'''
This Python code uses PyMyPDF to convert covers pages of PDF files in a folder.
Both this code and the folder containing PDF files should be in same directory.
There is an option to create HTML tags to add the PNG files into a web page. It
is stored in a text file named imageHTML.txt. The ALT tag for images are added
based on theme of the folder name with a suffix added for image number.

#--Syntax: py pdf-png-CoverPage.py folder_name
'''
import fitz
import glob, sys, os, subprocess
from PIL import Image
#------------------------------------------------------------------------------
if (len(sys.argv) < 2):
	print("\n Error! Usage: py {} folder")
	print("\n Please specify the folder name")
	sys.exit(1)
#------------------------------------------------------------------------------
root = str(sys.argv[1])
topic = root.split('/')[0][3:] #Remove first 3 characters from the folder name

# Make directory named by variable output_dir
output_dir = "PDF2PNG"
#------------------------------------------------------------------------------
#Check whether folder exists and clear contents if it does
if os.path.exists(output_dir):
	files = glob.glob(output_dir + "/" + "*.png")
	for f in files:
		os.remove(f)
else:
	os.makedirs(output_dir, exist_ok=True)

i = 0
for path, subdirs, files in os.walk(root):
	for name in files:
		i = i + 1

j = 0 
w = 600 #Desired width of all the images
out_f = "imageHTML.txt"
f = open(out_f, "w")

for path, subdirs, files in os.walk(root):
	for name in sorted(files):
		j = j + 1
		s = os.path.join(path, name)
		ext = os.path.splitext(s)[-1].lower()
		f1 = os.path.join(os.getcwd(),name)
		f2 = name.strip(".pdf") + ".png"
		with open(s, 'rb') as pfile:
			if ext == ".pdf":
				doc = fitz.open(pfile)
			page = doc.load_page(0)
			pg = page.get_pixmap()
			png_name = output_dir + "/" + f2
			pg.save(png_name)
			doc.close
		print("File "+'{0:03}'.format(j)+ " of " + str(i) + " processed!")

j = 0
for path, subdirs, files in os.walk(output_dir):
	for name in sorted(files):
		s = os.path.join(path, name)
		img = Image.open(s)
		wc = (w/float(img.size[0]))
		h = int((float(img.size[1])*float(wc)))
		#Caution about following statement. Image after resampling
		#should be stored in same or another variable before saving
		img = img.resize((w, h), Image.Resampling.LANCZOS)
		img.save(s)
		img.close()
		j = j + 1
		sen="<p><img src=\"Books/" + name + "\"> <alt=\"Books on "
		sen=sen + topic + " " + '{0:04}'.format(j) + "\"> </p> <hr>"
		f.write(sen+'\n')
		print("Image "+'{0:03}'.format(j)+ " of " + str(i) + " processed!")
f.close