fixup compute_dictionary

Author: sokra

turbopack/crates/turbo-persistence/src/static_sorted_file_builder.rs

@@ -183,6 +183,11 @@ fn compute_dictionary<E>(
         return Ok(Vec::new());
     }
 
+    let max_sample_size = max(
+        MIN_COMPRESSION_DICTIONARY_SAMPLE_PER_ENTRY,
+        key_compression_samples_size / 1024,
+    );
+
     let mut sample_sizes = Vec::new();
 
     for entry in entries {
@@ -192,7 +197,8 @@ fn compute_dictionary<E>(
         }
         let len = get_entry_size(entry);
         if len >= MIN_COMPRESSION_DICTIONARY_SAMPLE_PER_ENTRY {
-            let optimal_len = max(MIN_COMPRESSION_DICTIONARY_SAMPLE_PER_ENTRY, len / 8);
+            let optimal_len =
+                (len / 8).clamp(MIN_COMPRESSION_DICTIONARY_SAMPLE_PER_ENTRY, max_sample_size);
             let used_len = min(key_remaining, optimal_len);
             if len <= used_len {
                 sample_sizes.push(len);
@@ -204,12 +210,14 @@ fn compute_dictionary<E>(
             }
         }
     }
-    debug_assert!(buffer.len() == sample_sizes.iter().sum::<usize>());
-    let result = if buffer.len() > MIN_COMPRESSION_SAMPLES_SIZE && sample_sizes.len() > 5 {
-        zstd::dict::from_continuous(buffer, &sample_sizes, dictionary_size)?
-    } else {
-        Vec::new()
-    };
+    /// The zlib dict builder requires at least 7 samples
+    const MIN_SAMPLE_SIZE: usize = 7;
+    let result =
+        if buffer.len() > MIN_COMPRESSION_SAMPLES_SIZE && sample_sizes.len() > MIN_SAMPLE_SIZE {
+            zstd::dict::from_continuous(buffer, &sample_sizes, dictionary_size)?
+        } else {
+            Vec::new()
+        };
     buffer.clear();
     Ok(result)
 }
@@ -221,7 +229,7 @@ fn compute_key_compression_dictionary<E: Entry>(
     total_key_size: usize,
     buffer: &mut Vec<u8>,
 ) -> Result<Vec<u8>> {
-    compute_dictionary(
+    Ok(compute_dictionary(
         entries,
         total_key_size,
         KEY_COMPRESSION_SAMPLES_SIZE,
@@ -242,6 +250,7 @@ fn compute_key_compression_dictionary<E: Entry>(
         buffer,
     )
     .context("Key dictionary creation failed")
+    .unwrap())
 }
 
 /// Computes compression dictionaries from values of all entries