def process_bibtex(corpus_path):
import pybtex
from pybtex.database import parse_file
from topicexplorer.lib.util import overwrite_prompt, safe_symlink
print "Loading BibTeX from", corpus_path
bib = parse_file(corpus_path)
target_dir = os.path.basename(corpus_path).replace('.bib','')
if not os.path.exists(target_dir):
os.makedirs(target_dir)
elif overwrite_prompt(target_dir):
shutil.rmtree(target_dir)
os.makedirs(target_dir)
else:
raise IOError("Path exits: {}".format(target_dir))
for entry in bib.entries:
if bib.entries[entry].fields.get('file', None):
filename = '/' + bib.entries[entry].fields['file'].replace(':pdf','')[1:]
if 'C$\\backslash$:' in filename:
filename = filename.replace('C$\\backslash$:', '')
filename = filename[1:]
filename = os.path.normpath(filename)
filename = os.path.abspath(filename)
if not os.path.exists(filename):
print "Invalid 'file' field for BibTeX entry {}:\n\t({})".format(entry, filename)
else:
new_path = os.path.join(target_dir, os.path.basename(filename))
safe_symlink(filename, new_path)
else:
print "No 'file' field for BibTeX entry: {}".format(entry)
return target_dir
def convert_and_write(fname, output_dir=None, overwrite=False, verbose=False):
output = os.path.basename(fname)
output = output.replace('.pdf','.txt')
if output_dir:
output = os.path.join(output_dir, output)
if output_dir is not None and not os.path.exists(output_dir):
os.makedirs(output_dir)
if overwrite or util.overwrite_prompt(output):
with open(output, 'wb') as outfile:
outfile.write(convert(fname))
if verbose:
print "converted", fname, "->", output
def convert_and_write(fname, output_dir=None, overwrite=False, verbose=False):
output = os.path.basename(fname)
output = output.replace('.pdf', '.txt')
if output_dir:
output = os.path.join(output_dir, output)
if output_dir is not None and not os.path.exists(output_dir):
os.makedirs(output_dir)
if overwrite or util.overwrite_prompt(output):
with open(output, 'wb') as outfile:
outfile.write(convert(fname))
if verbose:
print "converted", fname, "->", output
def convert_and_write(fname, output_dir=None, overwrite=False, verbose=False,
tokenizer='modern'):
output = os.path.basename(fname)
if output_dir:
output = os.path.join(output_dir, output)
if output_dir is not None and not os.path.exists(output_dir):
os.makedirs(output_dir)
if overwrite or util.overwrite_prompt(output):
with open(output, 'wb', encoding='utf8') as outfile:
outfile.write(convert(fname, tokenizer=tokenizer))
if verbose:
print "converted", fname, "->", output
def convert_and_write(fname,
output_dir=None,
overwrite=False,
verbose=False,
tokenizer='modern'):
output = os.path.basename(fname)
if output_dir:
output = os.path.join(output_dir, output)
if output_dir is not None and not os.path.exists(output_dir):
os.makedirs(output_dir)
if overwrite or util.overwrite_prompt(output):
with open(output, 'wb', encoding='utf8') as outfile:
outfile.write(convert(fname, tokenizer=tokenizer))
if verbose:
print "converted", fname, "->", output
def process_bibtex(corpus_path, quiet):
import pybtex
from pybtex.database import parse_file
from topicexplorer.lib.util import overwrite_prompt, safe_symlink
print("Loading BibTeX from", corpus_path)
bib = parse_file(corpus_path)
target_dir = os.path.basename(corpus_path).replace('.bib', '')
if not os.path.exists(target_dir):
os.makedirs(target_dir)
elif overwrite_prompt(target_dir) and not quiet:
shutil.rmtree(target_dir)
os.makedirs(target_dir)
else:
raise IOError("Path exits: {}".format(target_dir))
for entry in bib.entries:
if bib.entries[entry].fields.get('file', None):
filename = '/' + bib.entries[entry].fields['file'].replace(
':pdf', '')[1:]
if 'C$\\backslash$:' in filename:
filename = filename.replace('C$\\backslash$:', '')
filename = filename[1:]
filename = os.path.normpath(filename)
filename = os.path.abspath(filename)
if not os.path.exists(filename):
print(
"Invalid 'file' field for BibTeX entry {}:\n\t({})".format(
entry, filename))
else:
new_path = os.path.join(target_dir, os.path.basename(filename))
try:
safe_symlink(filename, new_path)
except OSError:
print("Error linking file for BibTeX entry {}:\n\t({})".
format(entry, filename))
else:
print("No 'file' field for BibTeX entry: {}".format(entry))
return target_dir
def main(args):
args.config_file = os.path.abspath(args.config_file)
with open(get_static_resource_path('ipynb/corpus.tmpl.py')) as corpustmpl:
corpus_py = corpustmpl.read()
corpus_py = Template(corpus_py)
corpus_py = corpus_py.safe_substitute(config_file=args.config_file)
ipynb_path = os.path.join(os.path.dirname(args.config_file), "notebooks")
print(ipynb_path)
if not os.path.exists(ipynb_path):
os.makedirs(ipynb_path)
filename = os.path.join(ipynb_path, "corpus.py")
if overwrite_prompt(filename, default=True):
print("Writing", filename)
with open(filename, 'w') as corpusloader:
corpusloader.write(corpus_py)
pyflag = 'py2' if sys.version_info.major == 2 else 'py3'
glob_path = (get_static_resource_path('ipynb') + '/*.{}.ipynb').format(pyflag)
for notebook in glob(glob_path):
new_nb_name = os.path.basename(notebook).replace('.' +pyflag, '')
new_nb_path = os.path.join(ipynb_path, new_nb_name)
if overwrite_prompt(new_nb_path, default=False):
print("Copying", notebook)
shutil.copy(notebook, new_nb_path)
if args.launch:
import subprocess
os.chdir(ipynb_path)
try:
# TODO: Fix KeyboardInterrupt errors
try:
grp_fn = os.setsid
except AttributeError:
grp_fn = None
proc = subprocess.Popen("jupyter notebook", shell=True, preexec_fn=grp_fn)
# stdin=subprocess.PIPE, preexec_fn=grp_fn)
# stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
except OSError:
print("ERROR: Command `jupyter notebook` not found.")
print(" If IPython or Anaconda is installed, check your PATH variable.")
sys.exit(1)
# CLEAN EXIT AND SHUTDOWN OF IPYTHON NOTEBOOK
def signal_handler(signal, frame):
# Cross-Platform Compatability
try:
os.killpg(proc.pid, signal)
proc.communicate()
except AttributeError:
subprocess.call(['taskkill', '/F', '/T', '/PID', str(proc.pid)])
sys.exit(0)
signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGTERM, signal_handler)
print("\nPress Ctrl+C to shutdown the IPython notebook server\n")
# Cross-platform Compatability
try:
signal.pause()
except AttributeError:
# Windows hack
while True:
time.sleep(1)
def main(args):
args.config_file = os.path.abspath(args.config_file)
template_dir = os.path.dirname(__file__)
template_dir = os.path.join(template_dir, '../ipynb/')
template_dir = os.path.normpath(template_dir)
with open(os.path.join(template_dir, 'corpus.tmpl.py')) as corpustmpl:
corpus_py = corpustmpl.read()
corpus_py = Template(corpus_py)
corpus_py = corpus_py.safe_substitute(config_file=args.config_file)
ipynb_path = os.path.join(os.path.dirname(args.config_file), "notebooks")
print ipynb_path
if not os.path.exists(ipynb_path):
os.makedirs(ipynb_path)
filename = os.path.join(ipynb_path, "corpus.py")
if overwrite_prompt(filename, default=True):
print "Writing", filename
with open(filename, 'w') as corpusloader:
corpusloader.write(corpus_py)
for notebook in glob(template_dir + '/*.ipynb'):
new_nb_path = os.path.join(ipynb_path, os.path.basename(notebook))
if overwrite_prompt(new_nb_path, default=False):
print "Copying", notebook
shutil.copy(notebook, ipynb_path)
if args.launch:
import subprocess, sys
os.chdir(ipynb_path)
try:
# TODO: Fix KeyboardInterrupt errors
try:
grp_fn = os.setsid
except AttributeError:
grp_fn = None
proc = subprocess.Popen("ipython notebook",
shell=True,
preexec_fn=grp_fn)
#stdin=subprocess.PIPE, preexec_fn=grp_fn)
#stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
except OSError:
print "ERROR: Command `ipython notebook` not found."
print " If IPython or Anaconda is installed, check your PATH variable."
sys.exit(1)
# CLEAN EXIT AND SHUTDOWN OF IPYTHON NOTEBOOK
def signal_handler(signal, frame):
# Cross-Platform Compatability
try:
os.killpg(proc.pid, signal)
proc.communicate()
except AttributeError:
subprocess.call(
['taskkill', '/F', '/T', '/PID',
str(proc.pid)])
sys.exit(0)
signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGTERM, signal_handler)
print "\nPress Ctrl+C to shutdown the IPython notebook server\n"
# Cross-platform Compatability
try:
signal.pause()
except AttributeError:
# Windows hack
while True:
time.sleep(1)
def main(args):
args.config_file = os.path.abspath(args.config_file)
with open(get_static_resource_path('ipynb/corpus.tmpl.py')) as corpustmpl:
corpus_py = corpustmpl.read()
corpus_py = Template(corpus_py)
corpus_py = corpus_py.safe_substitute(config_file=args.config_file)
ipynb_path = os.path.join(os.path.dirname(args.config_file), "notebooks")
print(ipynb_path)
if not os.path.exists(ipynb_path):
os.makedirs(ipynb_path)
filename = os.path.join(ipynb_path, "corpus.py")
if overwrite_prompt(filename, default=True):
print("Writing", filename)
with open(filename, 'w') as corpusloader:
corpusloader.write(corpus_py)
pyflag = 'py2' if sys.version_info.major == 2 else 'py3'
glob_path = (get_static_resource_path('ipynb') +
'/*.{}.ipynb').format(pyflag)
for notebook in glob(glob_path):
new_nb_name = os.path.basename(notebook).replace('.' + pyflag, '')
new_nb_path = os.path.join(ipynb_path, new_nb_name)
if overwrite_prompt(new_nb_path, default=False):
print("Copying", notebook)
shutil.copy(notebook, new_nb_path)
if args.launch:
import subprocess
os.chdir(ipynb_path)
try:
# TODO: Fix KeyboardInterrupt errors
try:
grp_fn = os.setsid
except AttributeError:
grp_fn = None
proc = subprocess.Popen("jupyter notebook",
shell=True,
preexec_fn=grp_fn)
# stdin=subprocess.PIPE, preexec_fn=grp_fn)
# stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
except OSError:
print("ERROR: Command `jupyter notebook` not found.")
print(
" If IPython or Anaconda is installed, check your PATH variable."
)
sys.exit(1)
# CLEAN EXIT AND SHUTDOWN OF IPYTHON NOTEBOOK
def signal_handler(signal, frame):
# Cross-Platform Compatability
try:
os.killpg(proc.pid, signal)
proc.communicate()
except AttributeError:
subprocess.call(
['taskkill', '/F', '/T', '/PID',
str(proc.pid)])
sys.exit(0)
signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGTERM, signal_handler)
print("\nPress Ctrl+C to shutdown the IPython notebook server\n")
# Cross-platform Compatability
try:
signal.pause()
except AttributeError:
# Windows hack
while True:
time.sleep(1)
def main(args):
args.config_file = os.path.abspath(args.config_file)
template_dir = os.path.dirname(__file__)
template_dir = os.path.join(template_dir, '../ipynb/')
template_dir = os.path.normpath(template_dir)
with open(os.path.join(template_dir, 'corpus.tmpl.py')) as corpustmpl:
corpus_py = corpustmpl.read()
corpus_py = Template(corpus_py)
corpus_py = corpus_py.safe_substitute(config_file=args.config_file)
ipynb_path = os.path.join(os.path.dirname(args.config_file), "notebooks")
print ipynb_path
if not os.path.exists(ipynb_path):
os.makedirs(ipynb_path)
filename = os.path.join(ipynb_path, "corpus.py")
if overwrite_prompt(filename, default=True):
print "Writing", filename
with open(filename,'w') as corpusloader:
corpusloader.write(corpus_py)
for notebook in glob(os.path.join(template_dir, '*.ipynb')):
new_nb_path = os.path.join(ipynb_path, os.path.basename(notebook))
if overwrite_prompt(new_nb_path, default=False):
print "Copying", notebook
shutil.copy(notebook, ipynb_path)
if args.launch:
import subprocess, sys
os.chdir(ipynb_path)
try:
# TODO: Fix KeyboardInterrupt errors
try:
grp_fn = os.setsid
except AttributeError:
grp_fn = None
proc = subprocess.Popen("ipython notebook", shell=True, preexec_fn=grp_fn)
#stdin=subprocess.PIPE, preexec_fn=grp_fn)
#stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
except OSError:
print "ERROR: Command `ipython notebook` not found."
print " If IPython or Anaconda is installed, check your PATH variable."
sys.exit(1)
# CLEAN EXIT AND SHUTDOWN OF IPYTHON NOTEBOOK
def signal_handler(signal,frame):
# Cross-Platform Compatability
try:
os.killpg(proc.pid, signal)
proc.communicate()
except AttributeError:
subprocess.call(['taskkill', '/F', '/T', '/PID', str(proc.pid)])
sys.exit(0)
signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGTERM, signal_handler)
print "\nPress Ctrl+C to shutdown the IPython notebook server\n"
# Cross-platform Compatability
try:
signal.pause()
except AttributeError:
# Windows hack
while True:
time.sleep(1)
