def train_model_ui():
base_model_name = data_files_prompt(
name='base_model_name',
message='Select a model to train.',
dir_='models',
allow_custom_file=True,
custom_file_choice_name='SpaCy built-in or new...',
custom_file_message=
"What is the name of the model you want to train? If it doesn't exists it will be created."
)
annotations_loaders = get_annotations_loaders()
a_training = prompt([
data_files_question(
name='training_data_filename',
message=
"Select the annotated training data file you want to train this model against.",
dir_='annotated_training_data',
allow_custom_file=False), {
'type':
'list',
'name':
'annotations_loader',
'message':
'What annotation loader do you want to use to load the file?',
'choices':
map(lambda loader: loader[0], annotations_loaders),
'filter':
lambda loader_name: [
loader[1] for loader in annotations_loaders
if loader[0] == loader_name
][0]
},
{
'type': 'input',
'name': 'output_model_name',
'message':
"Where do you want to save the trained model? If the model already exists it will be overridden.",
'default': base_model_name,
'validate': 'required'
}, {
'type': 'input',
'name': 'num_iter',
'message': "How many iteration should this training last?",
'default': '100',
'validate': lambda a: a.isdigit() and int(a) > 0,
'filter': 'to_int'
}
])
annotations_loaders = a_training['annotations_loader']
training_data_filename = path.join(get_data_dir(),
'annotated_training_data',
a_training['training_data_filename'])
output_model_name = path.join(get_data_dir(), 'models',
a_training['output_model_name'])
train_ner(model=base_model_name,
training_data=annotations_loaders(training_data_filename),
output_model=output_model_name,
num_iter=a_training['num_iter'])
def get_browser_history_ui(browser: str):
# This is gonna exit in case it's not running on the evidence machine
show_running_on_same_machine_warning()
a_dataset = prompt([
{
'type':
'input',
'name':
'dataset_filename',
'message':
f"Where should the {browser.capitalize()} history get saved to? If the file does not exist it "
"will be created, otherwise it will be overwritten.",
'validate':
'required',
},
yes_no_question(
'load_history',
'Do you want to also crawl the content of the visited web pages? This operation will take a while...',
yes_next=
[{
'type':
'input',
'name':
'dataset_content_filename',
'message':
f"Where should the {browser.capitalize()} history content get saved to? If the "
"file doesn't exist it will be created, otherwise it will be overwritten.",
'validate':
'required'
},
datetime_question(
'dataset_content_from',
'From what date and time would you like to retrieve the history content (format: ISO 8601, blank '
'for no lower filtering)?'),
datetime_question(
name='dataset_content_to',
message=
'To what date and time would you like to retrieve the history content (format: ISO 8601, '
'blank for no upper filtering)?')]),
])
dataset_filepath = path.join(get_data_dir(), 'datasets', 'history_data',
a_dataset['dataset_filename'])
history = getattr(browser_history, f'get_{browser}_history')()
history.to_csv(dataset_filepath, index=False)
LOGGER.info(f'Saved dataset to {dataset_filepath}')
# Load history content
if a_dataset['load_history']:
dataset_content_filepath = path.join(
get_data_dir(), 'datasets', 'history_data',
a_dataset['load_history']['next']['dataset_content_filename'])
history_content = load_history(
history, a_dataset['load_history']['next']['dataset_content_from'],
a_dataset['load_history']['next']['dataset_content_to'])
history_content.to_csv(dataset_content_filepath, index=False)
LOGGER.info(f'Saved dataset to {dataset_content_filepath}')
def _process_load_file(df_a, column_name_a, df_b, column_name_b, column_date):
while True:
a_set_a_set_b = prompt(q_set_a_set_b)
if a_set_a_set_b['option'] == 'exit':
break
if a_set_a_set_b['option'] == 'set_A':
a_append_overwrite = prompt(q_append_overwrite)
# a_file_loc_generic = prompt(q_file_loc_generic)
a_file_loc_generic = file_loc_generic()
filename_a = path.join(get_data_dir(), 'datasets',
a_file_loc_generic)
with Halo(text=f"Loading load file function...", spinner='dots'):
read = read_file(filename_a)
if a_append_overwrite['integrate'] == 'overwrite' or df_a.empty:
# Filter out empty titles (stored as nan aka flat values) and
# get only content and date column
df_a = read.dropna()[[column_name_a, column_date]]
df_a = spacy_token(df_a, column_name_a)
df_a = spacy_label_token_full(df_a, column_name_a)
else:
df_temporary = read.dropna()[[column_name_a, column_date
]] # impt to match columns
df_temporary = spacy_token(df_temporary, column_name_a)
df_temporary = spacy_label_token_full(
df_temporary, column_name_a)
df_a = venn.venn_union(df_a, df_temporary)
if a_set_a_set_b['option'] == 'set_B':
a_append_overwrite = prompt(q_append_overwrite)
# a_file_loc_generic = prompt(q_file_loc_generic)
a_file_loc_generic = file_loc_generic()
filename_b = path.join(get_data_dir(), 'datasets',
a_file_loc_generic)
with Halo(text=f"Loading load file function...", spinner='dots'):
read = read_file(filename_b)
if a_append_overwrite['integrate'] == 'overwrite' or df_b.empty:
# Filter out empty titles (stored as nan aka flat values) and
# get only content and date column
df_b = read.dropna()[[column_name_b, column_date]]
df_b = spacy_token(df_b, column_name_b)
df_b = spacy_label_token_full(df_b, column_name_b)
else:
df_temporary = read.dropna()[[column_name_a, column_date
]] # impt to match columns
df_temporary = spacy_token(df_temporary, column_name_b)
df_temporary = spacy_label_token_full(
df_temporary, column_name_b)
df_b = venn.venn_union(df_b, df_temporary)
return df_a, df_b
def get_twitter_timeline_ui():
a_dataset = prompt([{
'type':
'input',
'name':
'profiled_screen_name',
'message':
"What's the Twitter screen name of the person you're profiling (string after '@')?"
}, {
'type':
'input',
'name':
'dataset_filename',
'message':
"Where should the Twitter timeline get saved to? If the file doesn't exist it "
"will be created, otherwise it will be overwritten.",
'validate':
'required'
}])
dataset_filepath = path.join(get_data_dir(), 'datasets', 'social_data',
a_dataset['dataset_filename'])
twitter_timeline = get_twitter_timeline(a_dataset['profiled_screen_name'])
twitter_timeline.to_csv(dataset_filepath, index=False)
LOGGER.info(f'Saved dataset to {dataset_filepath}')
def get_facebook_data_ui():
print(
'Refer to https://www.facebook.com/help/212802592074644 on how to download data from a '
'Facebook account. You will need the login credentials of the person you are profiling. '
)
input('Press enter when you are ready to proceed...')
a_dataset = prompt([{
'type': 'input',
'name': 'facebook_data_dir',
'message': "Where is your Facebook data directory located?",
'validate': 'required'
}, {
'type':
'input',
'name':
'dataset_filename',
'message':
"Where should the Facebook data get saved to? If the file doesn't exist it "
"will be created, otherwise it will be overwritten.",
'validate':
'required'
}])
dataset_filepath = path.join(get_data_dir(), 'datasets', 'social_data',
a_dataset['dataset_filename'])
posts = get_facebook_posts(a_dataset['facebook_data_dir'])
posts.to_csv(dataset_filepath, index=False)
LOGGER.info(f'Saved dataset to {dataset_filepath}')
def data_files_question(*,
name: str,
message: str,
dir_: str,
allow_custom_file: bool = True,
custom_file_choice_name: str = 'Other...',
custom_file_message: str = 'Filename:',
recursive=False):
"""
Returns a question dictionary containing a list of files in a directory relative to the data folder.
:param custom_file_choice_name:
:param recursive:
:param custom_file_message:
:param allow_custom_file:
:param name:
:param message:
:param dir_:
:return:
"""
dir_path = path.join(get_data_dir(), dir_)
# Get all files inside directory, except the directory itself, if using recursive, then make it relative to that
# folder (remove first part)
files: Iterator[str] = map(
lambda p: path.relpath(p, dir_path),
glob(path.join(dir_path, '**'),
recursive=recursive)[1 if recursive else 0:])
choices: List[Dict[str, Any]] = list(map(lambda f: {'name': f}, files))
if allow_custom_file:
choices.append({
'name': custom_file_choice_name,
'next': {
'type': 'input',
'name': 'filename',
'message': custom_file_message,
'validate': 'required'
}
})
return {
'type': 'list',
'name': name,
'message': message,
'choices': choices
}
def save_training_data(training_data):
training_data_filename = prompt({
'type':
'input',
'name':
'training_data_filename',
'message':
"Where should the training data get saved to? If the file doesn't exist it will be created, "
"otherwise this training data will be appended.",
'validate':
'required'
})
with open(path.join(get_data_dir(), 'training_data',
training_data_filename),
'a',
encoding='utf-8') as f:
f.writelines(map(lambda l: f'{l}\n', training_data))
LOGGER.info(f'Successfully saved training data {training_data_filename}.')
def get_File_dir():
return get_data_dir()
def write_file_csv(df, filename):
output_datasets_name = path.join(get_data_dir(), 'results', filename)
df.to_csv(output_datasets_name, index=False)
def write_file_json(content, filename): # write to file.
output_datasets_name = path.join(get_data_dir(), 'results', filename)
with open(output_datasets_name, 'w+') as fp:
json.dump(content, fp)
