From 399e1e54ea0a1770c58c116bc3826167eb20f1bc Mon Sep 17 00:00:00 2001
From: Jason Kim <git@jasonk.me>
Date: Fri, 25 Oct 2024 16:22:19 -0700
Subject: [PATCH] [#24671] YSQL: add postgres linter

Summary:
Add a basic linter for postgres code using [arc lint
script-and-regex][1].  Implement the script using a bash script that
runs some grep searches on the given file.  It is not powerful enough to
check for certain things such as whether indentation is aligned properly
with respect to other lines, but it is still better to have some basic
enforcement over nothing.

[1]: https://secure.phabricator.com/book/phabricator/article/arcanist_lint_script_and_regex/
Jira: DB-13743

Test Plan:
manually check arc lint on a few postgres files

Close: #24671

Reviewers: mihnea, tfoucher

Reviewed By: tfoucher

Subscribers: yql

Differential Revision: https://phorge.dev.yugabyte.com/D36535
---
 .arclint                     | 22 ++++++++++++++++++++--
 src/postgres/ybsimplelint.sh | 13 +++++++++++++
 2 files changed, 33 insertions(+), 2 deletions(-)
 create mode 100755 src/postgres/ybsimplelint.sh

diff --git a/.arclint b/.arclint
index b8766f017bae..17dc556b0380 100644
--- a/.arclint
+++ b/.arclint
@@ -8,7 +8,6 @@
      "([.]sql$)",
      "([.]txt$)",
      "(^src/odyssey/.*$)",
-     "(^src/postgres/.*$)",
      "(^src/yb/integration-tests/upgrade-tests/builds.xml$)",
      "(^[.]gitmodules$)",
      "(^managed/src/main/resources/alert/alert_templates[.]yml$)",
@@ -101,19 +100,25 @@
         "(^docs/.*[.]md$)",
         "(^[.]fossa[.]yml$)",
         "(^managed/.*[.]conf)",
+        "(^src/postgres/)",
         "(^[.]gitignore$)"
       ]
     },
     "pycodestyle": {
       "type": "pep8",
       "include": "([.]py$)",
-      "exclude": "(^thirdparty/|^[.]ycm_extra_conf[.]py$)",
+      "exclude": [
+        "(^[.]ycm_extra_conf[.]py$)",
+        "(^src/postgres/)",
+        "(^thirdparty/)"
+      ],
       "bin": "pycodestyle",
       "flags": ["--max-line-length=100"]
     },
     "googlecpplint": {
       "type": "googlecpplint",
       "exclude": [
+        "(^src/postgres/)",
         "(^src/yb/yql/cql/ql/kwlist[.]h$)",
         "(^src/yb/gutil/linux_syscall_support[.]h$)",
         "(^src/yb/gutil/cycleclock-inl[.]h$)",
@@ -140,6 +145,19 @@
         "(^build/c[+][+]11$)": "disabled",
         "(.*)": "error"
       }
+    },
+    "postgres": {
+      "type": "script-and-regex",
+      "exclude": [
+        "(^src/yb/)"
+      ],
+      "include": [
+        "([.]c$)",
+        "([.]h$)"
+      ],
+      "script-and-regex.script": "src/postgres/ybsimplelint.sh",
+      "script-and-regex.regex":
+        "/^(?P<severity>\\w+):(?P<message>\\w+):(?P<line>\\d+):(?P<original>.*)/m"
     }
   }
 }
diff --git a/src/postgres/ybsimplelint.sh b/src/postgres/ybsimplelint.sh
new file mode 100755
index 000000000000..b85393764442
--- /dev/null
+++ b/src/postgres/ybsimplelint.sh
@@ -0,0 +1,13 @@
+#!/usr/bin/env bash
+# Simple linter for postgres code.
+grep -nE '\s+$' "$1" | sed 's/^/error:trailing_whitespace:/'
+grep -nvE '^(	* {0,3}\S|$)' "$1" | sed 's/^/error:leading_whitespace:/'
+grep -nE '(\)|else|else if)\s+{$' "$1" | sed 's/^/warning:likely_bad_brace:/'
+
+grep -nE ',\s*errmsg(_plural)?\(' "$1" | sed 's/^/error:missing_linebreak_before_errmsg:/'
+grep -nE ',\s*errdetail(_plural)?\(' "$1" | sed 's/^/error:missing_linebreak_before_errdetail:/'
+grep -nE ',\s*errhint\(' "$1" | sed 's/^/error:missing_linebreak_before_errhint:/'
+
+grep -nE 'errmsg(_plural)?\("[A-Z][a-z]' "$1" | sed 's/^/warning:likely_bad_capitalization_in_errmsg:/'
+grep -nE 'errdetail(_plural)?\("[a-z]' "$1" | sed 's/^/warning:likely_bad_lowercase_in_errdetail:/'
+grep -nE 'errhint\("[a-z]' "$1" | sed 's/^/warning:likely_bad_lowercase_in_errhint:/'