feat: add a flag & config for skipping path glob patterns

cli/Cargo.toml

@@ -27,6 +27,7 @@ squawk-parser = { version = "0.0.0", path = "../parser" }
 squawk-linter = { version = "0.0.0", path = "../linter" }
 squawk-github = { version = "0.0.0", path = "../github" }
 toml = "0.5.9"
+glob = "0.3.1"
 
 [dev-dependencies]
 insta = "0.16.0"

cli/src/config.rs

@@ -38,6 +38,8 @@ impl std::fmt::Display for ConfigError {
 
 #[derive(Debug, Default, Deserialize)]
 pub struct Config {
+    #[serde(default)]
+    pub excluded_paths: Vec<String>,
     #[serde(default)]
     pub excluded_rules: Vec<RuleViolationKind>,
     #[serde(default)]
@@ -98,6 +100,7 @@ mod test_config {
         let squawk_toml = NamedTempFile::new().expect("generate tempFile");
         let file = r#"
 pg_version = "19.1"
+excluded_paths = ["example.sql"]
 excluded_rules = ["require-concurrent-index-creation"]
 assume_in_transaction = true
         
@@ -126,6 +129,16 @@ excluded_rules = ["require-concurrent-index-creation"]
         assert_debug_snapshot!(Config::parse(Some(squawk_toml.path().to_path_buf())));
     }
     #[test]
+    fn test_load_excluded_paths() {
+        let squawk_toml = NamedTempFile::new().expect("generate tempFile");
+        let file = r#"
+excluded_paths = ["example.sql"]
+        
+        "#;
+        fs::write(&squawk_toml, file).expect("Unable to write file");
+        assert_debug_snapshot!(Config::parse(Some(squawk_toml.path().to_path_buf())));
+    }
+    #[test]
     fn test_load_assume_in_transaction() {
         let squawk_toml = NamedTempFile::new().expect("generate tempFile");
         let file = r#"

cli/src/main.rs

@@ -14,6 +14,7 @@ use crate::reporter::{
 use crate::subcommand::{check_and_comment_on_pr, Command};
 use atty::Stream;
 use config::Config;
+use glob::Pattern;
 use log::info;
 use simplelog::CombinedLogger;
 use squawk_linter::versions::Version;
@@ -40,6 +41,12 @@ struct Opt {
     /// Paths to search
     #[structopt(value_name = "path")]
     paths: Vec<String>,
+    /// Paths to exclude
+    ///
+    /// For example:
+    /// --exclude-paths 202401*.sql
+    #[structopt(long = "exclude-paths", use_delimiter = true)]
+    excluded_paths: Option<Vec<String>>,
     /// Exclude specific warnings
     ///
     /// For example:
@@ -117,6 +124,22 @@ fn main() {
         conf.excluded_rules
     };
 
+    // the --exclude-paths flag completely overrides the configuration file.
+    let excluded_paths = if let Some(excluded_paths) = opts.excluded_paths {
+        excluded_paths
+    } else {
+        conf.excluded_paths
+    };
+    let excluded_path_patterns = excluded_paths
+        .iter()
+        .map(|excluded_path| {
+            Pattern::new(excluded_path).unwrap_or_else(|e| {
+                eprintln!("Pattern error: {e}");
+                process::exit(1);
+            })
+        })
+        .collect::<Vec<Pattern>>();
+
     let pg_version = if let Some(pg_version) = opts.pg_version {
         Some(pg_version)
     } else {
@@ -135,6 +158,7 @@ fn main() {
 
     info!("pg version: {:?}", pg_version);
     info!("excluded rules: {:?}", &excluded_rules);
+    info!("excluded paths: {:?}", &excluded_paths);
     info!("assume in a transaction: {:?}", assume_in_transaction);
 
     let mut clap_app = Opt::clap();
@@ -149,6 +173,7 @@ fn main() {
                 is_stdin,
                 opts.stdin_filepath,
                 &excluded_rules,
+                &excluded_path_patterns,
                 pg_version,
                 assume_in_transaction,
             ),
@@ -167,6 +192,7 @@ fn main() {
                 read_stdin,
                 opts.stdin_filepath,
                 &excluded_rules,
+                &excluded_path_patterns,
                 pg_version,
                 assume_in_transaction,
             ) {

cli/src/reporter.rs

@@ -1,5 +1,6 @@
 use console::strip_ansi_codes;
 use console::style;
+use glob::Pattern;
 use log::info;
 use serde::Serialize;
 use squawk_linter::errors::CheckSqlError;
@@ -164,6 +165,7 @@ pub fn check_files(
     read_stdin: bool,
     stdin_path: Option<String>,
     excluded_rules: &[RuleViolationKind],
+    excluded_paths: &[Pattern],
     pg_version: Option<Version>,
     assume_in_transaction: bool,
 ) -> Result<Vec<ViolationContent>, CheckFilesError> {
@@ -188,19 +190,37 @@ pub fn check_files(
     }
 
     for path in paths {
-        info!("checking file path: {}", path);
-        let sql = get_sql_from_path(path)?;
-        output_violations.push(process_violations(
-            &sql,
-            path,
-            excluded_rules,
-            pg_version,
-            assume_in_transaction,
-        ));
+        if has_match_in_paths(path, excluded_paths) {
+            info!("skipping exluded file path: {}", path);
+            output_violations.push(ViolationContent {
+                filename: path.into(),
+                sql: String::new(),
+                violations: vec![],
+            });
+        } else {
+            info!("checking file path: {}", path);
+            let sql = get_sql_from_path(path)?;
+            output_violations.push(process_violations(
+                &sql,
+                path,
+                excluded_rules,
+                pg_version,
+                assume_in_transaction,
+            ));
+        }
     }
     Ok(output_violations)
 }
 
+fn has_match_in_paths(path: &str, excluded_paths: &[Pattern]) -> bool {
+    for excluded in excluded_paths {
+        if excluded.matches(path) {
+            return true;
+        }
+    }
+    false
+}
+
 fn get_sql_from_stdin() -> Result<String, io::Error> {
     let mut buffer = String::new();
     let stdin = io::stdin();
@@ -520,7 +540,7 @@ pub fn get_comment_body(files: &[ViolationContent], version: &str) -> String {
         violation_count = violations_count,
         file_count = files.len(),
         sql_file_content = files
-            .into_iter()
+            .iter()
             .filter_map(|x| get_sql_file_content(x).ok())
             .collect::<Vec<String>>()
             .join("\n"),

cli/src/snapshots/squawk__config__test_config__load_assume_in_transaction.snap

@@ -1,10 +1,12 @@
 ---
 source: cli/src/config.rs
 expression: "Config::parse(Some(squawk_toml.path().to_path_buf()))"
+
 ---
 Ok(
     Some(
         Config {
+            excluded_paths: [],
             excluded_rules: [],
             pg_version: None,
             assume_in_transaction: Some(

cli/src/snapshots/squawk__config__test_config__load_cfg_full.snap

@@ -1,10 +1,14 @@
 ---
 source: cli/src/config.rs
 expression: "Config::parse(Some(squawk_toml.path().to_path_buf()))"
+
 ---
 Ok(
     Some(
         Config {
+            excluded_paths: [
+                "example.sql",
+            ],
             excluded_rules: [
                 RequireConcurrentIndexCreation,
             ],

cli/src/snapshots/squawk__config__test_config__load_excluded_paths.snap

@@ -0,0 +1,17 @@
+---
+source: cli/src/config.rs
+expression: "Config::parse(Some(squawk_toml.path().to_path_buf()))"
+
+---
+Ok(
+    Some(
+        Config {
+            excluded_paths: [
+                "example.sql",
+            ],
+            excluded_rules: [],
+            pg_version: None,
+            assume_in_transaction: None,
+        },
+    ),
+)

cli/src/snapshots/squawk__config__test_config__load_excluded_rules.snap

@@ -1,10 +1,12 @@
 ---
 source: cli/src/config.rs
 expression: "Config::parse(Some(squawk_toml.path().to_path_buf()))"
+
 ---
 Ok(
     Some(
         Config {
+            excluded_paths: [],
             excluded_rules: [
                 RequireConcurrentIndexCreation,
             ],

cli/src/snapshots/squawk__config__test_config__load_pg_version.snap

@@ -1,10 +1,12 @@
 ---
 source: cli/src/config.rs
 expression: "Config::parse(Some(squawk_toml.path().to_path_buf()))"
+
 ---
 Ok(
     Some(
         Config {
+            excluded_paths: [],
             excluded_rules: [],
             pg_version: Some(
                 Version {

cli/src/subcommand.rs

@@ -1,5 +1,6 @@
 use crate::reporter::{check_files, get_comment_body, CheckFilesError};
 
+use glob::Pattern;
 use log::info;
 use squawk_github::{actions, app, comment_on_pr, GitHubApi, GithubError};
 use squawk_linter::{versions::Version, violations::RuleViolationKind};
@@ -157,6 +158,7 @@ pub fn check_and_comment_on_pr(
     is_stdin: bool,
     stdin_path: Option<String>,
     root_cmd_exclude: &[RuleViolationKind],
+    root_cmd_exclude_paths: &[Pattern],
     pg_version: Option<Version>,
     assume_in_transaction: bool,
 ) -> Result<(), SquawkError> {
@@ -188,6 +190,7 @@ pub fn check_and_comment_on_pr(
         is_stdin,
         stdin_path,
         &concat(&exclude.unwrap_or_default(), root_cmd_exclude),
+        root_cmd_exclude_paths,
         pg_version,
         assume_in_transaction,
     )?;