Don't print non-utf8 bodies (#125)

* Don't print non-utf8 bodies

A lot of our services take in multipart data or binary data in the HTTP
body and when a wiremock predicate fails megabytes of binary data get
printed out to the console. Here I address this by not printing out a
string if it isn't valid utf8 and instead printing out the body length.

Another alternative may be to put in an upper limit to how large a body
will be printed and if it exceeds that size maybe saving the request to
a file for later analysis. But this change was so simple I figured I'd
open the PR first to start the discussion.

* Remove need for clone

* Apply some PR feature suggestions

1. Setting limit via env var or builder
2. Having a default length limit
3. Printed suggestion on upping the limit

* Update src/request.rs

Co-authored-by: Luca Palmieri <[email protected]>

* Apply feedback with BodyPrintLimit enum

* Expand to try and find valid printable byte

* Update src/request.rs

Co-authored-by: Luca Palmieri <[email protected]>

* Update src/mock_server/builder.rs

---------

Co-authored-by: Luca Palmieri <[email protected]>

src/mock_server/bare_server.rs

@@ -1,6 +1,7 @@
 use crate::mock_server::hyper::run_server;
 use crate::mock_set::MockId;
 use crate::mock_set::MountedMockSet;
+use crate::request::BodyPrintLimit;
 use crate::{mock::Mock, verification::VerificationOutcome, Request};
 use std::net::{SocketAddr, TcpListener, TcpStream};
 use std::pin::pin;
@@ -29,13 +30,15 @@ pub(crate) struct BareMockServer {
 pub(super) struct MockServerState {
     mock_set: MountedMockSet,
     received_requests: Option<Vec<Request>>,
+    body_print_limit: BodyPrintLimit,
 }
 
 impl MockServerState {
     pub(super) async fn handle_request(
         &mut self,
-        request: Request,
+        mut request: Request,
     ) -> (http_types::Response, Option<futures_timer::Delay>) {
+        request.body_print_limit = self.body_print_limit;
         // If request recording is enabled, record the incoming request
         // by adding it to the `received_requests` stack
         if let Some(received_requests) = &mut self.received_requests {
@@ -48,7 +51,11 @@ impl MockServerState {
 impl BareMockServer {
     /// Start a new instance of a `BareMockServer` listening on the specified
     /// [`TcpListener`](std::net::TcpListener).
-    pub(super) async fn start(listener: TcpListener, request_recording: RequestRecording) -> Self {
+    pub(super) async fn start(
+        listener: TcpListener,
+        request_recording: RequestRecording,
+        body_print_limit: BodyPrintLimit,
+    ) -> Self {
         let (shutdown_trigger, shutdown_receiver) = tokio::sync::oneshot::channel();
         let received_requests = match request_recording {
             RequestRecording::Enabled => Some(Vec::new()),
@@ -57,6 +64,7 @@ impl BareMockServer {
         let state = Arc::new(RwLock::new(MockServerState {
             mock_set: MountedMockSet::new(),
             received_requests,
+            body_print_limit,
         }));
         let server_address = listener
             .local_addr()

src/mock_server/builder.rs

@@ -1,20 +1,31 @@
 use crate::mock_server::bare_server::{BareMockServer, RequestRecording};
 use crate::mock_server::exposed_server::InnerServer;
+use crate::request::{BodyPrintLimit, BODY_PRINT_LIMIT};
 use crate::MockServer;
+use std::env;
 use std::net::TcpListener;
 
 /// A builder providing a fluent API to assemble a [`MockServer`] step-by-step.  
 /// Use [`MockServer::builder`] to get started.
 pub struct MockServerBuilder {
     listener: Option<TcpListener>,
     record_incoming_requests: bool,
+    body_print_limit: BodyPrintLimit,
 }
 
 impl MockServerBuilder {
     pub(super) fn new() -> Self {
+        let body_print_limit = match env::var("WIREMOCK_BODY_PRINT_LIMIT")
+            .ok()
+            .and_then(|x| x.parse::<usize>().ok())
+        {
+            Some(limit) => BodyPrintLimit::Limited(limit),
+            None => BodyPrintLimit::Limited(BODY_PRINT_LIMIT),
+        };
         Self {
             listener: None,
             record_incoming_requests: true,
+            body_print_limit,
         }
     }
 
@@ -76,6 +87,18 @@ impl MockServerBuilder {
         self
     }
 
+    /// The mock server prints the requests it received when one or more mocks have expectations that have not been satisfied.
+    /// By default, the size of the printed body is limited.
+    ///
+    /// You may want to change this if you're working with services with very large
+    /// bodies, or when printing wiremock output to a file where size matters
+    /// less than in a terminal window. You can configure this limit with
+    /// `MockServerBuilder::body_print_limit`.
+    pub fn body_print_limit(mut self, limit: BodyPrintLimit) -> Self {
+        self.body_print_limit = limit;
+        self
+    }
+
     /// Finalise the builder to get an instance of a [`BareMockServer`].
     pub(super) async fn build_bare(self) -> BareMockServer {
         let listener = if let Some(listener) = self.listener {
@@ -88,7 +111,7 @@ impl MockServerBuilder {
         } else {
             RequestRecording::Disabled
         };
-        BareMockServer::start(listener, recording).await
+        BareMockServer::start(listener, recording, self.body_print_limit).await
     }
 
     /// Finalise the builder and launch the [`MockServer`] instance!

src/request.rs

@@ -7,6 +7,18 @@ use http_types::convert::DeserializeOwned;
 use http_types::headers::{HeaderName, HeaderValue, HeaderValues};
 use http_types::{Method, Url};
 
+pub const BODY_PRINT_LIMIT: usize = 10_000;
+
+/// Specifies limitations on printing request bodies when logging requests. For some mock servers
+/// the bodies may be too large to reasonably print and it may be desireable to limit them.
+#[derive(Debug, Copy, Clone)]
+pub enum BodyPrintLimit {
+    /// Maximum length of a body to print in bytes.
+    Limited(usize),
+    /// There is no limit to the size of a body that may be printed.
+    Unlimited,
+}
+
 /// An incoming request to an instance of [`MockServer`].
 ///
 /// Each matcher gets an immutable reference to a `Request` instance in the [`matches`] method
@@ -31,6 +43,7 @@ pub struct Request {
     pub method: Method,
     pub headers: HashMap<HeaderName, HeaderValues>,
     pub body: Vec<u8>,
+    pub body_print_limit: BodyPrintLimit,
 }
 
 impl fmt::Display for Request {
@@ -44,7 +57,47 @@ impl fmt::Display for Request {
             let values = values.join(",");
             writeln!(f, "{}: {}", name, values)?;
         }
-        writeln!(f, "{}", String::from_utf8_lossy(&self.body))
+
+        match self.body_print_limit {
+            BodyPrintLimit::Limited(limit) if self.body.len() > limit => {
+                let mut written = false;
+                for end_byte in limit..(limit + 4).max(self.body.len()) {
+                    if let Ok(truncated) = std::str::from_utf8(&self.body[..end_byte]) {
+                        written = true;
+                        writeln!(f, "{}", truncated)?;
+                        if end_byte < self.body.len() {
+                            writeln!(
+                                f,
+                                "We truncated the body because it was too large: {} bytes (limit: {} bytes)",
+                                self.body.len(),
+                                limit
+                            )?;
+                            writeln!(f, "Increase this limit by setting `WIREMOCK_BODY_PRINT_LIMIT`, or calling `MockServerBuilder::body_print_limit` when building your MockServer instance")?;
+                        }
+                    }
+                }
+                if !written {
+                    writeln!(
+                        f,
+                        "Body is likely binary (invalid utf-8) size is {} bytes",
+                        self.body.len()
+                    )
+                } else {
+                    Ok(())
+                }
+            }
+            _ => {
+                if let Ok(body) = std::str::from_utf8(&self.body) {
+                    writeln!(f, "{}", body)
+                } else {
+                    writeln!(
+                        f,
+                        "Body is likely binary (invalid utf-8) size is {} bytes",
+                        self.body.len()
+                    )
+                }
+            }
+        }
     }
 }
 
@@ -75,6 +128,7 @@ impl Request {
             method,
             headers,
             body,
+            body_print_limit: BodyPrintLimit::Limited(BODY_PRINT_LIMIT),
         }
     }
 
@@ -119,6 +173,7 @@ impl Request {
             method,
             headers,
             body,
+            body_print_limit: BodyPrintLimit::Limited(BODY_PRINT_LIMIT),
         }
     }
 }