converting to and from octal escaped UTF--8

[Michael Goerz]

Hi,
I am writing unicode stings into a special text file that requires to
have non-ascii characters as as octal-escaped UTF-8 codes.
For example, the letter "?" (latin capital I with acute, code point 205)
would come out as "\303\215".
I will also have to read back from the file later on and convert the
escaped characters back into a unicode string.
Does anyone have any suggestions on how to go from "?" to "\303\215" and
vice versa?

... return "".join("\%o" % ord(c) for c in source.encode('utf8'))
...
... bytes = "".join(chr(int(c, 8)) for c in encoded.split('\\')[1:])
... return bytes.decode('utf8')
...

encode(u"?")

'\\303\\215'

print decode(_)

?
HTH
Michael

Nice one. :) If I might suggest a slight variation to handle cases
where the "encoded" string contains plain text as well as octal
escapes...
encoded = encoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return encoded.decode('utf8')
This way it can handle both "\\141\\144\\146\\303\\215\\141\\144\\146"
as well as "adf\\303\\215adf".
Regards,
Jordan

err...
encoded = encoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return encoded.decode('utf8')

Great suggestions from both of you! I came up with my "final" solution
based on them. It encodes only non-ascii and non-printables, and stays
in unicode strings for both input and output. Also, low ascii values now
encode into a 3-digit octal sequence also, so that decode can catch them
properly.

Thanks a lot,
Michael

____________

import re

def encode(source):
encoded = ""
for character in source:
if (ord(character) < 32) or (ord(character) > 128):
for byte in character.encode('utf8'):
encoded += ("\%03o" % ord(byte))
else:
encoded += character
return encoded.decode('utf-8')

def decode(encoded):
decoded = encoded.encode('utf-8')
for octc in re.findall(r'\\(\d{3})', decoded):
decoded = decoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return decoded.decode('utf8')


orig = u"bla?blub" + chr(10)
enc = encode(orig)
dec = decode(enc)
print orig
print enc
print dec

[Michael Goerz]

Hi,

I am writing unicode stings into a special text file that requires to
have non-ascii characters as as octal-escaped UTF-8 codes.

For example, the letter "?" (latin capital I with acute, code point 205)
would come out as "\303\215".

I will also have to read back from the file later on and convert the
escaped characters back into a unicode string.

Does anyone have any suggestions on how to go from "?" to "\303\215" and
vice versa?

I know I can get the code point by doing

"?".decode('utf-8').encode('unicode_escape')

but there doesn't seem to be any similar method for getting the octal
escaped version.

Thanks,
Michael

[MonkeeSage]

Hi,
I am writing unicode stings into a special text file that requires to
have non-ascii characters as as octal-escaped UTF-8 codes.
For example, the letter "?" (latin capital I with acute, code point 205)
would come out as "\303\215".
I will also have to read back from the file later on and convert the
escaped characters back into a unicode string.
Does anyone have any suggestions on how to go from "?" to "\303\215" and
vice versa?

... return "".join("\%o" % ord(c) for c in source.encode('utf8'))
...
... bytes = "".join(chr(int(c, 8)) for c in encoded.split('\\')[1:])
... return bytes.decode('utf8')
...

encode(u"?")

'\\303\\215'

print decode(_)

?
HTH
Michael

Nice one. :) If I might suggest a slight variation to handle cases
where the "encoded" string contains plain text as well as octal
escapes...

def decode(encoded):
for octc in (c for c in re.findall(r'\\(\d{3})', encoded)):
encoded = encoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return encoded.decode('utf8')

This way it can handle both "\\141\\144\\146\\303\\215\\141\\144\\146"
as well as "adf\\303\\215adf".

Regards,
Jordan

[MonkeeSage]

Hi,
I am writing unicode stings into a special text file that requires to
have non-ascii characters as as octal-escaped UTF-8 codes.
For example, the letter "?" (latin capital I with acute, code point 205)
would come out as "\303\215".
I will also have to read back from the file later on and convert the
escaped characters back into a unicode string.
Does anyone have any suggestions on how to go from "?" to "\303\215" and
vice versa?

... return "".join("\%o" % ord(c) for c in source.encode('utf8'))
...
... bytes = "".join(chr(int(c, 8)) for c in encoded.split('\\')[1:])
... return bytes.decode('utf8')
...

encode(u"?")

'\\303\\215'

print decode(_)

?
HTH
Michael

Nice one. :) If I might suggest a slight variation to handle cases
where the "encoded" string contains plain text as well as octal
escapes...
encoded = encoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return encoded.decode('utf8')
This way it can handle both "\\141\\144\\146\\303\\215\\141\\144\\146"
as well as "adf\\303\\215adf".
Regards,
Jordan

err...

def decode(encoded):
for octc in re.findall(r'\\(\d{3})', encoded):
encoded = encoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return encoded.decode('utf8')

[Michael Goerz]

Looks like escape() can be a bit simpler...
result = []
result.append("\%o" % ord(char))
return ''.join(result)
Regards,
Jordan

Very neat! Thanks a lot...
Michael

[MonkeeSage]

Hi,
I am writing unicode stings into a special text file that requires to
have non-ascii characters as as octal-escaped UTF-8 codes.
For example, the letter "?" (latin capital I with acute, code point 205)
would come out as "\303\215".
I will also have to read back from the file later on and convert the
escaped characters back into a unicode string.
Does anyone have any suggestions on how to go from "?" to "\303\215" and
vice versa?
I know I can get the code point by doing

"?".decode('utf-8').encode('unicode_escape')

but there doesn't seem to be any similar method for getting the octal
escaped version.
Thanks,
Michael

I've come up with the following solution. It's not very pretty, but it
works (no bugs, I hope). Can anyone think of a better way to do it?
Michael
_________
import binascii
hexstring = binascii.b2a_hex(s)
result = ""
(hexbyte, hexstring) = (hexstring[:2], hexstring[2:])
octbyte = oct(int(hexbyte, 16)).zfill(3)
result += "\\" + octbyte[-3:]
return result
result = ""
(octbyte, s) = (s[1:4], s[4:])
result += chr(int(octbyte, 8))
result += "\\"
s = octbyte + s
result += s[0]
s = s[1:]
return result
print escape("\303\215")
print unescape('adf\\303\\215adf')

Looks like escape() can be a bit simpler...

def escape(s):
result = []
for char in s:
result.append("\%o" % ord(char))
return ''.join(result)

Regards,
Jordan

[Michael Spencer]

Hi,
I am writing unicode stings into a special text file that requires to
have non-ascii characters as as octal-escaped UTF-8 codes.
For example, the letter "?" (latin capital I with acute, code point 205)
would come out as "\303\215".
I will also have to read back from the file later on and convert the
escaped characters back into a unicode string.
Does anyone have any suggestions on how to go from "?" to "\303\215" and
vice versa?

... return "".join("\%o" % ord(c) for c in source.encode('utf8'))
...
... bytes = "".join(chr(int(c, 8)) for c in encoded.split('\\')[1:])
... return bytes.decode('utf8')
...

encode(u"?")

'\\303\\215'

print decode(_)

?
HTH
Michael

[Michael Goerz]

Hi,
I am writing unicode stings into a special text file that requires to
have non-ascii characters as as octal-escaped UTF-8 codes.
For example, the letter "?" (latin capital I with acute, code point 205)
would come out as "\303\215".
I will also have to read back from the file later on and convert the
escaped characters back into a unicode string.
Does anyone have any suggestions on how to go from "?" to "\303\215" and
vice versa?
I know I can get the code point by doing

"?".decode('utf-8').encode('unicode_escape')

but there doesn't seem to be any similar method for getting the octal
escaped version.
Thanks,
Michael

I've come up with the following solution. It's not very pretty, but it
works (no bugs, I hope). Can anyone think of a better way to do it?

Michael
_________

import binascii

def escape(s):
hexstring = binascii.b2a_hex(s)
result = ""
while len(hexstring) > 0:
(hexbyte, hexstring) = (hexstring[:2], hexstring[2:])
octbyte = oct(int(hexbyte, 16)).zfill(3)
result += "\\" + octbyte[-3:]
return result

def unescape(s):
result = ""
while len(s) > 0:
if s[0] == "\\":
(octbyte, s) = (s[1:4], s[4:])
try:
result += chr(int(octbyte, 8))
except ValueError:
result += "\\"
s = octbyte + s
else:
result += s[0]
s = s[1:]
return result

print escape("\303\215")
print unescape('adf\\303\\215adf')

[Piet van Oostrum]

If you encode chars < 32 it seems more appropriate to also encode 127.

Moreover your code is quadratic in the size of the string so if you use
long strings it would be better to use join.

[MonkeeSage]

Post by Michael Goerz

Hi,
I am writing unicode stings into a special text file that requires to
have non-ascii characters as as octal-escaped UTF-8 codes.
For example, the letter "?" (latin capital I with acute, code point 205)
would come out as "\303\215".
I will also have to read back from the file later on and convert the
escaped characters back into a unicode string.
Does anyone have any suggestions on how to go from "?" to "\303\215" and
vice versa?

... return "".join("\%o" % ord(c) for c in source.encode('utf8'))
...
... bytes = "".join(chr(int(c, 8)) for c in encoded.split('\\')[1:])
... return bytes.decode('utf8')
...

encode(u"?")

'\\303\\215'

print decode(_)

?
HTH
Michael

Nice one. :) If I might suggest a slight variation to handle cases
where the "encoded" string contains plain text as well as octal
escapes...
encoded = encoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return encoded.decode('utf8')
This way it can handle both "\\141\\144\\146\\303\\215\\141\\144\\146"
as well as "adf\\303\\215adf".
Regards,
Jordan

err...
encoded = encoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return encoded.decode('utf8')

Great suggestions from both of you! I came up with my "final" solution
based on them. It encodes only non-ascii and non-printables, and stays
in unicode strings for both input and output. Also, low ascii values now
encode into a 3-digit octal sequence also, so that decode can catch them
properly.
Thanks a lot,
Michael
____________
import re
encoded = ""
encoded += ("\%03o" % ord(byte))
encoded += character
return encoded.decode('utf-8')
decoded = encoded.encode('utf-8')
decoded = decoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return decoded.decode('utf8')
orig = u"bla?blub" + chr(10)
enc = encode(orig)
dec = decode(enc)
print orig
print enc
print dec

An optimization...in decode() store matches as keys in a dict, so you
only do the string replacement once for each unique character...

def decode(encoded):
decoded = encoded.encode('utf-8')
matches = {}
for octc in re.findall(r'\\(\d{3})', decoded):
matches[octc] = None
for octc in matches:
decoded = decoded.replace(r'\%s' % octc, chr(int(octc, 8)))
return decoded.decode('utf8')

Untested...

Regards,
Jordan