From e657523a4514b5a2b37ba5c0e11fdfc69e8dee5b Mon Sep 17 00:00:00 2001
From: Kris Reeves <kris@pressbuttonllc.com>
Date: Thu, 29 Dec 2016 20:06:48 -0500
Subject: [PATCH] Calculate buffer length of utf-8 data for json correctly

---
 lib/bhttp.coffee      |  9 ++++++---
 lib/bhttp.js          |  6 ++++--
 test-utf8-json.coffee | 22 ++++++++++++++++++++++
 3 files changed, 32 insertions(+), 5 deletions(-)
 create mode 100644 test-utf8-json.coffee

diff --git a/lib/bhttp.coffee b/lib/bhttp.coffee
index 77331e2..f423ef6 100644
--- a/lib/bhttp.coffee
+++ b/lib/bhttp.coffee
@@ -226,16 +226,19 @@ preparePayload = (request, response, requestState) ->
 
 				if request.options.encodeJSON
 					debugRequest "... but encodeJSON was set, so we will send JSON instead"
-					request.options.headers["content-type"] = "application/json"
+					request.options.headers["content-type"] = "application/json; charset=utf-8"
 					request.payload = JSON.stringify request.options.formFields ? null
+					# strings are utf-16, so string.length can be wrong
+					request.options.headers["content-length"] = Buffer.byteLength(request.payload, "utf8")
 				else if not _.isEmpty request.options.formFields
 					# The `querystring` module copies the key name verbatim, even if the value is actually an array. Things like PHP don't understand this, and expect every array-containing key to be suffixed with []. We'll just append that ourselves, then.
 					request.options.headers["content-type"] = "application/x-www-form-urlencoded"
 					request.payload = querystring.stringify formFixArray(request.options.formFields)
+					# URI encoded string is assumed to be ascii
+					request.options.headers["content-length"] = request.payload.length
 				else
 					request.payload = ""
-
-				request.options.headers["content-length"] = request.payload.length
+					request.options.headers["content-length"] = 0
 
 				return Promise.resolve()
 			else if request.options.formFields? and multipart
diff --git a/lib/bhttp.js b/lib/bhttp.js
index 6f90a5f..3b3c6e2 100644
--- a/lib/bhttp.js
+++ b/lib/bhttp.js
@@ -257,15 +257,17 @@ preparePayload = function(request, response, requestState) {
         debugRequest("got url-encodable form-data");
         if (request.options.encodeJSON) {
           debugRequest("... but encodeJSON was set, so we will send JSON instead");
-          request.options.headers["content-type"] = "application/json";
+          request.options.headers["content-type"] = "application/json; charset=utf-8";
           request.payload = JSON.stringify((ref1 = request.options.formFields) != null ? ref1 : null);
+          request.options.headers["content-length"] = Buffer.byteLength(request.payload, "utf8");
         } else if (!_.isEmpty(request.options.formFields)) {
           request.options.headers["content-type"] = "application/x-www-form-urlencoded";
           request.payload = querystring.stringify(formFixArray(request.options.formFields));
+          request.options.headers["content-length"] = request.payload.length;
         } else {
           request.payload = "";
+          request.options.headers["content-length"] = 0;
         }
-        request.options.headers["content-length"] = request.payload.length;
         return Promise.resolve();
       } else if ((request.options.formFields != null) && multipart) {
         debugRequest("got multipart form-data");
diff --git a/test-utf8-json.coffee b/test-utf8-json.coffee
new file mode 100644
index 0000000..39e1b08
--- /dev/null
+++ b/test-utf8-json.coffee
@@ -0,0 +1,22 @@
+Promise = require "bluebird"
+bhttp = require "./"
+
+formatLine = (line) -> line.toString().replace(/\n/g, "\\n").replace(/\r/g, "\\r")
+
+# this just tests we can parse the data we sent. if ever we send
+# more data than the content-length it will probably be broken
+Promise.try ->
+    bhttp.post "http://posttestserver.com/post.php",
+        value: "hello \ud801\udc37",
+    ,
+        encodeJSON: true,
+        headers: {"user-agent": "bhttp/test POST UTF-8 JSON"}
+.then (response) ->
+    responseUrl = response.body.toString().split("\n")[1].replace(/^View it at /, "")
+    console.log responseUrl
+    bhttp.get responseUrl
+.then (response) ->
+    lines = response.body.toString().split("\n")
+    JSON.parse lines[lines.length-1]
+.then (obj) ->
+    console.log "POST UTF-8 string", obj