def test_guess_language_name_untrained_model():
with tempfile.TemporaryDirectory() as model_dir:
guess = Guess(model_dir)
assert not guess.is_trained
with pytest.raises(GuesslangError):
guess.language_name(C_CODE)
def test_guess_train_with_default_model():
guess = Guess()
with tempfile.TemporaryDirectory() as source_files_dir:
_create_training_files(source_files_dir)
with pytest.raises(GuesslangError):
guess.train(source_files_dir, max_steps=10)
def test_guess_train_without_subdirectories():
with tempfile.TemporaryDirectory() as model_dir:
guess = Guess(model_dir)
with tempfile.TemporaryDirectory() as source_files_dir:
with pytest.raises(GuesslangError):
guess.train(source_files_dir, max_steps=10)
def test_guess_probabilities():
guess = Guess()
scores = guess.probabilities(PYTHON_CODE)
assert len(scores) == len(guess.supported_languages)
for language, probability in scores:
assert language in guess.supported_languages
assert 0 <= probability <= 1
top_language = scores[0][0]
assert top_language == 'Python'
def is_source_code(data):
if len(data) < SOURCE_CODE_MIN_LEN:
return False
try:
guess = Guess()
lang = guess.language_name(data)
for l in C_SIMILAR_LANG:
if lang == l:
return True
return False
except GuesslangError as e:
print(color.red(e))
def scan_plain(crawler, url, text):
'''CPUTASK
scan text/plain content to be a known language. try to evaluate if
known language is found check for more urls'''
LOGGER.info('Checking plain text content from url: %s' % url)
langnames = Guess().probable_languages(text)
LOGGER.info('Guessed langnames from url: %s %s' % (url, langnames))
if 'Javascript' in langnames:
try:
json_from_plain = json.loads(text)
except Exception: # we dont care too much
json_from_plain = None
LOGGER.error('JSON evaluation failed on : %s' % text)
if json_from_plain:
LOGGER.info('Got JSON from plain: %s' % json_from_plain)
if 'Python' in langnames:
try:
# python_fom_plain = eval(text)
python_fom_plain = builtin_ast.literal_eval(text)
except Exception: # we dont care too much
LOGGER.error('Python evaluation failed on : %s' % text)
python_fom_plain = None
if python_fom_plain:
LOGGER.info('Got Python from plain: %s' % python_fom_plain)
if isinstance(python_fom_plain, dict):
urls = list(nested_find('url', python_fom_plain))
LOGGER.info('Got urls from python dict: %s' % urls)
for link in urls:
crawler.check_link(url, link)
def detect_filetype():
global guess
if guess is None:
from guesslang import Guess
guess = Guess()
buffer = vim.current.buffer
text = "".join(buffer)
ft = str(guess.language_name(text))
if lang_map.__contains__(ft):
ft = lang_map[ft]
ft = ft.lower()
vim.command("set filetype=" + ft)
vim.command("echon ', set filetype=" + ft + "'")
def guess_language(code_value):
print("code here")
lang = Guess().language_name(code_value)
channel_layer = get_channel_layer()
async_to_sync(channel_layer.group_send)("snippet_snippet", {
"type": "code_language",
"message": lang
})
def __init__(self, bot):
self.bot = bot
self.guild = bot.get_guild(constants.tortoise_guild_id)
self.session = aiohttp.ClientSession()
self.banned_words = ConfigHandler("banned_words.json")
self.trusted = self.guild.get_role(constants.trusted_role_id)
self.log_channel = bot.get_channel(constants.bot_log_channel_id)
self.guess_language = Guess()
def detectionLang(listOfFiles):#prend la liste des fichiers er retourner dict contenant fichier:prog langage
guess=Guess()
b = {'0','1'}
filesLangages=dict()
for i in listOfFiles:
if os.stat(i).st_size != 0:
with open(i,'r',encoding='latin1',errors='ignore') as f:
if b!=set(f.read()) and set(f.read())!={'0'} and set(f.read())!={'1'}:
#filesLangages[i]=guess.language_name(f.read())
filesLangages[i]=subprocess.check_output(['guesslang',i]).decode("utf-8")
return(filesLangages)
def login_page():
try:
if request.method == "POST":
json = request.get_json(silent=True)
code = json['code']
name = Guess().language_name(code)
language = lang_dict[name]
return jsonify({'language': language})
except Exception as e:
return jsonify({'status': "Error", 'message': "Some error occurred!"}), 500
def test_guess_train():
with tempfile.TemporaryDirectory() as model_dir:
guess = Guess(model_dir)
with tempfile.TemporaryDirectory() as source_files_dir:
_create_training_files(source_files_dir)
guess.train(source_files_dir, max_steps=10)
assert guess.language_name(PYTHON_CODE) == 'Python'
assert guess.language_name(C_CODE) == 'C'
def __init__(self,
api_key: str,
lang: Mapping[str, str],
theme_image_ids: tuple[str],
keyboards: Mapping[str, InlineKeyboardMarkup],
guesslang_syntaxes: Mapping[str, str],
*args: Any,
admin_chat_id: Optional[str] = None,
db_path: str = str(
local.path(__file__).up() / 'user_themes.sqlite'),
**kwargs: Any):
self.lang = lang
self.theme_image_ids = theme_image_ids
self.kb = keyboards
self.guesslang_syntaxes = guesslang_syntaxes
self.admin_chat_id = admin_chat_id
self.db_path = db_path
self.user_themes = KeyValue(key_field=IntegerField(primary_key=True),
value_field=CharField(),
database=APSWDatabase(db_path))
self.log = mk_logger()
self.bot = TeleBot(api_key, *args, **kwargs)
self.register_handlers()
self.guesser = Guess()
def fxn():
import warnings
warnings.warn("deprecated", DeprecationWarning)
import warnings
warnings.filterwarnings('ignore')
from guesslang import Guess
z=0
with open('Test1.java') as e:
z=e.read()
e.close()
name = Guess().language_name(z)
print(name)
def test_guess_supported_languages():
guess = Guess()
assert len(guess.supported_languages) >= 30
assert 'Python' in guess.supported_languages
assert 'C' in guess.supported_languages
from guesslang import Guess
guess = Guess()
name = guess.language_name("""
RuleSet{
name={}
readOnly={0}
origin={}
global={0}
comment={}
objrenamed={0}
baseid={0}
incid={1}
featureLevel={16}
useAppRules={0}
id={}
transobj={7.1.1.1}
creator={}
localCascade={0}
allowRID={0}
allowAppRules={0}
prefixmatch={
}
rulesettype={}
loadsets={}
name={}
readOnly={0}
origin={}
global={0}
comment={}
netprefixobj={
import os
import uuid
from guesslang import Guess
import requests
from app.event_handlers.stylecheck.linters import *
FILE_FOLDER = "./temp_files"
SUPPORTED_LANGUAGES = ["C", "C++", "Python"]
SUPPORTED_FILE_TYPES = ["c", "cpp", "py"]
TYPE_TO_LANGUAGE_MAPPING = dict(zip(SUPPORTED_FILE_TYPES, SUPPORTED_LANGUAGES))
LANGUAGE_TO_TYPE_MAPPING = dict(zip(SUPPORTED_LANGUAGES, SUPPORTED_FILE_TYPES))
language_guess = Guess()
class CodeEntry:
def __init__(self, language, code, file_path):
self.language = language
self.code = code
self.file_path = file_path
def run_linting(self):
"""
:return: Linting Result
"""
if self.language in SUPPORTED_LANGUAGES[:2]:
return run_c_lint(self)
elif self.language == SUPPORTED_LANGUAGES[2]:
return run_py_lint(self)
def Analytics(request):
posts = Post.objects.all()
text = ''
total = Post.objects.all().count()
anshm = {}
com = Comments.objects.all()
answered = 0
for i in com:
if i.post.id not in anshm:
answered += 1
anshm[i.post.id] = 1
r = Rake()
for i in posts:
titletext = i.title
r.extract_keywords_from_text(titletext)
r.get_ranked_phrases()
li = r.get_ranked_phrases_with_scores()
for h in li:
if h[0] <= 3:
text = text + (h[-1])
text = text + (h[-1])
else:
for k in range(int(h[0])):
text += (h[-1])
guess = Guess()
lang = {}
for i in posts:
code = i.content
language = guess.language_name(code)
if language in lang:
lang[language] += 1
else:
lang[language] = 1
# print(text)
# print(lang)
# number_of_colors = 8
# color = ["#"+''.join([random.choice('0123456789ABCDEF') for j in range(6)])
# for i in range(number_of_colors)]
# print(color)
sort_langs = sorted(lang.items(), key=lambda x: x[1], reverse=True)
print(sort_langs)
langlabels = list()
langvals = list()
for i in sort_langs:
langlabels.append(str(i[0]))
langvals.append(int(i[1]))
langlabels = langlabels[0:5]
# langvals = langvals[0:5]
l1 = langlabels[0]
l2 = langlabels[1]
l3 = langlabels[2]
l4 = langlabels[3]
l5 = langlabels[4]
v1 = langvals[0]
v2 = langvals[1]
v3 = langvals[2]
v4 = langvals[3]
v5 = langvals[4]
print(langlabels)
print(langvals)
return render(
request, 'blog/analytics.html', {
'text': text,
'langvals': langvals,
'l1': l1,
'l2': l2,
'l3': l3,
'l4': l4,
'l5': l5,
'v1': v1,
'v2': v2,
'v3': v3,
'v4': v4,
'v5': v5,
'total': total,
'answered': answered
})
# def some_view(request):
# # return render_to_response('../../../VidyoConnector/js/VidyoConnector.html')
# # return render(request, 'blog/test/VidyoConnector.html')
from collections import Counter
def word_count(fname):
with open(fname) as f:
return Counter(f.read().split())
print("Frequency of words in the file: ", word_count("testfile.txt"))
from guesslang import Guess
name = Guess().language_name("testo.py")
print(name)
class ColorCodeBot:
def __init__(self,
api_key: str,
lang: Mapping[str, str],
theme_image_ids: tuple[str],
keyboards: Mapping[str, InlineKeyboardMarkup],
guesslang_syntaxes: Mapping[str, str],
*args: Any,
admin_chat_id: Optional[str] = None,
db_path: str = str(
local.path(__file__).up() / 'user_themes.sqlite'),
**kwargs: Any):
self.lang = lang
self.theme_image_ids = theme_image_ids
self.kb = keyboards
self.guesslang_syntaxes = guesslang_syntaxes
self.admin_chat_id = admin_chat_id
self.db_path = db_path
self.user_themes = KeyValue(key_field=IntegerField(primary_key=True),
value_field=CharField(),
database=APSWDatabase(db_path))
self.log = mk_logger()
self.bot = TeleBot(api_key, *args, **kwargs)
self.register_handlers()
self.guesser = Guess()
def register_handlers(self):
self.welcome = self.bot.message_handler(commands=['start', 'help'])(
self.welcome)
self.browse_themes = self.bot.message_handler(
commands=['theme', 'themes'])(self.browse_themes)
self.mk_theme_previews = self.bot.message_handler(
commands=['previews'])(self.mk_theme_previews)
self.intake_snippet = self.bot.message_handler(
func=lambda m: m.content_type == 'text')(self.intake_snippet)
self.recv_photo = self.bot.message_handler(content_types=['photo'])(
self.recv_photo)
self.restore_kb = self.bot.callback_query_handler(
lambda q: yload(q.data)['action'] == 'restore')(self.restore_kb)
self.set_snippet_filetype = self.bot.callback_query_handler(
lambda q: yload(q.data)['action'] == 'set ext')(
self.set_snippet_filetype)
self.set_theme = self.bot.callback_query_handler(
lambda q: yload(q.data)['action'] == 'set theme')(self.set_theme)
self.send_photo_elsewhere = self.bot.inline_handler(
lambda q: q.query.startswith("img "))(self.send_photo_elsewhere)
self.switch_from_inline = self.bot.inline_handler(lambda q: True)(
self.switch_from_inline)
@retry
def switch_from_inline(self, inline_query: InlineQuery):
self.log.msg("receiving inline query",
user_id=inline_query.from_user.id,
user_first_name=inline_query.from_user.first_name,
query=inline_query.query)
self.bot.answer_inline_query(
inline_query.id, [],
switch_pm_text=self.lang['switch to direct'],
switch_pm_parameter='x')
@retry
def welcome(self, message: Message):
self.log.msg("introducing myself",
user_id=message.from_user.id,
user_first_name=message.from_user.first_name,
chat_id=message.chat.id)
self.bot.reply_to(
message,
self.lang['welcome'],
reply_markup=ForceReply(
input_field_placeholder=self.lang['input field placeholder']))
@retry
def mk_theme_previews(self, message: Message):
if not self.admin_chat_id or str(
message.chat.id) != self.admin_chat_id:
self.log.msg("naughty preview attempt",
user_id=message.from_user.id,
user_first_name=message.from_user.first_name,
chat_id=message.chat.id,
admin_chat_id=self.admin_chat_id)
return
sample_code = dedent("""
# palinDay :: Int -> [ISO Date]
def palinDay(y):
'''A possibly empty list containing the palindromic
date for the given year, if such a date exists.
'''
s = str(y)
r = s[::-1]
iso = '-'.join([s, r[0:2], r[2:]])
try:
datetime.strptime(iso, '%Y-%m-%d')
return [iso]
except ValueError:
return []
""")
for button in chain.from_iterable(self.kb['theme'].keyboard):
theme = button.text
html = mk_html(f"# {theme}{sample_code}", 'py', theme)
with local.tempdir() as folder:
png_path = mk_png(html, folder=folder)
send_image(bot=self.bot,
chat_id=message.chat.id,
png_path=png_path,
reply_msg_id=message.message_id)
@retry
def browse_themes(self, message: Message):
self.log.msg("browsing themes",
user_id=message.from_user.id,
user_first_name=message.from_user.first_name,
chat_id=message.chat.id)
albums = [
self.theme_image_ids[i:i + 10]
for i in range(0, len(self.theme_image_ids), 10)
]
for album in albums:
self.bot.send_media_group(message.chat.id,
map(InputMediaPhoto, album),
reply_to_message_id=message.message_id)
self.bot.reply_to(message,
self.lang['select theme'],
reply_markup=self.kb['theme'])
@retry
def set_theme(self, cb_query: CallbackQuery):
data = yload(cb_query.data)
user = cb_query.message.reply_to_message.from_user
self.log.msg("setting theme",
user_id=user.id,
user_first_name=user.first_name,
theme=data['theme'],
chat_id=cb_query.message.chat.id)
self.bot.edit_message_reply_markup(cb_query.message.chat.id,
cb_query.message.message_id,
reply_markup=minikb('theme'))
self.user_themes[user.id] = data['theme']
self.bot.answer_callback_query(
cb_query.id,
text=self.lang['acknowledge theme'].format(data['theme']))
if self.admin_chat_id:
with open(self.db_path, 'rb') as doc:
self.bot.send_document(self.admin_chat_id, doc)
def guess_ext(self,
code: str,
probability_min: float = .12) -> Optional[str]:
syntax, probability = self.guesser.probabilities(code)[0]
ext = self.guesslang_syntaxes.get(syntax)
self.log.msg("guessed syntax",
probability_min=probability_min,
probability=probability,
syntax=syntax,
ext=ext)
if probability >= probability_min:
return ext
for start, ext in {
'{': 'json',
'---\n': 'yaml',
'[[': 'toml',
'[': 'ini',
'<?php': 'php',
'<': 'xml',
'-- ': 'lua'
}.items():
if code.startswith(start):
return ext
@retry
def intake_snippet(self, message: Message):
self.log.msg("receiving code",
user_id=message.from_user.id,
user_first_name=message.from_user.first_name,
chat_id=message.chat.id)
ext = self.guess_ext(message.text)
if ext:
kb_msg = self.bot.reply_to(
message,
f"{self.lang['query ext']}\n\n{self.lang['guessed syntax'].format(ext)}",
reply_markup=minikb('syntax', self.lang['syntax picker']),
parse_mode='Markdown',
disable_web_page_preview=True)
self.set_snippet_filetype(cb_query=None,
query_message=kb_msg,
ext=ext)
else:
self.bot.reply_to(message,
self.lang['query ext'],
reply_markup=self.kb['syntax'],
parse_mode='Markdown',
disable_web_page_preview=True)
@retry
def send_photo_elsewhere(self, inline_query: InlineQuery):
file_id = inline_query.query.split('img ', 1)[-1]
self.log.msg("creating inline query result",
file_id=file_id,
file_info=self.bot.get_file(file_id))
self.bot.answer_inline_query(inline_query.id, [
InlineQueryResultCachedPhoto(
id=str(uuid4()), photo_file_id=file_id, title="Send Image")
],
is_personal=True)
@retry
def restore_kb(self, cb_query: CallbackQuery):
data = yload(cb_query.data)
self.bot.edit_message_reply_markup(
cb_query.message.chat.id,
cb_query.message.message_id,
reply_markup=self.kb[data['kb_name']])
self.bot.answer_callback_query(cb_query.id)
@retry
def set_snippet_filetype(self,
cb_query: Optional[CallbackQuery] = None,
query_message: Optional[Message] = None,
ext: Optional[str] = None):
if cb_query:
query_message = cb_query.message
ext = yload(cb_query.data)['ext']
elif not (query_message and ext):
raise Exception(
"Either cb_query or both query_message and ext are required")
self.log.msg("colorizing code",
user_id=query_message.reply_to_message.from_user.id,
user_first_name=query_message.reply_to_message.from_user.
first_name,
syntax=ext,
chat_id=query_message.chat.id)
if cb_query:
self.bot.edit_message_reply_markup(query_message.chat.id,
query_message.message_id,
reply_markup=minikb(
'syntax',
self.lang['syntax picker']))
snippet = query_message.reply_to_message
theme = self.user_themes.get(snippet.from_user.id,
'base16/gruvbox-dark-hard')
html = mk_html(snippet.text, ext, theme)
send_html(bot=self.bot,
chat_id=snippet.chat.id,
html=html,
reply_msg_id=snippet.message_id)
with local.tempdir() as folder:
png_path = mk_png(html, folder=folder)
did_send = False
if len(snippet.text.splitlines()) <= 30:
try:
photo_msg = send_image(bot=self.bot,
chat_id=snippet.chat.id,
png_path=png_path,
reply_msg_id=snippet.message_id)
except ApiException as e:
self.log.error("failed to send compressed image",
exc_info=e,
chat_id=snippet.chat.id)
else:
did_send = True
kb_to_chat = InlineKeyboardMarkup()
kb_to_chat.add(
InlineKeyboardButton(
self.lang['send to chat'],
switch_inline_query=
f"img {photo_msg.photo[-1].file_id}"))
self.bot.edit_message_reply_markup(photo_msg.chat.id,
photo_msg.message_id,
reply_markup=kb_to_chat)
if not did_send:
send_image(bot=self.bot,
chat_id=snippet.chat.id,
png_path=png_path,
reply_msg_id=snippet.message_id,
compress=False)
if cb_query:
self.bot.answer_callback_query(cb_query.id)
def recv_photo(self, message: Message):
self.log.msg('received photo',
file_id=message.photo[0].file_id,
user_id=message.from_user.id,
user_first_name=message.from_user.first_name,
chat_id=message.chat.id)
def test_guess_language_name_plain_text():
guess = Guess()
assert guess.language_name(PLAIN_TEXT) is None
for folder in folders:
files = [f for f in listdir(folder) if isfile(join(folder, f))]
for f in files:
file = join(folder, f)
description = ""
not_recognized = False
with open(file, 'r') as file_reader:
data = file_reader.read()
data = '\n'.join([x for x in data.split("\n") if x.strip() != ''])
# Stage 1 - recognize code vs human readable file.
coding_langs = Guess().scores(data)
k, v = sorted(coding_langs.items(), key=lambda p: p[1])[-1]
if k == "Markdown" or v < 0.9:
not_recognized = True
else:
description += "Code in " + k + "."
if not_recognized:
# Stage 2 - Check weird interpunction.
punctuation_coding = count(
data,
string.punctuation.replace("!",
"").replace(".", "").replace(
"?", "").replace(",", ""))
if punctuation_coding / len(data) > 0.06:
def _real_main():
# Get the arguments
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument(
'-i',
'--input-file',
type=argparse.FileType('r'),
default=sys.stdin,
help="source code file. Default is standard input (stdin)")
parser.add_argument('-a',
'--all',
default=False,
action='store_true',
help="print all matching languages when guessing")
parser.add_argument(
'--learn',
metavar='LEARN_DIR',
help="learn from a directory containing source code files")
parser.add_argument(
'--test',
metavar='TEST_DIR',
help="test Guesslang model accuracy using source code files directory")
parser.add_argument('--model',
metavar='MODEL_DIR',
help="custom Guesslang learning model directory")
parser.add_argument('-d',
'--debug',
default=False,
action='store_true',
help="show debug messages")
args = parser.parse_args()
if args.learn:
if not args.model:
parser.error("Argument --model is required when using --learn")
if args.all:
parser.error("Argument --all cannot be used with --learn")
config_logging(debug=args.debug)
LOGGER.debug("Run with args: %s", vars(args))
# Create a language guesser
guess = Guess(args.model)
if args.learn: # Learn from source files
accuracy = guess.learn(args.learn)
LOGGER.info("Guessing learning accuracy is %.2f%%", 100 * accuracy)
if args.test: # Test Guesslang model accuracy
results = guess.test(args.test)
percent = 100 * results['overall-accuracy']
LOGGER.info("The overall accuracy of the test is %.2f%%", percent)
LOGGER.info("Test report saved into '%s'", _save_report(results))
if not args.learn and not args.test: # Guess language
content = _read_file(args.input_file)
if args.all:
language_info = " or ".join(guess.probable_languages(content))
else:
language_info = guess.language_name(content)
LOGGER.info("The source code is written in %s", language_info)
LOGGER.debug("Exit OK")
from flask import Flask, render_template, request, redirect, abort
from flask_limiter import Limiter
from flask_limiter.util import get_remote_address
from guesslang import Guess
from os import environ
from distutils.util import strtobool
from threading import Thread
pastey_version = "0.4.2"
loaded_config = {}
loaded_themes = []
app = Flask(__name__)
limiter = Limiter(app, key_func=get_remote_address)
guess = Guess()
from pastey import config, common, routes, functions
# Check environment variable overrides
config.data_directory = environ[
"PASTEY_DATA_DIRECTORY"] if "PASTEY_DATA_DIRECTORY" in environ else config.data_directory
config.listen_address = environ[
"PASTEY_LISTEN_ADDRESS"] if "PASTEY_LISTEN_ADDRESS" in environ else config.listen_address
config.listen_port = environ[
"PASTEY_LISTEN_PORT"] if "PASTEY_LISTEN_PORT" in environ else config.listen_port
config.use_whitelist = bool(
strtobool(environ["PASTEY_USE_WHITELIST"]
)) if "PASTEY_USE_WHITELIST" in environ else config.use_whitelist
config.restrict_pasting = bool(
strtobool(environ["PASTEY_RESTRICT_PASTING"])
) if "PASTEY_RESTRICT_PASTING" in environ else config.restrict_pasting
def test_guess_init_with_model_dir():
with tempfile.TemporaryDirectory() as model_dir:
guess = Guess(model_dir)
assert not guess.is_trained
def test_guess_init():
guess = Guess()
assert guess.is_trained
def detect_lang(request):
code = request.POST['code']
lang = Guess().language_name(code)
return HttpResponse(lang)
def test_guess_language_name():
guess = Guess()
assert guess.language_name(PYTHON_CODE) == 'Python'
assert guess.language_name(C_CODE) == 'C'
#!/usr/bin/env python3
import os
from guesslang import Guess
import warnings
warnings.filterwarnings("ignore")
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
g = Guess()
def guess_lang(text):
"""
text: input text
return: predicted progamming language
"""
lang = g.language_name(text)
return lang
if __name__ == "__main__":
sample_lang = """
% Quick sort
-module (recursion).
-export ([qsort/1]).
qsort([]) -> [];
qsort([Pivot|T]) ->
qsort([X || X <- T, X < Pivot])
def test_guess_language_name_empty_code():
guess = Guess()
assert guess.language_name('') is None
assert guess.language_name(' \t \n ') is None
