frontend support for openstack logs

src/main.rs

@@ -7,8 +7,11 @@ mod packages;
 #[command(author, version, about, long_about = None)]
 struct Args {
    /// Name of the raw logfile to convert to a CSV
-   #[arg(short, long)]
-   raw: String,
+   #[arg(long)]
+   raw_linux: Option<String>,
+
+   #[arg(long)]
+   raw_openstack: Option<String>,
 }
 
 #[test]
@@ -28,15 +31,23 @@ fn derive_2grams(trigram:&str) -> Vec<String> {
 fn main() {
     let args = Args::parse();
 
-    let (double_dict, triple_dict, all_token_list) = packages::parser::parse_raw_linux(args.raw);
-    packages::parser::print_dict(&triple_dict);
+    let (mut double_dict, mut triple_dict, mut all_token_list) : (Option<HashMap<String, i32>>, Option<HashMap<String, i32>>, Option<String>) = (None, None, None);
+    if let Some(raw_linux) = args.raw_linux {
+        let (dd, td, atl) = packages::parser::parse_raw_linux(raw_linux);
+        packages::parser::print_dict(&td);
+        triple_dict = Some(td);
+    } else if let Some(raw_openstack) = args.raw_openstack {
+        let (dd, td, atl) = packages::parser::parse_raw_openstack(raw_openstack);
+        packages::parser::print_dict(&td);
+        triple_dict = Some(td);
+    }
 
     const CUTOFF : i32 = 72;
     // let's say that the cutoff is 72 for Linux2k.log.
     // check all of the 2-grams from the less-frequent 3-grams
 
     let mut two_grams = HashMap::new();
-    let (val_set, reverse_d) = packages::parser::reverse_dict(&triple_dict);
+    let (val_set, reverse_d) = packages::parser::reverse_dict(&triple_dict.unwrap());
     for val in val_set.iter().filter(|x| **x < CUTOFF) {
         for key in reverse_d.get(val).unwrap() {
             let val_2grams = derive_2grams(key);

src/packages/parser.rs

@@ -16,7 +16,7 @@ fn linux_regexps() -> Vec<Regex> {
 }
 
 fn openstack_format() -> String {
-    return r"(?P<Logrecord>.*?)\s+(?P<Date>.*?)\s+(?P<Time>.*?)\s+(?P<Pid>.*?)\s+(?P<Level>.*?)\s+(?P<Component>.*?)\s+(\[(?P<ADDR>.*?)\])?\s+(?P<Content>.*)".to_string();
+    return r"'<Logrecord> <Date> <Time> <Pid> <Level> <Component> \[<ADDR>\] <Content>'".to_string();
 }
 
 fn openstack_regexps() -> Vec<Regex> {