vector/enrichment_tables/
geoip.rs

1//! Handles enrichment tables for `type = geoip`.
2//! Enrichment data is loaded from one of the MaxMind GeoIP databases,
3//! [MaxMind GeoIP2][maxmind] or [GeoLite2 binary city database][geolite].
4//!
5//! [maxmind]: https://dev.maxmind.com/geoip/geoip2/downloadable
6//! [geolite]: https://dev.maxmind.com/geoip/geoip2/geolite2/#Download_Access
7use std::{fs, net::IpAddr, path::PathBuf, sync::Arc, time::SystemTime};
8
9use maxminddb::{
10    Reader,
11    geoip2::{AnonymousIp, City, ConnectionType, Isp, Names},
12};
13use ordered_float::NotNan;
14use serde::Deserialize;
15use vector_lib::{
16    configurable::configurable_component,
17    enrichment::{Case, Condition, Error, IndexHandle, Table},
18};
19use vrl::value::{ObjectMap, Value};
20
21use crate::config::{EnrichmentTableConfig, GenerateConfig};
22
23// MaxMind GeoIP database files have a type field we can use to recognize specific
24// products. If it is an unknown type, an error will be returned.
25#[derive(Copy, Clone, Debug)]
26#[allow(missing_docs)]
27pub enum DatabaseKind {
28    Asn,
29    Isp,
30    ConnectionType,
31    City,
32    AnonymousIp,
33}
34
35impl TryFrom<&str> for DatabaseKind {
36    type Error = ();
37
38    fn try_from(value: &str) -> Result<Self, Self::Error> {
39        match value {
40            "GeoLite2-ASN" => Ok(Self::Asn),
41            "GeoIP2-ISP" => Ok(Self::Isp),
42            "GeoIP2-Connection-Type" => Ok(Self::ConnectionType),
43            "GeoIP2-City" | "GeoLite2-City" => Ok(Self::City),
44            "GeoIP2-Anonymous-IP" => Ok(Self::AnonymousIp),
45            _ => Err(()),
46        }
47    }
48}
49
50/// Configuration for the `geoip` enrichment table.
51#[derive(Clone, Debug, Eq, PartialEq)]
52#[configurable_component(enrichment_table("geoip"))]
53pub struct GeoipConfig {
54    /// Path to the [MaxMind GeoIP2][geoip2] or [GeoLite2 binary city database file][geolite2]
55    /// (**GeoLite2-City.mmdb**).
56    ///
57    /// Other databases, such as the country database, are not supported.
58    /// `mmdb` enrichment table can be used for other databases.
59    ///
60    /// [geoip2]: https://dev.maxmind.com/geoip/geoip2/downloadable
61    /// [geolite2]: https://dev.maxmind.com/geoip/geoip2/geolite2/#Download_Access
62    pub path: PathBuf,
63
64    /// The locale to use when querying the database.
65    ///
66    /// MaxMind includes localized versions of some of the fields within their database, such as
67    /// country name. This setting can control which of those localized versions are returned by the
68    /// transform.
69    ///
70    /// More information on which portions of the geolocation data are localized, and what languages
71    /// are available, can be found [here][locale_docs].
72    ///
73    /// [locale_docs]: https://support.maxmind.com/hc/en-us/articles/4414877149467-IP-Geolocation-Data#h_01FRRGRYTGZB29ERDBZCX3MR8Q
74    #[serde(default = "default_locale")]
75    pub locale: String,
76}
77
78fn default_locale() -> String {
79    // Valid locales at the time of writing are: "de”, "en", “es”, “fr”, “ja”, “pt-BR”, “ru”, and
80    // “zh-CN”.
81    //
82    // More information, including the up-to-date list of locales, can be found at
83    // https://dev.maxmind.com/geoip/docs/databases/city-and-country?lang=en.
84
85    // TODO: could we detect the system locale and use that as the default locale if it matches one
86    // of the available locales in the dataset, and then fallback to "en" otherwise?
87    "en".to_string()
88}
89
90impl GenerateConfig for GeoipConfig {
91    fn generate_config() -> toml::Value {
92        toml::Value::try_from(Self {
93            path: "/path/to/GeoLite2-City.mmdb".into(),
94            locale: default_locale(),
95        })
96        .unwrap()
97    }
98}
99
100impl EnrichmentTableConfig for GeoipConfig {
101    async fn build(
102        &self,
103        _: &crate::config::GlobalOptions,
104    ) -> crate::Result<Box<dyn Table + Send + Sync>> {
105        Ok(Box::new(Geoip::new(self.clone())?))
106    }
107}
108
109#[derive(Clone)]
110/// A struct that implements [vector_lib::enrichment::Table] to handle loading enrichment data from a GeoIP database.
111pub struct Geoip {
112    config: GeoipConfig,
113    dbreader: Arc<maxminddb::Reader<Vec<u8>>>,
114    dbkind: DatabaseKind,
115    last_modified: SystemTime,
116}
117
118fn lookup_value<'de, A: Deserialize<'de>>(
119    dbreader: &'de Reader<Vec<u8>>,
120    address: IpAddr,
121) -> crate::Result<Option<(A, String)>> {
122    let result = dbreader.lookup(address)?;
123    match result.decode::<A>()? {
124        Some(data) => {
125            let network = result.network()?.to_string();
126            Ok(Some((data, network)))
127        }
128        None => Ok(None),
129    }
130}
131
132impl Geoip {
133    /// Creates a new GeoIP struct from the provided config.
134    pub fn new(config: GeoipConfig) -> crate::Result<Self> {
135        let dbreader = Arc::new(Reader::open_readfile(&config.path)?);
136        let dbkind =
137            DatabaseKind::try_from(dbreader.metadata.database_type.as_str()).map_err(|_| {
138                format!(
139                    "Unsupported MMDB database type ({}). Use `mmdb` enrichment table instead.",
140                    dbreader.metadata.database_type
141                )
142            })?;
143
144        // Check if we can read database with dummy Ip.
145        let ip = IpAddr::V4(std::net::Ipv4Addr::UNSPECIFIED);
146        match dbkind {
147            // Isp
148            DatabaseKind::Asn | DatabaseKind::Isp => lookup_value::<Isp>(&dbreader, ip).map(|_| ()),
149            DatabaseKind::ConnectionType => {
150                lookup_value::<ConnectionType>(&dbreader, ip).map(|_| ())
151            }
152            DatabaseKind::City => lookup_value::<City>(&dbreader, ip).map(|_| ()),
153            DatabaseKind::AnonymousIp => lookup_value::<AnonymousIp>(&dbreader, ip).map(|_| ()),
154        }?;
155
156        Ok(Geoip {
157            last_modified: fs::metadata(&config.path)?.modified()?,
158            dbreader,
159            dbkind,
160            config,
161        })
162    }
163
164    fn lookup(&self, ip: IpAddr, select: Option<&[String]>) -> Option<ObjectMap> {
165        let mut map = ObjectMap::new();
166        let mut add_field = |key: &str, value: Option<Value>| {
167            if select
168                .map(|fields| fields.iter().any(|field| field == key))
169                .unwrap_or(true)
170            {
171                map.insert(key.into(), value.unwrap_or(Value::Null));
172            }
173        };
174
175        macro_rules! add_field {
176            ($k:expr_2021, $v:expr_2021) => {
177                add_field($k, $v.map(Into::into))
178            };
179        }
180
181        match self.dbkind {
182            DatabaseKind::Asn | DatabaseKind::Isp => {
183                let (data, network) = lookup_value::<Isp>(&self.dbreader, ip).ok()??;
184
185                add_field!("autonomous_system_number", data.autonomous_system_number);
186                add_field!(
187                    "autonomous_system_organization",
188                    data.autonomous_system_organization
189                );
190                add_field!("isp", data.isp);
191                add_field!("organization", data.organization);
192                add_field!("network", Some(network));
193            }
194            DatabaseKind::City => {
195                let (data, network): (City, String) =
196                    lookup_value::<City>(&self.dbreader, ip).ok()??;
197
198                add_field!("city_name", self.take_translation(&data.city.names));
199
200                add_field!("continent_code", data.continent.code);
201
202                let country = data.country;
203                add_field!("country_code", country.iso_code);
204                add_field!("country_name", self.take_translation(&country.names));
205
206                let location = data.location;
207                add_field!("timezone", location.time_zone);
208                add_field!(
209                    "latitude",
210                    location.latitude.map(|latitude| Value::Float(
211                        NotNan::new(latitude).expect("latitude cannot be Nan")
212                    ))
213                );
214                add_field!(
215                    "longitude",
216                    location
217                        .longitude
218                        .map(|longitude| NotNan::new(longitude).expect("longitude cannot be Nan"))
219                );
220                add_field!("metro_code", location.metro_code);
221
222                // last subdivision is most specific per https://github.com/maxmind/GeoIP2-java/blob/39385c6ce645374039450f57208b886cf87ade47/src/main/java/com/maxmind/geoip2/model/AbstractCityResponse.java#L96-L107
223                let subdivision = data.subdivisions.last();
224                add_field!(
225                    "region_name",
226                    subdivision.map(|s| self.take_translation(&s.names))
227                );
228
229                add_field!(
230                    "region_code",
231                    subdivision.and_then(|subdivision| subdivision.iso_code)
232                );
233                add_field!("postal_code", data.postal.code);
234                add_field!("network", Some(network));
235            }
236            DatabaseKind::ConnectionType => {
237                let (data, network) = lookup_value::<ConnectionType>(&self.dbreader, ip).ok()??;
238
239                add_field!("connection_type", data.connection_type);
240                add_field!("network", Some(network));
241            }
242            DatabaseKind::AnonymousIp => {
243                let (data, network) = lookup_value::<AnonymousIp>(&self.dbreader, ip).ok()??;
244
245                add_field!("is_anonymous", data.is_anonymous);
246                add_field!("is_anonymous_vpn", data.is_anonymous_vpn);
247                add_field!("is_hosting_provider", data.is_hosting_provider);
248                add_field!("is_public_proxy", data.is_public_proxy);
249                add_field!("is_residential_proxy", data.is_residential_proxy);
250                add_field!("is_tor_exit_node", data.is_tor_exit_node);
251                add_field!("network", Some(network));
252            }
253        }
254
255        Some(map)
256    }
257
258    fn take_translation<'a>(&self, translations: &'a Names<'a>) -> Option<&'a str> {
259        match self.config.locale.as_ref() {
260            "en" => translations.english,
261            "de" => translations.german,
262            "es" => translations.spanish,
263            "fr" => translations.french,
264            "ja" => translations.japanese,
265            "pt-BR" => translations.brazilian_portuguese,
266            "ru" => translations.russian,
267            "zh-CN" => translations.simplified_chinese,
268            _ => None,
269        }
270    }
271}
272
273impl Table for Geoip {
274    /// Search the enrichment table data with the given condition.
275    /// All conditions must match (AND).
276    ///
277    /// # Errors
278    /// Errors if no rows, or more than 1 row is found.
279    fn find_table_row<'a>(
280        &self,
281        case: Case,
282        condition: &'a [Condition<'a>],
283        select: Option<&[String]>,
284        wildcard: Option<&Value>,
285        index: Option<IndexHandle>,
286    ) -> Result<ObjectMap, Error> {
287        let mut rows = self.find_table_rows(case, condition, select, wildcard, index)?;
288
289        match rows.pop() {
290            Some(row) if rows.is_empty() => Ok(row),
291            Some(_) => Err(Error::MoreThanOneRowFound),
292            None => Err(Error::NoRowsFound),
293        }
294    }
295
296    /// Search the enrichment table data with the given condition.
297    /// All conditions must match (AND).
298    /// Can return multiple matched records
299    fn find_table_rows<'a>(
300        &self,
301        _: Case,
302        condition: &'a [Condition<'a>],
303        select: Option<&[String]>,
304        _wildcard: Option<&Value>,
305        _: Option<IndexHandle>,
306    ) -> Result<Vec<ObjectMap>, Error> {
307        match condition.first() {
308            Some(_) if condition.len() > 1 => Err(Error::OnlyOneConditionAllowed),
309            Some(Condition::Equals { value, .. }) => {
310                let ip = value
311                    .to_string_lossy()
312                    .parse::<IpAddr>()
313                    .map_err(|source| Error::InvalidAddress { source })?;
314                Ok(self
315                    .lookup(ip, select)
316                    .map(|values| vec![values])
317                    .unwrap_or_default())
318            }
319            Some(_) => Err(Error::OnlyEqualityConditionAllowed),
320            None => Err(Error::MissingCondition { kind: "IP" }),
321        }
322    }
323
324    /// Hints to the enrichment table what data is going to be searched to allow it to index the
325    /// data in advance.
326    ///
327    /// # Errors
328    /// Errors if the fields are not in the table.
329    fn add_index(&mut self, _: Case, fields: &[&str]) -> Result<IndexHandle, Error> {
330        match fields.len() {
331            0 => Err(Error::MissingRequiredField { field: "IP" }),
332            1 => Ok(IndexHandle(0)),
333            _ => Err(Error::OnlyOneFieldAllowed),
334        }
335    }
336
337    /// Returns a list of the field names that are in each index
338    fn index_fields(&self) -> Vec<(Case, Vec<String>)> {
339        Vec::new()
340    }
341
342    /// Returns true if the underlying data has changed and the table needs reloading.
343    fn needs_reload(&self) -> bool {
344        matches!(fs::metadata(&self.config.path)
345            .and_then(|metadata| metadata.modified()),
346            Ok(modified) if modified > self.last_modified)
347    }
348}
349
350impl std::fmt::Debug for Geoip {
351    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
352        write!(
353            f,
354            "Geoip {} database {})",
355            self.config.locale,
356            self.config.path.display()
357        )
358    }
359}
360
361#[cfg(test)]
362mod tests {
363    use super::*;
364
365    #[test]
366    fn city_lookup() {
367        let values = find("2.125.160.216", "tests/data/GeoIP2-City-Test.mmdb").unwrap();
368
369        let mut expected = ObjectMap::new();
370        expected.insert("city_name".into(), "Boxford".into());
371        expected.insert("country_code".into(), "GB".into());
372        expected.insert("continent_code".into(), "EU".into());
373        expected.insert("country_name".into(), "United Kingdom".into());
374        expected.insert("region_code".into(), "WBK".into());
375        expected.insert("region_name".into(), "West Berkshire".into());
376        expected.insert("timezone".into(), "Europe/London".into());
377        expected.insert("latitude".into(), Value::from(51.75));
378        expected.insert("longitude".into(), Value::from(-1.25));
379        expected.insert("postal_code".into(), "OX1".into());
380        expected.insert("metro_code".into(), Value::Null);
381        expected.insert("network".into(), "2.125.160.216/29".into());
382
383        assert_eq!(values, expected);
384    }
385
386    #[test]
387    fn city_partial_lookup() {
388        let values = find_select(
389            "2.125.160.216",
390            "tests/data/GeoIP2-City-Test.mmdb",
391            Some(&["latitude".to_string(), "longitude".to_string()]),
392        )
393        .unwrap();
394
395        let mut expected = ObjectMap::new();
396        expected.insert("latitude".into(), Value::from(51.75));
397        expected.insert("longitude".into(), Value::from(-1.25));
398
399        assert_eq!(values, expected);
400    }
401
402    #[test]
403    fn city_lookup_partial_results() {
404        let values = find("67.43.156.9", "tests/data/GeoIP2-City-Test.mmdb").unwrap();
405
406        let mut expected = ObjectMap::new();
407        expected.insert("city_name".into(), Value::Null);
408        expected.insert("country_code".into(), "BT".into());
409        expected.insert("country_name".into(), "Bhutan".into());
410        expected.insert("continent_code".into(), "AS".into());
411        expected.insert("region_code".into(), Value::Null);
412        expected.insert("region_name".into(), Value::Null);
413        expected.insert("timezone".into(), "Asia/Thimphu".into());
414        expected.insert("latitude".into(), Value::from(27.5));
415        expected.insert("longitude".into(), Value::from(90.5));
416        expected.insert("postal_code".into(), Value::Null);
417        expected.insert("metro_code".into(), Value::Null);
418        expected.insert("network".into(), "67.43.156.0/24".into());
419
420        assert_eq!(values, expected);
421    }
422
423    #[test]
424    fn city_lookup_no_results() {
425        let values = find("10.1.12.1", "tests/data/GeoIP2-City-Test.mmdb");
426
427        assert!(values.is_none());
428    }
429
430    #[test]
431    fn isp_lookup() {
432        let values = find("208.192.1.2", "tests/data/GeoIP2-ISP-Test.mmdb").unwrap();
433
434        let mut expected = ObjectMap::new();
435        expected.insert("autonomous_system_number".into(), 701i64.into());
436        expected.insert(
437            "autonomous_system_organization".into(),
438            "MCI Communications Services, Inc. d/b/a Verizon Business".into(),
439        );
440        expected.insert("isp".into(), "Verizon Business".into());
441        expected.insert("organization".into(), "Verizon Business".into());
442        expected.insert("network".into(), "208.192.0.0/10".into());
443
444        assert_eq!(values, expected);
445    }
446
447    #[test]
448    fn isp_lookup_partial_results() {
449        let values = find("2600:7000::1", "tests/data/GeoLite2-ASN-Test.mmdb").unwrap();
450
451        let mut expected = ObjectMap::new();
452        expected.insert("autonomous_system_number".into(), 6939i64.into());
453        expected.insert(
454            "autonomous_system_organization".into(),
455            "Hurricane Electric, Inc.".into(),
456        );
457        expected.insert("isp".into(), Value::Null);
458        expected.insert("organization".into(), Value::Null);
459        expected.insert("network".into(), "2600:7000::/24".into());
460
461        assert_eq!(values, expected);
462    }
463
464    #[test]
465    fn isp_lookup_no_results() {
466        let values = find("10.1.12.1", "tests/data/GeoLite2-ASN-Test.mmdb");
467
468        assert!(values.is_none());
469    }
470
471    #[test]
472    fn connection_type_lookup_success() {
473        let values = find(
474            "201.243.200.1",
475            "tests/data/GeoIP2-Connection-Type-Test.mmdb",
476        )
477        .unwrap();
478
479        let mut expected = ObjectMap::new();
480        expected.insert("connection_type".into(), "Corporate".into());
481        expected.insert("network".into(), "201.243.200.0/24".into());
482
483        assert_eq!(values, expected);
484    }
485
486    #[test]
487    fn connection_type_lookup_missing() {
488        let values = find("10.1.12.1", "tests/data/GeoIP2-Connection-Type-Test.mmdb");
489
490        assert!(values.is_none());
491    }
492
493    #[test]
494    fn custom_mmdb_type_error() {
495        let result = Geoip::new(GeoipConfig {
496            path: "tests/data/custom-type.mmdb".into(),
497            locale: default_locale(),
498        });
499
500        assert!(result.is_err());
501    }
502    #[test]
503    fn anonymous_ip_lookup() {
504        let values = find("101.99.92.179", "tests/data/GeoIP2-Anonymous-IP-Test.mmdb").unwrap();
505
506        let mut expected = ObjectMap::new();
507        expected.insert("is_anonymous".into(), true.into());
508        expected.insert("is_anonymous_vpn".into(), true.into());
509        expected.insert("is_hosting_provider".into(), true.into());
510        expected.insert("is_tor_exit_node".into(), true.into());
511        expected.insert("is_public_proxy".into(), Value::Null);
512        expected.insert("is_residential_proxy".into(), Value::Null);
513        expected.insert("network".into(), "101.99.92.179/32".into());
514
515        assert_eq!(values, expected);
516    }
517
518    #[test]
519    fn anonymous_ip_lookup_no_results() {
520        let values = find("10.1.12.1", "tests/data/GeoIP2-Anonymous-IP-Test.mmdb");
521
522        assert!(values.is_none());
523    }
524
525    fn find(ip: &str, database: &str) -> Option<ObjectMap> {
526        find_select(ip, database, None)
527    }
528
529    fn find_select(ip: &str, database: &str, select: Option<&[String]>) -> Option<ObjectMap> {
530        Geoip::new(GeoipConfig {
531            path: database.into(),
532            locale: default_locale(),
533        })
534        .unwrap()
535        .find_table_rows(
536            Case::Insensitive,
537            &[Condition::Equals {
538                field: "ip",
539                value: ip.into(),
540            }],
541            select,
542            None,
543            None,
544        )
545        .unwrap()
546        .pop()
547    }
548}
vector/enrichment_tables/geoip.rs

vector/enrichment_tables/
geoip.rs
