pre-commit-bigfix 2.3.0__tar.gz → 2.4.0__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {pre_commit_bigfix-2.3.0/pre_commit_bigfix.egg-info → pre_commit_bigfix-2.4.0}/PKG-INFO +62 -8
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/README.md +61 -7
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/__init__.py +1 -1
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_siterepo_validate_layout.py +379 -33
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0/pre_commit_bigfix.egg-info}/PKG-INFO +62 -8
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_siterepo_validate_layout.py +446 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/LICENSE +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/MANIFEST.in +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_actionscript_lint_schclass.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_actionscript_validate_prefetch.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_actionscript_validate_script.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_common.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_content_score_risk.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_conventions_check.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_relevance_convert_group.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_relevance_lint.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/bes_schema_validate.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/run_all.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/schclass.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/schclass_data/ExpandedActionScript.schclass +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/schclass_data/bigfix_overrides.schclass +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix/schclass_tokenizer.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix.egg-info/SOURCES.txt +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix.egg-info/dependency_links.txt +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix.egg-info/entry_points.txt +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix.egg-info/requires.txt +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pre_commit_bigfix.egg-info/top_level.txt +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/pyproject.toml +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/setup.cfg +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/setup.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_actionscript_lint_schclass.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_actionscript_validate_prefetch.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_actionscript_validate_script.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_common.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_content_score_risk.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_conventions_check.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_relevance_convert_group.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_relevance_lint.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_bes_schema_validate.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_hook_declarations.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_package_metadata.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_release_workflow.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_run_all.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_schclass.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_schclass_tokenizer.py +0 -0
- {pre_commit_bigfix-2.3.0 → pre_commit_bigfix-2.4.0}/tests/test_schema_guard.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: pre-commit-bigfix
|
|
3
|
-
Version: 2.
|
|
3
|
+
Version: 2.4.0
|
|
4
4
|
Summary: pre-commit hooks for BigFix content
|
|
5
5
|
Home-page: https://github.com/jgstew/pre-commit-bigfix
|
|
6
6
|
Author: JGStew
|
|
@@ -837,13 +837,14 @@ warning (`W808`), since RHEL is case-sensitive.
|
|
|
837
837
|
| --- | --- |
|
|
838
838
|
| `E800` | a `.bes` file in the `.bes` folder whose name does not start with a numeric ID and a space or hyphen; `42- Name.bes` and `42 Name.bes` both work |
|
|
839
839
|
| `E801` | two or more `.bes` files in one site share an ID, compared as integers, so `0042` and `42` collide; each names the others |
|
|
840
|
-
| `E802` | a `.bes` file's server path is too long and the rename cannot fix it: even `<ID> X.bes` is too long (the folders are), it has no ID, or `E805` is disabled. Renaming a folder changes its `.fxf` or digest, so that is left to you |
|
|
840
|
+
| `E802` | a `.bes` file's server path is too long and the rename cannot fix it: even `<ID> X.bes` is too long (the folders are), it has no ID (and `E809` gives it none), or `E805` is disabled. Renaming a folder changes its `.fxf` or digest, so that is left to you |
|
|
841
841
|
| `E803` | fix: zero padding in the ID (`0042- X.bes` -> `42- X.bes`); done even when the ID is a duplicate (`E801`), since the ID does not change |
|
|
842
842
|
| `E804` | fix: non-ASCII characters in the name, folded to ASCII |
|
|
843
843
|
| `E805` | fix: the server path is too long, or the name is over 255 bytes, so the file is renamed `<ID> <Title>.bes` |
|
|
844
|
-
| `E806` | a rename was skipped: the new name is taken (ignoring case), the file has unstaged changes pre-commit set aside, the commit uses a temporary index (`git commit -a`, `git commit <paths>`),
|
|
844
|
+
| `E806` | a rename was skipped: the new name is taken (ignoring case), the file has unstaged changes pre-commit set aside, the commit uses a temporary index (`git commit -a`, `git commit <paths>`), the rename failed, or `E809` could not choose an ID safely (shallow clone, unknown base branch) |
|
|
845
845
|
| `E807` | another tracked file's server path is too long, inside the site roots or not (the server clones the whole repo). It is not renamed, since actions may refer to it by name |
|
|
846
846
|
| `E808` | a file or folder name over 255 bytes in UTF-8 (Linux `NAME_MAX`) that the rename cannot fix; each folder once |
|
|
847
|
+
| `E809` | fix: a file with no ID (`E800`), or a duplicate (`E801`) that is not the one keeping the ID, gets a new ID no file in git history has used; see **New IDs** |
|
|
847
848
|
| `W800` | a `.bes` file outside the `.bes` folder; the server does not gather it |
|
|
848
849
|
| `W801` | a `.bes` file directly in the `.bes` folder, or deeper than `<BesFileDir>/<fxf>/<digest>/` (how the server handles either is undocumented) |
|
|
849
850
|
| `W802` | a file in the `.bes` folder other than `.bes` or `digest.xml`; an `.xsd` too, since `bes-schema-validate` already has the schemas |
|
|
@@ -877,6 +878,8 @@ count in full. The options:
|
|
|
877
878
|
| `--repo-name` | the last part of the `origin` remote URL without `.git`, or the repo folder's name |
|
|
878
879
|
| `--branch` | `main` |
|
|
879
880
|
| `--max-path` | `259` |
|
|
881
|
+
| `--base-ref` | `PRE_COMMIT_FROM_REF`, else `origin/$GITHUB_BASE_REF`, else `HEAD` (see **New IDs**) |
|
|
882
|
+
| `--min-id` | `1` (see **New IDs**) |
|
|
880
883
|
|
|
881
884
|
**Renames.** By default each `.bes` file in the `.bes` folder is renamed at
|
|
882
885
|
most once, combining the fixes it needs: zero padding stripped from the ID
|
|
@@ -887,7 +890,7 @@ cut to fit both limits (`E805`). The title comes from the file's `<Title>`,
|
|
|
887
890
|
or from its current name if the title has no letter or digit; it is made
|
|
888
891
|
ASCII and safe on Windows, cut short, and any trailing space, dot or hyphen
|
|
889
892
|
is dropped. A name that already fits is left alone, even when it differs
|
|
890
|
-
from the title.
|
|
893
|
+
from the title. Only `E809` changes an ID, and no file moves between folders.
|
|
891
894
|
|
|
892
895
|
Tracked files are renamed with `git mv`, so the renames are staged and the
|
|
893
896
|
next commit carries them whole; a plain rename would leave an unstaged
|
|
@@ -907,10 +910,61 @@ renamed (`E806`) where the rename would be lost:
|
|
|
907
910
|
Any rename makes the hook exit 1, so review the renames and commit again.
|
|
908
911
|
`--check` is a dry run that reports what would be renamed.
|
|
909
912
|
|
|
910
|
-
|
|
911
|
-
|
|
912
|
-
|
|
913
|
-
|
|
913
|
+
**New IDs (`E809`).** A file with no ID, or a duplicate ID, gets a new one
|
|
914
|
+
in the same rename, which also applies `E804` and `E805`:
|
|
915
|
+
|
|
916
|
+
- No ID: `Name.bes` becomes `<ID> Name.bes`. The server never gathered it,
|
|
917
|
+
so nothing deployed refers to it.
|
|
918
|
+
- Duplicate ID: the one file on the base branch keeps the ID, and the others
|
|
919
|
+
get new ones (`42- X.bes` -> `1043- X.bes`). Renumbering only new content
|
|
920
|
+
keeps actions and baselines that refer to the old ID working. A file
|
|
921
|
+
renamed from one on the base branch counts as on it, as `git diff -M`
|
|
922
|
+
pairs them. When every file in the group is new, the first by path keeps
|
|
923
|
+
the ID. Nothing is renumbered, and `E806` says why, when two or more are
|
|
924
|
+
already on the base branch (which one keeps the ID is for a person to
|
|
925
|
+
decide), or when the ID is on the base branch but git sees no file here as
|
|
926
|
+
that file renamed (it may be one, changed too much to pair).
|
|
927
|
+
- The new ID is one more than the highest ID any `.bes` path has had in git
|
|
928
|
+
history (`git log --all --full-history -m`: every branch and tag, both
|
|
929
|
+
sides of every merge, every site in the repo) or has now, so it never
|
|
930
|
+
reuses the ID of a deleted or renamed file. It sees only the branches and
|
|
931
|
+
tags this clone has fetched: fetch first. Gaps are never filled. For a site whose older IDs predate the git
|
|
932
|
+
repo, set a floor with `--min-id`.
|
|
933
|
+
- The base branch is the first of `--base-ref`, `PRE_COMMIT_FROM_REF` (set by `pre-commit run --from-ref`)
|
|
934
|
+
and `origin/$GITHUB_BASE_REF` (set in a GitHub `pull_request` run) that is
|
|
935
|
+
a commit in the clone. Without any, it is `HEAD`: in a local commit, the
|
|
936
|
+
staged files are the new ones. Once both files of a duplicate are
|
|
937
|
+
committed on a feature branch, `HEAD` sees neither as new: run with
|
|
938
|
+
`--base-ref origin/main` (or the branch it merges into), as `E806` says.
|
|
939
|
+
- In a shallow clone, git history is incomplete, so no ID is assigned
|
|
940
|
+
(`E806`); likewise for duplicates when `--base-ref` is not a commit.
|
|
941
|
+
- `--disable E809` reports `E800` and `E801` only, as before. Disabling
|
|
942
|
+
`E800` or `E801` also stops `E809` for those files.
|
|
943
|
+
|
|
944
|
+
In GitHub Actions, check out the full history (the default is one commit),
|
|
945
|
+
and give pre-commit the PR's base so that it knows which files are new:
|
|
946
|
+
|
|
947
|
+
```yaml
|
|
948
|
+
- uses: actions/checkout@v7
|
|
949
|
+
with:
|
|
950
|
+
fetch-depth: 0
|
|
951
|
+
- uses: pre-commit/action@v3.0.1
|
|
952
|
+
with:
|
|
953
|
+
extra_args: >-
|
|
954
|
+
--from-ref ${{ github.event.pull_request.base.sha }}
|
|
955
|
+
--to-ref ${{ github.event.pull_request.head.sha }}
|
|
956
|
+
```
|
|
957
|
+
|
|
958
|
+
The hook renames the files on the runner and fails, and the log names each
|
|
959
|
+
new file name, for the author to apply (or run the hook locally with
|
|
960
|
+
`--base-ref` naming the PR's base branch). To commit
|
|
961
|
+
the fix back to a PR branch instead, check out `github.head_ref` with
|
|
962
|
+
`contents: write`, then commit and push any change. Fork PRs cannot be
|
|
963
|
+
pushed to, and a push with `GITHUB_TOKEN` starts no new workflow run. Two
|
|
964
|
+
open PRs can choose the same ID; the one merged second then has a duplicate
|
|
965
|
+
on the base branch and is renumbered when it runs again, so require branches
|
|
966
|
+
to be up to date before merging. A site that assigns IDs after merge from a
|
|
967
|
+
counter file, as CommunityContent does, should use `--disable E809`.
|
|
914
968
|
|
|
915
969
|
Not checked here: more than one object in a `.bes` file, or Baselines
|
|
916
970
|
(repository sites are thought to handle both); schemas, GroupRelevance and
|
|
@@ -805,13 +805,14 @@ warning (`W808`), since RHEL is case-sensitive.
|
|
|
805
805
|
| --- | --- |
|
|
806
806
|
| `E800` | a `.bes` file in the `.bes` folder whose name does not start with a numeric ID and a space or hyphen; `42- Name.bes` and `42 Name.bes` both work |
|
|
807
807
|
| `E801` | two or more `.bes` files in one site share an ID, compared as integers, so `0042` and `42` collide; each names the others |
|
|
808
|
-
| `E802` | a `.bes` file's server path is too long and the rename cannot fix it: even `<ID> X.bes` is too long (the folders are), it has no ID, or `E805` is disabled. Renaming a folder changes its `.fxf` or digest, so that is left to you |
|
|
808
|
+
| `E802` | a `.bes` file's server path is too long and the rename cannot fix it: even `<ID> X.bes` is too long (the folders are), it has no ID (and `E809` gives it none), or `E805` is disabled. Renaming a folder changes its `.fxf` or digest, so that is left to you |
|
|
809
809
|
| `E803` | fix: zero padding in the ID (`0042- X.bes` -> `42- X.bes`); done even when the ID is a duplicate (`E801`), since the ID does not change |
|
|
810
810
|
| `E804` | fix: non-ASCII characters in the name, folded to ASCII |
|
|
811
811
|
| `E805` | fix: the server path is too long, or the name is over 255 bytes, so the file is renamed `<ID> <Title>.bes` |
|
|
812
|
-
| `E806` | a rename was skipped: the new name is taken (ignoring case), the file has unstaged changes pre-commit set aside, the commit uses a temporary index (`git commit -a`, `git commit <paths>`),
|
|
812
|
+
| `E806` | a rename was skipped: the new name is taken (ignoring case), the file has unstaged changes pre-commit set aside, the commit uses a temporary index (`git commit -a`, `git commit <paths>`), the rename failed, or `E809` could not choose an ID safely (shallow clone, unknown base branch) |
|
|
813
813
|
| `E807` | another tracked file's server path is too long, inside the site roots or not (the server clones the whole repo). It is not renamed, since actions may refer to it by name |
|
|
814
814
|
| `E808` | a file or folder name over 255 bytes in UTF-8 (Linux `NAME_MAX`) that the rename cannot fix; each folder once |
|
|
815
|
+
| `E809` | fix: a file with no ID (`E800`), or a duplicate (`E801`) that is not the one keeping the ID, gets a new ID no file in git history has used; see **New IDs** |
|
|
815
816
|
| `W800` | a `.bes` file outside the `.bes` folder; the server does not gather it |
|
|
816
817
|
| `W801` | a `.bes` file directly in the `.bes` folder, or deeper than `<BesFileDir>/<fxf>/<digest>/` (how the server handles either is undocumented) |
|
|
817
818
|
| `W802` | a file in the `.bes` folder other than `.bes` or `digest.xml`; an `.xsd` too, since `bes-schema-validate` already has the schemas |
|
|
@@ -845,6 +846,8 @@ count in full. The options:
|
|
|
845
846
|
| `--repo-name` | the last part of the `origin` remote URL without `.git`, or the repo folder's name |
|
|
846
847
|
| `--branch` | `main` |
|
|
847
848
|
| `--max-path` | `259` |
|
|
849
|
+
| `--base-ref` | `PRE_COMMIT_FROM_REF`, else `origin/$GITHUB_BASE_REF`, else `HEAD` (see **New IDs**) |
|
|
850
|
+
| `--min-id` | `1` (see **New IDs**) |
|
|
848
851
|
|
|
849
852
|
**Renames.** By default each `.bes` file in the `.bes` folder is renamed at
|
|
850
853
|
most once, combining the fixes it needs: zero padding stripped from the ID
|
|
@@ -855,7 +858,7 @@ cut to fit both limits (`E805`). The title comes from the file's `<Title>`,
|
|
|
855
858
|
or from its current name if the title has no letter or digit; it is made
|
|
856
859
|
ASCII and safe on Windows, cut short, and any trailing space, dot or hyphen
|
|
857
860
|
is dropped. A name that already fits is left alone, even when it differs
|
|
858
|
-
from the title.
|
|
861
|
+
from the title. Only `E809` changes an ID, and no file moves between folders.
|
|
859
862
|
|
|
860
863
|
Tracked files are renamed with `git mv`, so the renames are staged and the
|
|
861
864
|
next commit carries them whole; a plain rename would leave an unstaged
|
|
@@ -875,10 +878,61 @@ renamed (`E806`) where the rename would be lost:
|
|
|
875
878
|
Any rename makes the hook exit 1, so review the renames and commit again.
|
|
876
879
|
`--check` is a dry run that reports what would be renamed.
|
|
877
880
|
|
|
878
|
-
|
|
879
|
-
|
|
880
|
-
|
|
881
|
-
|
|
881
|
+
**New IDs (`E809`).** A file with no ID, or a duplicate ID, gets a new one
|
|
882
|
+
in the same rename, which also applies `E804` and `E805`:
|
|
883
|
+
|
|
884
|
+
- No ID: `Name.bes` becomes `<ID> Name.bes`. The server never gathered it,
|
|
885
|
+
so nothing deployed refers to it.
|
|
886
|
+
- Duplicate ID: the one file on the base branch keeps the ID, and the others
|
|
887
|
+
get new ones (`42- X.bes` -> `1043- X.bes`). Renumbering only new content
|
|
888
|
+
keeps actions and baselines that refer to the old ID working. A file
|
|
889
|
+
renamed from one on the base branch counts as on it, as `git diff -M`
|
|
890
|
+
pairs them. When every file in the group is new, the first by path keeps
|
|
891
|
+
the ID. Nothing is renumbered, and `E806` says why, when two or more are
|
|
892
|
+
already on the base branch (which one keeps the ID is for a person to
|
|
893
|
+
decide), or when the ID is on the base branch but git sees no file here as
|
|
894
|
+
that file renamed (it may be one, changed too much to pair).
|
|
895
|
+
- The new ID is one more than the highest ID any `.bes` path has had in git
|
|
896
|
+
history (`git log --all --full-history -m`: every branch and tag, both
|
|
897
|
+
sides of every merge, every site in the repo) or has now, so it never
|
|
898
|
+
reuses the ID of a deleted or renamed file. It sees only the branches and
|
|
899
|
+
tags this clone has fetched: fetch first. Gaps are never filled. For a site whose older IDs predate the git
|
|
900
|
+
repo, set a floor with `--min-id`.
|
|
901
|
+
- The base branch is the first of `--base-ref`, `PRE_COMMIT_FROM_REF` (set by `pre-commit run --from-ref`)
|
|
902
|
+
and `origin/$GITHUB_BASE_REF` (set in a GitHub `pull_request` run) that is
|
|
903
|
+
a commit in the clone. Without any, it is `HEAD`: in a local commit, the
|
|
904
|
+
staged files are the new ones. Once both files of a duplicate are
|
|
905
|
+
committed on a feature branch, `HEAD` sees neither as new: run with
|
|
906
|
+
`--base-ref origin/main` (or the branch it merges into), as `E806` says.
|
|
907
|
+
- In a shallow clone, git history is incomplete, so no ID is assigned
|
|
908
|
+
(`E806`); likewise for duplicates when `--base-ref` is not a commit.
|
|
909
|
+
- `--disable E809` reports `E800` and `E801` only, as before. Disabling
|
|
910
|
+
`E800` or `E801` also stops `E809` for those files.
|
|
911
|
+
|
|
912
|
+
In GitHub Actions, check out the full history (the default is one commit),
|
|
913
|
+
and give pre-commit the PR's base so that it knows which files are new:
|
|
914
|
+
|
|
915
|
+
```yaml
|
|
916
|
+
- uses: actions/checkout@v7
|
|
917
|
+
with:
|
|
918
|
+
fetch-depth: 0
|
|
919
|
+
- uses: pre-commit/action@v3.0.1
|
|
920
|
+
with:
|
|
921
|
+
extra_args: >-
|
|
922
|
+
--from-ref ${{ github.event.pull_request.base.sha }}
|
|
923
|
+
--to-ref ${{ github.event.pull_request.head.sha }}
|
|
924
|
+
```
|
|
925
|
+
|
|
926
|
+
The hook renames the files on the runner and fails, and the log names each
|
|
927
|
+
new file name, for the author to apply (or run the hook locally with
|
|
928
|
+
`--base-ref` naming the PR's base branch). To commit
|
|
929
|
+
the fix back to a PR branch instead, check out `github.head_ref` with
|
|
930
|
+
`contents: write`, then commit and push any change. Fork PRs cannot be
|
|
931
|
+
pushed to, and a push with `GITHUB_TOKEN` starts no new workflow run. Two
|
|
932
|
+
open PRs can choose the same ID; the one merged second then has a duplicate
|
|
933
|
+
on the base branch and is renumbered when it runs again, so require branches
|
|
934
|
+
to be up to date before merging. A site that assigns IDs after merge from a
|
|
935
|
+
counter file, as CommunityContent does, should use `--disable E809`.
|
|
882
936
|
|
|
883
937
|
Not checked here: more than one object in a `.bes` file, or Baselines
|
|
884
938
|
(repository sites are thought to handle both); schemas, GroupRelevance and
|