|
| 1 | +//! Skip invalid addresses and don't pass them through to the next layer. |
| 2 | +
|
| 3 | +use async_trait::async_trait; |
| 4 | +use metrics::{counter, describe_counter}; |
| 5 | + |
| 6 | +use crate::addresses::Address; |
| 7 | + |
| 8 | +use super::{Geocoded, Geocoder, Result}; |
| 9 | + |
| 10 | +/// Skip invalid addresses and don't pass them through to the next layer. |
| 11 | +pub struct InvalidRecordSkipper { |
| 12 | + // Our inner geocoder that we're normalizing for. |
| 13 | + inner: Box<dyn Geocoder>, |
| 14 | +} |
| 15 | + |
| 16 | +impl InvalidRecordSkipper { |
| 17 | + /// Create a new `Normalizer` wrapping the specified geocoder. |
| 18 | + pub fn new(inner: Box<dyn Geocoder>) -> InvalidRecordSkipper { |
| 19 | + describe_counter!( |
| 20 | + "geocodecsv.invalid_records.total", |
| 21 | + "Invalid records which could not be geocoded" |
| 22 | + ); |
| 23 | + |
| 24 | + InvalidRecordSkipper { inner } |
| 25 | + } |
| 26 | +} |
| 27 | + |
| 28 | +#[async_trait] |
| 29 | +impl Geocoder for InvalidRecordSkipper { |
| 30 | + fn tag(&self) -> &str { |
| 31 | + // We don't change anything which could possibly affect caching, |
| 32 | + // so we can just use our inner tag. |
| 33 | + self.inner.tag() |
| 34 | + } |
| 35 | + |
| 36 | + fn configuration_key(&self) -> &str { |
| 37 | + self.inner.configuration_key() |
| 38 | + } |
| 39 | + |
| 40 | + fn column_names(&self) -> &[String] { |
| 41 | + self.inner.column_names() |
| 42 | + } |
| 43 | + |
| 44 | + async fn geocode_addresses( |
| 45 | + &self, |
| 46 | + addresses: &[Address], |
| 47 | + ) -> Result<Vec<Option<Geocoded>>> { |
| 48 | + // Extract our valid addresses, keeping track of their original |
| 49 | + // positions. |
| 50 | + let mut original_indices = vec![]; |
| 51 | + let mut valid_addresses = vec![]; |
| 52 | + for (i, address) in addresses.iter().enumerate() { |
| 53 | + if address.is_valid() { |
| 54 | + valid_addresses.push(address.clone()); |
| 55 | + original_indices.push(i); |
| 56 | + } else { |
| 57 | + counter!("geocodecsv.invalid_records.total", 1); |
| 58 | + } |
| 59 | + } |
| 60 | + |
| 61 | + // Geocode our valid addresses. |
| 62 | + let geocodeded = self.inner.geocode_addresses(&valid_addresses).await?; |
| 63 | + |
| 64 | + // Rebuild our geocoded addresses, inserting `None` for invalid. |
| 65 | + let mut result = vec![None; addresses.len()]; |
| 66 | + for (i, geocoded) in geocodeded.into_iter().enumerate() { |
| 67 | + let original_index = original_indices[i]; |
| 68 | + result[original_index] = geocoded; |
| 69 | + } |
| 70 | + Ok(result) |
| 71 | + } |
| 72 | +} |
0 commit comments