clean up

src/configure/mod.rs

@@ -77,14 +77,14 @@ pub fn from_json_file() -> Option<Config> {
 File signature and encapsulation type from file
  See: https://www.tcpdump.org/linktypes.html
 */
-// Futher:file.len() is included in metadata() but only shows up if called explicitly. Maybe this is not needed at all in the end
+// Further:file.len() is included in metadata() but only shows up if called explicitly. Maybe this is not needed at all in the end
 // This would be needed for comparability over time. print metadata and  you will see
 fn bytes_from_file(entry: std::path::PathBuf) -> Result<([u8; 4], u16, u16), std::io::Error> {
     let mut magic_number: [u8; 4] = [0; 4];
     let mut buffer: [u8; 32] = [0; 32];
     let mut _file = File::open(entry.to_owned())?;
     _file.read_exact(&mut buffer)?;
-    magic_number.clone_from_slice(&buffer[0..4]);
+    magic_number.copy_from_slice(&buffer[0..4]);
     let enc_pcap: u16 = LittleEndian::read_u16(&buffer[20..22]);
     let enc_pcapng: u16 = LittleEndian::read_u16(&buffer[12..14]);
 

src/main.rs

@@ -11,6 +11,8 @@ use tokio_postgres::{Error, NoTls};
 /* conditionals */
 const FROM_FILE: bool = false;
 const FROM_DEVICE: bool = true;
+const NON_CHUNKED: bool = true;
+const CHUNKED: bool = false;
 
 fn query_string(insert_max: &usize) -> String {
     let mut insert_template = String::with_capacity(insert_max * 8 + 43);
@@ -59,13 +61,13 @@ async fn main() -> Result<(), Error> {
         FROM_FILE => {
             for (_pcap_file, _pcap_info) in pcap_map.iter() {
                 println!("{:?}", &_pcap_file);
-                // TODO: Tuning vector capacity according to mean average & std dev of packet size
+                // TODO: Tuning vector capacity according to mean average & std dev of packet sizes
                 let v: Vec<parser::QryData> =
                     parser::parse(&_pcap_file, &config.filter, &config.regex_filter);
+                let packets_serialized = serializer::serialize_packets(v);
+
 //                let mut v = Vec::<parser::QryData>::with_capacity(100000);
 //                v.extend(parser::parse(&_pcap_file, &config.filter, &config.regex_filter));
-
-                let packets_serialized = serializer::serialize_packets(v);
 //                let mut packets_serialized = Vec::<serde_json::Value>::with_capacity(100000);
 //                packets_serialized.extend(serializer::serialize_packets(v));
 
@@ -74,18 +76,17 @@ async fn main() -> Result<(), Error> {
                 let remainder: usize = packets_serialized.len() % config.insert_max;
                 let chunker = &packets_serialized.len() < &config.insert_max;
                 match chunker {
-                    true => {
+                    NON_CHUNKED => {
                         let insert_str = query_string(&packets_serialized.len());
-                        let statement_false = client.prepare(&insert_str).await?;
+                        let statement = client.prepare(&insert_str).await?;
                         client
                             .query_raw(
-                                &statement_false,
+                                &statement,
                                 packets_serialized.iter().map(|p| p as &dyn ToSql),
                             )
                             .await?;
                     }
-
-                    false => {
+                    CHUNKED => {
                         let insert_str = query_string(&config.insert_max);
                         let statement = client.prepare(&insert_str).await?;
 
@@ -98,15 +99,14 @@ async fn main() -> Result<(), Error> {
                                 )
                                 .await?;
                         }
-
                         if  0 < remainder {
                             let rem_str = query_string(&remainder);
-                            let statement_remainder = client.prepare(&rem_str).await?;
+                            let statement = client.prepare(&rem_str).await?;
                             let (_garbage, _input) =
                                 packets_serialized.split_at(packets_serialized.len() - remainder);
                             client
                                 .query_raw(
-                                    &statement_remainder,
+                                    &statement,
                                     _input.to_vec().iter().map(|p| p as &dyn ToSql),
                                 )
                                 .await?;

src/parser.json

@@ -1,7 +1,7 @@
 {
     "insert_max": 20000,
-    "filter": "!ip6 && tcp",
-    "regex_filter": "http://",
+    "filter": "ip6 && tcp",
+    "regex_filter": "(?:http|https)[[::punct::]]//([[::word::]]+\\.)*",
     "from_device": false,
     "parse_device": "enp7s0",
     "pcap_file": "",

src/serializer/mod.rs

@@ -12,7 +12,7 @@ impl Serialize for parser::QryData {
     where
         S: Serializer,
     {
-        let mut state = serializer.serialize_struct("parser::QryData", 11)?;
+        let mut state = serializer.serialize_struct("parser::QryData", 9)?;
         state.serialize_field("time", &self.time)?;
         state.serialize_field("ether_header", &self.ether_header)?;
         state.serialize_field("ipv4_header", &self.ipv4_header)?;
@@ -28,12 +28,12 @@ impl Serialize for parser::QryData {
 
 pub fn serialize_packets(v: Vec<parser::QryData>) -> Vec<serde_json::Value> {
     /* rayon parallelized */
-    // TODO: Benchmark. As far as I tested, this reaps no benefit.
+    // TODO: Benchmark. As far as I've tested, using rayon reaps no benefit nor does it run any slower. I leave it in for now.
     let packets_serialized = v
         .par_iter()
         .map(|x| serde_json::to_value(x).unwrap())
         .collect();
-    //let packets_serialized: Vec<serde_json::Value> = v.par_iter().map(|x| json!(x)).collect();
+//  let packets_serialized: Vec<serde_json::Value> = v.par_iter().map(|x| json!(x)).collect();
 
     packets_serialized
 }
@@ -50,12 +50,3 @@ pub fn serialize_packets_as_string(v: Vec<parser::QryData>) -> Vec<serde_json::V
 
     packets_serialized
 }
-
-#[allow(dead_code)]
-fn smallest_prime_divisor(remainder: usize) -> usize {
-    let smallest_divisor: usize = (2..(remainder / 2))
-        .into_par_iter()
-        .find_first(|x| remainder % x == 0)
-        .unwrap();
-    smallest_divisor
-}