def main( ):
# Program data
pvars = {
'outputDir' : None,
'clobber' : False,
'executeScripts' : False,
'tokenFile' : None,
'inputScripts' : [],
'tokSuffix' : '.tok',
'breakOnError' : False
}
tokenizedFileNames = []
# Read input
#try:
readCommandLine( sys.argv[1:], pvars )
tokens = readTokenFile( pvars['tokenFile'] )
# Apply tokens to scripts
for scriptName in pvars[ 'inputScripts' ]:
tokenizedFileName, tokenizedFile = makeTokenizedFile( scriptName, pvars['clobber'], \
pvars['outputDir'], pvars['tokSuffix'] )
tokenizedFileNames.append( tokenizedFileName )
tokenize( tokenizedFile, tokens, scriptName )
tokenizedFile.close( )
# Run tokenized files
if( pvars['executeScripts'] ):
for tokenizedFileName in tokenizedFileNames:
ret = os.system( tokenizedFileName )
if ret and pvars[ 'breakOnError' ]:
print >> sys.stderr, "Script " + tokenizedFileName + \
" failed with exit code " + str( ret ) +". Aborting."
sys.exit( 1 )
def errors(self, input_str, pos):
try:
Parse(tokenize(input_str)).go()
self.fail('ParseError not raised: {0}'.format(input_str))
except ParseError, e:
if e.position != pos:
self.fail('ParseError at wrong position: expected {0}, got {1}'.format(pos, e.position))
def tokenize_fn(c):
try:
tokens = list(tokenize(BytesIO(c.encode('utf-8')).readline))
except Exception as e:
# print(e)
return None
return tokens
def decistmt(s):
"""Substitute Decimals for floats in a string of statements.
>>> from decimal import Decimal
>>> s = 'print(+21.3e-5*-.1234/81.7)'
>>> decistmt(s)
"print (+Decimal ('21.3e-5')*-Decimal ('.1234')/Decimal ('81.7'))"
The format of the exponent is inherited from the platform C library.
Known cases are "e-007" (Windows) and "e-07" (not Windows). Since
we're only showing 12 digits, and the 13th isn't close to 5, the
rest of the output should be platform-independent.
>>> exec(s) #doctest: +ELLIPSIS
-3.217160342717258e-0...7
Output from calculations with Decimal should be identical across all
platforms.
>>> exec(decistmt(s))
-3.217160342717258261933904529E-7
"""
result = []
g = tokenize(BytesIO(
s.encode('utf-8')).readline) # tokenize the string
for toknum, tokval, _, _, _ in g:
if toknum == NUMBER and '.' in tokval: # replace NUMBER tokens
result.extend([(NAME, 'Decimal'), (OP, '('),
(STRING, repr(tokval)), (OP, ')')])
else:
result.append((toknum, tokval))
return untokenize(result).decode('utf-8')
def decistmt(s):
"""Substitute Decimals for floats in a string of statements.
>>> from decimal import Decimal
>>> s = 'print(+21.3e-5*-.1234/81.7)'
>>> decistmt(s)
"print (+Decimal ('21.3e-5')*-Decimal ('.1234')/Decimal ('81.7'))"
The format of the exponent is inherited from the platform C library.
Known cases are "e-007" (Windows) and "e-07" (not Windows). Since
we're only showing 12 digits, and the 13th isn't close to 5, the
rest of the output should be platform-independent.
>>> exec(s) #doctest: +ELLIPSIS
-3.217160342717258e-0...7
Output from calculations with Decimal should be identical across all
platforms.
>>> exec(decistmt(s))
-3.217160342717258261933904529E-7
"""
result = []
g = tokenize(BytesIO(s.encode('utf-8')).readline) # tokenize the string
for toknum, tokval, _, _, _ in g:
if toknum == NUMBER and '.' in tokval: # replace NUMBER tokens
result.extend([
(NAME, 'Decimal'),
(OP, '('),
(STRING, repr(tokval)),
(OP, ')')
])
else:
result.append((toknum, tokval))
return untokenize(result).decode('utf-8')
def batch_tokenize_process(source_list):
tmp_sentences = []
tmp_index = [ele[0] for ele in source_list]
for index, ele in enumerate(source_list):
sentence = ele[1]
#print sentence
tmp_sentences.append(tokenize(sentence).strip() + "\n")
return tmp_index, tmp_sentences
def parse(program):
## Log("HERE:D.1", program = program)
global token, next
token = None
next = None
next = tokenize(program).next
token = next()
return expression()
def parse(v):
r = tokenize(v)
p = ParserCtx(r)
x = do_prog(p)
# except:
if x == None:
print(" at line " + str( p.token.pos ) + " unknown error")
return x
def source_to_code(self, data, path, *, _optimize=-1):
print(path)
source = importlib._bootstrap.decode_source(data)
tokens = tokenize(io.BytesIO(source.encode('utf-8')).readline)
tokens = retokenize(tokens)
source = untokenize(tokens).decode('utf-8')
return _call_with_frames_removed(compile, source, path, 'exec',
dont_inherit=True,
optimize=_optimize)
def parse(program): global token, next next = tokenize(program).next token = next() return expression()
def visit(self, featureset):
try:
_result = []
for text in featureset.get_column_values(self._column):
if isinstance(text, list):
_preprocessed = []
for word in text:
_preprocessed.append(tokenize(word))
_result.append(_preprocessed)
else:
_preprocessed = tokenize(text)
_result.append(_preprocessed)
_new_result = np.asarray(list(_result))[:, np.newaxis]
_new_result = _new_result.reshape(
featureset.get_column_values(self._column).shape)
featureset.set_featureset_column(self._column, _new_result)
except Exception as error:
util.print_error("Unable to tokenize column")
util.print_error(error)
def parse(self, equa):
self.token_generator = tokenize(equa, self.TOKENS_SPEC)
self.current_token = None
self.next_token = None
self._next()
self._tab(self._prob())
if self.next_token:
raise Exception(
'Wrong token sequence busted. Processing stopped at : ' +
self.next_token.value)
def feature_extract(self, tweets): preproc_tweets = map(lambda t: tokenize(t), tweets) model = Word2Vec(preproc_tweets) word2vec = dict(zip(model.wv.index2word, model.wv.syn0)) dim = len(word2vec.itervalues().next()) mean_embeds = np.array([ np.mean([self.word2vec[w] for w in words if w in self.word2vec] or [np.zeros(self.dim)], axis=0) for words in tweets ])
def parse(content):
r = tokenize(content)
p = ParserCtx(r, content)
p.next()
try:
while p.token.type != 'eof':
parse_block(p)
x = p.tree
# except:
if x == None:
p.error()
return x
except Exception as e:
# print(e, v)
compile_error("parse", content, p.token, str(e))
raise (e)
def parse_file(fname):
customize_symbols(
['function', 'var', 'end', 'append'], '-=[];,./!%*()+{}:<>@^$&', [
'-', '+', '*', '**', '/', '%', '<<', '>>', '-=', '+=', '*=', '/=',
'=', '==', '!=', '<', '>', '<=', '>=', '[', ']', '{', '}', '(',
')', '.', ':', ',', ';', '&', '|', '!', '@', '^', '$'
])
disable_tk_indent()
list = tokenize(load(fname))
list = list_to_chain(list)
if len(list) == 0: return
item = list[0]
#while item != None:
#print(item.val)
#item = item.next
parse_macro(item)
def read_input(f1_name,f2_name):
#read line by line from file 1 and file 2
with open(f1_name) as f1, open(f2_name) as f2:
for line1, line2 in zip(f1,f2):
#get the classes the line belongs to
class1 = line1.strip()
priors[class1]+=1
#tokenize each line
token_list = tokenize(line2)
#store the tokens in a dictionary
#update freq of token
for token in token_list:
if token not in tfreq_dict:
tfreq_dict[token] = {'1':0,'-1':0,'0':0}
tfreq_dict[token][class1]+=1
def cleanAndNormalizeText(data):
tokens = tokenize(data)
tokens = [
token if emoticon_re.search(token) else token.lower()
for token in tokens
]
filterText = [w for w in tokens if w not in stop]
filterText = [w for w in filterText if not len(w) <= 1]
# stem
ps = PorterStemmer()
for i in range(len(filterText) - 1):
if len(filterText[i]) > 1:
try:
filterText[i] = ps.stem(filterText[i])
except Exception as e:
filterText[i] = filterText[i]
return filterText
def main():
file = get_source(argv[INPUT_INDEX])
code = file.read()
check_parens(code)
tokenized = tokenize(code)
func_map = make_function_map(iter(tokenized), tokenized)
if False:
for f in func_map:
for t in func_map[f].def_block:
print(t.symbol)
# check if the parse succeeds;
# is_parsed fails with sys.exit() and message if parse error
if is_parsed(func_map):
#compile(func_map)
pass
file.close()
def transform(fname):
s = load(fname)
tokenList = tokenize(s)
indents = 0
idx = 0
lastisnl = False
needspace = False
while hasnext(tokenList, idx):
i = tokenList[idx]
next = getnext(tokenList, idx)
idx += 1
if i.type in _ws_after:
printf("%s ", i.val)
elif i.type in _ws_both:
printf(" %s ", i.val)
elif i.type == 'nl':
printf('\n')
if next == None:
pass
elif next.type == 'indent':
printf(' ' * (indents + 4))
elif next.type == 'dedent':
#printf(' '*(indents - 4))
pass
else:
printf(' ' * indents)
elif i.type == 'indent':
indents += 4
elif i.type == 'dedent':
indents -= 4
if next == None:
pass
elif next.type != 'dedent':
printf(' ' * indents)
elif i.type == 'string':
printf(get_printable_str(i.val))
elif i.type == 'notin':
printf(' not in ')
else:
printf(i.val)
def transform(fname):
s = load(fname)
tokenList = tokenize(s)
indents = 0
idx = 0
lastisnl = False
needspace = False
while hasnext(tokenList, idx):
i = tokenList[idx]
next = getnext(tokenList, idx)
idx += 1
if i.type in _ws_after:
printf("%s ", i.val)
elif i.type in _ws_both:
printf(" %s ", i.val)
elif i.type == 'nl':
printf('\n')
if next == None:
pass
elif next.type == 'indent':
printf(' '* (indents + 4))
elif next.type == 'dedent':
#printf(' '*(indents - 4))
pass
else:
printf(' '* indents)
elif i.type == 'indent':
indents += 4
elif i.type == 'dedent':
indents -= 4
if next == None:
pass
elif next.type != 'dedent':
printf(' ' * indents)
elif i.type == 'string':
printf(get_printable_str(i.val))
elif i.type == 'notin':
printf(' not in ')
else:
printf(i.val)
def calc_sim(query, threshold=0):
'''
calculate similarity scores between documents and the query
'''
query = clean_token(query)
file_list = get_file_names()
documents = {}
for i in range(len(file_list)):
documents[file_list[i]] = tokenize(convert(file_list[i]))
query_vec = vectorize(query)
results = {}
for name, doc in documents.items():
doc_vec = vectorize(doc)
sim_score = cos_sim(query_vec, doc_vec)
if sim_score > threshold:
results[name] = sim_score
sort_result = sorted(results.items(),
key=operator.itemgetter(1),
reverse=True)
return sort_result
def test1():
test("1")
test("+1")
test("-1")
test("1+2")
test("1+2+3")
test("1+2*3")
test("(1+2)*3")
test("()")
test("(1)")
test("(1,)")
test("(1, 2)")
test("[1, 2, 3]")
test("{}")
test("{1: 'one', 2: 'two'}")
test("1.0*2+3")
test("'hello'+'world'")
test("2**3**4")
test("1 and 2")
test("foo.bar")
test("1 + hello")
test("1 if 2 else 3")
test("'hello'[0]")
test("hello()")
test("hello(1,2,3)")
test("lambda: 1")
test("lambda a, b, c: a+b+c")
test("True")
test("True or False")
test("1 in 2")
test("1 not in 2")
test("1 is 2")
test("1 is not 2")
test("1 is (not 2)")
print
print list(tokenize("1 not in 2"))
arg = arg.replace(bracks, '')
arg = arg.strip()
arg = re.sub(' +', ' ', arg)
t = ' '.join(arg.split(' ')[:-1] + [bracks])
n = arg.split(' ')[-1]
types.append(t)
names.append(n)
return types, names
if __name__ == '__main__':
# parser
parser = argparse.ArgumentParser()
parser.add_argument('--input_file', default='',
help='The file to strip comments from.')
parser.add_argument('--l', default='python',
choices=['python', 'java'], help='language of input code')
args = parser.parse_args()
assert args.input_file == '' or os.path.isfile(args.input_file)
# read from standard input, or from input file
if args.input_file == '':
source = sys.stdin.read()
else:
with io.open(args.input_file, encoding='utf-8') as f:
source = f.read()
tokenize = globals()[f"tokenize_{args.l}"]
# tokenize
print(tokenize(source), end='')
def parse(string):
tokenize(string)
def file_elements(filename, filtering='normal'):
'''Take a Python file, return a tuple of contents.
Argument 'filterint' determines how much filtering is applied to symbols
that may be uninteresting. Possible values are 'minimal' or 'normal'.
'''
header = ''
comments = []
tmp_file = None
full_path = os.path.join(os.getcwd(), filename)
def cleanup():
stream.close()
if tmp_file:
log.debug('closing {}'.format(tmp_file))
tmp_file.close()
# Set up the dictionary. We may end up returning only part of this
# filled out, if we encounter errors along the way.
elements = {}
elements['header'] = ''
elements['comments'] = []
elements['docstrings'] = []
elements['imports'] = []
elements['classes'] = []
elements['functions'] = []
elements['variables'] = []
elements['strings'] = []
elements['calls'] = []
elements['parse_result'] = 'success'
# Open the file for reading. FileIO is needed for the Python 'ast' module.
log = Logger('file_parser').get_log()
log.info('parsing Python file {}'.format(full_path))
stream = io.FileIO(filename)
# Pass #0: account for Python 2 vs 3 syntax.
# I haven't found another way to detect whether a script uses Python 2 or
# 3 syntax other than to try to parse it and test for failure. We need
# to use ast later below, and if an input file needs Python 2, we have to
# convert it first. So we test first and convert at the beginning.
if assumes_python2(stream):
try:
# This creates a temporary file that must be deleted later.
log.debug('attempting to convert from Python 2')
tmp_file = convert_python2_file(filename)
if tmp_file:
log.debug('conversion successful'.format(full_path))
log.debug('closing file {}'.format(full_path))
stream.close()
log.debug('opening file {}'.format(tmp_file.name))
stream = io.FileIO(tmp_file.name)
else:
# We thought it was Python 2 but couldn't convert it.
# Something is wrong. Bail.
log.warn(
'conversion failed -- giving up on {}'.format(full_path))
# At this point, we still have an empty elements dictionary.
elements['parse_result'] = 'error'
return elements
except Exception as err:
log.error(
'error trying to detect if {} uses Python 2'.format(full_path))
log.error(err)
elements['parse_result'] = 'error'
cleanup()
return elements
# Pass #1: use tokenize to find and store headers and comments.
log.debug('tokenizing {}'.format(full_path))
try:
tokens = tokenize(stream.readline)
except Exception as err:
log.error('error trying to tokenize {}'.format(full_path))
log.error(err)
elements['parse_result'] = 'error'
cleanup()
return elements
# Look for a header at the top, if any. There are two common forms in
# Python: a string, and a comment block. The heuristic used here is that
# if the first thing after any ignorable comments is a string, it's
# assumed to be the doc string; else, any initial comments (after certain
# special case comments, such as Unix hash-bang lines) are taken to be
# the header; else, no header.
for kind, thing, _, _, line in tokens:
if kind == ENCODING:
continue
if ignorable_comment(thing):
continue
if kind != COMMENT and kind != NL:
break
header += strip_comment_char(thing)
# When the above ends, 'thing' & 'kind' will be the next values to examine.
# If it's a string, it's assumed to be the file doc string.
# Once we do this, we'll have read the header comment or the doc string and
# the file position will be immediately after that point. When we do our
# 2nd pass, we don't want to read that stuff again. Back up over the last
# non-string/comment thing we read, and remember where we are.
if kind == STRING:
restart_point = stream.tell()
header = header + ' ' + thing.replace('"', '')
(kind, thing, _, _, line) = next(tokens)
else:
restart_point = stream.tell() - len(line)
# Iterate through the rest of the file, looking for comments.
# This gathers consecutive comment lines together, on the premise that
# they may contain sentences split across multiple comment lines.
chunk = ''
while thing != ENDMARKER:
try:
if kind == NL:
pass
elif kind == COMMENT and not ignorable_comment(thing):
chunk = chunk + strip_comment_char(thing) + '\n'
elif chunk:
comments.append(chunk.strip())
chunk = ''
(kind, thing, _, _, _) = next(tokens)
except StopIteration:
break
except Exception:
# Unicode decoding problems can cause exceptions.
log.error('tokenization failed for {}'.format(full_path))
break
# This concludes what we gather without parsing the file into an AST.
# Store the header and comments, if any.
elements['header'] = clean_plain_text(header)
elements['comments'] = clean_plain_text_list(comments)
# Pass #2: pull out remaining elements separately using the AST. This is
# inefficient, because we're iterating over the file a 2nd time, but our
# efforts right now are about getting things to work any way possible.
# AST parsing failures are possible here, particularly if the file was
# converted from Python 2. Some programs do stuff you can't automatically
# convert with 2to3. If that happens, bail and return what we can.
stream.seek(restart_point)
try:
log.debug('parsing into AST')
tree = ast.parse(stream.read())
except Exception as err:
log.error('AST parsing failed; returning what we have so far'.format(
full_path))
cleanup()
elements['parse_result'] = 'error'
return elements
# We were able to parse the file into an AST.
try:
collector = ElementCollector(filtering)
collector.visit(tree)
except Exception as err:
log.error('internal AST code walking error'.format(full_path))
cleanup()
elements['parse_result'] = 'error'
return elements
# We store the names of variables we find temporarily as paths separated
# by '|' so that we can find unique variable name assignments within each
# function or class context. E.g., variable x in function foo is "foo|x".
# Remove the paths now, leaving just the variable names.
# Also filter the variables to remove things we don't bother with.
unique_var_paths = list(set(collector.variables))
collector.variables = [x[x.rfind('|') + 1:] for x in unique_var_paths]
filtered_calls = filter_variables(collector.calls, collector.variables)
# We are done. Do final cleanup and count up frequencies of some things.
# Note that docstrings don't get frequencies associated with them.
elements['docstrings'] = clean_plain_text_list(collector.docstrings)
# The rest are turned into ('string', frequency) tuples.
elements['imports'] = countify(collector.imports)
elements['classes'] = countify(collector.classes)
elements['functions'] = countify(collector.functions)
elements['variables'] = countify(collector.variables)
elements['strings'] = countify(clean_plain_text_list(collector.strings))
elements['calls'] = countify(filtered_calls)
cleanup()
return elements
"multiline
"multi
" Func x does stuff
" to x
func xx()
foo()
end
a += 3 >= 4
loop i in 1:10
foo()
'''
'''
"a
"b'''
EXAMPLE = EXAMPLE1
print('py')
for token in tokenize_py(EXAMPLE):
print(repr(token))
print('zoof')
for token in tokenize(EXAMPLE, __file__, 286):
print(token)
import sys
from tokenize import *
from dictionary_words_2 import *
from stochastic_sampling import *
# [brian] Usually `import *` is bad form in python.
# One of the best things that distinguishes it from
# ruby is that it's always easy to tell where some
# behavior comes from. If you `import *` you'll
# later have a hard time figuring out which module
# a given function lives in.
if __name__ == '__main__':
source = open(sys.argv[1]).read()
tokens = tokenize(source)
a_dictionary = list_to_dictionary(tokens)
stochastic_list = new_list(a_dictionary)
root_node = construct_tree(stochastic_list)
first_word = random_word(root_node)
printf("%-10s%-10s:debug source code\n", "-debug", "[file]")
printf("%-10s%-10s:disassemble builtin-func\n", "-dis-bf", "[file]")
argc = len(ARGV)
if argc == 1:
if ARGV[0] == '-help':
print_usage()
else:
print_usage()
elif argc > 2:
opt = ARGV[1]
name = ARGV[2]
if opt == '-tk':
from tokenize import *
r = tokenize(load(name))
for i in r:
printf("%s := %s\n", i.type, i.val)
elif opt == '-src':
printSource(name)
elif opt == '-p':
_execute_file(name)
input("press any key to quit")
elif opt == '-dis':
from dis import dissimple
argv = ARGV.clone()
del argv[0]
dissimple(argv)
elif opt == '-dump':
compilefile(name, name + '.bin')
elif opt == '-ast':
def parse(program):
global curr, next_token
next_token = tokenize(program).next
curr = next_token()
return expression()
from tokenize import *
from getLexicon import *
from symScoreClassify import *
from splitData import *
from naiveBayesClassify import *
import numpy as np
posindir = os.path.abspath('') + '\\POS'
negindir = os.path.abspath('') + '\\NEG'
posDocs = tokenize(posindir)
negDocs = tokenize(negindir)
nfold = 10
posLexicon, negLexicon, posLexiconWeights, negLexiconWeights = getLexicon()
len(posLexicon), len(negLexicon), len(posLexiconWeights), len(
negLexiconWeights)
resultsBow = np.zeros((10, 8))
resultsSig2nonW = np.zeros((10, 198))
resultsSig2W = np.zeros((10, 198))
for iteration in range(0, nfold):
print iteration
trainPosDocs, trainNegDocs, testPosDocs, testNegDocs = splitData(
posDocs, negDocs, nfold, iteration)
resultsIteration = symScoreClassify(testPosDocs, posLexicon, negLexicon,
posLexiconWeights, negLexiconWeights,
True)
print resultsIteration[4:12]
resultsBow[iteration, :] += np.array(resultsIteration[4:12])
def tk_test(string, types, vals):
r = tokenize(string)
assert tk_types(r) == types
assert tk_vals(r) == vals
printf("%-10s%-10s:print abstract syntax tree\n", "-printast", "[file]")
printf("%-10s%-10s:debug source code\n", "-debug", "[file]")
printf("%-10s%-10s:disassemble builtin-func\n", "-dis-bf", "[file]")
argc = len(ARGV)
if argc == 1:
if ARGV[0] == '-help':
print_usage()
else:
print_usage()
elif argc > 2:
opt = ARGV[1]
name = ARGV[2]
if opt == '-tk':
from tokenize import *
r = tokenize(load(name))
for i in r:
printf("%s := %s\n", i.type, i.val)
elif opt == '-src':
printSource(name)
elif opt == '-p':
_execute_file(name)
input("press any key to quit")
elif opt == '-dis':
from dis import dissimple
argv = ARGV.clone()
del argv[0]
dissimple(argv)
elif opt == '-dump':
compilefile(name, name + '.bin')
elif opt == '-ast':
def tk_test(string, types, vals):
r = tokenize(string)
assert tk_types(r) == types
assert tk_vals(r) == vals
pattern = r"\s*(?:(<=|>=|\W)|([a-zA-Z]\w*)|(\d+(?:\.\d*)?))"
for operator, name, literal in re.findall(pattern, program):
if operator:
yield "(operator)", operator
elif name:
yield "(name)", name
elif literal:
yield "(literal)", literal
else:
raise SyntaxError
yield "(end)", "(end)"
import time
print len(program), "bytes"
print len(list(tokenize(program))), "tokens"
def bench(name, func):
t0 = time.clock()
for i in xrange(1000):
func(program)
print name, time.clock() - t0
import parser, compiler
program_list = list(tokenize_python(program))
bench("topdown", parse)
bench("topdown pretokenized", lambda program: parse(program_list))
tokenize_python = custom_tokenize_python
def same(self, input_str1, input_str2):
self.assertEqual(str(Parse(tokenize(input_str1)).go()),
str(Parse(tokenize(input_str2)).go()))
def matches(self, input_str, desired_function):
self.assertEqual(str(Parse(tokenize(input_str)).go()),
str(desired_function))
pattern = r"\s*(?:(<=|>=|\W)|([a-zA-Z]\w*)|(\d+(?:\.\d*)?))"
for operator, name, literal in re.findall(pattern, program):
if operator:
yield "(operator)", operator
elif name:
yield "(name)", name
elif literal:
yield "(literal)", literal
else:
raise SyntaxError
yield "(end)", "(end)"
import time
print len(program), "bytes"
print len(list(tokenize(program))), "tokens"
def bench(name, func):
t0 = time.clock()
for i in xrange(1000):
func(program)
print name, time.clock() - t0
import parser, compiler
program_list = list(tokenize_python(program))
bench("topdown", parse)
bench("topdown pretokenized", lambda program: parse(program_list))
tokenize_python = custom_tokenize_python
def parse(program):
global token, next
next = tokenize(program).next
token = next()
return expression()
def parse(string):
tokenize(string)
