http: improve performance for incoming headers

PR-URL: https://github.com/nodejs/node/pull/26041
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Sakthipriyan Vairamani <thechargingvolcano@gmail.com>

3 files changed, 124 insertions, 118 deletions

diff --git a/benchmark/_http-benchmarkers.js b/benchmark/_http-benchmarkers.js
index f456628454..a4d6230039 100644
--- a/benchmark/_http-benchmarkers.js
+++ b/benchmark/_http-benchmarkers.js
@@ -25,8 +25,11 @@ class AutocannonBenchmarker {
       '-c', options.connections,
       '-j',
       '-n',
-      `http://127.0.0.1:${options.port}${options.path}`,
     ];
+    for (const field in options.headers) {
+      args.push('-H', `${field}=${options.headers[field]}`);
+    }
+    args.push(`http://127.0.0.1:${options.port}${options.path}`);
     const child = child_process.spawn(this.executable, args);
     return child;
   }
diff --git a/benchmark/http/incoming_headers.js b/benchmark/http/incoming_headers.js
new file mode 100644
index 0000000000..a1ab57e238
--- /dev/null
+++ b/benchmark/http/incoming_headers.js
@@ -0,0 +1,35 @@
+'use strict';
+const common = require('../common.js');
+const http = require('http');
+
+const bench = common.createBenchmark(main, {
+  // unicode confuses ab on os x.
+  c: [50, 500],
+  headerDuplicates: [0, 5, 20]
+});
+
+function main({ c, headerDuplicates }) {
+  const server = http.createServer((req, res) => {
+    res.end();
+  });
+
+  server.listen(common.PORT, () => {
+    const headers = {
+      'Content-Type': 'text/plain',
+      'Accept': 'text/plain',
+      'User-Agent': 'nodejs-benchmark',
+      'Date': new Date().toString(),
+      'Cache-Control': 'no-cache'
+    };
+    for (let i = 0; i < headerDuplicates; i++) {
+      headers[`foo${i}`] = `some header value ${i}`;
+    }
+    bench.http({
+      path: '/',
+      connections: c,
+      headers
+    }, () => {
+      server.close();
+    });
+  });
+}
diff --git a/lib/_http_incoming.js b/lib/_http_incoming.js
index 1e0c42f7bd..7ba4d366b2 100644
--- a/lib/_http_incoming.js
+++ b/lib/_http_incoming.js
@@ -134,133 +134,101 @@ function _addHeaderLines(headers, n) {
 // TODO: perhaps http_parser could be returning both raw and lowercased versions
 // of known header names to avoid us having to call toLowerCase() for those
 // headers.
-
-// 'array' header list is taken from:
-// https://mxr.mozilla.org/mozilla/source/netwerk/protocol/http/src/nsHttpHeaderArray.cpp
-function matchKnownFields(field) {
-  var low = false;
-  while (true) {
-    switch (field) {
-      case 'Content-Type':
-      case 'content-type':
-        return 'content-type';
-      case 'Content-Length':
-      case 'content-length':
-        return 'content-length';
-      case 'User-Agent':
-      case 'user-agent':
-        return 'user-agent';
-      case 'Referer':
-      case 'referer':
-        return 'referer';
-      case 'Host':
-      case 'host':
-        return 'host';
-      case 'Authorization':
-      case 'authorization':
-        return 'authorization';
-      case 'Proxy-Authorization':
-      case 'proxy-authorization':
-        return 'proxy-authorization';
-      case 'If-Modified-Since':
-      case 'if-modified-since':
-        return 'if-modified-since';
-      case 'If-Unmodified-Since':
-      case 'if-unmodified-since':
-        return 'if-unmodified-since';
-      case 'From':
-      case 'from':
-        return 'from';
-      case 'Location':
-      case 'location':
+function matchKnownFields(field, lowercased) {
+  switch (field.length) {
+    case 3:
+      if (field === 'Age' || field === 'age') return 'age';
+      break;
+    case 4:
+      if (field === 'Host' || field === 'host') return 'host';
+      if (field === 'From' || field === 'from') return 'from';
+      if (field === 'ETag' || field === 'etag') return 'etag';
+      if (field === 'Date' || field === 'date') return '\u0000date';
+      if (field === 'Vary' || field === 'vary') return '\u0000vary';
+      break;
+    case 6:
+      if (field === 'Server' || field === 'server') return 'server';
+      if (field === 'Cookie' || field === 'cookie') return '\u0002cookie';
+      if (field === 'Origin' || field === 'origin') return '\u0000origin';
+      if (field === 'Expect' || field === 'expect') return '\u0000expect';
+      if (field === 'Accept' || field === 'accept') return '\u0000accept';
+      break;
+    case 7:
+      if (field === 'Referer' || field === 'referer') return 'referer';
+      if (field === 'Expires' || field === 'expires') return 'expires';
+      if (field === 'Upgrade' || field === 'upgrade') return '\u0000upgrade';
+      break;
+    case 8:
+      if (field === 'Location' || field === 'location')
         return 'location';
-      case 'Max-Forwards':
-      case 'max-forwards':
-        return 'max-forwards';
-      case 'Retry-After':
-      case 'retry-after':
-        return 'retry-after';
-      case 'ETag':
-      case 'etag':
-        return 'etag';
-      case 'Last-Modified':
-      case 'last-modified':
-        return 'last-modified';
-      case 'Server':
-      case 'server':
-        return 'server';
-      case 'Age':
-      case 'age':
-        return 'age';
-      case 'Expires':
-      case 'expires':
-        return 'expires';
-      case 'Set-Cookie':
-      case 'set-cookie':
+      if (field === 'If-Match' || field === 'if-match')
+        return '\u0000if-match';
+      break;
+    case 10:
+      if (field === 'User-Agent' || field === 'user-agent')
+        return 'user-agent';
+      if (field === 'Set-Cookie' || field === 'set-cookie')
         return '\u0001';
-      case 'Cookie':
-      case 'cookie':
-        return '\u0002cookie';
-      // The fields below are not used in _addHeaderLine(), but they are common
-      // headers where we can avoid toLowerCase() if the mixed or lower case
-      // versions match the first time through.
-      case 'Transfer-Encoding':
-      case 'transfer-encoding':
-        return '\u0000transfer-encoding';
-      case 'Date':
-      case 'date':
-        return '\u0000date';
-      case 'Connection':
-      case 'connection':
+      if (field === 'Connection' || field === 'connection')
         return '\u0000connection';
-      case 'Cache-Control':
-      case 'cache-control':
+      break;
+    case 11:
+      if (field === 'Retry-After' || field === 'retry-after')
+        return 'retry-after';
+      break;
+    case 12:
+      if (field === 'Content-Type' || field === 'content-type')
+        return 'content-type';
+      if (field === 'Max-Forwards' || field === 'max-forwards')
+        return 'max-forwards';
+      break;
+    case 13:
+      if (field === 'Authorization' || field === 'authorization')
+        return 'authorization';
+      if (field === 'Last-Modified' || field === 'last-modified')
+        return 'last-modified';
+      if (field === 'Cache-Control' || field === 'cache-control')
         return '\u0000cache-control';
-      case 'Vary':
-      case 'vary':
-        return '\u0000vary';
-      case 'Content-Encoding':
-      case 'content-encoding':
-        return '\u0000content-encoding';
-      case 'Origin':
-      case 'origin':
-        return '\u0000origin';
-      case 'Upgrade':
-      case 'upgrade':
-        return '\u0000upgrade';
-      case 'Expect':
-      case 'expect':
-        return '\u0000expect';
-      case 'If-Match':
-      case 'if-match':
-        return '\u0000if-match';
-      case 'If-None-Match':
-      case 'if-none-match':
+      if (field === 'If-None-Match' || field === 'if-none-match')
         return '\u0000if-none-match';
-      case 'Accept':
-      case 'accept':
-        return '\u0000accept';
-      case 'Accept-Encoding':
-      case 'accept-encoding':
+      break;
+    case 14:
+      if (field === 'Content-Length' || field === 'content-length')
+        return 'content-length';
+      break;
+    case 15:
+      if (field === 'Accept-Encoding' || field === 'accept-encoding')
         return '\u0000accept-encoding';
-      case 'Accept-Language':
-      case 'accept-language':
+      if (field === 'Accept-Language' || field === 'accept-language')
         return '\u0000accept-language';
-      case 'X-Forwarded-For':
-      case 'x-forwarded-for':
+      if (field === 'X-Forwarded-For' || field === 'x-forwarded-for')
         return '\u0000x-forwarded-for';
-      case 'X-Forwarded-Host':
-      case 'x-forwarded-host':
+      break;
+    case 16:
+      if (field === 'Content-Encoding' || field === 'content-encoding')
+        return '\u0000content-encoding';
+      if (field === 'X-Forwarded-Host' || field === 'x-forwarded-host')
         return '\u0000x-forwarded-host';
-      case 'X-Forwarded-Proto':
-      case 'x-forwarded-proto':
+      break;
+    case 17:
+      if (field === 'If-Modified-Since' || field === 'if-modified-since')
+        return 'if-modified-since';
+      if (field === 'Transfer-Encoding' || field === 'transfer-encoding')
+        return '\u0000transfer-encoding';
+      if (field === 'X-Forwarded-Proto' || field === 'x-forwarded-proto')
         return '\u0000x-forwarded-proto';
-      default:
-        if (low)
-          return '\u0000' + field;
-        field = field.toLowerCase();
-        low = true;
-    }
+      break;
+    case 19:
+      if (field === 'Proxy-Authorization' || field === 'proxy-authorization')
+        return 'proxy-authorization';
+      if (field === 'If-Unmodified-Since' || field === 'if-unmodified-since')
+        return 'if-unmodified-since';
+      break;
+  }
+  if (lowercased) {
+    return '\u0000' + field;
+  } else {
+    return matchKnownFields(field.toLowerCase(), true);
   }
 }
 // Add the given (field, value) pair to the message