Форум программистов [Powered by Invision Power Board]

Код


>>> f.readline()
'<DIV style="position:absolute;top:416;left:58"><nobr><span class="ft10"><b>
\xc3\x8e\xc3\x80\xc3\x8e &quot;\xc3\x82\xc3\xbb\xc3\xac\xc3\xaf\xc3\xa5\xc3\xab\xc3\xaa\xc3\xae\xc3\xac&quot;, \xc3\x82\xc3\xa5\xc3\xb0\xc3\xad\xc3\xa0\xc3\xa4\xc3\xb1\xc3\xaa\xc3\xae\xc3\xa5 \xc3\x8e\xc3\x91\xc3\x81 \xc2\xb9 7970, 
\xc3\xa3. \xc3\x8c\xc3\xae\xc3\xb1\xc3\xaa\xc3\xa2\xc3\xa0<br>\xc3\x91\xc3\xa1\xc3
\xa5\xc3\xb0\xc3\xa1\xc3\xa0\xc3\xad\xc3\xaa \xc3\x90\xc3\xae\xc3\xb1\xc3\xb1\xc3\xa8
\xc3\xa8 \xc3\x8e\xc3\x80\xc3\x8e, \xc3\xa3. \xc3\x8c\xc3\xae\xc3\xb1\xc3\xaa\xc3\xa2
\xc3\xa0<br>\xc3\x90/\xc3\xb1 40702810138180121008<br>\xc3\x8a/\xc3\xb1 30101810400000000225
<br>\xc3\x81\xc3\x88\xc3\x8a 044525225 \xc3\x88\xc3\x8d\xc3\x8d 7713076301 \xc3\x8a\xc3\x8f\xc3\x8f 997750001</b></span></nobr></DIV>\n'
>>>
>>> def enc(name,file=f):
...     f.seek(0)
...     i=f.readline()
...     print f.readline().decode(name)
... 
>>> enc('utf-8')
<DIV style="position:absolute;top:416;left:58"><nobr><span class="ft10"><b>ÎÀÎ &quot;Âûìïåëêîì&quot;, Âåðíàäñêîå ÎÑÁ ¹ 7970,
 ã. Ìîñêâà<br>Ñáåðáàíê Ðîññèè ÎÀÎ, ã. Ìîñêâà<br>Ð/ñ 40702810138180121008<br>Ê/ñ 30101810400000000225
<br>ÁÈÊ 044525225 ÈÍÍ 7713076301 ÊÏÏ 997750001</b></span></nobr></DIV>

>>> enc('koi8_r')
<DIV style="position:absolute;top:416;left:58"><nobr><span class="ft10"><b>ц▌ц─ц▌ &quot;ц┌ц╩ц╛ц╞ц╔ц╚ц╙ц╝ц╛&quot;, ц┌ц╔ц╟ц╜ц═ц╓ц╠ц╙ц╝ц╔ ц▌ц▒ц│ б╧ 7970, 
цё. ц▄ц╝ц╠ц╙ц╒ц═<br>ц▒ц║ц╔ц╟ц║ц═ц╜ц╙ ц░ц╝ц╠ц╠ц╗ц╗ ц▌ц─ц▌, цё. ц▄ц╝ц╠ц╙ц╒ц═
<br>ц░/ц╠ 40702810138180121008<br>ц┼/ц╠ 30101810400000000225<br>ц│ц┬ц┼ 044525225 ц┬ц█ц█ 7713076301 ц┼ц▐ц▐ 
997750001</b></span></nobr></DIV>

>>> enc('cp1251')
<DIV style="position:absolute;top:416;left:58"><nobr><span class="ft10"><b>ГЋГЂГЋ &quot;Г‚Г»Г¬ГЇГҐГ«ГЄГ®Г¬&quot;, 
Г‚ГҐГ°ГГ Г¤Г±ГЄГ®ГҐ ГЋГ‘ГЃ В№ 7970, ГЈ. ГЊГ®Г±ГЄГўГ <br>
Г‘ГЎГҐГ°ГЎГ ГГЄ ГђГ®Г±Г±ГЁГЁ ГЋГЂГЋ, ГЈ. ГЊГ®Г±ГЄГўГ <br>Гђ/Г± 40702810138180121008<br>
ГЉ/Г± 30101810400000000225<br>ГЃГ€ГЉ 044525225 Г€ГЌГЌ 7713076301 ГЉГЏГЏ 997750001</b></span></nobr></DIV>

>>> chardet.detect(a)
{'confidence': 0.98999999999999999, 'encoding': 'utf-8'}
>>>

Код


>>> print f.readline().decode('utf16')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python2.5/encodings/utf_16.py", line 16, in decode
    return codecs.utf_16_decode(input, errors, True)
UnicodeDecodeError: 'utf16' codec can't decode byte 0x0a in position 364: truncated data
>>>

Цитата

>>> print f.readline().decode('utf16','ignore')
䐼噉猠祴敬∽潰楳楴湯愺獢汯瑵㭥潴㩰ㄴ㬶敬瑦㔺∸㰾潮牢㰾灳湡挠慬獳∽瑦〱㸢戼쌾쎎쎀₎焦潵㭴苃믃곃꿃ꗃꯃ꫃껃곃焦潵㭴‬苃ꗃ냃귃ꃃ꓃뇃꫃껃ꗃ쌠쎎쎑₁맂㜠㜹ⰰ쌠⺣쌠쎌쎮쎱쎪쎢㲠牢쌾쎑쎡쎥쎰쎡쎠쎭₪郃껃뇃뇃ꣃꣃ쌠쎎쎀Ⲏ쌠⺣쌠쎌쎮쎱쎪쎢㲠牢쌾⾐뇃㐠㜰㈰ㄸ㄰㠳㠱㄰ㄲ〰㰸牢쌾⾊뇃㌠㄰㄰ㄸ㐰〰〰〰〰㈲㰵牢쌾쎁쎈₊㐰㔴㔲㈲‵裃跃跃㜠ㄷ〳㘷〳‱諃迃迃㤠㜹㔷〰㄰⼼㹢⼼灳湡㰾港扯㹲⼼䥄㹖
>>>

Цитата

>>> ss
'\'<DIV style="position:absolute;top:416;left:58">\n\xc3\x8e\xc3\x80\xc3\x8e "\xc3\x82\xc3\xbb\xc3\xac\xc3\xaf\xc3\xa5\xc3\xab\xc3\xaa\xc3\xae\xc3\xac", \xc3\x82\xc3\xa5\xc3\xb0\xc3\xad\xc3\xa0\xc3\xa4\xc3\xb1\xc3\xaa\xc3\xae\xc3\xa5 \xc3\x8e\xc3\x91\xc3\x81 \xc2\xb9 7970, \n\xc3\xa3. \xc3\x8c\xc3\xae\xc3\xb1\xc3\xaa\xc3\xa2\xc3\xa0 \xc3\x91\xc3\xa1\xc3\n\xa5\xc3\xb0\xc3\xa1\xc3\xa0\xc3\xad\xc3\xaa \xc3\x90\xc3\xae\xc3\xb1\xc3\xb1\xc3\xa8\n\xc3\xa8 \xc3\x8e\xc3\x80\xc3\x8e, \xc3\xa3. \xc3\x8c\xc3\xae\xc3\xb1\xc3\xaa\xc3\xa2\n\xc3\xa0 \xc3\x90/\xc3\xb1 40702810138180121008 \xc3\x8a/\xc3\xb1 30101810400000000225\n \xc3\x81\xc3\x88\xc3\x8a 044525225 \xc3\x88\xc3\x8d\xc3\x8d 7713076301 \xc3\x8a\xc3\x8f\xc3\x8f 997750001</DIV>\n'
>>> import chardet
>>> chardet.detect
<function detect at 0x8df956c>
>>> chardet.detect(ss)
{'confidence': 0.26190629807642268, 'encoding': 'ISO-8859-2'}

Код


>>> enc('utf-8')
<DIV style="position:absolute;top:416;left:58">
<nobr>
<span class="ft10">
<b>ÎÀÎ &quot;Âûìïåëêîì&quot;, Âåðíàäñêîå ÎÑÁ ¹ 7970, ã. Ìîñêâà<br>Ñáåðáàíê Ðîññèè ÎÀÎ, ã. Ìîñêâà
<br>Ð/ñ 40702810138180121008
<br>Ê/ñ30101810400000000225
<br>ÁÈÊ 044525225 ÈÍÍ 7713076301
 ÊÏÏ 997750001</b>
</span></nobr></DIV>