def parse(sentences, weights):
results = []
edge_feats_list, featdict = c.extract(sentences)
size = len(sentences)
count = 0
correct = 0
total = 0
for sentence, edge_feats in zip(sentences, edge_feats_list):
count = count + 1
Log.i("Parsing sentence: %d of %d" % (count, size))
result = []
heads = eisner.parse(sentence, weights, edge_feats)
for index, word in enumerate(sentence):
if index == 0:
continue
token = [
str(word.id),
word.form,
word.lemma,
word.cpostag,
word.postag,
word.feats,
str(heads[index]),
word.deprel
]
line = "\t".join(token)
if heads[index] == word.head:
correct = correct + 1
total = total + 1
result.append(line)
results.append(result)
return results, float(correct / total)
def downloadQtPackge():
qt_version_dotless = qt_version.replace('.', '')
base_url = 'https://download.qt.io/online/qtsdkrepository/{}/{}/qt5_{}' \
.format(os_url, target_platform ,qt_version_dotless)
updates_file = 'Updates-{}-{}.xml'.format(qt_version, os_name)
c.download(base_url + '/Updates.xml', updates_file)
updates = ET.parse(updates_file)
updates_root = updates.getroot()
all_modules = {}
for i in updates_root.iter('PackageUpdate'):
name = i.find('Name').text
if 'debug' in name or not kit_arch in name:
continue
archives = i.find('DownloadableArchives')
if archives.text is None:
continue
archives_parts = archives.text.split(',')
version = i.find('Version').text
for archive in archives_parts:
archive = archive.strip()
parts = archive.split('-')
module_name = parts[0]
all_modules[module_name] = {
'package': name,
'file': version + archive
}
if len(sys.argv) > 1: # handle subcommand
if sys.argv[1] == 'list':
c.print('Available modules:')
for k in iter(sorted(all_modules.keys())):
c.print(k, '---', all_modules[k]['file'])
exit(0)
for module in qt_modules_list:
if module not in all_modules:
c.print('>> Required module {} not available'.format(module))
continue
file_name = all_modules[module]['file']
package = all_modules[module]['package']
c.download(base_url + '/' + package + '/' + file_name, file_name)
c.extract(file_name, '.')
def download_mingw_tool():
qt_version_dotless = qt_version.replace('.', '')
base_url = 'https://download.qt.io/online/qtsdkrepository/windows_x86/desktop/tools_mingw/'
updates_file = 'Updates-{}-{}-{}.xml'.format(qt_version, os_name, 'qttool')
c.download(base_url + '/Updates.xml', updates_file)
updates = ET.parse(updates_file)
updates_root = updates.getroot()
all_modules = {}
for i in updates_root.iter('PackageUpdate'):
name = i.find('Name').text
if 'debug' in name or not kit_arch in name:
continue
archives = i.find('DownloadableArchives')
if archives.text is None:
continue
c.print(' archives: {}'.format(archives))
archives_parts = archives.text.split(',')
version = i.find('Version').text
c.print(' version: {}'.format(version))
for archive in archives_parts:
archive = archive.strip()
parts = archive.split('-')
module_name = parts[0]
all_modules[module_name] = {
'package': name,
'file': version + archive
}
if len(sys.argv) > 1: # handle subcommand
if sys.argv[1] == 'list':
c.print('Available modules:')
for k in iter(sorted(all_modules.keys())):
c.print(k, '---', all_modules[k]['file'])
exit(0)
file_name = all_modules[module_name]['file']
package = all_modules[module_name]['package']
c.print('download url: {}'.format(base_url + '/' + package + '/' +
file_name))
c.download(base_url + '/' + package + '/' + file_name, file_name)
c.extract(file_name, '.')
def train(sentences, iteration=10):
edge_feats_list, featdict = c.extract(sentences)
weights = np.zeros(len(featdict))
size = len(sentences)
for i in range(1, iteration + 1):
Log.i("Training iteration: %d of %d" % (i, iteration))
count = 0
for sentence, edge_feats in zip(sentences, edge_feats_list):
count = count + 1
Log.i("\tTraining sentence: %d of %d" % (count, size))
predicts = eisner.parse(sentence, weights, edge_feats)
weights = update(weights, predicts, sentence, edge_feats)
"""
print(predicts)
for feat, index in featdict.items():
print(index, feat, weights[index])
"""
return featdict, weights
includes_path = install_dir + '/include/tesseract'
if len(c.get_folder_files(includes_path)) == 0:
return False
return True
if check_existing() and not 'FORCE' in os.environ:
c.print('>> Using cached')
exit(0)
archive = 'tesseract-' + os.path.basename(url)
c.download(url, archive)
src_dir = os.path.abspath('tesseract_src')
c.extract(archive, '.')
c.symlink(c.get_archive_top_dir(archive), src_dir)
c.ensure_got_path(install_dir)
c.recreate_dir(build_dir)
os.chdir(build_dir)
cmake_args = '"{0}" -DCMAKE_INSTALL_PREFIX="{1}" -DLeptonica_DIR="{1}/cmake" \
-DBUILD_TRAINING_TOOLS=OFF -DBUILD_TESTS=OFF'.format(src_dir, install_dir)
if platform.system() == "Windows":
env_cmd = c.get_msvc_env_cmd(bitness=bitness, msvc_version=msvc_version)
c.apply_cmd_env(env_cmd)
cmake_args += ' ' + c.get_cmake_arch_args(bitness=bitness)
def main():
extract(get_info, 'gendriver.tpl')
result_str = '{}'.format(url_table.get(ver, ""))
c.print('>>> linux url: {}'.format(result_str))
return result_str
src_base = 'qt-creator-opensource-src-' + qtc_version
qtc_dir = src_base
qtc_main_version = qtc_version[:qtc_version.rindex('.')]
base_url = 'https://download.qt.io/official_releases/qtcreator/{}/{}'\
.format(qtc_main_version, qtc_version)
src_ext = 'zip' if os_name.startswith('win') else 'tar.xz'
src_archive = src_base + '.' + src_ext
c.download(base_url + '/' + src_archive, src_archive)
c.extract(src_archive, '.')
if os_name == 'linux':
os_url = get_linux_url(qtc_main_version)
elif os_name == 'win32_msvc':
os_url = get_windows_url(32, qtc_main_version)
elif os_name == 'win64_msvc':
os_url = get_windows_url(64, qtc_main_version)
elif os_name == 'macos':
os_url = 'mac_x64'
if os_name == 'macos':
c.print('>> Workaround')
shutil.rmtree(
qtc_dir +
'/src/shared/qbs/examples/cocoa-application/CocoaApplication/en_US.lproj',
from bs4 import BeautifulSoup
from urllib.request import urlopen
import re
from common import extract, openBS, trimBio
actorName = input("Enter actor name to find:\t").strip()
actorName = actorName.replace(" ", "+")
# we extract the actual actor's url (finalUrl) from this url below
url = "https://www.imdb.com/find?ref_=nv_sr_fn&q=" + actorName + "&s=all"
pageSoup = openBS(url)
strng = str(pageSoup.find(
"table", {"class": "findList"
})) # findAll returns a list. find returns first element
uri = re.findall(r'<a href="\/name[^>]*>', strng)[0]
finalUrl = "http://imdb.com" + extract(uri)
pageSoup = openBS(finalUrl)
bio = str(pageSoup.find("div", {"class": "name-trivia-bio-text"}).div)
print(trimBio(bio))
if len(sys.argv) > 1: # handle subcommand
if sys.argv[1] == 'list':
c.print('Available modules:')
for k in iter(sorted(all_modules.keys())):
c.print(k, '---', all_modules[k]['file'])
exit(0)
for module in qt_modules:
if module not in all_modules:
c.print('>> Required module {} not available'.format(module))
continue
file_name = all_modules[module]['file']
package = all_modules[module]['package']
c.download(base_url + '/' + package + '/' + file_name, file_name)
c.extract(file_name, '.')
c.symlink(qt_dir_prefix, qt_dir)
c.print('>> Updating license')
config_name = qt_dir + '/mkspecs/qconfig.pri'
config = ''
with open(config_name, 'r') as f:
config = f.read()
config = config.replace('Enterprise', 'OpenSource')
config = config.replace('licheck.exe', '')
config = config.replace('licheck64', '')
config = config.replace('licheck_mac', '')
with open(config_name, 'w') as f:
import common as c
from config import qtc_version, os_name
import os
c.print('>> Downloading Qt Creator {} for {}'.format(qtc_version, os_name))
if os_name == 'linux':
os_url = 'linux_gcc_64_rhel72'
elif os_name == 'win32':
os_url = 'windows_msvc2017_x86'
elif os_name == 'win64':
os_url = 'windows_msvc2017_x64'
elif os_name == 'macos':
os_url = 'mac_x64'
qtc_main_version = qtc_version[:qtc_version.rindex('.')]
base_url = 'https://download.qt.io/official_releases/qtcreator/{}/{}/installer_source/{}'\
.format(qtc_main_version, qtc_version, os_url)
qtc_dir = 'qtcreator-{}'.format(qtc_version)
for archive in ['qtcreator.7z', 'qtcreator_dev.7z']:
c.download(base_url + '/' + archive, archive)
c.extract(archive, qtc_dir)
c.symlink(qtc_dir, 'qtcreator')
if os_name == 'macos':
os.makedirs('qtcreator/bin', exist_ok=True)
c.symlink(os.path.abspath('qtcreator/Qt Creator.app'),
'qtcreator/bin/Qt Creator.app')
def main():
extract(get_info, 'genvmi.tpl')
