benchmarking

benches/common/data_generator.rs

@@ -0,0 +1,131 @@
+// benches/common/data_generator.rs
+use rand::{Rng, SeedableRng};
+use rand::rngs::StdRng;
+
+/// Deterministic data generator for benchmarks
+pub struct DataGenerator {
+    rng: StdRng,
+}
+
+impl DataGenerator {
+    /// Create a new data generator with a fixed seed for reproducibility
+    pub fn new(seed: u64) -> Self {
+        Self {
+            rng: StdRng::seed_from_u64(seed),
+        }
+    }
+
+    /// Generate a single key with the given prefix and ID
+    pub fn generate_key(&self, prefix: &str, id: usize) -> String {
+        format!("{}:{:08}", prefix, id)
+    }
+
+    /// Generate a random string value of the specified size
+    pub fn generate_value(&mut self, size: usize) -> String {
+        const CHARSET: &[u8] = b"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
+        (0..size)
+            .map(|_| {
+                let idx = self.rng.gen_range(0..CHARSET.len());
+                CHARSET[idx] as char
+            })
+            .collect()
+    }
+
+    /// Generate a batch of key-value pairs
+    pub fn generate_string_pairs(&mut self, count: usize, value_size: usize) -> Vec<(String, String)> {
+        (0..count)
+            .map(|i| {
+                let key = self.generate_key("bench:key", i);
+                let value = self.generate_value(value_size);
+                (key, value)
+            })
+            .collect()
+    }
+
+    /// Generate hash data (key -> field-value pairs)
+    pub fn generate_hash_data(&mut self, num_hashes: usize, fields_per_hash: usize, value_size: usize) 
+        -> Vec<(String, Vec<(String, String)>)> {
+        (0..num_hashes)
+            .map(|i| {
+                let hash_key = self.generate_key("bench:hash", i);
+                let fields: Vec<(String, String)> = (0..fields_per_hash)
+                    .map(|j| {
+                        let field = format!("field{}", j);
+                        let value = self.generate_value(value_size);
+                        (field, value)
+                    })
+                    .collect();
+                (hash_key, fields)
+            })
+            .collect()
+    }
+
+    /// Generate list data (key -> list of elements)
+    pub fn generate_list_data(&mut self, num_lists: usize, elements_per_list: usize, element_size: usize) 
+        -> Vec<(String, Vec<String>)> {
+        (0..num_lists)
+            .map(|i| {
+                let list_key = self.generate_key("bench:list", i);
+                let elements: Vec<String> = (0..elements_per_list)
+                    .map(|_| self.generate_value(element_size))
+                    .collect();
+                (list_key, elements)
+            })
+            .collect()
+    }
+
+    /// Generate keys for pattern matching tests
+    pub fn generate_pattern_keys(&mut self, count: usize) -> Vec<String> {
+        let mut keys = Vec::new();
+        
+        // Generate keys with different patterns
+        for i in 0..count / 3 {
+            keys.push(format!("user:{}:profile", i));
+        }
+        for i in 0..count / 3 {
+            keys.push(format!("session:{}:data", i));
+        }
+        for i in 0..count / 3 {
+            keys.push(format!("cache:{}:value", i));
+        }
+        
+        keys
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_deterministic_generation() {
+        let mut generator1 = DataGenerator::new(42);
+        let mut generator2 = DataGenerator::new(42);
+
+        let pairs1 = generator1.generate_string_pairs(10, 50);
+        let pairs2 = generator2.generate_string_pairs(10, 50);
+
+        assert_eq!(pairs1, pairs2, "Same seed should produce same data");
+    }
+
+    #[test]
+    fn test_value_size() {
+        let mut generator = DataGenerator::new(42);
+        let value = generator.generate_value(100);
+        assert_eq!(value.len(), 100);
+    }
+
+    #[test]
+    fn test_hash_generation() {
+        let mut generator = DataGenerator::new(42);
+        let hashes = generator.generate_hash_data(5, 10, 50);
+        
+        assert_eq!(hashes.len(), 5);
+        for (_, fields) in hashes {
+            assert_eq!(fields.len(), 10);
+            for (_, value) in fields {
+                assert_eq!(value.len(), 50);
+            }
+        }
+    }
+}