def split_pdf_to_png_files (pdf_file_spec,output_dir):
"""
Split the PDF file specified by PDF_FILE_SPEC into a series of
files, each representing a single page as a PNG image. Write files
to directory specified by OUTPUT_DIR.
"""
png_files = None
try:
# sanity check
if not os.path.isabs(pdf_file_spec):
msg = "The input PDF must be specified as an absolute file path"
lg.error(json1.json_msg(108,[msg],False,files=[pdf_file_spec]))
sys.exit(msg)
else:
# array of (<file_name>,<page_number>) tuples
png_specs = pdf.pdf_to_pngs(pdf_file_spec,output_dir)
except Exception as e:
msg = json1.json_failed_to_convert_pdf(e,pdf_file_spec)
lg.error(msg)
print "failed to convert PDF file(s): %s" % pdf_file_spec
print "e: %s" % e
lg.info(json1.json_last_log_msg())
sys.exit(msg)
else:
lg.info(json1.json_pdf_to_pngs_success(pdf_file_spec,png_specs))
return png_specs
def invoke_pdfimages_on (pdf_file_spec,output_dir):
"""
Extract images in PDF file specified by PDF_FILE_SPEC into a
series of files, each representing a single PNG image. Write files
to directory specified by OUTPUT_DIR.
Returns a list of tuples where each tuple has the structure
(png_file,png_file_page_number) png_file_page_number is an
integer. The list is an ordered sequence with respect to page
number - low to high.
"""
png_file_page_number_tuples = None
try:
# sanity check
if not os.path.isabs(pdf_file_spec):
msg = "The input PDF must be specified as an absolute file path"
lg.error(json1.json_msg(108,[msg],False,files=[pdf_file_spec]))
sys.exit(msg)
else:
png_file_page_number_tuples = pdf.pdfimages(pdf_file_spec,output_dir)
except Exception as e:
lg.debug(str(e))
msg = json1.json_failed_to_convert_pdf(e,pdf_file_spec)
lg.error(msg)
lg.info(json1.json_last_log_msg())
sys.exit(msg)
else:
# Is it really import to log png files? (Need to dig them out of tuples...)
lg.info(json1.json_pdf_to_pngs_success(pdf_file_spec,
None #png_files
))
return png_file_page_number_tuples
def barcodeScan(imagePNGPath, scan_region):
"""
Return None if a barcode was not found. If a barcode was found,
return a string corresponding to the barcode-encoded data.
Search within the region defined by SCAN_REGION when SCAN_REGION
is a list. When SCAN_REGION is a list, it specifies two points as
[x1,y1,x2,y2]. These two points (x1,y1) and (x2,y2) are pairs
(x,y) of percentages (each expressed as a value between 0 and 1.0)
relative to the dimensions of the image; they define the box
within which the barcode scan occurs.
If SCAN_REGION is not a list, the full image is analyzed. If
analysis of the full image is desirable, do not set SCAN_REGION to
[0,0,1,1] but instead set it to None or some other non-list value.
"""
# sanity check(s)
if not isinstance(scan_region,list):
scan_region = None
else:
for value in scan_region:
if (value < 0 or value > 1):
msg = json1.json_msg(999,"insane scan region value",False,None)
lg.error(msg)
lg.info(json1.json_last_log_msg())
sys.exit(msg)
# obtain image data either via PIL or CV2/numpy
# 1. using pil
# PIL origin (0,0) is top left corner
pil = Image.open(imagePNGPath).convert('L') # 'L' is "black and white mode": converts to 8-bit pixels B/W
# 2. using cv2/numpy
#pil_1 = Image.open(imagePNGPath)
#frame = pil_1.convert("RGB")
#pil_gray = cv2.cvtColor(numpy.array(frame), cv2.COLOR_BGR2GRAY, dstCn=0)
#pil = Image.fromarray(pil_gray)
pilCropped = pil
width, height = pil.size
lg.debug("width: %s height: %s",width,height)
if scan_region:
# relative (percentage) values between 0 and 1
x_crop_min = min(scan_region[0],scan_region[2])
x_crop_max = max(scan_region[0],scan_region[2])
y_crop_min = min(scan_region[1],scan_region[3])
y_crop_max = max(scan_region[1],scan_region[3])
cropTop=int(height*y_crop_min)
cropBottom=int(height*y_crop_max)
cropLeft=int(height*x_crop_min)
cropRight=int(height*x_crop_max)
# crop box is 4-tuple: left,upper,right,lower
pilCropBox = [cropLeft,cropTop,cropRight,cropBottom]
pilCropped = pil.crop(pilCropBox)
# zbar sometimes catches a barcode at a lower resolution but misses it at a higher resolution. Scan for barcode with several variants of image specified by IMAGE_FILE_SPEC.
barcodeString = barcode_scan_at_resolutions(pilCropped,None)
if ( not barcodeString ):
lg.warn(json1.json_barcode_not_found_msg([imagePNGPath],""))
return barcodeString
def module_sanity_check (module_name,exitp):
"""MODULE_NAME is a string"""
# check for presence of module which might not be installed/accessible
try:
imp.find_module(module_name)
except ImportError:
msg = json1.json_msg_module_not_accessible(module_name)
lg.error(msg)
lg.info(json1.json_last_log_msg())
if exitp:
sys.exit(msg)
def executable_sanity_checks (executables):
"""
Check for availability of executables specified in the list of strings EXECUTABLES.
"""
lg.debug(util)
for executable_spec in executables:
if not util.which(executable_spec):
msg = json1.json_msg_executable_not_accessible(executable_spec)
lg.error(msg)
lg.info(json1.json_last_log_msg())
sys.exit(msg)
def signal_handler(signal, frame):
# Ensure receipt of signal is logged prior to terminating
msg = json1.json_exit_on_external_request_msg()
lg.error(msg)
lg.info(json1.json_last_log_msg())
sys.exit()
def main():
"""Handle command-line invocation of pdfxcb.py."""
global lg
parser = argparse.ArgumentParser(description="This is pdfxcb")
parser.add_argument("-f",
help="absolute path to log file",
action="store",
dest="log_file",
type=str)
parser.add_argument("-d",
help="absolute path to output directory",
action="store",
dest="output_dir",
type=str)
parser.add_argument("-m",
help="match barcodes to regex (ignore if no match)",
action="store",
dest="match_re_string",
type=str)
parser.add_argument("-p",
help="identifier for a specific instance of pdfxcb",
action="store",
dest="identifier",
type=str)
parser.add_argument("-l",
help="integer between 0 (verbose) and 51 (terse) defining logging",
action="store",
dest="log_level",
type=int)
#parser.add_argument('-v', '--version', action='version', version=version.version)
parser.add_argument("input_files", help="an input (PDF) file",
# keep nargs as we may want to accept multiple PDFs as input at some point
nargs=1,
type=str)
args = parser.parse_args()
#
# define logging (level, file, message format, ...)
#
log_level = args.log_level
if isinstance(log_level, int) and log_level >= 0 and log_level <= 51:
log_level = log_level
else:
# since this function doesn't necessarily exit quickly
log_level = logging.INFO
if args.log_file:
logfile = args.log_file
else:
logfile = 'busca.log'
json_log_format = '%(message)s'
for handler in lg.getLogger().handlers:
lg.getLogger().removeHandler(handler)
formatter = logging.Formatter(json_log_format)
# sanity check for existence of log file directory
if (os.path.dirname(logfile) and
not os.path.exists(os.path.dirname(logfile))):
raise Exception(str.format("log file directory {0} not present",
os.path.dirname(logfile)))
file_handler = logging.FileHandler(logfile,'w')
file_handler.setFormatter(formatter)
lg.getLogger().addHandler(file_handler)
lg.getLogger().setLevel(log_level)
lg.debug("args: %s", args)
lg.debug("sys.argv: %s",sys.argv)
if args.identifier:
identifier = args.identifier
else:
identifier = str(uuid.uuid1())
# 1000[0-9][0-9][0-9]$ matches on tt user id
match_re_string = args.match_re_string
lg.debug(match_re_string)
match_re = None
if match_re_string:
match_re = re.compile(match_re_string)
pdf_file_spec = args.input_files[0]
lg.debug(pdf_file_spec)
lg.info(json1.json_first_log_msg(identifier, files = [pdf_file_spec] ))
rasterize_p = False
# generic debugging
lg.debug(os.getcwd()) # current/working directory
# might also want to import platform to get architecture, other details...
try:
pdfxcb(pdf_file_spec,args.output_dir,match_re,rasterize_p)
except Exception as e:
lg.error("Crash and burn")
lg.error(sys.exc_info()[0])
raise
lg.info(json1.json_last_log_msg())
def file_sanity_check (file,exitp):
if not os.path.isfile(file):
lg.error(json1.json_file_not_found(file))
lg.info(json1.json_last_log_msg())
if exitp:
sys.exit("File " + file + " not found.")
def directory_sanity_check (directory_spec,exitp):
if not os.path.isdir(directory_spec):
lg.error(json1.json_file_not_found(directory_spec))
lg.info(json1.json_last_log_msg())
if exitp:
sys.exit("Directory " + directory_spec + " not found.")
# bubbles.py, deskew.py, feature_detect.py, and util.py use numpy
# try:
# imp.find_module('numpy')
# except ImportError:
# msg = json1.json_msg_module_not_accessible('numpy')
# lg.error(msg)
# lg.info(json1.json_last_log_msg())
# sys.exit(msg)
try:
imp.find_module('PyPDF2')
except ImportError:
msg = json1.json_msg_module_not_accessible('PyPDF2')
lg.error(msg)
lg.info(json1.json_last_log_msg())
sys.exit(msg)
import PyPDF2
# handle external signals requesting termination
def signal_handler(signal, frame):
# Ensure receipt of signal is logged prior to terminating
msg = json1.json_exit_on_external_request_msg()
lg.error(msg)
lg.info(json1.json_last_log_msg())
sys.exit()
signal.signal(signal.SIGHUP, signal_handler)
signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGTERM, signal_handler)