Skip to contentSkip to conversation

check-package: fix Python3 support

PATCH Superseded Ricardo Martincoski 4 participants 8 messages Jul 23, 2018 – Jan 28, 2019

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 check-package: fix Python3 support

This script currently uses "/usr/bin/env python" as shebang but it does not really support Python3. Instead of limiting the script to Python2, fix it to support both versions.

Do this by first using an automated tool:

$ python -m modernize -w -p <files>
and then manually fixing what remains:
 - keep the imports in alphabetical order;
 - fix handling of files that cannot be decoded by 'utf-8' codec, such
   as package/urg/0002-urg-gcc6-fix-narrowing-conversion.patch .

In order to avoid errors when decoding files, use errors="surrogateescape" when opening files, the docs for open() states:
"This is useful for processing files in an unknown encoding.". This argument is not compatible with Python2 open() so import 'six' to use it only when running in Python3.
As a consequence the file handler becomes explicit, so use it to close() the file after it got processed.

---
Using the default docker (debian) image which uses Python2:
before this patch:
https://gitlab.com/buildroot.org/buildroot/-/jobs/83454639
after this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/83542815

Using an arch docker image which uses Python3:
before this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/81892607
after this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/83543363
The arch image for docker was generated using:
http://patchwork.ozlabs.org/patch/943301/

---
 utils/check-package                 |  9 ++++++++-
 utils/checkpackagelib/lib.py        |  3 ++-
 utils/checkpackagelib/lib_config.py | 13 +++++++------
 utils/checkpackagelib/lib_hash.py   | 13 +++++++------
 utils/checkpackagelib/lib_mk.py     | 11 ++++++-----
 utils/checkpackagelib/lib_patch.py  |  5 +++--
 6 files changed, 33 insertions(+), 21 deletions(-)
utils/check-package +8 -1
index 3dbc28b0a2..a8f5ce6157 100755
@@ -1,11 +1,13 @@
 #!/usr/bin/env python
 # See utils/checkpackagelib/readme.txt before editing this file.
 
+from __future__ import absolute_import
 from __future__ import print_function
 import argparse
 import inspect
 import os
 import re
+import six
 import sys
 
 import checkpackagelib.lib_config
@@ -127,10 +129,15 @@ def check_file_using_lib(fname):
 
     for cf in objects:
         nwarnings += print_warnings(cf.before())
-    for lineno, text in enumerate(open(fname, "r").readlines()):
+    if six.PY3:
+        f = open(fname, "r", errors="surrogateescape")
+    else:
+        f = open(fname, "r")
+    for lineno, text in enumerate(f.readlines()):
         nlines += 1
         for cf in objects:
             nwarnings += print_warnings(cf.check_line(lineno + 1, text))
+    f.close()
     for cf in objects:
         nwarnings += print_warnings(cf.after())
 
utils/checkpackagelib/lib.py +2 -1
index 91f4ad49b7..61fa46125d 100644
@@ -1,6 +1,7 @@
 # See utils/checkpackagelib/readme.txt before editing this file.
 
-from base import _CheckFunction
+from __future__ import absolute_import
+from .base import _CheckFunction
 
 
 class ConsecutiveEmptyLines(_CheckFunction):
utils/checkpackagelib/lib_config.py +7 -6
index 1d273f1c5f..5f316a6d5b 100644
@@ -3,13 +3,14 @@
 # "bool", so below check functions don't need to check for things already
 # checked by running "make menuconfig".
 
+from __future__ import absolute_import
 import re
 
-from base import _CheckFunction
-from lib import ConsecutiveEmptyLines  # noqa: F401
-from lib import EmptyLastLine          # noqa: F401
-from lib import NewlineAtEof           # noqa: F401
-from lib import TrailingSpace          # noqa: F401
+from .base import _CheckFunction
+from .lib import ConsecutiveEmptyLines  # noqa: F401
+from .lib import EmptyLastLine          # noqa: F401
+from .lib import NewlineAtEof           # noqa: F401
+from .lib import TrailingSpace          # noqa: F401
 
 
 def _empty_or_comment(text):
@@ -45,7 +46,7 @@ class AttributesOrder(_CheckFunction):
         if attribute in entries_that_should_not_be_indented:
             self.state = 0
             return
-        if attribute not in self.attributes_order_convention.keys():
+        if attribute not in list(self.attributes_order_convention.keys()):
             return
         new_state = self.attributes_order_convention[attribute]
         wrong_order = self.state > new_state
utils/checkpackagelib/lib_hash.py +7 -6
index 6d4cc9fd62..8ca92a5a90 100644
@@ -3,13 +3,14 @@
 # functions don't need to check for things already checked by running
 # "make package-dirclean package-source".
 
+from __future__ import absolute_import
 import re
 
-from base import _CheckFunction
-from lib import ConsecutiveEmptyLines  # noqa: F401
-from lib import EmptyLastLine          # noqa: F401
-from lib import NewlineAtEof           # noqa: F401
-from lib import TrailingSpace          # noqa: F401
+from .base import _CheckFunction
+from .lib import ConsecutiveEmptyLines  # noqa: F401
+from .lib import EmptyLastLine          # noqa: F401
+from .lib import NewlineAtEof           # noqa: F401
+from .lib import TrailingSpace          # noqa: F401
 
 
 def _empty_line_or_comment(text):
@@ -43,7 +44,7 @@ class HashType(_CheckFunction):
         htype, hexa = fields[:2]
         if htype == "none":
             return
-        if htype not in self.len_of_hash.keys():
+        if htype not in list(self.len_of_hash.keys()):
             return ["{}:{}: unexpected type of hash ({}#adding-packages-hash)"
                     .format(self.filename, lineno, self.url_to_manual),
                     text]
utils/checkpackagelib/lib_mk.py +6 -5
index 86e9aa2d97..9c17a3cfd2 100644
@@ -4,13 +4,14 @@
 # menu options using "make menuconfig" and by running "make" with appropriate
 # packages enabled.
 
+from __future__ import absolute_import
 import re
 
-from base import _CheckFunction
-from lib import ConsecutiveEmptyLines  # noqa: F401
-from lib import EmptyLastLine          # noqa: F401
-from lib import NewlineAtEof           # noqa: F401
-from lib import TrailingSpace          # noqa: F401
+from .base import _CheckFunction
+from .lib import ConsecutiveEmptyLines  # noqa: F401
+from .lib import EmptyLastLine          # noqa: F401
+from .lib import NewlineAtEof           # noqa: F401
+from .lib import TrailingSpace          # noqa: F401
 
 
 class Indent(_CheckFunction):
utils/checkpackagelib/lib_patch.py +3 -2
index 555621afa1..4b0d32268f 100644
@@ -3,10 +3,11 @@
 # functions don't need to check for things already checked by running
 # "make package-dirclean package-patch".
 
+from __future__ import absolute_import
 import re
 
-from base import _CheckFunction
-from lib import NewlineAtEof           # noqa: F401
+from .base import _CheckFunction
+from .lib import NewlineAtEof           # noqa: F401
 
 
 class ApplyOrder(_CheckFunction):
Signature · 1 line
2.17.1
#1 lore raw

PATCH 1/1 check-package: fix Python3 support

Quoted 6 lines · Ricardo Martincoski

This script currently uses "/usr/bin/env python" as shebang but it does not really support Python3. Instead of limiting the script to Python2, fix it to support both versions.

Do this by first using an automated tool:

$ python -m modernize -w -p <files>

In general (if it doesn't break things), I prefer to do the automatic stuff in
a separate patch from the manual stuff. That makes review easier (because for the automatic stuff, you mainly review the automation itself rather than the result).

Quoted 10 lines

and then manually fixing what remains:
- keep the imports in alphabetical order;
- fix handling of files that cannot be decoded by 'utf-8' codec, such as package/urg/0002-urg-gcc6-fix-narrowing-conversion.patch .

In order to avoid errors when decoding files, use errors="surrogateescape" when opening files, the docs for open() states:
"This is useful for processing files in an unknown encoding.". This argument is not compatible with Python2 open() so import 'six' to use it only when running in Python3.

I would prefer to open files with "rb" instead. However, that's a much bigger
change because all checks should then use bytes objects instead of strings.

Quoted 32 lines

As a consequence the file handler becomes explicit, so use it to close() the file after it got processed.

---
Using the default docker (debian) image which uses Python2:
before this patch:
https://gitlab.com/buildroot.org/buildroot/-/jobs/83454639
after this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/83542815

Using an arch docker image which uses Python3:
before this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/81892607
after this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/83543363
The arch image for docker was generated using:
http://patchwork.ozlabs.org/patch/943301/

---
 utils/check-package                 |  9 ++++++++-
 utils/checkpackagelib/lib.py        |  3 ++-
 utils/checkpackagelib/lib_config.py | 13 +++++++------
 utils/checkpackagelib/lib_hash.py   | 13 +++++++------
 utils/checkpackagelib/lib_mk.py     | 11 ++++++-----
 utils/checkpackagelib/lib_patch.py  |  5 +++--
 6 files changed, 33 insertions(+), 21 deletions(-)
utils/check-package +0 -0
index 3dbc28b0a2..a8f5ce6157 100755

utils/check-package line 4

Quoted 4 lines
utils/check-package +1 -0
@@ -1,11 +1,13 @@
 #!/usr/bin/env python
 # See utils/checkpackagelib/readme.txt before editing this file.
 
+from __future__ import absolute_import

This doesn't make sense in a top-level script, does it?

Quoted 31 lines
utils/check-package +7 -1
 from __future__ import print_function
 import argparse
 import inspect
 import os
 import re
+import six
 import sys
 
 import checkpackagelib.lib_config
@@ -127,10 +129,15 @@ def check_file_using_lib(fname):
 
     for cf in objects:
         nwarnings += print_warnings(cf.before())
-    for lineno, text in enumerate(open(fname, "r").readlines()):
+    if six.PY3:
+        f = open(fname, "r", errors="surrogateescape")
+    else:
+        f = open(fname, "r")
+    for lineno, text in enumerate(f.readlines()):
         nlines += 1
         for cf in objects:
             nwarnings += print_warnings(cf.check_line(lineno + 1, text))
+    f.close()
     for cf in objects:
         nwarnings += print_warnings(cf.after())
 
utils/checkpackagelib/lib.py +0 -0
index 91f4ad49b7..61fa46125d 100644

utils/checkpackagelib/lib.py line 4

Quoted 5 lines
utils/checkpackagelib/lib.py +2 -1
@@ -1,6 +1,7 @@
 # See utils/checkpackagelib/readme.txt before editing this file.
 
-from base import _CheckFunction
+from __future__ import absolute_import
+from .base import _CheckFunction

Why is this changed? Will Python3 look for base anywhere else than in the
checkpackagelib directory?

Same below.

Quoted 28 lines
utils/checkpackagelib/lib.py +0 -0
 class ConsecutiveEmptyLines(_CheckFunction):
utils/checkpackagelib/lib_config.py +6 -5
index 1d273f1c5f..5f316a6d5b 100644
@@ -3,13 +3,14 @@
 # "bool", so below check functions don't need to check for things already
 # checked by running "make menuconfig".
 
+from __future__ import absolute_import
 import re
 
-from base import _CheckFunction
-from lib import ConsecutiveEmptyLines  # noqa: F401
-from lib import EmptyLastLine          # noqa: F401
-from lib import NewlineAtEof           # noqa: F401
-from lib import TrailingSpace          # noqa: F401
+from .base import _CheckFunction
+from .lib import ConsecutiveEmptyLines  # noqa: F401
+from .lib import EmptyLastLine          # noqa: F401
+from .lib import NewlineAtEof           # noqa: F401
+from .lib import TrailingSpace          # noqa: F401
 
 
 def _empty_or_comment(text):

utils/checkpackagelib/lib_config.py class AttributesOrder(_CheckFunction): · line 49

Quoted 5 lines
utils/checkpackagelib/lib_config.py +1 -1
@@ -45,7 +46,7 @@ class AttributesOrder(_CheckFunction):
         if attribute in entries_that_should_not_be_indented:
             self.state = 0
             return
-        if attribute not in self.attributes_order_convention.keys():
+        if attribute not in list(self.attributes_order_convention.keys()):

Why is this needed? 'not in' works on an iterator, doesn't it?

Same below.

Bottom line: I don't think modernize does anything useful.

Regards,
Arnout

Quoted 79 lines
utils/checkpackagelib/lib_config.py +0 -0
             return
         new_state = self.attributes_order_convention[attribute]
         wrong_order = self.state > new_state
utils/checkpackagelib/lib_hash.py +7 -6
index 6d4cc9fd62..8ca92a5a90 100644
@@ -3,13 +3,14 @@
 # functions don't need to check for things already checked by running
 # "make package-dirclean package-source".
 
+from __future__ import absolute_import
 import re
 
-from base import _CheckFunction
-from lib import ConsecutiveEmptyLines  # noqa: F401
-from lib import EmptyLastLine          # noqa: F401
-from lib import NewlineAtEof           # noqa: F401
-from lib import TrailingSpace          # noqa: F401
+from .base import _CheckFunction
+from .lib import ConsecutiveEmptyLines  # noqa: F401
+from .lib import EmptyLastLine          # noqa: F401
+from .lib import NewlineAtEof           # noqa: F401
+from .lib import TrailingSpace          # noqa: F401
 
 
 def _empty_line_or_comment(text):
@@ -43,7 +44,7 @@ class HashType(_CheckFunction):
         htype, hexa = fields[:2]
         if htype == "none":
             return
-        if htype not in self.len_of_hash.keys():
+        if htype not in list(self.len_of_hash.keys()):
             return ["{}:{}: unexpected type of hash ({}#adding-packages-hash)"
                     .format(self.filename, lineno, self.url_to_manual),
                     text]
utils/checkpackagelib/lib_mk.py +6 -5
index 86e9aa2d97..9c17a3cfd2 100644
@@ -4,13 +4,14 @@
 # menu options using "make menuconfig" and by running "make" with appropriate
 # packages enabled.
 
+from __future__ import absolute_import
 import re
 
-from base import _CheckFunction
-from lib import ConsecutiveEmptyLines  # noqa: F401
-from lib import EmptyLastLine          # noqa: F401
-from lib import NewlineAtEof           # noqa: F401
-from lib import TrailingSpace          # noqa: F401
+from .base import _CheckFunction
+from .lib import ConsecutiveEmptyLines  # noqa: F401
+from .lib import EmptyLastLine          # noqa: F401
+from .lib import NewlineAtEof           # noqa: F401
+from .lib import TrailingSpace          # noqa: F401
 
 
 class Indent(_CheckFunction):
utils/checkpackagelib/lib_patch.py +3 -2
index 555621afa1..4b0d32268f 100644
@@ -3,10 +3,11 @@
 # functions don't need to check for things already checked by running
 # "make package-dirclean package-patch".
 
+from __future__ import absolute_import
 import re
 
-from base import _CheckFunction
-from lib import NewlineAtEof           # noqa: F401
+from .base import _CheckFunction
+from .lib import NewlineAtEof           # noqa: F401
 
 
 class ApplyOrder(_CheckFunction):
Signature · 6 lines
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7493 020B C7E3 8618 8DEC 222C 82EB F404 F9AC 0DDF
#2 lore raw

PATCH 1/1 check-package: fix Python3 support

Hello,

Thank you for your review.
I don't usually write Python3 code, and I still beginning to use modernize. Sorry I did the newbie mistake of trusting too much in the defaults of the tool.

Quoted 5 lines · Arnout Vandecappelle
Quoted 4 lines · Ricardo Martincoski

This script currently uses "/usr/bin/env python" as shebang but it does not really support Python3. Instead of limiting the script to Python2, fix it to support both versions.

Quoted 7 lines
Quoted 2 lines

Do this by first using an automated tool:

$ python -m modernize -w -p <files>

In general (if it doesn't break things), I prefer to do the automatic stuff in
a separate patch from the manual stuff. That makes review easier (because for the automatic stuff, you mainly review the automation itself rather than the result).

OK. In this case things are already broken. The automatic stuff would unbreak 99% of them, and the manual stuff would unbreak the remaining 1%. So separate patches makes sense.
I will not use the tool for the new iteration, but I will keep in mind for future patch series.

Quoted 6 lines
Quoted 6 lines

and then manually fixing what remains:
- keep the imports in alphabetical order;
- fix handling of files that cannot be decoded by 'utf-8' codec, such as package/urg/0002-urg-gcc6-fix-narrowing-conversion.patch .

Quoted 8 lines
Quoted 5 lines

In order to avoid errors when decoding files, use errors="surrogateescape" when opening files, the docs for open() states:
"This is useful for processing files in an unknown encoding.". This argument is not compatible with Python2 open() so import 'six' to use it only when running in Python3.

I would prefer to open files with "rb" instead. However, that's a much bigger
change because all checks should then use bytes objects instead of strings.

Indeed it would change all functions in the lib*.py files.
It seems to me it's not worth the effort in this case. All functions are there to check text files. The only case we can have non-ascii/non-utf-8 files being checked by the script are for patch files when the file to be patched is not ascii or utf-8. We have currently one case in the tree right now.

[snip]

Quoted 8 lines

utils/check-package line 4

Quoted 6 lines

+++ b/utils/check-package

utils/check-package +1 -0
@@ -1,11 +1,13 @@
 #!/usr/bin/env python
 # See utils/checkpackagelib/readme.txt before editing this file.
 
+from __future__ import absolute_import

This doesn't make sense in a top-level script, does it?

I am not sure in the general sense. I am still trying to understand what it really does. It's not easy to get a docker with Python 2.5 to test (the official docker images from python start from 2.7). I will probably fallback to setup a VM.

But it is certainly not needed here. It was a result from the automatic stuff. I will remove it. Actually from everywhere.

[snip]

Quoted 10 lines

utils/checkpackagelib/lib.py line 4

Quoted 7 lines
utils/checkpackagelib/lib.py +3 -1
+++ b/utils/checkpackagelib/lib.py
@@ -1,6 +1,7 @@
 # See utils/checkpackagelib/readme.txt before editing this file.
 
-from base import _CheckFunction
+from __future__ import absolute_import
+from .base import _CheckFunction

Why is this changed? Will Python3 look for base anywhere else than in the
checkpackagelib directory?

Python3 follows PEP328 and dropped implicit relative imports.

But maybe a better approach is to use the absolute import:
from checkpackagelib.base import _CheckFunction

I will do this, everywhere, if you agree.

Quoted 3 lines

Same below.

[snip]

Quoted 8 lines

utils/checkpackagelib/lib_config.py class AttributesOrder(_CheckFunction): · line 49

Quoted 6 lines
utils/checkpackagelib/lib_config.py +1 -1
@@ -45,7 +46,7 @@ class AttributesOrder(_CheckFunction):
         if attribute in entries_that_should_not_be_indented:
             self.state = 0
             return
-        if attribute not in self.attributes_order_convention.keys():
+        if attribute not in list(self.attributes_order_convention.keys()):

Why is this needed? 'not in' works on an iterator, doesn't it?

You are right. It does work. It seems modernize adds it always, when it is needed (when a index will be used in the result for example) and when it is not needed.

So I probably won't use the 'hash' transformation from modernize anymore, it is enabled by default.

Quoted 4 lines

Same below.

Bottom line: I don't think modernize does anything useful.

For this script I must agree.

Good examples (not applicable to this script) seem to be the conversion of print, exception, and urllib2.

Regards,
Ricardo

#3 lore raw

PATCH 1/1 check-package: fix Python3 support

Quoted 9 lines · Ricardo Martincoski

Hello,

Thank you for your review.
I don't usually write Python3 code, and I still beginning to use modernize. Sorry I did the newbie mistake of trusting too much in the defaults of the tool.

Quoted 1 line · Arnout Vandecappelle

On 23-07-18 02:41, Ricardo Martincoski wrote:

[snip]

Quoted 6 lines
Quoted 3 lines
Quoted 1 line

+from __future__ import absolute_import

This doesn't make sense in a top-level script, does it?

I am not sure in the general sense. I am still trying to understand what it really does.

So am I. As I understand it, in modern Python, for scripts that are part of a
package, the import path is relative to the top level of the package rather than the directory of the module itself. For scripts outside of a package, nothing has changed AFAIK.

Quoted 2 lines

It's not easy to get a docker with Python 2.5 to test (the official docker images from python start from 2.7).

Since we're targetting RedHat 6 here, Python 2.6 should be sufficient, no?

Quoted 17 lines

I will probably fallback to setup a VM.

But it is certainly not needed here. It was a result from the automatic stuff. I will remove it. Actually from everywhere.

[snip]

Quoted 10 lines

utils/checkpackagelib/lib.py line 4

Quoted 7 lines

+++ b/utils/checkpackagelib/lib.py

utils/checkpackagelib/lib.py +2 -1
@@ -1,6 +1,7 @@
 # See utils/checkpackagelib/readme.txt before editing this file.
 
-from base import _CheckFunction
+from __future__ import absolute_import
+from .base import _CheckFunction

Why is this changed? Will Python3 look for base anywhere else than in the
checkpackagelib directory?

Quoted 6 lines

Python3 follows PEP328 and dropped implicit relative imports.

But maybe a better approach is to use the absolute import:
from checkpackagelib.base import _CheckFunction

I will do this, everywhere, if you agree.

Yes, the .xxx approach only makes sense in cases where you want a module in the
package be callable directly as a script.

[snip]

Quoted 2 lines

Good examples (not applicable to this script) seem to be the conversion of print, exception, and urllib2.

print() was already handled before. We're lacking tests of python3 compliance
of this script to make sure there are no regression. Heck, we're lacking tests of this script, period.

Regards,
Arnout

Signature · 6 lines
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7493 020B C7E3 8618 8DEC 222C 82EB F404 F9AC 0DDF
#4 lore raw

PATCH v2 1/1 check-package: fix Python3 support

In reply to Ricardo Martincoski

This script currently uses "/usr/bin/env python" as shebang but it does not really support Python3. Instead of limiting the script to Python2, fix it to support both versions.

So change all imports to absolute imports because Python3 follows PEP328 and dropped implicit relative imports.

In order to avoid errors when decoding files with the default 'utf-8' codec, use errors="surrogateescape" when opening files, the docs for open() states: "This is useful for processing files in an unknown encoding.". This argument is not compatible with Python2 open() so import 'six' to use it only when running in Python3.
As a consequence the file handler becomes explicit, so use it to close() the file after it got processed.

This "surrogateescape" is a simple alternative to the complete solution of opening files with "rb" and changing all functions in the lib*.py files to use bytes objects instead of strings. The only case we can have non-ascii/non-utf-8 files being checked by the script are for patch files when the upstream file to be patched is not ascii or utf-8. There is currently one case in the tree:
package/urg/0002-urg-gcc6-fix-narrowing-conversion.patch.

Cc Arnout Vandecappelle <[email protected]>

---
Changes v1 -> v2:
- do not use an automated tool (modernize);
- use absolute import;
- drop list() added to .keys(), it's not needed (Arnout);
- drop 'from __future__ import absolute_import' everywhere;
- list the (not used) approach of opening files in binary mode in the

    commit message;

Using the default docker (debian) image which uses Python2:
before this patch:
https://gitlab.com/buildroot.org/buildroot/-/jobs/88548503
after this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/88561994

Using an arch docker image which uses Python3:
before this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/88561488
after this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/88561704
The arch image for docker was generated using:
http://patchwork.ozlabs.org/patch/943301/

---
 utils/check-package                 |  8 +++++++-
 utils/checkpackagelib/lib.py        |  2 +-
 utils/checkpackagelib/lib_config.py | 10 +++++-----
 utils/checkpackagelib/lib_hash.py   | 10 +++++-----
 utils/checkpackagelib/lib_mk.py     | 10 +++++-----
 utils/checkpackagelib/lib_patch.py  |  4 ++--
 6 files changed, 25 insertions(+), 19 deletions(-)
utils/check-package +7 -1
index 3dbc28b0a2..26439f08eb 100755
@@ -6,6 +6,7 @@ import argparse
 import inspect
 import os
 import re
+import six
 import sys
 
 import checkpackagelib.lib_config
@@ -127,10 +128,15 @@ def check_file_using_lib(fname):
 
     for cf in objects:
         nwarnings += print_warnings(cf.before())
-    for lineno, text in enumerate(open(fname, "r").readlines()):
+    if six.PY3:
+        f = open(fname, "r", errors="surrogateescape")
+    else:
+        f = open(fname, "r")
+    for lineno, text in enumerate(f.readlines()):
         nlines += 1
         for cf in objects:
             nwarnings += print_warnings(cf.check_line(lineno + 1, text))
+    f.close()
     for cf in objects:
         nwarnings += print_warnings(cf.after())
 
utils/checkpackagelib/lib.py +1 -1
index 91f4ad49b7..6afe1aabce 100644
@@ -1,6 +1,6 @@
 # See utils/checkpackagelib/readme.txt before editing this file.
 
-from base import _CheckFunction
+from checkpackagelib.base import _CheckFunction
 
 
 class ConsecutiveEmptyLines(_CheckFunction):
utils/checkpackagelib/lib_config.py +5 -5
index 1d273f1c5f..89d44da57e 100644
@@ -5,11 +5,11 @@
 
 import re
 
-from base import _CheckFunction
-from lib import ConsecutiveEmptyLines  # noqa: F401
-from lib import EmptyLastLine          # noqa: F401
-from lib import NewlineAtEof           # noqa: F401
-from lib import TrailingSpace          # noqa: F401
+from checkpackagelib.base import _CheckFunction
+from checkpackagelib.lib import ConsecutiveEmptyLines  # noqa: F401
+from checkpackagelib.lib import EmptyLastLine          # noqa: F401
+from checkpackagelib.lib import NewlineAtEof           # noqa: F401
+from checkpackagelib.lib import TrailingSpace          # noqa: F401
 
 
 def _empty_or_comment(text):
utils/checkpackagelib/lib_hash.py +5 -5
index 6d4cc9fd62..3e381119a5 100644
@@ -5,11 +5,11 @@
 
 import re
 
-from base import _CheckFunction
-from lib import ConsecutiveEmptyLines  # noqa: F401
-from lib import EmptyLastLine          # noqa: F401
-from lib import NewlineAtEof           # noqa: F401
-from lib import TrailingSpace          # noqa: F401
+from checkpackagelib.base import _CheckFunction
+from checkpackagelib.lib import ConsecutiveEmptyLines  # noqa: F401
+from checkpackagelib.lib import EmptyLastLine          # noqa: F401
+from checkpackagelib.lib import NewlineAtEof           # noqa: F401
+from checkpackagelib.lib import TrailingSpace          # noqa: F401
 
 
 def _empty_line_or_comment(text):
utils/checkpackagelib/lib_mk.py +5 -5
index 86e9aa2d97..9bf417fb73 100644
@@ -6,11 +6,11 @@
 
 import re
 
-from base import _CheckFunction
-from lib import ConsecutiveEmptyLines  # noqa: F401
-from lib import EmptyLastLine          # noqa: F401
-from lib import NewlineAtEof           # noqa: F401
-from lib import TrailingSpace          # noqa: F401
+from checkpackagelib.base import _CheckFunction
+from checkpackagelib.lib import ConsecutiveEmptyLines  # noqa: F401
+from checkpackagelib.lib import EmptyLastLine          # noqa: F401
+from checkpackagelib.lib import NewlineAtEof           # noqa: F401
+from checkpackagelib.lib import TrailingSpace          # noqa: F401
 
 
 class Indent(_CheckFunction):
utils/checkpackagelib/lib_patch.py +2 -2
index 555621afa1..453b782e6c 100644
@@ -5,8 +5,8 @@
 
 import re
 
-from base import _CheckFunction
-from lib import NewlineAtEof           # noqa: F401
+from checkpackagelib.base import _CheckFunction
+from checkpackagelib.lib import NewlineAtEof           # noqa: F401
 
 
 class ApplyOrder(_CheckFunction):
Signature · 1 line
2.17.1
#5 lore raw

PATCH v2 1/1 check-package: fix Python3 support

El s?b., 11 ago. 2018 a las 5:49, Ricardo Martincoski (<[email protected]>) escribi?:

Quoted 48 lines

This script currently uses "/usr/bin/env python" as shebang but it does not really support Python3. Instead of limiting the script to Python2, fix it to support both versions.

So change all imports to absolute imports because Python3 follows PEP328 and dropped implicit relative imports.

In order to avoid errors when decoding files with the default 'utf-8' codec, use errors="surrogateescape" when opening files, the docs for open() states: "This is useful for processing files in an unknown encoding.". This argument is not compatible with Python2 open() so import 'six' to use it only when running in Python3.
As a consequence the file handler becomes explicit, so use it to close() the file after it got processed.

This "surrogateescape" is a simple alternative to the complete solution of opening files with "rb" and changing all functions in the lib*.py files to use bytes objects instead of strings. The only case we can have non-ascii/non-utf-8 files being checked by the script are for patch files when the upstream file to be patched is not ascii or utf-8. There is currently one case in the tree:
package/urg/0002-urg-gcc6-fix-narrowing-conversion.patch.

Cc Arnout Vandecappelle <[email protected]>

---
Changes v1 -> v2:
- do not use an automated tool (modernize);
- use absolute import;
- drop list() added to .keys(), it's not needed (Arnout);
- drop 'from __future__ import absolute_import' everywhere;
- list the (not used) approach of opening files in binary mode in the

    commit message;

Using the default docker (debian) image which uses Python2:
before this patch:
https://gitlab.com/buildroot.org/buildroot/-/jobs/88548503
after this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/88561994

Using an arch docker image which uses Python3:
before this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/88561488
after this patch:
https://gitlab.com/RicardoMartincoski/buildroot/-/jobs/88561704
The arch image for docker was generated using:
http://patchwork.ozlabs.org/patch/943301/

Quoted 9 lines
---
 utils/check-package                 |  8 +++++++-
 utils/checkpackagelib/lib.py        |  2 +-
 utils/checkpackagelib/lib_config.py | 10 +++++-----
 utils/checkpackagelib/lib_hash.py   | 10 +++++-----
 utils/checkpackagelib/lib_mk.py     | 10 +++++-----
 utils/checkpackagelib/lib_patch.py  |  4 ++--
 6 files changed, 25 insertions(+), 19 deletions(-)

This patch looks good to me. I don't really know anything about surrogateescape, but the script works on the urg package so seems ok.

[Tested with Python 2.7.x and Python 3.5.x]
#6 lore raw

PATCH v2 1/1 check-package: fix Python3 support

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas De Schampheleire <[email protected]> writes:

Hi,

Quoted 37 lines

El s?b., 11 ago. 2018 a las 5:49, Ricardo Martincoski (<[email protected]>) escribi?:

Quoted 26 lines

This script currently uses "/usr/bin/env python" as shebang but it does not really support Python3. Instead of limiting the script to Python2, fix it to support both versions.

So change all imports to absolute imports because Python3 follows PEP328 and dropped implicit relative imports.

In order to avoid errors when decoding files with the default 'utf-8' codec, use errors="surrogateescape" when opening files, the docs for open() states: "This is useful for processing files in an unknown encoding.". This argument is not compatible with Python2 open() so import 'six' to use it only when running in Python3.
As a consequence the file handler becomes explicit, so use it to close() the file after it got processed.

This "surrogateescape" is a simple alternative to the complete solution of opening files with "rb" and changing all functions in the lib*.py files to use bytes objects instead of strings. The only case we can have non-ascii/non-utf-8 files being checked by the script are for patch files when the upstream file to be patched is not ascii or utf-8. There is currently one case in the tree:
package/urg/0002-urg-gcc6-fix-narrowing-conversion.patch.

Cc Arnout Vandecappelle <[email protected]>
Quoted 8 lines

---
Changes v1 -> v2:
- do not use an automated tool (modernize);
- use absolute import;
- drop list() added to .keys(), it's not needed (Arnout);
- drop 'from __future__ import absolute_import' everywhere;
- list the (not used) approach of opening files in binary mode in the commit message;

Quoted 6 lines

This patch looks good to me. I don't really know anything about surrogateescape, but the script works on the urg package so seems ok.

[Tested with Python 2.7.x and Python 3.5.x]

Committed, thanks both.

Signature · 1 line
Bye, Peter Korsgaard
#7 lore raw

PATCH v2 1/1 check-package: fix Python3 support

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Peter" == Peter Korsgaard <[email protected]> writes:

Hi,

Quoted 34 lines
Quoted 33 lines
Quoted 33 lines

This script currently uses "/usr/bin/env python" as shebang but it does not really support Python3. Instead of limiting the script to Python2, fix it to support both versions.

So change all imports to absolute imports because Python3 follows PEP328 and dropped implicit relative imports.

In order to avoid errors when decoding files with the default 'utf-8' codec, use errors="surrogateescape" when opening files, the docs for open() states: "This is useful for processing files in an unknown encoding.". This argument is not compatible with Python2 open() so import 'six' to use it only when running in Python3.
As a consequence the file handler becomes explicit, so use it to close() the file after it got processed.

This "surrogateescape" is a simple alternative to the complete solution of opening files with "rb" and changing all functions in the lib*.py files to use bytes objects instead of strings. The only case we can have non-ascii/non-utf-8 files being checked by the script are for patch files when the upstream file to be patched is not ascii or utf-8. There is currently one case in the tree:
package/urg/0002-urg-gcc6-fix-narrowing-conversion.patch.

Cc Arnout Vandecappelle <[email protected]>

---
Changes v1 -> v2:
- do not use an automated tool (modernize);
- use absolute import;
- drop list() added to .keys(), it's not needed (Arnout);
- drop 'from __future__ import absolute_import' everywhere;
- list the (not used) approach of opening files in binary mode in the commit message;

Quoted 6 lines
Quoted 6 lines

This patch looks good to me. I don't really know anything about surrogateescape, but the script works on the urg package so seems ok.

[Tested with Python 2.7.x and Python 3.5.x]

Committed to 2018.02.x and 2018.11.x, thanks.

Signature · 1 line
Bye, Peter Korsgaard