feat(err): get resolver trait working (#25420)

Co-authored-by: David Newell <[email protected]>

rust/cymbal/Cargo.toml

@@ -16,10 +16,16 @@ common-metrics = { path = "../common/metrics" }
 common-alloc = { path = "../common/alloc" }
 common-kafka = { path = "../common/kafka" }
 common-types = { path = "../common/types" }
+common-dns = { path = "../common/dns" }
 thiserror = { workspace = true }
 sqlx = { workspace = true }
 serde_json = { workspace = true }
 serde = { workspace = true }
+sourcemap = "9.0.0"
+reqwest = { workspace = true }
+
+[dev-dependencies]
+httpmock = { workspace = true }
 
 [lints]
 workspace = true

rust/cymbal/src/app_context.rs

@@ -5,13 +5,19 @@ use health::{HealthHandle, HealthRegistry};
 use sqlx::{postgres::PgPoolOptions, PgPool};
 use tracing::info;
 
-use crate::{config::Config, error::Error};
+use crate::{
+    config::Config,
+    error::Error,
+    resolver::{Resolver, ResolverImpl},
+    symbol_store::{basic::BasicStore, caching::CachingStore},
+};
 
 pub struct AppContext {
     pub health_registry: HealthRegistry,
     pub worker_liveness: HealthHandle,
     pub consumer: SingleTopicConsumer,
     pub pool: PgPool,
+    pub resolver: Box<dyn Resolver>,
 }
 
 impl AppContext {
@@ -31,11 +37,22 @@ impl AppContext {
             config.consumer.kafka_consumer_topic
         );
 
+        // We're going to make heavy use of this "layering" pattern with stores, e.g. a we'll add an s3
+        // store that wraps an underlying basic one and stores the returned values in s3, and looks in s3
+        // before making fetches, etc.
+        let symbol_store = BasicStore::new(config)?;
+        let symbol_store =
+            CachingStore::new(Box::new(symbol_store), config.symbol_store_cache_max_bytes);
+
+        // Box box, box box
+        let resolver = Box::new(ResolverImpl::new(Box::new(symbol_store)));
+
         Ok(Self {
             health_registry,
             worker_liveness,
             consumer,
             pool,
+            resolver,
         })
     }
 }

rust/cymbal/src/config.rs

@@ -28,6 +28,16 @@ pub struct Config {
 
     #[envconfig(default = "true")]
     pub skip_reads: bool,
+
+    // cymbal makes HTTP get requests to auto-resolve sourcemaps - and follows redirects. To protect against SSRF, we only allow requests to public URLs by default
+    #[envconfig(default = "false")]
+    pub allow_internal_ips: bool,
+
+    #[envconfig(default = "30")]
+    pub sourcemap_timeout_seconds: u64,
+
+    #[envconfig(default = "100000000")] // 100MB - in prod, we should use closer to 1-10GB
+    pub symbol_store_cache_max_bytes: usize,
 }
 
 impl Config {

rust/cymbal/src/error.rs

@@ -9,4 +9,14 @@ pub enum Error {
     KafkaError(#[from] KafkaError),
     #[error("Sqlx error: {0}")]
     SqlxError(#[from] sqlx::Error),
+    #[error("Reqwest error: {0}")]
+    ReqwestError(#[from] reqwest::Error),
+    #[error("Not implemented error: {0}")]
+    NotImplementedError(String),
+    #[error("Lookup failed: {0}")]
+    LookupFailed(String),
+    #[error("Could not get source ref from: {0}")]
+    InvalidSourceRef(String),
+    #[error("sourcemap error: {0}")]
+    SourceMapError(#[from] sourcemap::Error),
 }