From 83e25ad17f47543d361f3812de765d1adc582db1 Mon Sep 17 00:00:00 2001
From: Spenser Black <spenserblack01@gmail.com>
Date: Fri, 18 Aug 2023 10:10:04 -0400
Subject: [PATCH] Detect files with really long lines as generated (#59)

Files with absurdly long lines are most likely generated, after going
through a minifier or a code generator that produces unformatted code.
---
 gengo/src/generated.rs | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/gengo/src/generated.rs b/gengo/src/generated.rs
index ba912551..1c238871 100644
--- a/gengo/src/generated.rs
+++ b/gengo/src/generated.rs
@@ -20,8 +20,16 @@ impl Generated {
         self.globs.iter().any(|g| g.matches_path(filepath.as_ref()))
     }
 
-    fn is_generated_with_read<P: AsRef<Path>>(&self, _filepath: P, _contents: &[u8]) -> bool {
-        false
+    fn is_generated_with_read<P: AsRef<Path>>(&self, _filepath: P, contents: &[u8]) -> bool {
+        self.likely_minified(contents)
+    }
+
+    fn likely_minified(&self, contents: &[u8]) -> bool {
+        // NOTE If the first 10 lines are really long, it's probably minified.
+        contents
+            .split(|&b| b == b'\n')
+            .take(10)
+            .any(|line| line.len() > 250)
     }
 
     fn globs() -> Vec<Pattern> {
@@ -52,4 +60,13 @@ mod tests {
         let generated = Generated::new();
         assert_eq!(generated.is_generated_no_read(filepath), expected);
     }
+
+    #[test]
+    fn test_likely_minified() {
+        let generated = Generated::new();
+        let header: Vec<u8> = b"/*!\n  * This is my license etc etc\n */".to_vec();
+        let contents = b"console.log('hello, world!');".repeat(50);
+        let contents = [header, contents].concat();
+        assert!(generated.likely_minified(&contents));
+    }
 }