Details

Reviewers

douardda
vlorentz

Group Reviewers

Reviewers

Commits

rDLS89b409d30f95: gnu.lister: Update docstrings and fix type annotations
rDLSc6372eea7e96: gnu.lister: Unify timestamp formats to isoformat date in model
rDLS93e7afda3b5c: gnu.lister: Format timestamp to isoformat string for the tar loader
rDLS9fd648987eda: gnu.tree: Move helper utilities function to the bottom of the file
rDLS821f3f1cc28e: lister.gnu: Move version parsing logic to the lister

Summary

This:

Moves version parsing logic from loader-gnu to lister
Parses the timestamps provided by gnu tree to isoformat dates (to provide to loader)
(Aligns the timestamps stored in the lister model to be also a date in the right format to clarify reading from it)

This allows to have a generic "tar" loader.
That way, the loader expects some required arguments in the right type already.
No need for it to try and do specific computations depending on the origin of the archives.

Related D2145

Note:
This actually implies a schema migration (timestamp model migration)
As the loader gnu never ran though, we might as well drop the existing data and trigger a new run when time comes.
So it's not a blocker.

Test Plan

tox

Diff Detail

Repository

rDLS Listers

Branch

gnu-adaptations

Lint

No Linters Available

Unit

No Unit Test Coverage

Build Status

Buildable 8540
Build 12409: tox-on-jenkins	Jenkins
Build 12408: arc lint + arc unit

Event Timeline

ardumont created this revision.Oct 16 2019, 1:31 PM

Build has FAILED

Link to build: https://jenkins.softwareheritage.org/job/DLS/job/tox/442/
See console output for more information: https://jenkins.softwareheritage.org/job/DLS/job/tox/442/console

Harbormaster failed remote builds in B8480: Diff 7264!Oct 16 2019, 1:35 PM

Fix using set instead of list

Build is green
See https://jenkins.softwareheritage.org/job/DLS/job/tox/443/ for more details.

Harbormaster completed remote builds in B8481: Diff 7265.Oct 16 2019, 2:37 PM

gnu.tree: Move helper utilities function to the bottom of the file

Build is green
See https://jenkins.softwareheritage.org/job/DLS/job/tox/444/ for more details.

Harbormaster completed remote builds in B8482: Diff 7266.Oct 16 2019, 2:49 PM

ardumont mentioned this in D2145: package.tar: Add a generic archive loader implementation (merge with gnu's).Oct 17 2019, 5:14 AM

ardumont edited the summary of this revision. (Show Details)Oct 17 2019, 9:41 AM

ardumont added projects: Lister, Origin-GNU.

Rebase to latest master

Build is green
See https://jenkins.softwareheritage.org/job/DLS/job/tox/446/ for more details.

Harbormaster completed remote builds in B8505: Diff 7307.Oct 17 2019, 2:20 PM

ardumont edited the summary of this revision. (Show Details)Oct 17 2019, 6:14 PM

ardumont retitled this revision from lister.gnu: Move version parsing logic from loader-gnu to lister to lister.gnu: Standardize arguments to pass to the loader.Oct 18 2019, 3:23 AM

ardumont edited the summary of this revision. (Show Details)

ardumont retitled this revision from lister.gnu: Standardize arguments to pass to the loader to lister.gnu: Standardize arguments to pass to the loader tar.

ardumont edited the summary of this revision. (Show Details)Oct 18 2019, 3:29 AM

gnu.lister: Format timestamp to isoformat string for the tar loader
gnu.lister: Unify timestamp formats to isoformat date in model
gnu.tree: Drop debug statement
gnu.lister: Update docstring samples

ardumont edited the summary of this revision. (Show Details)Oct 18 2019, 3:32 AM

Build has FAILED

Link to build: https://jenkins.softwareheritage.org/job/DLS/job/tox/448/
See console output for more information: https://jenkins.softwareheritage.org/job/DLS/job/tox/448/console

Harbormaster failed remote builds in B8531: Diff 7332!Oct 18 2019, 3:34 AM

Build has FAILED

Yes, i did not parse the timestamp into an isoformat datestring correctly.
I'm fixing that.

Fix parsing timestamps

Build has FAILED

Link to build: https://jenkins.softwareheritage.org/job/DLS/job/tox/449/
See console output for more information: https://jenkins.softwareheritage.org/job/DLS/job/tox/449/console

Harbormaster failed remote builds in B8540: Diff 7342!Oct 18 2019, 1:07 PM

Build has FAILED

Yes, it now depends on D2158

I don't think this regexp is the right solution, it's quite fragile in case a package name uses one of the keywords. A possible solution is for get_version to only accept URLs, and split the path to find the package name.

swh/lister/gnu/tests/test_tree.py
152	I don't understand this sentence
swh/lister/gnu/tree.py
188–196	You forgot `.7z`, eg. https://ftp.gnu.org/gnu/octave/windows/
234	shouldn't the final `+` be non-greedy as well? (Don't forget to add a test for this, if relevant) I'm guessing you'll need it for `.7z`.
242–245	.format( extensions=... vkeywords=...

This revision now requires changes to proceed.Oct 18 2019, 2:25 PM

ardumont added inline comments.Oct 18 2019, 2:44 PM

swh/lister/gnu/tests/test_tree.py
152	"""Parsing version from url should yield some form of "sensible" version Given the dataset, it's not a simple task to extract correctly version. """ is it better?

vlorentz added inline comments.Oct 18 2019, 2:45 PM

swh/lister/gnu/tests/test_tree.py
152	yes

I don't think this regexp is the right solution, it's quite fragile in case a package name uses one of the keywords.

Yes, it's fragile and we don't see any better solution right now.
That's how we did it initially, we improved the current version to be a tad better though.
And also it's tested even if it's limited.

A possible solution is for get_version to only accept URLs, and split the path to find the package name.

I'm not getting the difference with the current implem.
filename is extracted from the uri in any case...

Do you have an example in mind, that'd help, i think.

TIA

swh/lister/gnu/tree.py
188–196	thx!

ardumont added inline comments.Oct 18 2019, 3:00 PM

swh/lister/gnu/tree.py
234	I added a test with a .7z and in current form, that's enough.

In D2147#50755, @ardumont wrote:

I don't think this regexp is the right solution, it's quite fragile in case a package name uses one of the keywords.

Yes, it's fragile and we don't see any better solution right now.
That's how we did it initially, we improved the current version to be a tad better though.
And also it's tested even if it's limited.

A possible solution is for get_version to only accept URLs, and split the path to find the package name.

I'm not getting the difference with the current implem.
filename is extracted from the uri in any case...

I meant splitting on / so you get the package name in the folder name. But I just realized it's not a reliable method, so nvm

ardumont added inline comments.Oct 18 2019, 3:03 PM

swh/lister/gnu/tree.py
242–245	Well, i tried and it does not work. That fails the tests somehow...

ardumont added inline comments.Oct 18 2019, 3:06 PM

swh/lister/gnu/tree.py

242–245

pattern1 is the current implem' (i renamed the global variable to use upper case in their name):

collecting ... ('pattern1: \n'
 '^\n'
 '(?:\n'
 '    # We have a software name and a release number, separated with a\n'
 '    # -, _ or dot.\n'
 '    (?P<software_name1>.+?[-_.])\n'
 '    '
 '(?P<release_number>(cygwin_me[-]?|w32[-]?|win32[-]?|nt[-]?|cygwin[-]?|mingw[-]?|latest[-]?|alpha[-]?|beta[-]?|release[-]?|stable[-]?|hppa[-]?|solaris[-]?|sunos[-]?|sun4u[-]?|sparc[-]?|sun[-]?|aix[-]?|ibm[-]?|rs6000[-]?|i386[-]?|i686[-]?|linux[-]?|redhat[-]?|linuxlibc[-]?|mips[-]?|powerpc[-]?|macos[-]?|apple[-]?|darwin[-]?|macosx[-]?|powermacintosh[-]?|unknown[-]?|netbsd[-]?|freebsd[-]?|sgi[-]?|irix[-]?|[0-9][0-9a-zA-Z_.+:~-]*?)+)\n'
 '|\n'
 "    # We couldn't match a release number, put everything in the\n"
 '    # software name.\n'
 '    (?P<software_name2>.+?)\n'
 ')\n'

pattern2 is the format call:

PATTERN2 = r'''
^
(?:
    # We have a software name and a release number, separated with a
    # -, _ or dot.
    (?P<software_name1>.+?[-_.])
    (?P<release_number>(%(vkeywords)s|[0-9][0-9a-zA-Z_.+:~-]*?)+)
|
    # We couldn't match a release number, put everything in the
    # software name.
    (?P<software_name2>.+?)
)
(?P<extension>(?:\.(?:%(extensions)s))+)
$
'''.format(
    extensions='|'.join(EXTENSIONS),
    vkeywords='|'.join('%s[-]?' % k for k in VERSION_KEYWORDS)
)

pprint(PATTERN2)

gives:

 '(?P<extension>(?:\\.(?:zip|tar|gz|tgz|bz2|bzip2|lzma|lz|xz|Z|7z))+)\n'
 '$\n')
('pattern2: \n'
 '^\n'
 '(?:\n'
 '    # We have a software name and a release number, separated with a\n'
 '    # -, _ or dot.\n'
 '    (?P<software_name1>.+?[-_.])\n'
 '    (?P<release_number>(%(vkeywords)s|[0-9][0-9a-zA-Z_.+:~-]*?)+)\n'
 '|\n'
 "    # We couldn't match a release number, put everything in the\n"
 '    # software name.\n'
 '    (?P<software_name2>.+?)\n'
 ')\n'
 '(?P<extension>(?:\\.(?:%(extensions)s))+)\n'
 '$\n')

it seems it does not work as expected...

Adapt according to what's possible from the review