chriswarbo-net: b571a435d3e2d7b186eacd1f66a2c18a2a2ca9f0

     1: #!/usr/bin/env python3
     2: 
     3: # Update a HTML file to use relative links, e.g. '../css/foo.css' rather than
     4: # '//chriswarbo.net/css/foo.css' or '/css/foo.css'
     5: 
     6: import sys
     7: err = sys.stderr.write
     8: 
     9: import os
    10: to_root = os.getenv('TO_ROOT')
    11: del(os)
    12: 
    13: if to_root is None:
    14:     err('No TO_ROOT given, not relativising\n')
    15:     sys.exit(0)
    16: 
    17: import fileinput
    18: stdin = ""
    19: for line in fileinput.input():
    20:     stdin += line
    21: del(fileinput)
    22: 
    23: from bs4 import BeautifulSoup
    24: doc = BeautifulSoup(stdin, 'html.parser')
    25: del(BeautifulSoup, stdin)
    26: 
    27: def prefix(attr):
    28:     def f(e):
    29:         if attr in e.attrs and e[attr].startswith(u'/'):
    30:             e[attr] = to_root + e[attr]
    31:     return f
    32: 
    33: for tag_attr in [
    34:         ('a'     , 'href'),
    35:         ('link'  , 'href'),
    36:         ('img'   , 'src' ),
    37:         ('script', 'src' )
    38: ]:
    39:     tag, attr = tag_attr
    40:     f = prefix(attr)
    41:     for elem in doc.find_all(tag):
    42:         f(elem)
    43: 
    44: print(doc)