From d2a90591bf53dd75fa719a295dc5819846fe90cd Mon Sep 17 00:00:00 2001
From: Vivek Chavan <111511821+vivekchavan14@users.noreply.github.com>
Date: Wed, 22 Oct 2025 11:51:38 +0530
Subject: [PATCH] fix(parser): Handle CRLF line endings in bash and PowerShell
 parsers (#6889)

Fixes #6867

The bash and PowerShell parsers were not properly handling CRLF (Windows-style)
line endings when parsing script arguments. The regex patterns were only
matching LF line endings, causing scripts with CRLF to fail metadata generation.

Changes:
- Updated RE_BASH regex to optionally match \r before line end (\r?$)
- Updated RE_POWERSHELL_ARGS regex to optionally match \r (\r?)
- Added test case test_parse_bash_sig_with_crlf() to verify CRLF handling

This fix ensures that bash scripts created on Windows (or with editors
using CRLF) will correctly parse arguments and generate proper metadata
via 'wmill script generate-metadata' command.

Python parser was verified to work correctly as it uses an AST parser
that inherently handles line endings properly.
---
 .../parsers/windmill-parser-bash/src/lib.rs   | 63 ++++++++++++++++++-
 1 file changed, 61 insertions(+), 2 deletions(-)

diff --git a/backend/parsers/windmill-parser-bash/src/lib.rs b/backend/parsers/windmill-parser-bash/src/lib.rs
index f179243d68..4e3d70c038 100644
--- a/backend/parsers/windmill-parser-bash/src/lib.rs
+++ b/backend/parsers/windmill-parser-bash/src/lib.rs
@@ -45,10 +45,10 @@ pub fn parse_powershell_sig(code: &str) -> anyhow::Result<MainArgSignature> {
 }
 
 lazy_static::lazy_static! {
-    static ref RE_BASH: Regex = Regex::new(r#"(?m)^(\w+)="\$(?:(\d+)|\{(\d+)\}|\{(\d+):-(.*)\})"(?:[\t ]*)?(?:#.*)?$"#).unwrap();
+    static ref RE_BASH: Regex = Regex::new(r#"(?m)^(\w+)="\$(?:(\d+)|\{(\d+)\}|\{(\d+):-(.*)\})"(?:[\t ]*)?(?:#.*)?\r?$"#).unwrap();
 
     pub static ref RE_POWERSHELL_PARAM: Regex = Regex::new(r#"(?m)param[\t ]*\(([^)]*)\)"#).unwrap();
-    static ref RE_POWERSHELL_ARGS: Regex = Regex::new(r#"(?:\[([\w\[\]]+)\])?\$(\w+)[\t ]*(?:=[\t ]*(?:(?:(?:"|')([^"\n\r\$]*)(?:"|'))|([\d.]+)))?"#).unwrap();
+    static ref RE_POWERSHELL_ARGS: Regex = Regex::new(r#"(?:\[([\w\[\]]+)\])?\$(\w+)[\t ]*(?:=[\t ]*(?:(?:(?:"|')([^"\n\r\$]*)(?:"|'))|([\d.]+)))?\r?"#).unwrap();
 }
 
 fn parse_bash_file(code: &str) -> anyhow::Result<Option<Vec<Arg>>> {
@@ -352,4 +352,63 @@ non_required="${5:-}"
         );
         Ok(())
     }
+
+    #[test]
+    fn test_parse_bash_sig_with_crlf() -> anyhow::Result<()> {
+        // Test with CRLF line endings (Windows-style)
+        let code = "\r\ntoken=\"$1\"\r\nimage=\"$2\"\r\ndigest=\"${3:-latest with spaces}\"\r\ntext=\"$4\" # with comment\r\nnon_required=\"${5:-}\"\r\n\r\n\r\n";
+        assert_eq!(
+            parse_bash_sig(code)?,
+            MainArgSignature {
+                star_args: false,
+                star_kwargs: false,
+                args: vec![
+                    Arg {
+                        otyp: None,
+                        name: "token".to_string(),
+                        typ: Typ::Str(None),
+                        default: None,
+                        has_default: false,
+                        oidx: None
+                    },
+                    Arg {
+                        otyp: None,
+                        name: "image".to_string(),
+                        typ: Typ::Str(None),
+                        default: None,
+                        has_default: false,
+                        oidx: None
+                    },
+                    Arg {
+                        otyp: None,
+                        name: "digest".to_string(),
+                        typ: Typ::Str(None),
+                        default: Some(json!("latest with spaces")),
+                        has_default: true,
+                        oidx: None
+                    },
+                    Arg {
+                        otyp: None,
+                        name: "text".to_string(),
+                        typ: Typ::Str(None),
+                        default: None,
+                        has_default: false,
+                        oidx: None
+                    },
+                    Arg {
+                        otyp: None,
+                        name: "non_required".to_string(),
+                        typ: Typ::Str(None),
+                        default: Some(json!("")),
+                        has_default: true,
+                        oidx: None
+                    }
+                ],
+                no_main_func: None,
+                has_preprocessor: None
+            }
+        );
+
+        Ok(())
+    }
 }