def test_insert_command_builder(monkeypatch):
DUMMY_EXIV2_PATH = "/usr/bin/exiv2"
src_file = "/Users/Documents/sample.jp2"
expected_cmd = ['/usr/bin/exiv2', '-iaX', '/Users/Documents/sample.jp2']
# Because the location of exiv2 is faked for this example, we need to
# ignore the checks
monkeypatch.setattr(os.path, "exists", lambda x: True)
command_builder = Exiv2CommandBuilder(builder=builders.InsertIPTCCommand(),
program_path=DUMMY_EXIV2_PATH)
command = command_builder.build_command(src=src_file)
assert command == expected_cmd
def convert_tiff_access_folder2(source: str,
destination,
overwrite_existing=True,
remove_on_success=False):
logger = logging.getLogger(__name__)
image_convert_command = ImagemagickCommandBuilder(
builder=im_cb.Standard())
metadata_extractor = Exiv2CommandBuilder(exi2_cb.ExtractIPTCCommand())
metadata_injector = Exiv2CommandBuilder(exi2_cb.InsertIPTCCommand())
with tempfile.TemporaryDirectory(prefix="convert_") as tmp_folder:
command_runner = CommandRunner()
tiffs = list(get_tiffs(source))
total_files_converted = 0
for i, tiff in enumerate(tiffs):
tmp_access_tif = os.path.join(tmp_folder,
os.path.basename(tiff))
tmp_access_jp2 = os.path.splitext(tmp_access_tif)[0] + ".jp2"
final_access_jp2 = os.path.join(
destination, os.path.basename(tmp_access_jp2))
if not overwrite_existing:
if os.path.exists(final_access_jp2):
logger.warning("{} already exists. skipping".format(
final_access_jp2))
continue
logger.info("Converting part {} of {}: From {} to {} ".format(
i + 1, len(tiffs), os.path.basename(tiff),
os.path.basename(final_access_jp2)))
logger.debug("Copying \"{}\" to temp folder \"{}\"".format(
tiff, tmp_folder))
shutil.copyfile(tiff, tmp_access_tif)
logger.debug("Using \"{}\" to create \"{}\"".format(
tmp_access_tif, tmp_access_jp2))
im_command = image_convert_command.build_command(
tmp_access_tif, tmp_access_jp2)
try:
command_runner.run(im_command)
if remove_on_success:
# remove the read only attribute first
os.chmod(tiff, stat.S_IWRITE)
logger.info("Deleting file, \"{}\".".format(tiff))
os.remove(tiff)
except RuntimeError as e:
logger.error(str(e))
raise
finally:
stdout, stderr = command_runner.get_output()
if stdout:
logger.info(stdout)
if stderr:
logger.warning(stderr)
# HACK: Clean up from Imagemagick because it seems to have a
# problem with embedded thumbnails
_cleanup_multiple(os.path.basename(tmp_access_jp2), tmp_folder)
# Create sidecar metadata files
logger.debug(
"Extracting metadata from \"{}\"".format(tmp_access_tif))
mde_command = metadata_extractor.build_command(tmp_access_tif)
try:
command_runner.run(mde_command)
except RuntimeError as e:
logger.error(str(e))
raise
finally:
stdout, stderr = command_runner.get_output()
if stdout:
logger.info(stdout)
if stderr:
logger.warning(stderr)
# Insert sidecar metadata files into jp2
logger.debug(
"Injecting metadata into \"{}\"".format(tmp_access_jp2))
mdi_command = metadata_injector.build_command(tmp_access_jp2)
try:
command_runner.run(mdi_command)
except RuntimeError as e:
logger.error(str(e))
raise
finally:
stdout, stderr = command_runner.get_output()
if stdout:
logger.info(stdout)
if stderr:
logger.warning(stderr)
logger.debug("Moving \"{}\" into \"{}\"".format(
tmp_access_jp2, destination))
if not os.path.exists(destination):
os.makedirs(destination)
shutil.move(tmp_access_jp2, final_access_jp2)
total_files_converted += 1
logger.info("Created {} file(s) in {}.".format(
total_files_converted, destination))
def convert_tiff_access_folder(path: str,
overwrite_existing=True,
remove_on_success=False):
"""
Converts all tiff files located in that folder into JP2000 .jp2 files and
migrated all the metadata from the tiff file into the newly produced jp2
file. All new files are saved in the same place that as their source files.
Args:
path: The path to the folder containing tiff files to be converter.
overwrite_existing: If an existing jp2 file already exists in the same
folder, overwrite it with a new file.
remove_on_success: Delete access tiff file afterwards if successfully converted.
"""
warn("Use convert_tiff_access_folder2 instead", DeprecationWarning)
logger = logging.getLogger(__name__)
image_convert_command = ImagemagickCommandBuilder(builder=im_cb.Standard())
metadata_extractor = Exiv2CommandBuilder(exi2_cb.ExtractIPTCCommand())
metadata_injector = Exiv2CommandBuilder(exi2_cb.InsertIPTCCommand())
with tempfile.TemporaryDirectory(prefix="convert_") as tmp_folder:
command_runner = CommandRunner()
tiffs = list(get_tiffs(path))
total_files_converted = 0
for i, tiff in enumerate(tiffs):
tmp_access_tif = os.path.join(tmp_folder, os.path.basename(tiff))
tmp_access_jp2 = os.path.splitext(tmp_access_tif)[0] + ".jp2"
final_access_jp2 = os.path.join(path,
os.path.basename(tmp_access_jp2))
if not overwrite_existing:
if os.path.exists(final_access_jp2):
logger.warning(
"{} already exists. skipping".format(final_access_jp2))
continue
logger.info("Converting part {} of {}: From {} to {} ".format(
i + 1, len(tiffs), os.path.basename(tiff),
os.path.basename(final_access_jp2)))
logger.debug("Copying \"{}\" to temp folder \"{}\"".format(
tiff, path))
shutil.copyfile(tiff, tmp_access_tif)
logger.debug("Using \"{}\" to create \"{}\"".format(
tmp_access_tif, tmp_access_jp2))
im_command = image_convert_command.build_command(
tmp_access_tif, tmp_access_jp2)
try:
command_runner.run(im_command)
if remove_on_success:
# remove the read only attribute first
os.chmod(tiff, stat.S_IWRITE)
logger.info("Deleting file, \"{}\".".format(tiff))
os.remove(tiff)
except RuntimeError as e:
logger.error(str(e))
raise
finally:
stdout, stderr = command_runner.get_output()
if stdout:
logger.info(stdout)
if stderr:
logger.warning(stderr)
# HACK: Clean up from Imagemagick because it seems to have a
# problem with embedded thumbnails
_cleanup_multiple(os.path.basename(tmp_access_jp2), tmp_folder)
# Create sidecar metadata files
logger.debug(
"Extracting metadata from \"{}\"".format(tmp_access_tif))
mde_command = metadata_extractor.build_command(tmp_access_tif)
try:
command_runner.run(mde_command)
except RuntimeError as e:
logger.error(str(e))
raise
finally:
stdout, stderr = command_runner.get_output()
if stdout:
logger.info(stdout)
if stderr:
logger.warning(stderr)
# Insert sidecar metadata files into jp2
logger.debug(
"Injecting metadata into \"{}\"".format(tmp_access_jp2))
mdi_command = metadata_injector.build_command(tmp_access_jp2)
try:
command_runner.run(mdi_command)
except RuntimeError as e:
logger.error(str(e))
raise
finally:
stdout, stderr = command_runner.get_output()
if stdout:
logger.info(stdout)
if stderr:
logger.warning(stderr)
logger.debug("Moving \"{}\" into \"{}\"".format(
tmp_access_jp2, path))
shutil.move(tmp_access_jp2, final_access_jp2)
total_files_converted += 1
logger.info("Converted {} file(s) in {}.".format(
total_files_converted, path))
def convert_tiff_access_folder(path: str,
overwrite_existing=True,
remove_on_success=False):
warn("Use convert_tiff_access_folder instead", DeprecationWarning)
logger = logging.getLogger(__name__)
total_files_converted = 0
image_convert_command = KakaduCommandBuilder(
builder=kd_cb.HathiPreset())
metadata_extractor = Exiv2CommandBuilder(exi2_cb.ExtractIPTCCommand())
metadata_injector = Exiv2CommandBuilder(exi2_cb.InsertIPTCCommand())
with tempfile.TemporaryDirectory(prefix="convert_") as tmp_folder:
command_runner = CommandRunner()
tiffs = list(get_tiffs(path))
for i, tiff in enumerate(tiffs):
tmp_access_tif = os.path.join(tmp_folder,
os.path.basename(tiff))
tmp_access_jp2 = os.path.splitext(tmp_access_tif)[0] + ".jp2"
final_access_jp2 = os.path.join(
path, os.path.basename(tmp_access_jp2))
if not overwrite_existing:
if os.path.exists(final_access_jp2):
logger.warning("{} already exists. skipping".format(
final_access_jp2))
continue
logger.info("Converting part {} of {}: From {} to {} ".format(
i + 1, len(tiffs), os.path.basename(tiff),
os.path.basename(final_access_jp2)))
logger.debug("Copying \"{}\" to temp folder \"{}\"".format(
tiff, path))
shutil.copyfile(tiff, tmp_access_tif)
logger.debug("Using \"{}\" to create \"{}\"".format(
tmp_access_tif, tmp_access_jp2))
im_command = image_convert_command.build_command(
tmp_access_tif, tmp_access_jp2)
try:
command_runner.run(im_command)
if remove_on_success:
logger.info("Deleting file, \"{}\".".format(tiff))
os.remove(tiff)
except RuntimeError as e:
logger.error(str(e))
raise
finally:
pass
stdout, stderr = command_runner.get_output()
if stdout:
logger.info(stdout)
if stderr:
logger.warning(stderr)
logger.debug(
"Extracting metadata from \"{}\"".format(tmp_access_tif))
mde_command = metadata_extractor.build_command(tmp_access_tif)
try:
command_runner.run(mde_command)
except RuntimeError as e:
logger.error(str(e))
raise
finally:
stdout, stderr = command_runner.get_output()
if stdout:
logger.info(stdout)
if stderr:
logger.warning(stderr)
# Insert sidecar metadata files into jp2
logger.debug(
"Injecting metadata into \"{}\"".format(tmp_access_jp2))
mdi_command = metadata_injector.build_command(tmp_access_jp2)
try:
command_runner.run(mdi_command)
except RuntimeError as e:
logger.error(str(e))
raise
finally:
stdout, stderr = command_runner.get_output()
if stdout:
logger.info(stdout)
if stderr:
logger.warning(stderr)
logger.debug("Moving \"{}\" into \"{}\"".format(
tmp_access_jp2, path))
shutil.move(tmp_access_jp2, final_access_jp2)
total_files_converted += 1
logger.info("Converted {} file(s) in {}.".format(
total_files_converted, path))