def deleteWorkingFolders(self):
'''
Delete tempoary folders.
'''
fs.clear_folder(self.temp_folder, also_folder=True)
if self.settings['output_pdf']:
fs.clear_folder(self.temp_pdf_folder, also_folder=True)
def step(self):
'''
This script's role is to wait until
ocr processing is complete before finishing.
In the event of a timeout, it reports back to
previous step before exiting.
'''
error = None
try:
#===================================================================
# Get and set variables
#===================================================================
self.getVariables()
#===================================================================
# Delete existing bw-pdf file
#===================================================================
fs.clear_folder(self.goobi_pdf)
#===================================================================
# Wait for PDF-file to be ready on OCR-server
#===================================================================
error = self.waitForOcr()
except IOError as e:
# if we get an IO error we need to crash
error = ('Error reading from directory {0}')
error = error.format(e.strerror)
return error
except ValueError as e:
# caused by conversion of non-numeric strings in config to nums
error = "Invalid config data supplied, error: {0}"
error = error.format(e.strerror)
return error
# if we've gotten this far, we've timed out and need to go back to the previous step
return error
def step(self):
error = None
try:
#===================================================================
# Get and set variables
#===================================================================
self.getVariables()
#===================================================================
# Create folder for temporary pdf-files
#===================================================================
fs.create_folder(self.temp_folder)
#===================================================================
# Delete previously created color pdf
#===================================================================
fs.clear_folder(self.pdf_color_folder_path)
#===================================================================
# Convert input images to one pdf
#===================================================================
msg = ('Creating PDF-file from images in "{0}". Outputting to temp '
'folder "{1}" and creates one PDF with path "{2}". Resize '
'PDF to {3}% and compress with quality {4}%')
msg = msg.format(self.input_folder,self.temp_folder,
self.color_pdf_path,self.resize,self.quality)
self.debug_message(msg)
convert.createPdfFromFolder(src = self.input_folder,
file_dest = self.color_pdf_path,
temp_folder = self.temp_folder,
quality = self.quality,
resize_pct = self.resize,
valid_exts = self.valid_exts)
except image_tools.ConvertError as e:
error = str(e)
return error
def addFrontispiecesToPdfs(self):
# Create temp folder for temp pdf-files
temp_folder = os.path.join(self.temp_root,self.process_title)
tools.create_folder(temp_folder)
self.addFrontispiecesToPdf(self.frontispieces_bw,self.pdf_bw_path,temp_folder)
self.addFrontispiecesToPdf(self.frontispieces,self.pdf_color_path,temp_folder)
# Delete temp_folder
fs.clear_folder(temp_folder, also_folder=True)
def processFiles(self):
'''
Process all the files
'''
file_paths = sorted(self.img_proc_info['images'].keys())
for file_path in file_paths:
info = self.img_proc_info['images'][file_path]
proc_time_stat = self.processFile(file_path,info)
self.add_to_avg_time_stat(proc_time_stat)
fs.clear_folder(self.temp_folder)
if self.settings['has_binding'] and not self.settings['remove_binding']:
for b in self.bindings:
file_name,_ = os.path.splitext(os.path.basename(b.rstrip(os.sep)))
b_pdf_dest = os.path.join(self.temp_pdf_folder,file_name+'.pdf')
if self.settings['output_images']: shutil.copy2(b,self.output_image_location)
if self.settings['output_pdf']: image_tools.compressFile(b,b_pdf_dest,resize=50,quality=33)
def addBindingsToPdf(self):
#=======================================================================
# Get density for bw-pdf (i.e. DPI/PixelsPerInch)
#=======================================================================
density = pdf_tools.getDensity(src=self.pdf_bw_path,layer=0)
#=======================================================================
# Create temp folder for temp pdf-files
#=======================================================================
temp_folder = os.path.join(self.temp_root,self.process_title)
tools.create_folder(temp_folder)
#=======================================================================
# Get path for first and last image
#=======================================================================
images = fs.getFilesInFolderWithExts(self.img_master_path,
self.valid_exts,
absolute=True)
#=======================================================================
# Create PDF of bindings (first and last image in master image folder)
#=======================================================================
front_image_path = images[0]
end_image_path = images[-1]
front_pdf_path = os.path.join(temp_folder,'front.pdf')
end_pdf_path = os.path.join(temp_folder,'end.pdf')
image_tools.compressFile(input_file = front_image_path,
output_file = front_pdf_path,
quality = self.quality,
resize = self.resize,
density = density)
image_tools.compressFile(input_file = end_image_path,
output_file = end_pdf_path,
quality = self.quality,
resize = self.resize,
density = density)
#=======================================================================
# Add front and back-binding to pdf
#=======================================================================
pdf_list = [front_pdf_path,self.pdf_bw_path,end_pdf_path]
temp_dest = os.path.join(temp_folder,self.process_title+'.pdf')
pdf_tools.joinPdfFiles(pdf_list, temp_dest)
#=======================================================================
# Move new pdf from temp to bw-pdf location (overwrite)
#=======================================================================
shutil.move(temp_dest, self.pdf_bw_path)
#=======================================================================
# Delete temp_folder
#=======================================================================
fs.clear_folder(temp_folder, also_folder=True)
def getCropCoordinates(self):
image_paths = sorted(self.img_proc_info['images'].keys())
debug_pivot = self.settings['debug_pivot']
if self.debug: self.logger.debug('Get crop coordinates')
for image_path in image_paths:
w = self.img_proc_info['images'][image_path]['image_width']
h = self.img_proc_info['images'][image_path]['image_height']
time_stat = {}
t = time.time()
if self.settings['bw_for_innercrop']:
threshold = self.settings['innercrop_bw_src_threshold']
file_name,_ = os.path.splitext(os.path.basename(image_path))
dest = os.path.join(self.temp_folder,file_name+'_bw_for_innercrop.tif')
src = image_tools.convertToBw(image_path,dest,threshold=threshold)
else:
src = image_path
time_stat['BW to get crop coordinates'] = time.time()-t
t = time.time()
fuzzval = self.settings['innercrop_fuzzval']
mode = self.settings['innercrop_mode']
#===================================================================
# TODO: Add an try-except here. If non-valid output, raise error
# except error and set crop to False (e.g. use mean/avg crops later)
#===================================================================
_,coordinates = image_tools.innercrop(src,self.temp_folder,w=w,h=h,
innercrop_path=self.innercrop_exe_path,
mode=mode,fuzzval=fuzzval)
self.img_proc_info['images'][image_path]['crop_coordinates'] = coordinates
time_stat['Get crop coordinates'] = time.time()-t
fs.clear_folder(self.temp_folder)
self.add_to_avg_time_stat(time_stat)
if self.debug:
count = self.img_proc_info['avg_time_stat']['Get crop coordinates'][1]
avg = self.img_proc_info['avg_time_stat']['Get crop coordinates'][2]
if (count%debug_pivot) == 0: # log for every 10 processed iamges
left = len(image_paths)-count
time_used = get_delta_time(count*avg)
time_left = get_delta_time(left*avg)
msg = ('\t{0} images cropped, {1} images left, '
'{2} time elapsed, {3} est. time left.')
self.logger.debug(msg.format(count,left,time_used,time_left))
def createPdfFromFolder(src, file_dest,temp_folder,
quality=50,resize_pct=50,valid_exts=['jpg','tif']):
'''
Use ImageMagick to create one pdf from all the images in a folder and
output to a given destination.
Create a pdf of each image and place in temp folder. Merge output pdf-files
to pdf-dest and remove temp folder.
'''
image_paths = fs.getFilesInFolderWithExts(src, valid_exts)
for image in image_paths:
# Handle spaces in filenames
image = '"' + image + '"'
input_path = os.path.join(src,image)
file_name,_ = os.path.splitext(image)
output_file_name = file_name+'.pdf'
output_path = os.path.join(temp_folder,output_file_name)
image_tools.compressFile(input_path, output_path, quality, resize_pct)
pdf_misc.mergePdfFilesInFolder(temp_folder,file_dest)
fs.clear_folder(temp_folder,also_folder=True)
def step(self):
error = None
try:
#===================================================================
# Get and set variables and settings for preprocessing
# self.settings contains settigs for preprocessing script
#===================================================================
self.getVariables()
#===================================================================
# Remove previously preprocessed images
#===================================================================
fs.clear_folder(self.img_pre_processed_path)
#===================================================================
# Preprocess images
#===================================================================
ip = image_preprocessor.ImagePreprocessor(self.img_master_path,
self.settings,
self.glogger,
self.debug)
ip.processFolder()
except image_tools.ConvertError as e:
error = str(e)
return error
def step(self):
error = None
try:
self.get_variables()
# Check if we have a METS file (meta.xml)
self.check_paths()
# Clean up the METS file (meta.xml)
reset.reset_mets_file(self.command_line.process_path)
# Clear files from Limb and Goobi (if any)
fs.clear_folder(self.goobi_altos)
fs.clear_folder(self.goobi_toc)
fs.clear_folder(self.goobi_ojs)
fs.clear_folder(self.goobi_pdf)
fs.clear_folder(self.splitted_pdfs)
fs.clear_folder(self.thumbnails)
except ValueError as e:
error = e
except IOError as e:
error = e.strerror
return error