def waitForPreprocessedImages(self):
retry = 0
while retry <= self.pp_retry_num:
# ==================================================================
# Get current number of preprocessed images
# ==================================================================
pp_files = fs.getFilesInFolderWithExts(self.source_folder, self.valid_exts)
# ==================================================================
# Exit loop when preprocessed images are ready
# ==================================================================
if len(pp_files) == self.expected_image_count:
# ==============================================================
# Wait 30 sec to make sure images are completely copied
# ==============================================================
time.sleep(30)
return True
# ==================================================================
# This shouldn't happen, but we have seen pdf's with duplicate pages, so better check
# ==================================================================
if len(pp_files) > self.expected_image_count:
if len(pp_files) > 0:
self.debug_message("Der er flere preprocesserede billeder ({}) end scannede billeder ({})"
.format(pp_files, self.expected_image_count))
return False
# ==================================================================
# Wait "self.pp_retry_wait" seconds
# ==================================================================
retry += 1
self.debug_message("Preprocesserede billeder ikke klar, venter {} sek".format(self.pp_retry_wait))
self.debug_message("Retry {} of {}".format(retry, self.pp_retry_num))
time.sleep(self.pp_retry_wait)
return False
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 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:
self.getVariables()
msg = 'Copying files from {0} to {1} via transit {2}.'
msg = msg.format(self.source_folder, self.hotfolder_dir, self.transit_dir)
self.debug_message(msg)
# ==================================================================
# Wait for preprocessed images to be ready
# Returns false if it times out
# ==================================================================
if not self.waitForPreprocessedImages():
pp_files = fs.getFilesInFolderWithExts(self.source_folder, self.valid_exts)
raise Exception('Timed out or count error while waiting for pre-processing of '
'images. Current number of processed images: '
'{0}. Expected amount: {1}'.format(pp_files, self.expected_image_count))
# ==================================================================
# Copy files to OCR-server
# ==================================================================
self.debug_message("Start copy of preprocessed images to OCR-server")
tools.copy_files(source = self.source_folder,
dest = self.hotfolder_dir,
transit = self.transit_dir,
delete_original = False,
wait_interval = self.retry_wait,
max_retries = self.retry_num,
logger = self.glogger,
valid_exts = self.valid_exts)
self.debug_message("Finished copy of preprocessed images to OCR-server")
except errors.TransferError as e:
error = e.strerror
except errors.TransferTimedOut as e:
error = e.strerror
except Exception as e:
error = str(e)
return error
def getVariables(self):
"""
Get all required vars from command line + config
and confirm their existence.
"""
process_title = self.command_line.process_title
process_path = self.command_line.process_path
# ======================================================================
# Path to folder with master image files
# ======================================================================
mi_img = self.getConfigItem('img_master_path',
section=self.folder_structure_section)
self.master_folder = os.path.join(process_path, mi_img)
# ======================================================================
# Path to folder with preprocessed files
# ======================================================================
pp_img = self.getConfigItem('img_pre_processed_path',
section=self.folder_structure_section)
self.source_folder = os.path.join(process_path, pp_img)
# ======================================================================
# legr: Get the correct OCR server for the process - antikva or fraktur
# Break if argument somehow is missing or have an invalid name
# ======================================================================
try:
ocr_workflow_type = self.getSetting('ocr_workflow_type').lower()
except KeyError:
self.error_message('{0} er ikke givet med som variabel til scriptet.'.format('ocr_workflow_type'))
if ocr_workflow_type == 'antikva':
# legr: currently antikva on ocr-01
ocr_transitfolder = self.getSetting('ocr_antikva_transit')
ocr_hotfolder = self.getSetting('ocr_antikva_hotfolder')
elif ocr_workflow_type == 'fraktur':
# legr: currently fraktur on ocr-02
ocr_transitfolder = self.getSetting('ocr_fraktur_transit')
ocr_hotfolder = self.getSetting('ocr_fraktur_hotfolder')
else:
err = ('Variablen "{0}" fra kaldet af "{1}" skal enten vaere '
'"fraktur" eller "antikva", men er pt. "{2}".')
err = err.format('ocr_workflow_type', self.name, ocr_workflow_type)
self.error_message(err)
self.transit_dir = os.path.join(ocr_transitfolder, process_title)
self.hotfolder_dir = os.path.join(ocr_hotfolder, process_title)
# ======================================================================
# Set retry wait time and retry count for copying files
# ======================================================================
self.retry_wait = int(self.getConfigItem('retry_wait'))
self.retry_num = int(self.getConfigItem('retry_num'))
# ======================================================================
# Set valid extensions for image files to check as preprocessed
# ======================================================================
self.valid_exts = self.getConfigItem('valid_file_exts', None, self.valid_exts_section).split(';')
# ======================================================================
# Set variables for waiting for preprocessed images to be ready
# ======================================================================
self.pp_retry_wait = int(self.getConfigItem('preprocess_retry_wait'))
self.pp_retry_num = int(self.getConfigItem('preprocess_retry_num'))
img_list = fs.getFilesInFolderWithExts(self.master_folder, self.valid_exts)
# Source images miunus first and last image
self.expected_image_count = len(img_list)-2
