put_on_one_page.py

import sys, os, re
from subprocess import check_output

if len(sys.argv)<2:
  print("USAGE: python put_on_one_page.py page_to_put_on_one_page")
  sys.exit(0)

if not os.path.isfile(sys.argv[1]):
  print("given path %s is not a file" % (sys.argv[1],))
  sys.exit(0)

with open(sys.argv[1], "r") as f:
  content = f.read()

# remove <?php...?>
content = re.sub(r'<\?php[\s\S]*?\?>','',content)

uploaded_image = {}
def upload_image(rel_path):
  global uploaded_image
  if rel_path not in uploaded_image:
    try:
      out = check_output(["bash", "/tmp/imgurbash.sh", rel_path])
      uploaded_image[rel_path] =  str(out, 'UTF-8')[:-1]
    except:
      uploaded_image[rel_path] = ""
  return uploaded_image[rel_path]

# replace <link rel="stylesheet" /> by file content
for i in set(re.findall(r'<link [^>]*?\brel="stylesheet" type="text/css" href="([^:"]+)" />', content)):
  if os.path.isfile(i):
    with open(i) as s:
      sub = s.read()
      for img in re.findall(r'["\'\(]([^\'(:"]+\.(?:png|jpeg|jpg|gif))["\'\)]', sub):
        realimg = os.path.normpath("./"+os.path.dirname(i)+'/'+img)
        sub = sub.replace(img, realimg)
      content = re.sub(r'<link [^>]*?\brel="stylesheet" type="text/css" href="'+i+'" />', '<style type="text/css">'+sub+'</style>', content)

# replace <script type="text/javascript" src="..." /> by file content
for i in set(re.findall(r'<script type="text/javascript" src="([^:"]+)"></script>', content)):
  if os.path.isfile(i):
    with open(i) as s:
      sub = s.read()
      content = content.replace('<script type="text/javascript" src="'+i+'"></script>', '<script type="text/javascript">'+sub+'</script>')

for img in re.findall(r'["\'\(]([^\'(:"]+\.(?:png|jpeg|jpg|gif))["\'\)]', content):
  if os.path.isfile(img):
    url = upload_image(img)
    if url != "":
      content = content.replace(img, url)

print(content)