def write_metadata_local(inputfile, outputfile, metadata_dictionary, verbose):
"""
EXIF and IPTC metadata writing, previous tag printing, to
images. If some tag not set, it is auto-added, but be a valid exif
or iptc tag.
@param inputfile: path to the image
@type inputfile: string
@param outputfile: path to the resulting image
@type outputfile: string
@param verbose: verbosity
@type verbose: int
@param metadata_dictionary: metadata information to update inputfile
@rtype: dict
"""
if inputfile != outputfile:
# Create copy of inputfile
try:
shutil.copy2(inputfile, outputfile)
except Exception, err:
raise InvenioWebSubmitFileMetadataRuntimeError(err)
def write_metadata_local(inputfile, outputfile, metadata_dictionary, verbose):
"""
Metadata write method, takes the .pdf as input and creates a new
one with the new info.
@param inputfile: path to the pdf
@type inputfile: string
@param outputfile: path to the resulting pdf
@type outputfile: string
@param verbose: verbosity
@type verbose: int
@param metadata_dictionary: metadata information to update inputfile
@type metadata_dictionary: dict
"""
# Take the file name (0 base, 1 name, 2 ext)
filename = decompose_file(inputfile)[1]
# Print pdf metadata
if verbose > 1:
print 'Metadata information in the PDF file ' + filename + ': \n'
try:
os.system(CFG_PATH_PDFTK + ' ' + inputfile + ' dump_data')
except Exception:
print 'Problem with inputfile to PDFTK'
# Info file for pdftk
(fd, path_to_info) = tempfile.mkstemp(prefix="wsm_pdf_plugin_info_", \
dir=CFG_TMPDIR)
os.close(fd)
file_in = open(path_to_info, 'w')
if verbose > 5:
print "Saving PDFTK info file to %s" % path_to_info
# User interaction to form the info file
# Main Case: Dictionary received through option -d
if not metadata_dictionary == {}:
for tag in metadata_dictionary:
line = 'InfoKey: ' + tag + '\nInfoValue: ' + \
metadata_dictionary[tag] + '\n'
if verbose > 0:
print line
file_in.writelines(line)
else:
data_modified = False
user_input = 'user_input'
print "Entering interactive mode. Choose what you want to do:"
while (user_input):
if not data_modified:
try:
user_input = raw_input('[w]rite / [q]uit\n')
except:
print "Aborting"
return
else:
try:
user_input = raw_input(
'[w]rite / [q]uit and apply / [a]bort \n')
except:
print "Aborting"
return
if user_input == 'q':
if not data_modified:
return
break
elif user_input == 'w':
try:
tag = raw_input('Tag to update:\n')
value = raw_input('With value:\n')
except:
print "Aborting"
return
# Write to info file
line = 'InfoKey: ' + tag + '\nInfoValue: ' + value + '\n'
data_modified = True
file_in.writelines(line)
elif user_input == 'a':
return
else:
print "Invalid option: "
file_in.close()
(fd, pdf_temp_path) = tempfile.mkstemp(prefix="wsm_pdf_plugin_pdf_", \
dir=CFG_TMPDIR)
os.close(fd)
# Now we call pdftk tool to update the info on a pdf
#try:
cmd_pdftk = '%s %s update_info %s output %s'
(exit_status, output_std, output_err) = \
run_shell_command(cmd_pdftk,
args=(CFG_PATH_PDFTK, inputfile,
path_to_info, pdf_temp_path))
if verbose > 5:
print output_std, output_err
if os.path.exists(pdf_temp_path):
# Move to final destination if exist
try:
shutil.move(pdf_temp_path, outputfile)
except Exception, err:
raise InvenioWebSubmitFileMetadataRuntimeError("Could not move %s to %s" % \
(pdf_temp_path, outputfile))
return
else:
print "Invalid option: "
file_in.close()
(fd, pdf_temp_path) = tempfile.mkstemp(prefix="wsm_pdf_plugin_pdf_", \
dir=CFG_TMPDIR)
os.close(fd)
# Now we call pdftk tool to update the info on a pdf
#try:
cmd_pdftk = '%s %s update_info %s output %s'
(exit_status, output_std, output_err) = \
run_shell_command(cmd_pdftk,
args=(CFG_PATH_PDFTK, inputfile,
path_to_info, pdf_temp_path))
if verbose > 5:
print output_std, output_err
if os.path.exists(pdf_temp_path):
# Move to final destination if exist
try:
shutil.move(pdf_temp_path, outputfile)
except Exception, err:
raise InvenioWebSubmitFileMetadataRuntimeError("Could not move %s to %s" % \
(pdf_temp_path, outputfile))
else:
# Something bad happened
raise InvenioWebSubmitFileMetadataRuntimeError(
"Could not update metadata " + output_err)
def read_metadata_remote(inputfile, loginpw, verbose):
"""
EXIF and IPTC metadata extraction and printing from remote images
@param inputfile: path to the remote image
@type inputfile: string
@param verbose: verbosity
@type verbose: int
@param loginpw: credentials to access secure servers (username:password)
@type loginpw: string
@return: dictionary with metadata
@rtype: dict
"""
# Check that inputfile is an URL
secure = False
pos = inputfile.lower().find('http://')
if pos < 0:
secure = True
pos = inputfile.lower().find('https://')
if pos < 0:
raise InvenioWebSubmitFileMetadataRuntimeError("Inputfile (" + inputfile + ") is " + \
"not an URL, nor remote resource.")
# Check if there is login and password
if loginpw != None:
(userid, passwd) = loginpw.split(':')
# Make HTTPS Connection
domain = inputfile.split('/')[2]
if verbose > 3:
print 'Domain: ', domain
url = inputfile.split(domain)[1]
if verbose > 3:
print 'URL: ', url
# Establish headers
if loginpw != None:
_headers = {"Accept": "*/*",
"Authorization": "Basic " + \
base64.encodestring(userid + ':' + passwd).strip()}
else:
_headers = {"Accept": "*/*"}
conn = None
# Establish connection
# Case HTTPS
if secure:
try:
conn = httplib.HTTPSConnection(domain)
## Request a connection
conn.request("GET", url,
headers = _headers)
except Exception:
# Cannot connect
print 'Could not connect'
# Case HTTP
else:
try:
conn = httplib.HTTPConnection(domain)
## Request a connection
conn.request("GET", url,
headers = _headers)
except Exception:
# Cannot connect
print 'Could not connect'
# Get response
if verbose > 5:
print "Fetching data from remote server."
response = conn.getresponse()
if verbose > 2:
print response.status, response.reason
if response.status == 401:
# Authentication required
raise InvenioWebSubmitFileMetadataRuntimeError("URL requires authentication. Use --loginpw option")
# Read first marker from image
data = response.read(2)
# Check if it is a valid image
if data[0:2] != '\xff\xd8':
raise InvenioWebSubmitFileMetadataRuntimeError("URL does not brings to a valid image file.")
else:
if verbose > 5:
print 'Valid JPEG Standard-based image'
# Start the fake image
path_to_fake = fake_image_init(verbose)
# Continue reading
data = response.read(2)
# Check if we find metadata (EXIF or IPTC)
while data[0:2] != '\xff\xdb':
if data[0:2] == '\xff\xe1' or data[0:2] == '\xff\xed':
marker = data
if verbose > 5:
print 'Metadata Marker->', repr(marker), '\nGetting data'
size = response.read(2)
length = ord(size[0]) * 256 + ord(size[1])
meta = response.read(length-2)
insert_metadata(path_to_fake, marker, size, meta, verbose)
break
else:
data = response.read(2)
# Close connection
conn.close()
# Close fake image
fake_image_close(path_to_fake, verbose)
# Extract metadata once fake image is done
return read_metadata_local(path_to_fake, verbose)
data_modified = True
except:
print "Aborting"
return
try:
image[tag] = value
except Exception, err:
print 'Tag or Value incorrect'
elif user_input == 'a':
return
else:
print "Invalid option: "
try:
image.writeMetadata()
except Exception, err:
raise InvenioWebSubmitFileMetadataRuntimeError("Could not update metadata: " + err)
def read_metadata_remote(inputfile, loginpw, verbose):
"""
EXIF and IPTC metadata extraction and printing from remote images
@param inputfile: path to the remote image
@type inputfile: string
@param verbose: verbosity
@type verbose: int
@param loginpw: credentials to access secure servers (username:password)
@type loginpw: string
@return: dictionary with metadata
@rtype: dict
"""
# Check that inputfile is an URL