fix(cli): validate transform VRL and conditions with --no-environment

vector validate --no-environment now catches VRL compilation errors and
invalid condition expressions in transforms before any environment
resources are available.

Design:
- Add validate(&TransformContext) to TransformConfig; remap and filter
  implement it to compile VRL programs and conditions respectively
- Introduce validate_transforms(), called in both --no-environment and
  normal validate paths, which builds stub enrichment tables from the
  config so VRL can resolve table name references without loading data
- Remove the validate() call from check_outputs: that function is for
  graph/wiring checks only; transform compilation belongs in
  validate_transforms()
- Convert all CLI integration test configs to YAML; add create_yaml_file
  helper so test files carry the .yaml extension Vector uses for format
  detection

Fixes #15037

Author: pront

changelog.d/15037_validate_no_environment_vrl.fix.md

@@ -0,0 +1,3 @@
+Fixed `vector validate --no-environment` so it reports VRL and condition compilation errors for transforms without requiring full environment-dependent component initialization.
+
+authors: pront

lib/codecs/src/decoding/mod.rs

@@ -311,7 +311,7 @@ pub enum DeserializerConfig {
     /// [influxdb]: https://docs.influxdata.com/influxdb/cloud/reference/syntax/line-protocol
     Influxdb(InfluxdbDeserializerConfig),
 
-    /// Decodes the raw bytes as an [Apache Avro][apache_avro] message.
+    /// Decodes the raw bytes as as an [Apache Avro][apache_avro] message.
     ///
     /// [apache_avro]: https://avro.apache.org/
     Avro {

src/conditions/mod.rs

@@ -210,6 +210,14 @@ impl AnyCondition {
             AnyCondition::Map(m) => m.build(enrichment_tables, metrics_storage),
         }
     }
+
+    pub fn validate(
+        &self,
+        enrichment_tables: &vector_lib::enrichment::TableRegistry,
+        metrics_storage: &MetricsStorage,
+    ) -> crate::Result<()> {
+        self.build(enrichment_tables, metrics_storage).map(|_| ())
+    }
 }
 
 impl From<ConditionConfig> for AnyCondition {

src/config/transform.rs

@@ -250,7 +250,7 @@ pub trait TransformConfig: DynClone + NamedComponent + core::fmt::Debug + Send +
     ///
     /// If validation does not succeed, an error variant containing a list of all validation errors
     /// is returned.
-    fn validate(&self, _merged_definition: &schema::Definition) -> Result<(), Vec<String>> {
+    fn validate(&self, _context: &TransformContext) -> Result<(), Vec<String>> {
         Ok(())
     }
 

src/config/validation.rs

@@ -9,7 +9,6 @@ use super::{
     ComponentKey, Config, OutputId, Resource, builder::ConfigBuilder,
     transform::get_transform_output_ids,
 };
-use crate::config::schema;
 
 /// Minimum value (exclusive) for EWMA alpha options.
 /// The alpha value must be strictly greater than this value.
@@ -234,13 +233,6 @@ pub fn check_outputs(config: &ConfigBuilder) -> Result<(), Vec<String>> {
     }
 
     for (key, transform) in config.transforms.iter() {
-        // use the most general definition possible, since the real value isn't known yet.
-        let definition = schema::Definition::any();
-
-        if let Err(errs) = transform.inner.validate(&definition) {
-            errors.extend(errs.into_iter().map(|msg| format!("Transform {key} {msg}")));
-        }
-
         if get_transform_output_ids(
             transform.inner.as_ref(),
             key.clone(),

src/transforms/exclusive_route/config.rs

@@ -101,7 +101,7 @@ impl TransformConfig for ExclusiveRouteConfig {
         Input::all()
     }
 
-    fn validate(&self, _: &schema::Definition) -> Result<(), Vec<String>> {
+    fn validate(&self, _: &TransformContext) -> Result<(), Vec<String>> {
         let mut errors = Vec::new();
 
         let mut counts = std::collections::HashMap::new();

src/transforms/filter.rs

@@ -50,6 +50,12 @@ impl TransformConfig for FilterConfig {
         )?)))
     }
 
+    fn validate(&self, context: &TransformContext) -> Result<(), Vec<String>> {
+        self.condition
+            .validate(&context.enrichment_tables, &context.metrics_storage)
+            .map_err(|e| vec![e.to_string()])
+    }
+
     fn input(&self) -> Input {
         Input::all()
     }

src/transforms/remap.rs

@@ -277,6 +277,16 @@ impl TransformConfig for RemapConfig {
         Ok(transform)
     }
 
+    fn validate(&self, context: &TransformContext) -> std::result::Result<(), Vec<String>> {
+        self.compile_vrl_program(
+            context.enrichment_tables.clone(),
+            context.metrics_storage.clone(),
+            context.merged_schema_definition.clone(),
+        )
+        .map(|_| ())
+        .map_err(|e| vec![e.to_string()])
+    }
+
     fn input(&self) -> Input {
         Input::all()
     }

src/transforms/route.rs

@@ -131,7 +131,7 @@ impl TransformConfig for RouteConfig {
         Input::all()
     }
 
-    fn validate(&self, _: &schema::Definition) -> Result<(), Vec<String>> {
+    fn validate(&self, _: &TransformContext) -> Result<(), Vec<String>> {
         if self.route.contains_key(UNMATCHED_ROUTE) {
             Err(vec![format!(
                 "cannot have a named output with reserved name: `{UNMATCHED_ROUTE}`"

src/transforms/sample/config.rs

@@ -222,7 +222,7 @@ impl TransformConfig for SampleConfig {
         Input::new(DataType::Log | DataType::Trace)
     }
 
-    fn validate(&self, _: &schema::Definition) -> Result<(), Vec<String>> {
+    fn validate(&self, _: &TransformContext) -> Result<(), Vec<String>> {
         self.sample_rate()
             .map(|_| ())
             .map_err(|e| vec![e.to_string()])
@@ -317,7 +317,11 @@ mod tests {
             exclude: None,
         };
 
-        assert!(config.validate(&crate::schema::Definition::any()).is_ok());
+        assert!(
+            config
+                .validate(&crate::config::TransformContext::default())
+                .is_ok()
+        );
     }
 
     #[test]