handle non-ascii header values

Author: fredr

runtimes/core/src/api/reqauth/meta.rs

@@ -1,5 +1,19 @@
 use std::str::{self, FromStr};
 
+use http::HeaderValue;
+
+pub trait HeaderValueExt {
+    fn to_utf8_str(&self) -> Result<&str, std::str::Utf8Error>;
+}
+
+impl HeaderValueExt for HeaderValue {
+    // Some header values may contain utf8 characters (e.g authdata) so we use `str::from_utf8`
+    // rather than using `HeaderValues::to_str` which errors on non-visible ASCII characters.
+    fn to_utf8_str(&self) -> Result<&str, std::str::Utf8Error> {
+        std::str::from_utf8(self.as_bytes())
+    }
+}
+
 #[derive(Debug, Copy, Clone, Hash, Eq, PartialEq)]
 pub enum MetaKey {
     TraceParent,
@@ -78,17 +92,15 @@ impl MetaMapMut for reqwest::header::HeaderMap {
 
 impl MetaMap for axum::http::HeaderMap {
     fn get_meta(&self, key: MetaKey) -> Option<&str> {
-        // Some meta values may contain utf8 characters (e.g authdata) so we use `str::from_utf8`
-        // rather than using `HeaderValues::to_str` which errors on non-visible ASCII characters.
         self.get(key.header_key())
-            .and_then(|val| std::str::from_utf8(val.as_bytes()).ok())
+            .and_then(|val| val.to_utf8_str().ok())
     }
 
     fn meta_values<'a>(&'a self, key: MetaKey) -> Box<dyn Iterator<Item = &'a str> + 'a> {
         Box::new(
             self.get_all(key.header_key())
                 .iter()
-                .filter_map(|v| v.to_str().ok()),
+                .filter_map(|v| v.to_utf8_str().ok()),
         )
     }
 
@@ -113,15 +125,15 @@ impl MetaMap for pingora::http::RequestHeader {
     fn get_meta(&self, key: MetaKey) -> Option<&str> {
         self.headers
             .get(key.header_key())
-            .and_then(|v| v.to_str().ok())
+            .and_then(|v| v.to_utf8_str().ok())
     }
 
     fn meta_values<'a>(&'a self, key: MetaKey) -> Box<dyn Iterator<Item = &'a str> + 'a> {
         Box::new(
             self.headers
                 .get_all(key.header_key())
                 .iter()
-                .filter_map(|v| v.to_str().ok()),
+                .filter_map(|v| v.to_utf8_str().ok()),
         )
     }
 

runtimes/core/src/api/schema/header.rs

@@ -1,3 +1,4 @@
+use crate::api::reqauth::meta::HeaderValueExt;
 use crate::api::schema::{JSONPayload, ParseResponse, ToOutgoingRequest, ToResponse};
 use crate::api::{self, PValue, PValues};
 use crate::api::{jsonschema, APIResult};
@@ -74,10 +75,10 @@ impl AsStr for &axum::http::header::HeaderName {
 }
 
 impl ToHeaderStr for &axum::http::header::HeaderValue {
-    type Error = axum::http::header::ToStrError;
+    type Error = std::str::Utf8Error;
 
     fn to_str(&self) -> Result<&str, Self::Error> {
-        <axum::http::header::HeaderValue>::to_str(self)
+        <axum::http::header::HeaderValue>::to_utf8_str(self)
     }
     fn is_empty(&self) -> bool {
         <axum::http::header::HeaderValue>::is_empty(self)

runtimes/core/src/trace/protocol.rs

@@ -3,6 +3,7 @@
 
 use std::sync::atomic::{AtomicU64, Ordering};
 
+use crate::api::reqauth::meta::HeaderValueExt;
 use crate::api::{self, PValue};
 use crate::model::{LogField, LogFieldValue, Request, TraceEventId};
 use crate::trace::eventbuf::EventBuffer;
@@ -888,7 +889,7 @@ impl EventBuffer {
         self.uvarint(headers.len() as u64);
         for (k, v) in headers.iter() {
             self.str(k.as_str());
-            self.str(v.to_str().unwrap_or(""));
+            self.str(v.to_utf8_str().unwrap_or(""));
         }
     }
 }

runtimes/js/src/request_meta.rs

@@ -1,7 +1,7 @@
 use chrono::{DateTime, SecondsFormat, Utc};
 use napi_derive::napi;
 
-use encore_runtime_core::model;
+use encore_runtime_core::{api::reqauth::meta::HeaderValueExt, model};
 
 use crate::pvalue::PVals;
 
@@ -149,7 +149,7 @@ fn serialize_headers(
     let mut map = Map::with_capacity(headers.len());
 
     for (k, v) in headers {
-        let Ok(v) = v.to_str() else {
+        let Ok(v) = v.to_utf8_str() else {
             continue;
         };