forked from kennethreitz/dive-into-python3
-
Notifications
You must be signed in to change notification settings - Fork 0
/
porting-code-to-python-3-with-2to3.html
1348 lines (1150 loc) · 70.8 KB
/
porting-code-to-python-3-with-2to3.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE html>
<meta charset=utf-8>
<title>Porting code to Python 3 with 2to3 - Dive Into Python 3</title>
<!--[if IE]><script src=j/html5.js></script><![endif]-->
<link rel=stylesheet href=dip3.css>
<style>
</style>
<link rel=stylesheet media='only screen and (max-device-width: 480px)' href=mobile.css>
<link rel=stylesheet media=print href=print.css>
<meta content='initial-scale=1.0' name=viewport>
<body id=appa>
<form action=http://www.google.com/cse><div><input name=cx type=hidden value=014021643941856155761:l5eihuescdw><input name=ie type=hidden value=UTF-8> <input type=search name=q size=25 placeholder="powered by Google™"> <input name=sa type=submit value=Search></div></form>
<p>You are here: <a href=index.html>Home</a> <span class=u>‣</span> <a href=table-of-contents.html#porting-code-to-python-3-with-2to3>Dive Into Python 3</a> <span class=u>‣</span>
<p id=level>Difficulty level: <span class=u title=pro>♦♦♦♦♦</span>
<h1>Porting Code to Python 3 with <code>2to3</code></h1>
<blockquote class=q>
<p><span class=u>❝</span> Life is pleasant. Death is peaceful. It’s the transition that’s troublesome. <span class=u>❞</span><br>— Isaac Asimov (attributed)
</blockquote>
<p id=toc>
<h2 id=divingin>Diving In</h2>
<p class=f>So much has changed between Python 2 and Python 3, there are vanishingly few programs that will run unmodified under both. But don’t despair! To help with this transition, Python 3 comes with a utility script called <code>2to3</code>, which takes your actual Python 2 source code as input and auto-converts as much as it can to Python 3. <a href=case-study-porting-chardet-to-python-3.html#running2to3>Case study: porting <code>chardet</code> to Python 3</a> describes how to run the <code>2to3</code> script, then shows some things it can’t fix automatically. This appendix documents what it <em>can</em> fix automatically.
<h2 id=print><code>print</code> statement</h2>
<p>In Python 2, <code><dfn>print</dfn></code> was a statement. Whatever you wanted to print simply followed the <code>print</code> keyword. In Python 3, <a href=your-first-python-program.html#divingin><code>print()</code> is a function</a>. Whatever you want to print, pass it to <code>print()</code> like any other function.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>print</code>
<td><code class=pp>print()</code>
<tr><th>②
<td><code class=pp>print 1</code>
<td><code class=pp>print(1)</code>
<tr><th>③
<td><code class=pp>print 1, 2</code>
<td><code class=pp>print(1, 2)</code>
<tr><th>④
<td><code class=pp>print 1, 2,</code>
<td><code class=pp>print(1, 2, end=' ')</code>
<tr><th>⑤
<td><code class=pp>print >>sys.stderr, 1, 2, 3</code>
<td><code class=pp>print(1, 2, 3, file=sys.stderr)</code>
</table>
<ol>
<li>To print a blank line, call <code>print()</code> without any arguments.
<li>To print a single value, call <code>print()</code> with one argument.
<li>To print two values separated by a space, call <code>print()</code> with two arguments.
<li>This one is a little tricky. In Python 2, if you ended a <code>print</code> statement with a comma, it would print the values separated by spaces, then print a trailing space, then stop without printing a carriage return. (Technically, it’s a little more complicated than that. The <code>print</code> statement in Python 2 used a now-deprecated attribute called <var>softspace</var>. Instead of printing a space, Python 2 would set <code>sys.stdout.softspace</code> to 1. The space character wasn’t really printed until something else got printed on the same line. If the next <code>print</code> statement printed a carriage return, <code>sys.stdout.softspace</code> would be set to 0 and the space would never be printed. You probably never noticed the difference unless your application was sensitive to the presence or absence of trailing whitespace in <code>print</code>-generated output.) In Python 3, the way to do this is to pass <code>end=' '</code> as a keyword argument to the <code>print()</code> function. The <code>end</code> argument defaults to <code>'\n'</code> (a carriage return), so overriding it will suppress the carriage return after printing the other arguments.
<li>In Python 2, you could redirect the output to a pipe — like <code>sys.stderr</code> — by using the <code>>>pipe_name</code> syntax. In Python 3, the way to do this is to pass the pipe in the <code>file</code> keyword argument. The <code>file</code> argument defaults to <code>sys.stdout</code> (standard out), so overriding it will output to a different pipe instead.
</ol>
<h2 id=unicodeliteral>Unicode string literals</h2>
<p>Python 2 had two string types: <dfn>Unicode</dfn> strings and non-Unicode strings. Python 3 has one string type: <a href=strings.html#divingin>Unicode strings</a>.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>u'PapayaWhip'</code>
<td><code class=pp>'PapayaWhip'</code>
<tr><th>②
<td><code class=pp>ur'PapayaWhip\foo'</code>
<td><code class=pp>r'PapayaWhip\foo'</code>
</table>
<ol>
<li>Unicode string literals are simply converted into string literals, which, in Python 3, are always Unicode.
<li>Unicode raw strings (in which Python does not auto-escape backslashes) are converted to raw strings. In Python 3, raw strings are always Unicode.
</ol>
<h2 id=unicode><code>unicode()</code> global function</h2>
<p>Python 2 had two global functions to coerce objects into strings: <code>unicode()</code> to coerce them into Unicode strings, and <code>str()</code> to coerce them into non-Unicode strings. Python 3 has only one string type, <a href=strings.html#divingin>Unicode strings</a>, so the <code>str()</code> function is all you need. (The <code>unicode()</code> function no longer exists.)
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>
<td><code class=pp>unicode(anything)</code>
<td><code class=pp>str(anything)</code>
</table>
<h2 id=long><code>long</code> data type</h2>
<p>Python 2 had separate <code>int</code> and <code><dfn>long</dfn></code> types for non-floating-point numbers. An <code>int</code> could not be any larger than <a href=#renames><code>sys.maxint</code></a>, which varied by platform. Longs were defined by appending an <code>L</code> to the end of the number, and they could be, well, longer than ints. In Python 3, <a href=native-datatypes.html#numbers>there is only one integer type</a>, called <code>int</code>, which mostly behaves like the <code>long</code> type in Python 2. Since there are no longer two types, there is no need for special syntax to distinguish them.
<p>Further reading: <a href=http://www.python.org/dev/peps/pep-0237/><abbr>PEP</abbr> 237: Unifying Long Integers and Integers</a>.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>x = 1000000000000L</code>
<td><code class=pp>x = 1000000000000</code>
<tr><th>②
<td><code class=pp>x = 0xFFFFFFFFFFFFL</code>
<td><code class=pp>x = 0xFFFFFFFFFFFF</code>
<tr><th>③
<td><code class=pp>long(x)</code>
<td><code class=pp>int(x)</code>
<tr><th>④
<td><code class=pp>type(x) is long</code>
<td><code class=pp>type(x) is int</code>
<tr><th>⑤
<td><code class=pp>isinstance(x, long)</code>
<td><code class=pp>isinstance(x, int)</code>
</table>
<ol>
<li>Base 10 long integer literals become base 10 integer literals.
<li>Base 16 long integer literals become base 16 integer literals.
<li>In Python 3, the old <code>long()</code> function no longer exists, since longs don’t exist. To coerce a variable to an integer, use the <code>int()</code> function.
<li>To check whether a variable is an integer, get its type and compare it to <code>int</code>, not <code>long</code>.
<li>You can also use the <code>isinstance()</code> function to check data types; again, use <code>int</code>, not <code>long</code>, to check for integers.
</ol>
<h2 id=ne><> comparison</h2>
<p>Python 2 supported <code><></code> as a synonym for <code>!=</code>, the not-equals comparison operator. Python 3 supports the <code>!=</code> operator, but not <code><></code>.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>if x <> y:</code>
<td><code class=pp>if x != y:</code>
<tr><th>②
<td><code class=pp>if x <> y <> z:</code>
<td><code class=pp>if x != y != z:</code>
</table>
<ol>
<li>A simple comparison.
<li>A more complex comparison between three values.
</ol>
<h2 id=has_key><code>has_key()</code> dictionary method</h2>
<p>In Python 2, dictionaries had a <code><dfn>has_key</dfn>()</code> method to test whether the dictionary had a certain key. In Python 3, this method no longer exists. Instead, you need to use <a href=native-datatypes.html#mixed-value-dictionaries>the <code>in</code> operator</a>.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>a_dictionary.has_key('PapayaWhip')</code>
<td><code class=pp>'PapayaWhip' in a_dictionary</code>
<tr><th>②</th >
<td><code class=pp>a_dictionary.has_key(x) or a_dictionary.has_key(y)</code>
<td><code class=pp>x in a_dictionary or y in a_dictionary</code>
<tr><th>③
<td><code class=pp>a_dictionary.has_key(x or y)</code>
<td><code class=pp>(x or y) in a_dictionary</code>
<tr><th>④
<td><code class=pp>a_dictionary.has_key(x + y)</code>
<td><code class=pp>(x + y) in a_dictionary</code>
<tr><th>⑤
<td><code class=pp>x + a_dictionary.has_key(y)</code>
<td><code class=pp>x + (y in a_dictionary)</code>
</table>
<ol>
<li>The simplest form.
<li>The <code>in</code> operator takes precedence over the <code>or</code> operator, so there is no need for parentheses around <code>x in a_dictionary</code> or around <code>y in a_dictionary</code>.
<li>On the other hand, you <em>do</em> need parentheses around <code>x or y</code> here, for the same reason — <code>in</code> takes precedence over <code>or</code>. (Note: this code is completely different from the previous line. Python interprets <code>x or y</code> first, which results in either <var>x</var> (if <var>x</var> is <a href=native-datatypes.html#booleans>true in a boolean context</a>) or <var>y</var>. Then it takes that singular value and checks whether it is a key in <var>a_dictionary</var>.)
<li>The <code>+</code> operator takes precedence over the <code>in</code> operator, so this form technically doesn’t need parentheses around <code>x + y</code>, but <code>2to3</code> includes them anyway.
<li>This form definitely needs parentheses around <code>y in a_dictionary</code>, since the <code>+</code> operator takes precedence over the <code>in</code> operator.
</ol>
<h2 id=dict>Dictionary methods that return lists</h2>
<p>In Python 2, many dictionary methods returned lists. The most frequently used methods were <code><dfn>keys</dfn>()</code>, <code><dfn>items</dfn>()</code>, and <code><dfn>values</dfn>()</code>. In Python 3, all of these methods return dynamic <dfn>views</dfn>. In some contexts, this is not a problem. If the method’s return value is immediately passed to another function that iterates through the entire sequence, it makes no difference whether the actual type is a list or a view. In other contexts, it matters a great deal. If you were expecting a complete list with individually addressable elements, your code will choke, because views do not support indexing.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>a_dictionary.keys()</code>
<td><code class=pp>list(a_dictionary.keys())</code>
<tr><th>②
<td><code class=pp>a_dictionary.items()</code>
<td><code class=pp>list(a_dictionary.items())</code>
<tr><th>③
<td><code class=pp>a_dictionary.iterkeys()</code>
<td><code class=pp>iter(a_dictionary.keys())</code>
<tr><th>④
<td><code class=pp>[i for i in a_dictionary.iterkeys()]</code>
<td><code class=pp>[i for i in a_dictionary.keys()]</code>
<tr><th>⑤
<td><code class=pp>min(a_dictionary.keys())</code>
<td><i>no change</i>
</table>
<ol>
<li><code>2to3</code> errs on the side of safety, converting the return value from <code>keys()</code> to a static list with the <code>list()</code> function. This will always work, but it will be less efficient than using a view. You should examine the converted code to see if a list is absolutely necessary, or if a view would do.
<li>Another view-to-list conversion, with the <code>items()</code> method. <code>2to3</code> will do the same thing with the <code>values()</code> method.
<li>Python 3 does not support the <code>iterkeys()</code> method anymore. Use <code>keys()</code>, and if necessary, convert the view to an iterator with the <code>iter()</code> function.
<li><code>2to3</code> recognizes when the <code>iterkeys()</code> method is used inside a list comprehension, and converts it to the <code>keys()</code> method (without wrapping it in an extra call to <code>iter()</code>). This works because views are iterable.
<li><code>2to3</code> recognizes that the <code>keys()</code> method is immediately passed to a function which iterates through an entire sequence, so there is no need to convert the return value to a list first. The <code>min()</code> function will happily iterate through the view instead. This applies to <code>min()</code>, <code>max()</code>, <code>sum()</code>, <code>list()</code>, <code>tuple()</code>, <code>set()</code>, <code>sorted()</code>, <code>any()</code>, and <code>all()</code>.
</ol>
<h2 id=imports>Modules that have been renamed or reorganized</h2>
<p>Several modules in the Python Standard Library have been renamed. Several other modules which are related to each other have been combined or reorganized to make their association more logical.
<h3 id=http><code>http</code></h3>
<p>In Python 3, several related <abbr>HTTP</abbr> modules have been combined into a single package, <code>http</code>.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>import <dfn>httplib</dfn></code>
<td><code class=pp>import http.client</code>
<tr><th>②
<td><code class=pp>import <dfn>Cookie</dfn></code>
<td><code class=pp>import http.cookies</code>
<tr><th>③
<td><code class=pp>import <dfn>cookielib</dfn></code>
<td><code class=pp>import http.cookiejar</code>
<tr><th>④
<td><pre class=pp><code>import <dfn>BaseHTTPServer</dfn>
import <dfn>SimpleHTTPServer</dfn>
import <dfn>CGIHttpServer</dfn></code></pre>
<td><code class=pp>import http.server</code>
</table>
<ol>
<li>The <code>http.client</code> module implements a low-level library that can request <abbr>HTTP</abbr> resources and interpret <abbr>HTTP</abbr> responses.
<li>The <code>http.cookies</code> module provides a Pythonic interface to browser cookies that are sent in a <code>Set-Cookie:</code> <abbr>HTTP</abbr> header.
<li>The <code>http.cookiejar</code> module manipulates the actual files on disk that popular web browsers use to store cookies.
<li>The <code>http.server</code> module provides a basic <abbr>HTTP</abbr> server.
</ol>
<h3 id=urllib><code>urllib</code></h3>
<p>Python 2 had a rat’s nest of overlapping modules to parse, encode, and fetch <abbr>URLs</abbr>. In Python 3, these have all been refactored and combined in a single package, <code>urllib</code>.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>import <dfn>urllib</dfn></code>
<td><code class=pp>import urllib.request, urllib.parse, urllib.error</code>
<tr><th>②
<td><code class=pp>import <dfn>urllib2</dfn></code>
<td><code class=pp>import urllib.request, urllib.error</code>
<tr><th>③
<td><code class=pp>import <dfn>urlparse</dfn></code>
<td><code class=pp>import urllib.parse</code>
<tr><th>④
<td><code class=pp>import <dfn>robotparser</dfn></code>
<td><code class=pp>import urllib.robotparser</code>
<tr><th>⑤
<td><pre class=pp><code>from urllib import <dfn>FancyURLopener</dfn>
from urllib import urlencode</code></pre>
<td><pre class=pp><code>from urllib.request import FancyURLopener
from urllib.parse import urlencode</code></pre>
<tr><th>⑥
<td><pre class=pp><code>from urllib2 import <dfn>Request</dfn>
from urllib2 import <dfn>HTTPError</dfn></code></pre>
<td><pre class=pp><code>from urllib.request import Request
from urllib.error import HTTPError</code></pre>
</table>
<ol>
<li>The old <code>urllib</code> module in Python 2 had a variety of functions, including <code>urlopen()</code> for fetching data and <code>splittype()</code>, <code>splithost()</code>, and <code>splituser()</code> for splitting a <abbr>URL</abbr> into its constituent parts. These functions have been reorganized more logically within the new <code>urllib</code> package. <code>2to3</code> will also change all calls to these functions so they use the new naming scheme.
<li>The old <code>urllib2</code> module in Python 2 has been folded into the <code>urllib</code> package in Python 3. All your <code>urllib2</code> favorites — the <code>build_opener()</code> method, <code>Request</code> objects, and <code>HTTPBasicAuthHandler</code> and friends — are still available.
<li>The <code>urllib.parse</code> module in Python 3 contains all the parsing functions from the old <code>urlparse</code> module in Python 2.
<li>The <code>urllib.robotparser</code> module parses <a href=http://www.robotstxt.org/><code>robots.txt</code> files</a>.
<li>The <code>FancyURLopener</code> class, which handles <abbr>HTTP</abbr> redirects and other status codes, is still available in the new <code>urllib.request</code> module. The <code>urlencode()</code> function has moved to <code>urllib.parse</code>.
<li>The <code>Request</code> object is still available in <code>urllib.request</code>, but constants like <code>HTTPError</code> have been moved to <code>urllib.error</code>.
</ol>
<p>Did I mention that <code>2to3</code> will rewrite your function calls too? For example, if your Python 2 code imports the <code>urllib</code> module and calls <code>urllib.urlopen()</code> to fetch data, <code>2to3</code> will fix both the import statement and the function call.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>
<td><pre class=pp><code>import urllib
print urllib.urlopen('http://diveintopython3.org/').read()</code></pre>
<td><pre class=pp><code>import urllib.request, urllib.parse, urllib.error
print(urllib.request.urlopen('http://diveintopython3.org/').read())</code></pre>
</table>
<h3 id=dbm><code>dbm</code></h3>
<p>All the various <abbr>DBM</abbr> clones are now in a single package, <code>dbm</code>. If you need a specific variant like <abbr>GNU</abbr> <abbr>DBM</abbr>, you can import the appropriate module within the <code>dbm</code> package.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>
<td><code class=pp>import <dfn>dbm</dfn></code>
<td><code class=pp>import dbm.ndbm</code>
<tr><th>
<td><code class=pp>import <dfn>gdbm</dfn></code>
<td><code class=pp>import dbm.gnu</code>
<tr><th>
<td><code class=pp>import <dfn>dbhash</dfn></code>
<td><code class=pp>import dbm.bsd</code>
<tr><th>
<td><code class=pp>import <dfn>dumbdbm</dfn></code>
<td><code class=pp>import dbm.dumb</code>
<tr><th>
<td><pre class=pp><code>import <dfn>anydbm</dfn>
import whichdb</code></pre>
<td><code class=pp>import dbm</code>
</table>
<h3 id=xmlrpc><code>xmlrpc</code></h3>
<p><abbr>XML-RPC</abbr> is a lightweight method of performing remote <abbr>RPC</abbr> calls over <abbr>HTTP</abbr>. The <abbr>XML-RPC</abbr> client library and several <abbr>XML-RPC</abbr> server implementations are now combined in a single package, <code>xmlrpc</code>.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>
<td><code class=pp>import <dfn>xmlrpclib</dfn></code>
<td><code class=pp>import xmlrpc.client</code>
<tr><th>
<td><pre class=pp><code>import <dfn>DocXMLRPCServer</dfn>
import <dfn>SimpleXMLRPCServer</dfn></code></pre>
<td><code class=pp>import xmlrpc.server</code>
</table>
<h3 id=othermodules>Other modules</h3>
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><pre class=pp><code>try:
import <dfn>cStringIO</dfn> as <dfn>StringIO</dfn>
except ImportError:
import StringIO</code></pre>
<td><code class=pp>import io</code>
<tr><th>②
<td><pre class=pp><code>try:
import cPickle as pickle
except ImportError:
import pickle</code></pre>
<td><code class=pp>import pickle</code>
<tr><th>③
<td><code class=pp>import <dfn>__builtin__</dfn></code>
<td><code class=pp>import builtins</code>
<tr><th>④
<td><code class=pp>import <dfn>copy_reg</dfn></code>
<td><code class=pp>import copyreg</code>
<tr><th>⑤
<td><code class=pp>import <dfn>Queue</dfn></code>
<td><code class=pp>import queue</code>
<tr><th>⑥
<td><code class=pp>import <dfn>SocketServer</dfn></code>
<td><code class=pp>import socketserver</code>
<tr><th>⑦
<td><code class=pp>import <dfn>ConfigParser</dfn></code>
<td><code class=pp>import configparser</code>
<tr><th>⑧
<td><code class=pp>import repr</code>
<td><code class=pp>import reprlib</code>
<tr><th>⑨
<td><code class=pp>import <dfn>commands</dfn></code>
<td><code class=pp>import subprocess</code>
</table>
<ol>
<li>A common idiom in Python 2 was to try to import <code>cStringIO as StringIO</code>, and if that failed, to import <code>StringIO</code> instead. Do not do this in Python 3; the <code>io</code> module does it for you. It will find the fastest implementation available and use it automatically.
<li>A similar idiom was used to import the fastest pickle implementation. Do not do this in Python 3; the <code>pickle</code> module does it for you.
<li>The <code>builtins</code> module contains the global functions, classes, and constants used throughout the Python language. Redefining a function in the <code>builtins</code> module will redefine the global function everywhere. That is exactly as powerful and scary as it sounds.
<li>The <code>copyreg</code> module adds pickle support for custom types defined in C.
<li>The <code>queue</code> module implements a multi-producer, multi-consumer queue.
<li>The <code>socketserver</code> module provides generic base classes for implementing different kinds of socket servers.
<li>The <code>configparser</code> module parses <abbr>INI</abbr>-style configuration files.
<li>The <code>reprlib</code> module reimplements the built-in <code>repr()</code> function, with additional controls on how long the representations can be before they are truncated.
<li>The <code>subprocess</code> module allows you to spawn processes, connect to their pipes, and obtain their return codes.
</ol>
<h2 id=import>Relative imports within a package</h2>
<p>A package is a group of related modules that function as a single entity. In Python 2, when modules within a package need to reference each other, you use <code>import foo</code> or <code>from foo import Bar</code>. The Python 2 interpreter first searches within the current package to find <code>foo.py</code>, and then moves on to the other directories in the Python search path (<code>sys.path</code>). Python 3 works a bit differently. Instead of searching the current package, it goes directly to the Python search path. If you want one module within a package to import another module in the same package, you need to explicitly provide the relative path between the two modules.
<p>Suppose you had this package, with multiple files in the same directory:
<pre>chardet/
|
+--__init__.py
|
+--constants.py
|
+--mbcharsetprober.py
|
+--universaldetector.py</pre>
<p>Now suppose that <code>universaldetector.py</code> needs to import the entire <code>constants.py</code> file and one class from <code>mbcharsetprober.py</code>. How do you do it?
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>import constants</code>
<td><code class=pp>from . import constants</code>
<tr><th>②
<td><code class=pp>from mbcharsetprober import MultiByteCharSetProber</code>
<td><code class=pp>from .mbcharsetprober import MultiByteCharsetProber</code>
</table>
<ol>
<li>When you need to import an entire module from elsewhere in your package, use the new <code>from . import</code> syntax. The period is actually a relative path from this file (<code>universaldetector.py</code>) to the file you want to import (<code>constants.py</code>). In this case, they are in the same directory, thus the single period. You can also import from the parent directory (<code>from .. import anothermodule</code>) or a subdirectory.
<li>To import a specific class or function from another module directly into your module’s namespace, prefix the target module with a relative path, minus the trailing slash. In this case, <code>mbcharsetprober.py</code> is in the same directory as <code>universaldetector.py</code>, so the path is a single period. You can also import form the parent directory (<code>from ..anothermodule import AnotherClass</code>) or a subdirectory.
</ol>
<h2 id=next><code>next()</code> iterator method</h2>
<p>In Python 2, iterators had a <code><dfn>next</dfn>()</code> method which returned the next item in the sequence. That’s still true in Python 3, but there is now also <a href=generators.html#generators>a global <code>next()</code> function</a> that takes an iterator as an argument.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>anIterator.next()</code>
<td><code class=pp>next(anIterator)</code>
<tr><th>②
<td><code class=pp>a_function_that_returns_an_iterator().next()</code>
<td><code class=pp>next(a_function_that_returns_an_iterator())</code>
<tr><th>③
<td><pre class=pp><code>class A:
def next(self):
pass</code></pre>
<td><pre class=pp><code>class A:
def __next__(self):
pass</code></pre>
<tr><th>④
<td><pre class=pp><code>class A:
def next(self, x, y):
pass</code></pre>
<td><i>no change</i>
<tr><th>⑤
<td><pre class=pp><code>next = 42
for an_iterator in a_sequence_of_iterators:
an_iterator.next()</code></pre>
<td><pre class=pp><code>next = 42
for an_iterator in a_sequence_of_iterators:
an_iterator.__next__()</code></pre>
</table>
<ol>
<li>In the simplest case, instead of calling an iterator’s <code>next()</code> method, you now pass the iterator itself to the global <code>next()</code> function.
<li>If you have a function that returns an iterator, call the function and pass the result to the <code>next()</code> function. (The <code>2to3</code> script is smart enough to convert this properly.)
<li>If you define your own class and mean to use it as an iterator, define the <code>__next__()</code> special method.
<li>If you define your own class and just happen to have a method named <code>next()</code> that takes one or more arguments, <code>2to3</code> will not touch it. This class can not be used as an iterator, because its <code>next()</code> method takes arguments.
<li>This one is a bit tricky. If you have a local variable named <var>next</var>, then it takes precedence over the new global <code>next()</code> function. In this case, you need to call the iterator’s special <code>__next__()</code> method to get the next item in the sequence. (Alternatively, you could also refactor the code so the local variable wasn’t named <var>next</var>, but <code>2to3</code> will not do that for you automatically.)
</ol>
<h2 id=filter><code>filter()</code> global function</h2>
<p>In Python 2, the <code><dfn>filter</dfn>()</code> function returned a list, the result of filtering a sequence through a function that returned <code>True</code> or <code>False</code> for each item in the sequence. In Python 3, the <code>filter()</code> function returns an iterator, not a list.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>filter(a_function, a_sequence)</code>
<td><code class=pp>list(filter(a_function, a_sequence))</code>
<tr><th>②
<td><code class=pp>list(filter(a_function, a_sequence))</code>
<td><i>no change</i>
<tr><th>③
<td><code class=pp>filter(None, a_sequence)</code>
<td><code class=pp>[i for i in a_sequence if i]</code>
<tr><th>④
<td><code class=pp>for i in filter(None, a_sequence):</code>
<td><i>no change</i>
<tr><th>⑤
<td><code class=pp>[i for i in filter(a_function, a_sequence)]</code>
<td><i>no change</i>
</table>
<ol>
<li>In the most basic case, <code>2to3</code> will wrap a call to <code>filter()</code> with a call to <code>list()</code>, which simply iterates through its argument and returns a real list.
<li>However, if the call to <code>filter()</code> is <em>already</em> wrapped in <code>list()</code>, <code>2to3</code> will do nothing, since the fact that <code>filter()</code> is returning an iterator is irrelevant.
<li>For the special syntax of <code>filter(None, ...)</code>, <code>2to3</code> will transform the call into a semantically equivalent list comprehension.
<li>In contexts like <code>for</code> loops, which iterate through the entire sequence anyway, no changes are necessary.
<li>Again, no changes are necessary, because the list comprehension will iterate through the entire sequence, and it can do that just as well if <code>filter()</code> returns an iterator as if it returns a list.
</ol>
<h2 id=map><code>map()</code> global function</h2>
<p>In much the same way as <a href=#filter><code>filter()</code></a>, the <code><dfn>map</dfn>()</code> function now returns an iterator. (In Python 2, it returned a list.)
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>map(a_function, 'PapayaWhip')</code>
<td><code class=pp>list(map(a_function, 'PapayaWhip'))</code>
<tr><th>②
<td><code class=pp>map(None, 'PapayaWhip')</code>
<td><code class=pp>list('PapayaWhip')</code>
<tr><th>③
<td><code class=pp>map(lambda x: x+1, range(42))</code>
<td><code class=pp>[x+1 for x in range(42)]</code>
<tr><th>④
<td><code class=pp>for i in map(a_function, a_sequence):</code>
<td><i>no change</i>
<tr><th>⑤
<td><code class=pp>[i for i in map(a_function, a_sequence)]</code>
<td><i>no change</i>
</table>
<ol>
<li>As with <code>filter()</code>, in the most basic case, <code>2to3</code> will wrap a call to <code>map()</code> with a call to <code>list()</code>.
<li>For the special syntax of <code>map(None, ...)</code>, the identity function, <code>2to3</code> will convert it to an equivalent call to <code>list()</code>.
<li>If the first argument to <code>map()</code> is a lambda function, <code>2to3</code> will convert it to an equivalent list comprehension.
<li>In contexts like <code>for</code> loops, which iterate through the entire sequence anyway, no changes are necessary.
<li>Again, no changes are necessary, because the list comprehension will iterate through the entire sequence, and it can do that just as well if <code>map()</code> returns an iterator as if it returns a list.
</ol>
<h2 id=reduce><code>reduce()</code> global function</h2>
<p>In Python 3, the <code><dfn>reduce</dfn>()</code> function has been removed from the global namespace and placed in the <code>functools</code> module.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>
<td><code class=pp>reduce(a, b, c)</code>
<td><pre class=pp><code>from functools import reduce
reduce(a, b, c)</code></pre>
</table>
<h2 id=apply><code>apply()</code> global function</h2>
<p>Python 2 had a global function called <code><dfn>apply</dfn>()</code>, which took a function <var>f</var> and a list <code>[a, b, c]</code> and returned <code>f(a, b, c)</code>. You can accomplish the same thing by calling the function directly and passing it the list of arguments preceded by an asterisk. In Python 3, the <code>apply()</code> function no longer exists; you must use the asterisk notation.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>apply(a_function, a_list_of_args)</code>
<td><code class=pp>a_function(*a_list_of_args)</code>
<tr><th>②
<td><code class=pp>apply(a_function, a_list_of_args, a_dictionary_of_named_args)</code>
<td><code class=pp>a_function(*a_list_of_args, **a_dictionary_of_named_args)</code>
<tr><th>③
<td><code class=pp>apply(a_function, a_list_of_args + z)</code>
<td><code class=pp>a_function(*a_list_of_args + z)</code>
<tr><th>④
<td><code class=pp>apply(aModule.a_function, a_list_of_args)</code>
<td><code class=pp>aModule.a_function(*a_list_of_args)</code>
</table>
<ol>
<li>In the simplest form, you can call a function with a list of arguments (an actual list like <code>[a, b, c]</code>) by prepending the list with an asterisk (<code>*</code>). This is exactly equivalent to the old <code>apply()</code> function in Python 2.
<li>In Python 2, the <code>apply()</code> function could actually take three parameters: a function, a list of arguments, and a dictionary of named arguments. In Python 3, you can accomplish the same thing by prepending the list of arguments with an asterisk (<code>*</code>) and the dictionary of named arguments with two asterisks (<code>**</code>).
<li>The <code>+</code> operator, used here for list concatenation, takes precedence over the <code>*</code> operator, so there is no need for extra parentheses around <code>a_list_of_args + z</code>.
<li>The <code>2to3</code> script is smart enough to convert complex <code>apply()</code> calls, including calling functions within imported modules.
</ol>
<h2 id=intern><code>intern()</code> global function</h2>
<p>In Python 2, you could call the <code><dfn>intern</dfn>()</code> function on a string to intern it as a performance optimization. In Python 3, the <code>intern()</code> function has been moved to the <code>sys</code> module.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>
<td><code class=pp>intern(aString)</code>
<td><code class=pp>sys.intern(aString)</code>
</table>
<h2 id=exec><code>exec</code> statement</h2>
<p>Just as <a href=#print>the <code>print</code> statement</a> became a function in Python 3, so too has the <code><dfn>exec</dfn></code> statement. The <code>exec()</code> function takes a string which contains arbitrary Python code and executes it as if it were just another statement or expression. <code>exec()</code> is like <a href=advanced-iterators.html#eval><code>eval()</code></a>, but even more powerful and evil. The <code>eval()</code> function can only evaluate a single expression, but <code>exec()</code> can execute multiple statements, imports, function declarations — essentially an entire Python program in a string.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>exec codeString</code>
<td><code class=pp>exec(codeString)</code>
<tr><th>②
<td><code class=pp>exec codeString in a_global_namespace</code>
<td><code class=pp>exec(codeString, a_global_namespace)</code>
<tr><th>③
<td><code class=pp>exec codeString in a_global_namespace, a_local_namespace</code>
<td><code class=pp>exec(codeString, a_global_namespace, a_local_namespace)</code>
</table>
<ol>
<li>In the simplest form, the <code>2to3</code> script simply encloses the code-as-a-string in parentheses, since <code>exec()</code> is now a function instead of a statement.
<li>The old <code>exec</code> statement could take a namespace, a private environment of globals in which the code-as-a-string would be executed. Python 3 can also do this; just pass the namespace as the second argument to the <code>exec()</code> function.
<li>Even fancier, the old <code>exec</code> statement could also take a local namespace (like the variables defined within a function). In Python 3, the <code>exec()</code> function can do that too.
</ol>
<h2 id=execfile><code>execfile</code> statement</h2>
<p>Like the old <a href=#exec><code>exec</code> statement</a>, the old <code>execfile</code> statement will execute strings as if they were Python code. Where <code>exec</code> took a string, <code>execfile</code> took a filename. In Python 3, the <code>execfile</code> statement has been eliminated. If you really need to take a file of Python code and execute it (but you’re not willing to simply import it), you can accomplish the same thing by opening the file, reading its contents, calling the global <code>compile()</code> function to force the Python interpreter to compile the code, and then call the new <code>exec()</code> function.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>
<td><code class=pp><dfn>execfile</dfn>('a_filename')</code>
<td><code class=pp>exec(compile(open('a_filename').read(), 'a_filename', 'exec'))</code>
</table>
<h2 id=repr><code>repr</code> literals (backticks)</h2>
<p>In Python 2, there was a special syntax of wrapping any object in <dfn>backticks</dfn> (like <code>`x`</code>) to get a representation of the object. In Python 3, this capability still exists, but you can no longer use backticks to get it. Instead, use the global <code>repr()</code> function.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>`x`</code>
<td><code class=pp>repr(x)</code>
<tr><th>②
<td><code class=pp>`'PapayaWhip' + `2``</code>
<td><code class=pp>repr('PapayaWhip' + repr(2))</code>
</table>
<ol>
<li>Remember, <var>x</var> can be anything — a class, a function, a module, a primitive data type, <i class=baa>&</i>c. The <code>repr()</code> function works on everything.
<li>In Python 2, backticks could be nested, leading to this sort of confusing (but valid) expression. The <code>2to3</code> tool is smart enough to convert this into nested calls to <code>repr()</code>.
</ol>
<h2 id=except><code>try...except</code> statement</h2>
<p>The syntax for <a href=your-first-python-program.html#exceptions>catching <dfn>exceptions</dfn></a> has changed slightly between Python 2 and Python 3.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><pre class=pp><code>try:
import mymodule
<dfn>except</dfn> ImportError, e
pass</code></pre>
<td><pre class=pp><code>try:
import mymodule
except ImportError as e:
pass</code></pre>
<tr><th>②
<td><pre class=pp><code>try:
import mymodule
except (RuntimeError, ImportError), e
pass</code></pre>
<td><pre class=pp><code>try:
import mymodule
except (RuntimeError, ImportError) as e:
pass</code></pre>
<tr><th>③
<td><pre class=pp><code>try:
import mymodule
except ImportError:
pass</code></pre>
<td><i>no change</i>
<tr><th>④
<td><pre class=pp><code>try:
import mymodule
except:
pass</code></pre>
<td><i>no change</i>
</table>
<ol>
<li>Instead of a comma after the exception type, Python 3 uses a new keyword, <code>as</code>.
<li>The <code>as</code> keyword also works for catching multiple types of exceptions at once.
<li>If you catch an exception but don’t actually care about accessing the <dfn>exception</dfn> object itself, the syntax is identical between Python 2 and Python 3.
<li>Similarly, if you use a fallback to catch <em>all</em> exceptions, the syntax is identical.
</ol>
<blockquote class=note>
<p><span class=u>☞</span>You should never use a fallback to catch <em>all</em> exceptions when importing modules (or most other times). Doing so will catch things like <code>KeyboardInterrupt</code> (if the user pressed <kbd>Ctrl-C</kbd> to interrupt the program) and can make it more difficult to debug errors.
</blockquote>
<h2 id=raise><code>raise</code> statement</h2>
<p>The syntax for <a href=your-first-python-program.html#exceptions>raising your own exceptions</a> has changed slightly between Python 2 and Python 3.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp><dfn>raise</dfn> MyException</code>
<td><i>unchanged</i>
<tr><th>②
<td><code class=pp>raise MyException, 'error message'</code>
<td><code class=pp>raise MyException('error message')</code>
<tr><th>③
<td><code class=pp>raise MyException, 'error message', a_traceback</code>
<td><code class=pp>raise MyException('error message').with_traceback(a_traceback)</code>
<tr><th>④
<td><code class=pp>raise 'error message'</code>
<td><i>unsupported</i>
</table>
<ol>
<li>In the simplest form, raising an exception without a custom error message, the syntax is unchanged.
<li>The change becomes noticeable when you want to raise an exception with a custom error message. Python 2 separated the exception class and the message with a comma; Python 3 passes the error message as a parameter.
<li>Python 2 supported a more complex syntax to raise an exception with a custom traceback (stack trace). You can do this in Python 3 as well, but the syntax is quite different.
<li>In Python 2, you could raise an exception with no exception class, just an error message. In Python 3, this is no longer possible. <code>2to3</code> will warn you that it was unable to fix this automatically.
</ol>
<h2 id=throw><code>throw</code> method on generators</h2>
<p>In Python 2, generators have a <code><dfn>throw</dfn>()</code> method. Calling <code>a_generator.throw()</code> raises an exception at the point where the generator was paused, then returns the next value yielded by the generator function. In Python 3, this functionality is still available, but the syntax is slightly different.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>a_generator.throw(MyException)</code>
<td><i>no change</i>
<tr><th>②
<td><code class=pp>a_generator.throw(MyException, 'error message')</code>
<td><code class=pp>a_generator.throw(MyException('error message'))</code>
<tr><th>③
<td><code class=pp>a_generator.throw('error message')</code>
<td><i>unsupported</i>
</table>
<ol>
<li>In the simplest form, a generator throws an exception without a custom error message. In this case, the syntax has not changed between Python 2 and Python 3.
<li>If the generator throws an exception <em>with</em> a custom error message, you need to pass the error string to the exception when you create it.
<li>Python 2 also supported throwing an exception with <em>only</em> a custom error message. Python 3 does not support this, and the <code>2to3</code> script will display a warning telling you that you will need to fix this code manually.
</ol>
<h2 id=xrange><code>xrange()</code> global function</h2>
<p>In Python 2, there were two ways to get a range of numbers: <code><dfn>range</dfn>()</code>, which returned a list, and <code><dfn>xrange</dfn>()</code>, which returned an iterator. In Python 3, <code>range()</code> returns an iterator, and <code>xrange()</code> doesn’t exist.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>xrange(10)</code>
<td><code class=pp>range(10)</code>
<tr><th>②
<td><code class=pp>a_list = range(10)</code>
<td><code class=pp>a_list = list(range(10))</code>
<tr><th>③
<td><code class=pp>[i for i in xrange(10)]</code>
<td><code class=pp>[i for i in range(10)]</code>
<tr><th>④
<td><code class=pp>for i in range(10):</code>
<td><i>no change</i>
<tr><th>⑤
<td><code class=pp>sum(range(10))</code>
<td><i>no change</i>
</table>
<ol>
<li>In the simplest case, the <code>2to3</code> script will simply convert <code>xrange()</code> to <code>range()</code>.
<li>If your Python 2 code used <code>range()</code>, the <code>2to3</code> script does not know whether you needed a list, or whether an iterator would do. It errs on the side of caution and coerces the return value into a list by calling the <code>list()</code> function.
<li>If the <code>xrange()</code> function was inside a list comprehension, the <code>2to3</code> script is clever enough <em>not</em> to wrap the <code>range()</code> function with a call to <code>list()</code>. The list comprehension will work just fine with the iterator that the <code>range()</code> function returns.
<li>Similarly, a <code>for</code> loop will work just fine with an iterator, so there is no need to change anything here.
<li>The <code>sum()</code> function will also work with an iterator, so <code>2to3</code> makes no changes here either. Like <a href=#dict>dictionary methods that return views instead of lists</a>, this applies to <code>min()</code>, <code>max()</code>, <code>sum()</code>, <code>list()</code>, <code>tuple()</code>, <code>set()</code>, <code>sorted()</code>, <code>any()</code>, and <code>all()</code>.
</ol>
<h2 id=raw_input><code>raw_input()</code> and <code>input()</code> global functions</h2>
<p>Python 2 had two global functions for asking the user for input on the command line. The first, called <code>input()</code>, expected the user to enter a Python expression (and returned the result). The second, called <code><dfn>raw_input</dfn>()</code>, just returned whatever the user typed. This was wildly confusing for beginners and widely regarded as a “wart” in the language. Python 3 excises this wart by renaming <code>raw_input()</code> to <code>input()</code>, so it works the way everyone naively expects it to work.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>raw_input()</code>
<td><code class=pp>input()</code>
<tr><th>②
<td><code class=pp>raw_input('prompt')</code>
<td><code class=pp>input('prompt')</code>
<tr><th>③
<td><code class=pp>input()</code>
<td><code class=pp>eval(input())</code>
</table>
<ol>
<li>In the simplest form, <code>raw_input()</code> becomes <code>input()</code>.
<li>In Python 2, the <code>raw_input()</code> function could take a prompt as a parameter. This has been retained in Python 3.
<li>If you actually need to ask the user for a Python expression to evaluate, use the <code>input()</code> function and pass the result to <code>eval()</code>.
</ol>
<h2 id=funcattrs><code>func_*</code> function attributes</h2>
<p>In Python 2, code within functions can access special attributes about the function itself. In Python 3, these special function attributes have been renamed for consistency with other attributes.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>a_function.<dfn>func_name</dfn></code>
<td><code class=pp>a_function.__name__</code>
<tr><th>②
<td><code class=pp>a_function.<dfn>func_doc</dfn></code>
<td><code class=pp>a_function.__doc__</code>
<tr><th>③
<td><code class=pp>a_function.<dfn>func_defaults</dfn></code>
<td><code class=pp>a_function.__defaults__</code>
<tr><th>④
<td><code class=pp>a_function.<dfn>func_dict</dfn></code>
<td><code class=pp>a_function.__dict__</code>
<tr><th>⑤
<td><code class=pp>a_function.<dfn>func_closure</dfn></code>
<td><code class=pp>a_function.__closure__</code>
<tr><th>⑥
<td><code class=pp>a_function.<dfn>func_globals</dfn></code>
<td><code class=pp>a_function.__globals__</code>
<tr><th>⑦
<td><code class=pp>a_function.<dfn>func_code</dfn></code>
<td><code class=pp>a_function.__code__</code>
</table>
<ol>
<li>The <code>__name__</code> attribute (previously <code>func_name</code>) contains the function’s name.
<li>The <code>__doc__</code> attribute (previously <code>func_doc</code>) contains the <i>docstring</i> that you defined in the function’s source code.
<li>The <code>__defaults__</code> attribute (previously <code>func_defaults</code>) is a tuple containing default argument values for those arguments that have default values.
<li>The <code>__dict__</code> attribute (previously <code>func_dict</code>) is the namespace supporting arbitrary function attributes.
<li>The <code>__closure__</code> attribute (previously <code>func_closure</code>) is a tuple of cells that contain bindings for the function’s free variables.
<li>The <code>__globals__</code> attribute (previously <code>func_globals</code>) is a reference to the global namespace of the module in which the function was defined.
<li>The <code>__code__</code> attribute (previously <code>func_code</code>) is a code object representing the compiled function body.
</ol>
<h2 id=xreadlines><code>xreadlines()</code> I/O method</h2>
<p>In Python 2, file objects had an <code><dfn>xreadlines</dfn>()</code> method which returned an iterator that would read the file one line at a time. This was useful in <code>for</code> loops, among other places. In fact, it was so useful, later versions of Python 2 added the capability to file objects themselves.
<p>In Python 3, the <code>xreadlines()</code> method no longer exists. <code>2to3</code> can fix the simple cases, but some edge cases will require manual intervention.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>for line in a_file.xreadlines():</code>
<td><code class=pp>for line in a_file:</code>
<tr><th>②
<td><code class=pp>for line in a_file.xreadlines(5):</code>
<td><i>no change (broken)</i>
</table>
<ol>
<li>If you used to call <code>xreadlines()</code> with no arguments, <code>2to3</code> will convert it to just the file object. In Python 3, this will accomplish the same thing: read the file one line at a time and execute the body of the <code>for</code> loop.
<li>If you used to call <code>xreadlines()</code> with an argument (the number of lines to read at a time), <code>2to3</code> will not fix it, and your code will fail with an <code>AttributeError: '_io.TextIOWrapper' object has no attribute 'xreadlines'</code>. You can manually change <code>xreadlines()</code> to <code>readlines()</code> to get it to work in Python 3. (The <code>readlines()</code> method now returns an iterator, so it is just as efficient as <code>xreadlines()</code> was in Python 2.)
</ol>
<p class=c><span style='font-size:56px;line-height:0.88'>☃</span>
<h2 id=tuple_params><code>lambda</code> functions that take a tuple instead of multiple parameters</h2>
<p>In Python 2, you could define anonymous <code><dfn>lambda</dfn></code> functions which took multiple parameters by defining the function as taking a tuple with a specific number of items. In effect, Python 2 would “unpack” the tuple into named arguments, which you could then reference (by name) within the <code>lambda</code> function. In Python 3, you can still pass a tuple to a <code>lambda</code> function, but the Python interpreter will not unpack the tuple into named arguments. Instead, you will need to reference each argument by its positional index.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>lambda (x,): x + f(x)</code>
<td><code class=pp>lambda x1: x1[0] + f(x1[0])</code>
<tr><th>②
<td><code class=pp>lambda (x, y): x + f(y)</code>
<td><code class=pp>lambda x_y: x_y[0] + f(x_y[1])</code>
<tr><th>③
<td><code class=pp>lambda (x, (y, z)): x + y + z</code>
<td><code class=pp>lambda x_y_z: x_y_z[0] + x_y_z[1][0] + x_y_z[1][1]</code>
<tr><th>④
<td><code class=pp>lambda x, y, z: x + y + z</code>
<td><i>unchanged</i>
</table>
<ol>
<li>If you had defined a <code>lambda</code> function that took a tuple of one item, in Python 3 that would become a <code>lambda</code> with references to <var>x1[0]</var>. The name <var>x1</var> is autogenerated by the <code>2to3</code> script, based on the named arguments in the original tuple.
<li>A <code>lambda</code> function with a two-item tuple <var>(x, y)</var> gets converted to <var>x_y</var> with positional arguments <var>x_y[0]</var> and <var>x_y[1]</var>.
<li>The <code>2to3</code> script can even handle <code>lambda</code> functions with nested tuples of named arguments. The resulting Python 3 code is a bit unreadable, but it works the same as the old code did in Python 2.
<li>You can define <code>lambda</code> functions that take multiple arguments. Without parentheses around the arguments, Python 2 just treats it as a <code>lambda</code> function with multiple arguments; within the <code>lambda</code> function, you simply reference the arguments by name, just like any other function. This syntax still works in Python 3.
</ol>
<h2 id=methodattrs>Special method attributes</h2>
<p>In Python 2, class methods can reference the class object in which they are defined, as well as the method object itself. <code>im_self</code> is the class instance object; <code>im_func</code> is the function object; <code>im_class</code> is the class of <code>im_self</code>. In Python 3, these special method attributes have been renamed to follow the naming conventions of other attributes.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>
<td><code class=pp>aClassInstance.aClassMethod.<dfn>im_func</dfn></code>
<td><code class=pp>aClassInstance.aClassMethod.__func__</code>
<tr><th>
<td><code class=pp>aClassInstance.aClassMethod.<dfn>im_self</dfn></code>
<td><code class=pp>aClassInstance.aClassMethod.__self__</code>
<tr><th>
<td><code class=pp>aClassInstance.aClassMethod.<dfn>im_class</dfn></code>
<td><code class=pp>aClassInstance.aClassMethod.__self__.__class__</code>
</table>
<h2 id=nonzero><code>__nonzero__</code> special method</h2>
<p>In Python 2, you could build your own classes that could be used in a boolean context. For example, you could instantiate the class and then use the instance in an <code>if</code> statement. To do this, you defined a special <code>__nonzero__()</code> method which returned <code>True</code> or <code>False</code>, and it was called whenever the instance was used in a boolean context. In Python 3, you can still do this, but the name of the method has changed to <code>__bool__()</code>.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><pre class=pp><code>class A:
def <dfn>__nonzero__</dfn>(self):
pass</code></pre>
<td><pre class=pp><code>class A:
def <dfn>__bool__</dfn>(self):
pass</code></pre>
<tr><th>②
<td><pre class=pp><code>class A:
def __nonzero__(self, x, y):
pass</code></pre>
<td><i>no change</i>
</table>
<ol>
<li>Instead of <code>__nonzero__()</code>, Python 3 calls the <code>__bool__()</code> method when evaluating an instance in a boolean context.
<li>However, if you have a <code>__nonzero__()</code> method that takes arguments, the <code>2to3</code> tool will assume that you were using it for some other purpose, and it will not make any changes.
</ol>
<h2 id=numliterals>Octal literals</h2>
<p>The syntax for defining base 8 (<dfn>octal</dfn>) numbers has changed slightly between Python 2 and Python 3.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>
<td><code class=pp>x = 0755</code>
<td><code class=pp>x = 0o755</code>
</table>
<h2 id=renames><code>sys.maxint</code></h2>
<p>Due to the <a href=#long>integration of the <code>long</code> and <code>int</code> types</a>, the <code>sys.maxint</code> constant is no longer accurate. Because the value may still be useful in determining platform-specific capabilities, it has been retained but renamed as <code>sys.maxsize</code>.
<table>
<tr><th>Notes
<th>Python 2
<th>Python 3
<tr><th>①
<td><code class=pp>from sys import <dfn>maxint</dfn></code>
<td><code class=pp>from sys import <dfn>maxsize</dfn></code>
<tr><th>②
<td><code class=pp>a_function(<dfn>sys.maxint</dfn>)</code>
<td><code class=pp>a_function(<dfn>sys.maxsize</dfn>)</code>
</table>
<ol>
<li><code>maxint</code> becomes <code>maxsize</code>.
<li>Any usage of <code>sys.maxint</code> becomes <code>sys.maxsize</code>.
</ol>
<h2 id=callable><code>callable()</code> global function</h2>
<p>In Python 2, you could check whether an object was callable (like a function) with the global <code><dfn>callable</dfn>()</code> function. In Python 3, this global function has been eliminated. To check whether an object is callable, check for the existence of the <code>__call__()</code> special method.
<table>
<tr><th>Notes
<th>Python 2