ class VideoPage(BaseYoutubePage):
|
||||||
if expr.isalpha():
|
if expr.isalpha():
|
||||||
return local_vars[expr]
|
return local_vars[expr]
|
||||||
|
|
||||||
m = re.match(r'^(?P<in>[a-z]+)\.(?P<member>.*)$', expr)
|
try:
|
||||||
|
return json.loads(expr)
|
||||||
|
except ValueError:
|
||||||
|
pass
|
||||||
|
|
||||||
|
m = re.match(r'^(?P<var>[a-zA-Z0-9_]+)\.(?P<member>[^(]+)(?:\(+(?P<args>[^()]*)\))?$', expr)
|
||||||
if m:
|
if m:
|
||||||
|
variable = m.group('var')
|
||||||
member = m.group('member')
|
member = m.group('member')
|
||||||
val = local_vars[m.group('in')]
|
arg_str = m.group('args')
|
||||||
if member == 'split("")':
|
|
||||||
return list(val)
|
if variable in local_vars:
|
||||||
if member == 'join("")':
|
obj = local_vars[variable]
|
||||||
return u''.join(val)
|
else:
|
||||||
if member == 'length':
|
if variable not in objects:
|
||||||
return len(val)
|
objects[variable] = extract_object(variable)
|
||||||
if member == 'reverse()':
|
obj = objects[variable]
|
||||||
return val[::-1]
|
|
||||||
slice_m = re.match(r'slice\((?P<idx>.*)\)', member)
|
if arg_str is None:
|
||||||
if slice_m:
|
# Member access
|
||||||
idx = interpret_expression(
|
if member == 'length':
|
||||||
slice_m.group('idx'), local_vars, allow_recursion-1)
|
return len(obj)
|
||||||
return val[idx:]
|
return obj[member]
|
||||||
|
|
||||||
|
assert expr.endswith(')')
|
||||||
|
# Function call
|
||||||
|
if arg_str == '':
|
||||||
|
argvals = tuple()
|
||||||
|
else:
|
||||||
|
argvals = tuple([
|
||||||
|
interpret_expression(v, local_vars, allow_recursion)
|
||||||
|
for v in arg_str.split(',')])
|
||||||
|
|
||||||
|
if member == 'split':
|
||||||
|
assert argvals == ('',)
|
||||||
|
return list(obj)
|
||||||
|
if member == 'join':
|
||||||
|
assert len(argvals) == 1
|
||||||
|
return argvals[0].join(obj)
|
||||||
|
if member == 'reverse':
|
||||||
|
assert len(argvals) == 0
|
||||||
|
obj.reverse()
|
||||||
|
return obj
|
||||||
|
if member == 'slice':
|
||||||
|
assert len(argvals) == 1
|
||||||
|
return obj[argvals[0]:]
|
||||||
|
if member == 'splice':
|
||||||
|
assert isinstance(obj, list)
|
||||||
|
index, howMany = argvals
|
||||||
|
res = []
|
||||||
|
for i in range(index, min(index + howMany, len(obj))):
|
||||||
|
res.append(obj.pop(index))
|
||||||
|
return res
|
||||||
|
|
||||||
|
return obj[member](argvals)
|
||||||
|
|
||||||
m = re.match(
|
m = re.match(
|
||||||
r'^(?P<in>[a-z]+)\[(?P<idx>.+)\]$', expr)
|
r'^(?P<in>[a-z]+)\[(?P<idx>.+)\]$', expr)
|
||||||
if m:
|
if m:
|
||||||
val = local_vars[m.group('in')]
|
val = local_vars[m.group('in')]
|
||||||
idx = interpret_expression(m.group('idx'), local_vars,
|
idx = interpret_expression(
|
||||||
allow_recursion-1)
|
m.group('idx'), local_vars, allow_recursion - 1)
|
||||||
return val[idx]
|
return val[idx]
|
||||||
|
|
||||||
m = re.match(r'^(?P<a>.+?)(?P<op>[%])(?P<b>.+?)$', expr)
|
m = re.match(r'^(?P<a>.+?)(?P<op>[%])(?P<b>.+?)$', expr)
|
||||||
if m:
|
if m:
|
||||||
a = interpret_expression(m.group('a'),
|
a = interpret_expression(
|
||||||
local_vars, allow_recursion)
|
m.group('a'), local_vars, allow_recursion)
|
||||||
b = interpret_expression(m.group('b'),
|
b = interpret_expression(
|
||||||
local_vars, allow_recursion)
|
m.group('b'), local_vars, allow_recursion)
|
||||||
return a % b
|
return a % b
|
||||||
|
|
||||||
m = re.match(
|
m = re.match(
|
||||||
r'^(?P<func>[a-zA-Z$]+)\((?P<args>[a-z0-9,]+)\)$', expr)
|
r'^(?P<func>[a-zA-Z$]+)\((?P<args>[a-z0-9,]+)\)$', expr)
|
||||||
if m:
|
if m:
|
||||||
fname = m.group('func')
|
fname = m.group('func')
|
||||||
|
argvals = tuple([
|
||||||
|
int(v) if v.isdigit() else local_vars[v]
|
||||||
|
for v in m.group('args').split(',')])
|
||||||
if fname not in functions:
|
if fname not in functions:
|
||||||
functions[fname] = extract_function(fname)
|
functions[fname] = extract_function(fname)
|
||||||
argvals = [int(v) if v.isdigit() else local_vars[v]
|
|
||||||
for v in m.group('args').split(',')]
|
|
||||||
return functions[fname](argvals)
|
return functions[fname](argvals)
|
||||||
raise BrokenPageError(u'Unsupported JS expression %r' % expr)
|
raise BrokenPageError(u'Unsupported JS expression %r' % expr)
|
||||||
|
|
||||||
|
|
|
||||||
Loading…
Add table
Add a link
Reference in a new issue