add extended network sibling check based on sibling orgs

src/asinfo/mod.rs

@@ -62,12 +62,13 @@
 
 mod hegemony;
 mod population;
+mod sibling_orgs;
 
 pub use crate::asinfo::hegemony::HegemonyData;
 pub use crate::asinfo::population::AsnPopulationData;
+use crate::asinfo::sibling_orgs::SiblingOrgsUtils;
 use crate::BgpkitCommons;
 use anyhow::{anyhow, Result};
-use oneio::OneIoError;
 use serde::{Deserialize, Serialize};
 use std::collections::HashMap;
 use tracing::info;
@@ -95,6 +96,7 @@ const BGPKIT_ASN_TXT_MIRROR_URL: &str = "https://data.bgpkit.com/commons/asn.txt
 
 pub struct AsInfoUtils {
     pub asinfo_map: HashMap<u32, AsInfo>,
+    pub sibling_orgs: Option<SiblingOrgsUtils>,
     pub load_as2org: bool,
     pub load_population: bool,
     pub load_hegemony: bool,
@@ -103,8 +105,14 @@ pub struct AsInfoUtils {
 impl AsInfoUtils {
     pub fn new(load_as2org: bool, load_population: bool, load_hegemony: bool) -> Result<Self> {
         let asinfo_map = get_asinfo_map(load_as2org, load_population, load_hegemony)?;
+        let sibling_orgs = if load_as2org {
+            Some(SiblingOrgsUtils::new()?)
+        } else {
+            None
+        };
         Ok(AsInfoUtils {
             asinfo_map,
+            sibling_orgs,
             load_as2org,
             load_population,
             load_hegemony,
@@ -225,7 +233,18 @@ impl BgpkitCommons {
             let org_1_opt = info_1_opt.unwrap().as2org;
             let org_2_opt = info_2_opt.unwrap().as2org;
             if org_1_opt.is_some() && org_2_opt.is_some() {
-                return Ok(org_1_opt.unwrap().org_id == org_2_opt.unwrap().org_id);
+                let org_id_1 = org_1_opt.unwrap().org_id;
+                let org_id_2 = org_2_opt.unwrap().org_id;
+
+                return Ok(org_id_1 == org_id_2
+                    || self
+                        .asinfo
+                        .as_ref()
+                        .unwrap()
+                        .sibling_orgs
+                        .as_ref()
+                        .unwrap()
+                        .are_sibling_orgs(org_id_1.as_str(), org_id_2.as_str()));
             }
         }
         Ok(false)

src/asinfo/sibling_orgs.rs

@@ -0,0 +1,72 @@
+use anyhow::Result;
+use std::collections::{HashMap, HashSet};
+use tracing::info;
+
+const BGPKIT_SIBLING_ORGS_URL: &str = "https://data.bgpkit.com/commons/sibling-orgs.txt";
+
+pub struct SiblingOrgsUtils {
+    sibling_orgs_map: HashMap<String, HashSet<String>>,
+}
+
+impl SiblingOrgsUtils {
+    pub fn new() -> Result<Self> {
+        info!(
+            "loading sibling orgs information from {}",
+            BGPKIT_SIBLING_ORGS_URL
+        );
+        let mut sibling_orgs = vec![];
+        for line in oneio::read_lines(BGPKIT_SIBLING_ORGS_URL)? {
+            let line_str = line?.trim().to_string();
+            if line_str.is_empty() || line_str.starts_with('#') {
+                // skip empty line or line started with #
+                continue;
+            }
+            let orgs: Vec<String> = line_str.split_whitespace().map(|x| x.to_owned()).collect();
+            sibling_orgs.push(orgs);
+        }
+
+        let mut res_map = HashMap::new();
+        for sibling_lst in sibling_orgs {
+            let mut org_set: HashSet<String> = HashSet::new();
+            sibling_lst.iter().for_each(|org| {
+                org_set.insert(org.to_lowercase());
+            });
+
+            sibling_lst.iter().for_each(|org| {
+                let org_id = org.to_owned();
+                res_map.insert(org_id.to_lowercase(), org_set.clone());
+            });
+        }
+
+        Ok(SiblingOrgsUtils {
+            sibling_orgs_map: res_map,
+        })
+    }
+
+    pub fn are_sibling_orgs(&self, org_1: &str, org_2: &str) -> bool {
+        if let Some(s) = self.sibling_orgs_map.get(org_1.to_lowercase().as_str()) {
+            if s.contains(org_2.to_lowercase().as_str()) {
+                return true;
+            }
+        }
+
+        false
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_sibling_orgs() {
+        let utils = SiblingOrgsUtils::new().unwrap();
+
+        // GTT
+        assert!(utils.are_sibling_orgs("GC-494-ARIN", "ORG-GCI2-RIPE"));
+        // GTT with random cases
+        assert!(utils.are_sibling_orgs("Gc-494-ArIn", "OrG-gCi2-RiPe"));
+        // GTT and Cogent (not sibling)
+        assert!(!utils.are_sibling_orgs("GC-494-ARIN", "COGC-ARIN"));
+    }
+}

src/lib.rs

@@ -148,3 +148,15 @@ impl BgpkitCommons {
         Ok(())
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_siblings() {
+        let mut commons = BgpkitCommons::new();
+        commons.load_asinfo(true, false, false).unwrap();
+        assert!(commons.asinfo_are_siblings(174, 1239).unwrap());
+    }
+}