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, 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>> {
122    let result = dbreader.lookup(address)?;
123    Ok(result.decode::<A>()?)
124}
125
126impl Geoip {
127    /// Creates a new GeoIP struct from the provided config.
128    pub fn new(config: GeoipConfig) -> crate::Result<Self> {
129        let dbreader = Arc::new(Reader::open_readfile(&config.path)?);
130        let dbkind =
131            DatabaseKind::try_from(dbreader.metadata.database_type.as_str()).map_err(|_| {
132                format!(
133                    "Unsupported MMDB database type ({}). Use `mmdb` enrichment table instead.",
134                    dbreader.metadata.database_type
135                )
136            })?;
137
138        // Check if we can read database with dummy Ip.
139        let ip = IpAddr::V4(std::net::Ipv4Addr::UNSPECIFIED);
140        match dbkind {
141            // Isp
142            DatabaseKind::Asn | DatabaseKind::Isp => lookup_value::<Isp>(&dbreader, ip).map(|_| ()),
143            DatabaseKind::ConnectionType => {
144                lookup_value::<ConnectionType>(&dbreader, ip).map(|_| ())
145            }
146            DatabaseKind::City => lookup_value::<City>(&dbreader, ip).map(|_| ()),
147            DatabaseKind::AnonymousIp => lookup_value::<AnonymousIp>(&dbreader, ip).map(|_| ()),
148        }?;
149
150        Ok(Geoip {
151            last_modified: fs::metadata(&config.path)?.modified()?,
152            dbreader,
153            dbkind,
154            config,
155        })
156    }
157
158    fn lookup(&self, ip: IpAddr, select: Option<&[String]>) -> Option<ObjectMap> {
159        let mut map = ObjectMap::new();
160        let mut add_field = |key: &str, value: Option<Value>| {
161            if select
162                .map(|fields| fields.iter().any(|field| field == key))
163                .unwrap_or(true)
164            {
165                map.insert(key.into(), value.unwrap_or(Value::Null));
166            }
167        };
168
169        macro_rules! add_field {
170            ($k:expr_2021, $v:expr_2021) => {
171                add_field($k, $v.map(Into::into))
172            };
173        }
174
175        match self.dbkind {
176            DatabaseKind::Asn | DatabaseKind::Isp => {
177                let data = lookup_value::<Isp>(&self.dbreader, ip).ok()??;
178
179                add_field!("autonomous_system_number", data.autonomous_system_number);
180                add_field!(
181                    "autonomous_system_organization",
182                    data.autonomous_system_organization
183                );
184                add_field!("isp", data.isp);
185                add_field!("organization", data.organization);
186            }
187            DatabaseKind::City => {
188                let data: City = lookup_value::<City>(&self.dbreader, ip).ok()??;
189
190                add_field!("city_name", self.take_translation(&data.city.names));
191
192                add_field!("continent_code", data.continent.code);
193
194                let country = data.country;
195                add_field!("country_code", country.iso_code);
196                add_field!("country_name", self.take_translation(&country.names));
197
198                let location = data.location;
199                add_field!("timezone", location.time_zone);
200                add_field!(
201                    "latitude",
202                    location.latitude.map(|latitude| Value::Float(
203                        NotNan::new(latitude).expect("latitude cannot be Nan")
204                    ))
205                );
206                add_field!(
207                    "longitude",
208                    location
209                        .longitude
210                        .map(|longitude| NotNan::new(longitude).expect("longitude cannot be Nan"))
211                );
212                add_field!("metro_code", location.metro_code);
213
214                // 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
215                let subdivision = data.subdivisions.last();
216                add_field!(
217                    "region_name",
218                    subdivision.map(|s| self.take_translation(&s.names))
219                );
220
221                add_field!(
222                    "region_code",
223                    subdivision.and_then(|subdivision| subdivision.iso_code)
224                );
225                add_field!("postal_code", data.postal.code);
226            }
227            DatabaseKind::ConnectionType => {
228                let data = lookup_value::<ConnectionType>(&self.dbreader, ip).ok()??;
229
230                add_field!("connection_type", data.connection_type);
231            }
232            DatabaseKind::AnonymousIp => {
233                let data = lookup_value::<AnonymousIp>(&self.dbreader, ip).ok()??;
234
235                add_field!("is_anonymous", data.is_anonymous);
236                add_field!("is_anonymous_vpn", data.is_anonymous_vpn);
237                add_field!("is_hosting_provider", data.is_hosting_provider);
238                add_field!("is_public_proxy", data.is_public_proxy);
239                add_field!("is_residential_proxy", data.is_residential_proxy);
240                add_field!("is_tor_exit_node", data.is_tor_exit_node);
241            }
242        }
243
244        Some(map)
245    }
246
247    fn take_translation<'a>(&self, translations: &'a Names<'a>) -> Option<&'a str> {
248        match self.config.locale.as_ref() {
249            "en" => translations.english,
250            "de" => translations.german,
251            "es" => translations.spanish,
252            "fr" => translations.french,
253            "ja" => translations.japanese,
254            "pt-BR" => translations.brazilian_portuguese,
255            "ru" => translations.russian,
256            "zh-CN" => translations.simplified_chinese,
257            _ => None,
258        }
259    }
260}
261
262impl Table for Geoip {
263    /// Search the enrichment table data with the given condition.
264    /// All conditions must match (AND).
265    ///
266    /// # Errors
267    /// Errors if no rows, or more than 1 row is found.
268    fn find_table_row<'a>(
269        &self,
270        case: Case,
271        condition: &'a [Condition<'a>],
272        select: Option<&[String]>,
273        wildcard: Option<&Value>,
274        index: Option<IndexHandle>,
275    ) -> Result<ObjectMap, String> {
276        let mut rows = self.find_table_rows(case, condition, select, wildcard, index)?;
277
278        match rows.pop() {
279            Some(row) if rows.is_empty() => Ok(row),
280            Some(_) => Err("More than 1 row found".to_string()),
281            None => Err("IP not found".to_string()),
282        }
283    }
284
285    /// Search the enrichment table data with the given condition.
286    /// All conditions must match (AND).
287    /// Can return multiple matched records
288    fn find_table_rows<'a>(
289        &self,
290        _: Case,
291        condition: &'a [Condition<'a>],
292        select: Option<&[String]>,
293        _wildcard: Option<&Value>,
294        _: Option<IndexHandle>,
295    ) -> Result<Vec<ObjectMap>, String> {
296        match condition.first() {
297            Some(_) if condition.len() > 1 => Err("Only one condition is allowed".to_string()),
298            Some(Condition::Equals { value, .. }) => {
299                let ip = value
300                    .to_string_lossy()
301                    .parse::<IpAddr>()
302                    .map_err(|_| "Invalid IP address".to_string())?;
303                Ok(self
304                    .lookup(ip, select)
305                    .map(|values| vec![values])
306                    .unwrap_or_default())
307            }
308            Some(_) => Err("Only equality condition is allowed".to_string()),
309            None => Err("IP condition must be specified".to_string()),
310        }
311    }
312
313    /// Hints to the enrichment table what data is going to be searched to allow it to index the
314    /// data in advance.
315    ///
316    /// # Errors
317    /// Errors if the fields are not in the table.
318    fn add_index(&mut self, _: Case, fields: &[&str]) -> Result<IndexHandle, String> {
319        match fields.len() {
320            0 => Err("IP field is required".to_string()),
321            1 => Ok(IndexHandle(0)),
322            _ => Err("Only one field is allowed".to_string()),
323        }
324    }
325
326    /// Returns a list of the field names that are in each index
327    fn index_fields(&self) -> Vec<(Case, Vec<String>)> {
328        Vec::new()
329    }
330
331    /// Returns true if the underlying data has changed and the table needs reloading.
332    fn needs_reload(&self) -> bool {
333        matches!(fs::metadata(&self.config.path)
334            .and_then(|metadata| metadata.modified()),
335            Ok(modified) if modified > self.last_modified)
336    }
337}
338
339impl std::fmt::Debug for Geoip {
340    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
341        write!(
342            f,
343            "Geoip {} database {})",
344            self.config.locale,
345            self.config.path.display()
346        )
347    }
348}
349
350#[cfg(test)]
351mod tests {
352    use super::*;
353
354    #[test]
355    fn city_lookup() {
356        let values = find("2.125.160.216", "tests/data/GeoIP2-City-Test.mmdb").unwrap();
357
358        let mut expected = ObjectMap::new();
359        expected.insert("city_name".into(), "Boxford".into());
360        expected.insert("country_code".into(), "GB".into());
361        expected.insert("continent_code".into(), "EU".into());
362        expected.insert("country_name".into(), "United Kingdom".into());
363        expected.insert("region_code".into(), "WBK".into());
364        expected.insert("region_name".into(), "West Berkshire".into());
365        expected.insert("timezone".into(), "Europe/London".into());
366        expected.insert("latitude".into(), Value::from(51.75));
367        expected.insert("longitude".into(), Value::from(-1.25));
368        expected.insert("postal_code".into(), "OX1".into());
369        expected.insert("metro_code".into(), Value::Null);
370
371        assert_eq!(values, expected);
372    }
373
374    #[test]
375    fn city_partial_lookup() {
376        let values = find_select(
377            "2.125.160.216",
378            "tests/data/GeoIP2-City-Test.mmdb",
379            Some(&["latitude".to_string(), "longitude".to_string()]),
380        )
381        .unwrap();
382
383        let mut expected = ObjectMap::new();
384        expected.insert("latitude".into(), Value::from(51.75));
385        expected.insert("longitude".into(), Value::from(-1.25));
386
387        assert_eq!(values, expected);
388    }
389
390    #[test]
391    fn city_lookup_partial_results() {
392        let values = find("67.43.156.9", "tests/data/GeoIP2-City-Test.mmdb").unwrap();
393
394        let mut expected = ObjectMap::new();
395        expected.insert("city_name".into(), Value::Null);
396        expected.insert("country_code".into(), "BT".into());
397        expected.insert("country_name".into(), "Bhutan".into());
398        expected.insert("continent_code".into(), "AS".into());
399        expected.insert("region_code".into(), Value::Null);
400        expected.insert("region_name".into(), Value::Null);
401        expected.insert("timezone".into(), "Asia/Thimphu".into());
402        expected.insert("latitude".into(), Value::from(27.5));
403        expected.insert("longitude".into(), Value::from(90.5));
404        expected.insert("postal_code".into(), Value::Null);
405        expected.insert("metro_code".into(), Value::Null);
406
407        assert_eq!(values, expected);
408    }
409
410    #[test]
411    fn city_lookup_no_results() {
412        let values = find("10.1.12.1", "tests/data/GeoIP2-City-Test.mmdb");
413
414        assert!(values.is_none());
415    }
416
417    #[test]
418    fn isp_lookup() {
419        let values = find("208.192.1.2", "tests/data/GeoIP2-ISP-Test.mmdb").unwrap();
420
421        let mut expected = ObjectMap::new();
422        expected.insert("autonomous_system_number".into(), 701i64.into());
423        expected.insert(
424            "autonomous_system_organization".into(),
425            "MCI Communications Services, Inc. d/b/a Verizon Business".into(),
426        );
427        expected.insert("isp".into(), "Verizon Business".into());
428        expected.insert("organization".into(), "Verizon Business".into());
429
430        assert_eq!(values, expected);
431    }
432
433    #[test]
434    fn isp_lookup_partial_results() {
435        let values = find("2600:7000::1", "tests/data/GeoLite2-ASN-Test.mmdb").unwrap();
436
437        let mut expected = ObjectMap::new();
438        expected.insert("autonomous_system_number".into(), 6939i64.into());
439        expected.insert(
440            "autonomous_system_organization".into(),
441            "Hurricane Electric, Inc.".into(),
442        );
443        expected.insert("isp".into(), Value::Null);
444        expected.insert("organization".into(), Value::Null);
445
446        assert_eq!(values, expected);
447    }
448
449    #[test]
450    fn isp_lookup_no_results() {
451        let values = find("10.1.12.1", "tests/data/GeoLite2-ASN-Test.mmdb");
452
453        assert!(values.is_none());
454    }
455
456    #[test]
457    fn connection_type_lookup_success() {
458        let values = find(
459            "201.243.200.1",
460            "tests/data/GeoIP2-Connection-Type-Test.mmdb",
461        )
462        .unwrap();
463
464        let mut expected = ObjectMap::new();
465        expected.insert("connection_type".into(), "Corporate".into());
466
467        assert_eq!(values, expected);
468    }
469
470    #[test]
471    fn connection_type_lookup_missing() {
472        let values = find("10.1.12.1", "tests/data/GeoIP2-Connection-Type-Test.mmdb");
473
474        assert!(values.is_none());
475    }
476
477    #[test]
478    fn custom_mmdb_type_error() {
479        let result = Geoip::new(GeoipConfig {
480            path: "tests/data/custom-type.mmdb".into(),
481            locale: default_locale(),
482        });
483
484        assert!(result.is_err());
485    }
486    #[test]
487    fn anonymous_ip_lookup() {
488        let values = find("101.99.92.179", "tests/data/GeoIP2-Anonymous-IP-Test.mmdb").unwrap();
489
490        let mut expected = ObjectMap::new();
491        expected.insert("is_anonymous".into(), true.into());
492        expected.insert("is_anonymous_vpn".into(), true.into());
493        expected.insert("is_hosting_provider".into(), true.into());
494        expected.insert("is_tor_exit_node".into(), true.into());
495        expected.insert("is_public_proxy".into(), Value::Null);
496        expected.insert("is_residential_proxy".into(), Value::Null);
497
498        assert_eq!(values, expected);
499    }
500
501    #[test]
502    fn anonymous_ip_lookup_no_results() {
503        let values = find("10.1.12.1", "tests/data/GeoIP2-Anonymous-IP-Test.mmdb");
504
505        assert!(values.is_none());
506    }
507
508    fn find(ip: &str, database: &str) -> Option<ObjectMap> {
509        find_select(ip, database, None)
510    }
511
512    fn find_select(ip: &str, database: &str, select: Option<&[String]>) -> Option<ObjectMap> {
513        Geoip::new(GeoipConfig {
514            path: database.into(),
515            locale: default_locale(),
516        })
517        .unwrap()
518        .find_table_rows(
519            Case::Insensitive,
520            &[Condition::Equals {
521                field: "ip",
522                value: ip.into(),
523            }],
524            select,
525            None,
526            None,
527        )
528        .unwrap()
529        .pop()
530    }
531}
vector/enrichment_tables/geoip.rs

vector/enrichment_tables/
geoip.rs
