picohttp 0.1.1 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8a11d937822d71f87cd78734c038f1a73a25af902b142958208af81b3397f3be
-  data.tar.gz: 21d3337883257c321b37c602d52c7ab68896343d856bfafaec329476b43c35fe
+  metadata.gz: 3b357892a265429f3731c5cc68f56aaa3649b1f346a9c86f021bb67be2e601ac
+  data.tar.gz: 7a499b9fe596ee583c238956d66f05f3ff6caeabd0594a5db1d0f3c1b6d3e023
 SHA512:
-  metadata.gz: 5e73eead0ddf6378830a445562e827a77f235e1f26d37d18c7956ec790652c9f8efb23b82385d90c512dc47a864b796b8f9f7c2826e5f724dcaae63d45c39d43
-  data.tar.gz: 92893c74ac27005fc38891c49ce4c2a5c72e0f7fd453fce38871ab5b626db8cc10f000263c01573b7cceac331dbf4f5089ee413ec71f3839cede7fa0bc29693e
+  metadata.gz: 704efdbcbd2daab576977d4f2857307a7de44a41b5785cc6d215886777b7d97e3e3ba257670e40afd4c0b49f05024c8e1b26f1a8a41864e5b70393736a26fe6a
+  data.tar.gz: 7318f991db2981ed09349edfd907141342121fa65f3a45fa0e51dc2503b973b1c83681356eacad29fd1f95c44af130ad1eb0952e52ae04e4a139678bf172d016

data/ext/picohttp/picohttp.c

@@ -2,7 +2,7 @@
 #include "picohttpparser.h"
 
 #define MAX_HEADER_NAME_LEN 256
-#define ENV_HASH_INITIAL_CAPACITY 64
+#define MAX_HEADERS 100
 
 VALUE rb_mPicohttp;
 VALUE rb_ePicohttpParseError;
@@ -16,6 +16,8 @@ static VALUE rb_str_empty;
 static VALUE rb_str_http_1_0;
 static VALUE rb_str_http_1_1;
 
+#include "string_lookup.inc"
+
 static VALUE
 http_version_string(int minor_version)
 {
@@ -28,6 +30,14 @@ http_version_string(int minor_version)
     }
 }
 
+static VALUE
+http_method_string(const char *method, size_t method_len)
+{
+    VALUE str = lookup_method(method, method_len);
+    if (str == Qnil) str = rb_str_new(method, method_len);
+    return str;
+}
+
 static VALUE
 header_name_to_env_key(const char *name, size_t name_len)
 {
@@ -35,12 +45,9 @@ header_name_to_env_key(const char *name, size_t name_len)
         rb_raise(rb_ePicohttpParseError, "Header name too long");
     }
 
-    // Special cases for Content-Type and Content-Length (no HTTP_ prefix)
-    if (name_len == 12 && strncasecmp(name, "content-type", 12) == 0) {
-        return rb_interned_str_cstr("CONTENT_TYPE");
-    }
-    if (name_len == 14 && strncasecmp(name, "content-length", 14) == 0) {
-        return rb_interned_str_cstr("CONTENT_LENGTH");
+    VALUE str = lookup_header(name, name_len);
+    if (str != Qnil) {
+        return str;
     }
 
     char env_name[MAX_HEADER_NAME_LEN + 6]; // "HTTP_" + name + null terminator
@@ -71,7 +78,7 @@ picohttp_parse_request(VALUE self, VALUE str)
 
     const char *method, *path;
     int minor_version;
-    struct phr_header headers[100];
+    struct phr_header headers[MAX_HEADERS];
     size_t method_len, path_len, num_headers = sizeof(headers) / sizeof(headers[0]);
 
     int result = phr_parse_request(buf, len, &method, &method_len, &path, &path_len,
@@ -112,7 +119,7 @@ picohttp_parse_request_env(VALUE self, VALUE str)
 
     const char *method, *path;
     int minor_version;
-    struct phr_header headers[100];
+    struct phr_header headers[MAX_HEADERS];
     size_t method_len, path_len, num_headers = sizeof(headers) / sizeof(headers[0]);
 
     int result = phr_parse_request(buf, len, &method, &method_len, &path, &path_len,
@@ -125,26 +132,33 @@ picohttp_parse_request_env(VALUE self, VALUE str)
         rb_raise(rb_ePicohttpParseError, "Invalid HTTP request");
     }
 
-#ifdef HAVE_RB_HASH_NEW_CAPA
-    VALUE env = rb_hash_new_capa(ENV_HASH_INITIAL_CAPACITY);
-#else
-    VALUE env = rb_hash_new();
-#endif
+    VALUE header_values[MAX_HEADERS * 2];
+    int idx = 0;
 
     // Standard CGI/Rack environment variables
-    rb_hash_aset(env, rb_str_request_method, rb_str_new(method, method_len));
-    rb_hash_aset(env, rb_str_server_protocol, http_version_string(minor_version));
+    header_values[idx++] = rb_str_request_method;
+    header_values[idx++] = http_method_string(method, method_len);
+
+    header_values[idx++] = rb_str_server_protocol;
+    header_values[idx++] = http_version_string(minor_version);
 
     // Parse path and query string in C
     const char *query_start = memchr(path, '?', path_len);
     if (query_start) {
         size_t path_info_len = query_start - path;
         size_t query_len = path_len - path_info_len - 1;
-        rb_hash_aset(env, rb_str_path_info, rb_str_new(path, path_info_len));
-        rb_hash_aset(env, rb_str_query_string, rb_str_new(query_start + 1, query_len));
+
+        header_values[idx++] = rb_str_path_info;
+        header_values[idx++] = rb_str_new(path, path_info_len);
+
+        header_values[idx++] = rb_str_query_string;
+        header_values[idx++] = rb_str_new(query_start + 1, query_len);
     } else {
-        rb_hash_aset(env, rb_str_path_info, rb_str_new(path, path_len));
-        rb_hash_aset(env, rb_str_query_string, rb_str_empty);
+        header_values[idx++] = rb_str_path_info;
+        header_values[idx++] = rb_str_new(path, path_len);
+
+        header_values[idx++] = rb_str_query_string;
+        header_values[idx++] = rb_str_empty;
     }
 
     // Convert headers to HTTP_ prefixed environment variables
@@ -153,11 +167,18 @@ picohttp_parse_request_env(VALUE self, VALUE str)
             rb_raise(rb_ePicohttpParseError, "HTTP line folding not supported");
         }
 
-        VALUE header_name = header_name_to_env_key(headers[i].name, headers[i].name_len);
-        VALUE header_value = rb_str_new(headers[i].value, headers[i].value_len);
-        rb_hash_aset(env, header_name, header_value);
+        header_values[idx++] = header_name_to_env_key(headers[i].name, headers[i].name_len);
+        header_values[idx++] = rb_str_new(headers[i].value, headers[i].value_len);
     }
 
+#ifdef HAVE_RB_HASH_NEW_CAPA
+    VALUE env = rb_hash_new_capa(idx / 2);
+#else
+    VALUE env = rb_hash_new();
+#endif
+
+    rb_hash_bulk_insert(idx, header_values, env);
+
     return env;
 }
 
@@ -174,6 +195,7 @@ Init_picohttp(void)
     rb_define_module_function(rb_mPicohttp, "parse_request_env", picohttp_parse_request_env, 1);
 
     // Initialize interned string constants
+    init_string_lookup();
     rb_str_request_method = rb_interned_str_cstr("REQUEST_METHOD");
     rb_str_server_protocol = rb_interned_str_cstr("SERVER_PROTOCOL");
     rb_str_path_info = rb_interned_str_cstr("PATH_INFO");

data/ext/picohttp/string_lookup.inc

@@ -0,0 +1,120 @@
+/* This file is auto-generated by tool/generate_lookup.rb */
+
+static VALUE rb_s_HTTP_HOST; /* "HTTP_HOST" */
+static VALUE rb_s_HTTP_ACCEPT; /* "HTTP_ACCEPT" */
+static VALUE rb_s_HTTP_COOKIE; /* "HTTP_COOKIE" */
+static VALUE rb_s_HTTP_REFERER; /* "HTTP_REFERER" */
+static VALUE rb_s_HTTP_USER_AGENT; /* "HTTP_USER_AGENT" */
+static VALUE rb_s_HTTP_CONNECTION; /* "HTTP_CONNECTION" */
+static VALUE rb_s_CONTENT_TYPE; /* "CONTENT_TYPE" */
+static VALUE rb_s_HTTP_CACHE_CONTROL; /* "HTTP_CACHE_CONTROL" */
+static VALUE rb_s_HTTP_AUTHORIZATION; /* "HTTP_AUTHORIZATION" */
+static VALUE rb_s_CONTENT_LENGTH; /* "CONTENT_LENGTH" */
+static VALUE rb_s_HTTP_ACCEPT_ENCODING; /* "HTTP_ACCEPT_ENCODING" */
+static VALUE rb_s_HTTP_ACCEPT_LANGUAGE; /* "HTTP_ACCEPT_LANGUAGE" */
+static VALUE rb_s_GET; /* "GET" */
+static VALUE rb_s_PUT; /* "PUT" */
+static VALUE rb_s_POST; /* "POST" */
+static VALUE rb_s_HEAD; /* "HEAD" */
+static VALUE rb_s_PATCH; /* "PATCH" */
+static VALUE rb_s_DELETE; /* "DELETE" */
+static VALUE rb_s_OPTIONS; /* "OPTIONS" */
+
+static VALUE lookup_header(const char *s, size_t len) {
+  switch (len) {
+  case 4:
+    if ((s[0] | 32) == 'h' && (s[1] | 32) == 'o' && (s[2] | 32) == 's' && (s[3] | 32) == 't') return rb_s_HTTP_HOST;
+    break;
+  case 6:
+    if ((s[0] | 32) == 'a' && (s[1] | 32) == 'c' && (s[2] | 32) == 'c' && (s[3] | 32) == 'e' && (s[4] | 32) == 'p' && (s[5] | 32) == 't') return rb_s_HTTP_ACCEPT;
+    if ((s[0] | 32) == 'c' && (s[1] | 32) == 'o' && (s[2] | 32) == 'o' && (s[3] | 32) == 'k' && (s[4] | 32) == 'i' && (s[5] | 32) == 'e') return rb_s_HTTP_COOKIE;
+    break;
+  case 7:
+    if ((s[0] | 32) == 'r' && (s[1] | 32) == 'e' && (s[2] | 32) == 'f' && (s[3] | 32) == 'e' && (s[4] | 32) == 'r' && (s[5] | 32) == 'e' && (s[6] | 32) == 'r') return rb_s_HTTP_REFERER;
+    break;
+  case 10:
+    if ((s[0] | 32) == 'u' && (s[1] | 32) == 's' && (s[2] | 32) == 'e' && (s[3] | 32) == 'r' && (s[4] | 32) == '-' && (s[5] | 32) == 'a' && (s[6] | 32) == 'g' && (s[7] | 32) == 'e' && (s[8] | 32) == 'n' && (s[9] | 32) == 't') return rb_s_HTTP_USER_AGENT;
+    if ((s[0] | 32) == 'c' && (s[1] | 32) == 'o' && (s[2] | 32) == 'n' && (s[3] | 32) == 'n' && (s[4] | 32) == 'e' && (s[5] | 32) == 'c' && (s[6] | 32) == 't' && (s[7] | 32) == 'i' && (s[8] | 32) == 'o' && (s[9] | 32) == 'n') return rb_s_HTTP_CONNECTION;
+    break;
+  case 12:
+    if ((s[0] | 32) == 'c' && (s[1] | 32) == 'o' && (s[2] | 32) == 'n' && (s[3] | 32) == 't' && (s[4] | 32) == 'e' && (s[5] | 32) == 'n' && (s[6] | 32) == 't' && (s[7] | 32) == '-' && (s[8] | 32) == 't' && (s[9] | 32) == 'y' && (s[10] | 32) == 'p' && (s[11] | 32) == 'e') return rb_s_CONTENT_TYPE;
+    break;
+  case 13:
+    if ((s[0] | 32) == 'c' && (s[1] | 32) == 'a' && (s[2] | 32) == 'c' && (s[3] | 32) == 'h' && (s[4] | 32) == 'e' && (s[5] | 32) == '-' && (s[6] | 32) == 'c' && (s[7] | 32) == 'o' && (s[8] | 32) == 'n' && (s[9] | 32) == 't' && (s[10] | 32) == 'r' && (s[11] | 32) == 'o' && (s[12] | 32) == 'l') return rb_s_HTTP_CACHE_CONTROL;
+    if ((s[0] | 32) == 'a' && (s[1] | 32) == 'u' && (s[2] | 32) == 't' && (s[3] | 32) == 'h' && (s[4] | 32) == 'o' && (s[5] | 32) == 'r' && (s[6] | 32) == 'i' && (s[7] | 32) == 'z' && (s[8] | 32) == 'a' && (s[9] | 32) == 't' && (s[10] | 32) == 'i' && (s[11] | 32) == 'o' && (s[12] | 32) == 'n') return rb_s_HTTP_AUTHORIZATION;
+    break;
+  case 14:
+    if ((s[0] | 32) == 'c' && (s[1] | 32) == 'o' && (s[2] | 32) == 'n' && (s[3] | 32) == 't' && (s[4] | 32) == 'e' && (s[5] | 32) == 'n' && (s[6] | 32) == 't' && (s[7] | 32) == '-' && (s[8] | 32) == 'l' && (s[9] | 32) == 'e' && (s[10] | 32) == 'n' && (s[11] | 32) == 'g' && (s[12] | 32) == 't' && (s[13] | 32) == 'h') return rb_s_CONTENT_LENGTH;
+    break;
+  case 15:
+    if ((s[0] | 32) == 'a' && (s[1] | 32) == 'c' && (s[2] | 32) == 'c' && (s[3] | 32) == 'e' && (s[4] | 32) == 'p' && (s[5] | 32) == 't' && (s[6] | 32) == '-' && (s[7] | 32) == 'e' && (s[8] | 32) == 'n' && (s[9] | 32) == 'c' && (s[10] | 32) == 'o' && (s[11] | 32) == 'd' && (s[12] | 32) == 'i' && (s[13] | 32) == 'n' && (s[14] | 32) == 'g') return rb_s_HTTP_ACCEPT_ENCODING;
+    if ((s[0] | 32) == 'a' && (s[1] | 32) == 'c' && (s[2] | 32) == 'c' && (s[3] | 32) == 'e' && (s[4] | 32) == 'p' && (s[5] | 32) == 't' && (s[6] | 32) == '-' && (s[7] | 32) == 'l' && (s[8] | 32) == 'a' && (s[9] | 32) == 'n' && (s[10] | 32) == 'g' && (s[11] | 32) == 'u' && (s[12] | 32) == 'a' && (s[13] | 32) == 'g' && (s[14] | 32) == 'e') return rb_s_HTTP_ACCEPT_LANGUAGE;
+    break;
+  }
+  return Qnil;
+}
+
+static VALUE lookup_method(const char *s, size_t len) {
+  switch (len) {
+  case 3:
+    if (s[0] == 'G' && s[1] == 'E' && s[2] == 'T') return rb_s_GET;
+    if (s[0] == 'P' && s[1] == 'U' && s[2] == 'T') return rb_s_PUT;
+    break;
+  case 4:
+    if (s[0] == 'P' && s[1] == 'O' && s[2] == 'S' && s[3] == 'T') return rb_s_POST;
+    if (s[0] == 'H' && s[1] == 'E' && s[2] == 'A' && s[3] == 'D') return rb_s_HEAD;
+    break;
+  case 5:
+    if (s[0] == 'P' && s[1] == 'A' && s[2] == 'T' && s[3] == 'C' && s[4] == 'H') return rb_s_PATCH;
+    break;
+  case 6:
+    if (s[0] == 'D' && s[1] == 'E' && s[2] == 'L' && s[3] == 'E' && s[4] == 'T' && s[5] == 'E') return rb_s_DELETE;
+    break;
+  case 7:
+    if (s[0] == 'O' && s[1] == 'P' && s[2] == 'T' && s[3] == 'I' && s[4] == 'O' && s[5] == 'N' && s[6] == 'S') return rb_s_OPTIONS;
+    break;
+  }
+  return Qnil;
+}
+
+static void init_string_lookup(void) {
+  rb_gc_register_address(&rb_s_HTTP_HOST);
+  rb_s_HTTP_HOST = rb_interned_str_cstr("HTTP_HOST");
+  rb_gc_register_address(&rb_s_HTTP_ACCEPT);
+  rb_s_HTTP_ACCEPT = rb_interned_str_cstr("HTTP_ACCEPT");
+  rb_gc_register_address(&rb_s_HTTP_COOKIE);
+  rb_s_HTTP_COOKIE = rb_interned_str_cstr("HTTP_COOKIE");
+  rb_gc_register_address(&rb_s_HTTP_REFERER);
+  rb_s_HTTP_REFERER = rb_interned_str_cstr("HTTP_REFERER");
+  rb_gc_register_address(&rb_s_HTTP_USER_AGENT);
+  rb_s_HTTP_USER_AGENT = rb_interned_str_cstr("HTTP_USER_AGENT");
+  rb_gc_register_address(&rb_s_HTTP_CONNECTION);
+  rb_s_HTTP_CONNECTION = rb_interned_str_cstr("HTTP_CONNECTION");
+  rb_gc_register_address(&rb_s_CONTENT_TYPE);
+  rb_s_CONTENT_TYPE = rb_interned_str_cstr("CONTENT_TYPE");
+  rb_gc_register_address(&rb_s_HTTP_CACHE_CONTROL);
+  rb_s_HTTP_CACHE_CONTROL = rb_interned_str_cstr("HTTP_CACHE_CONTROL");
+  rb_gc_register_address(&rb_s_HTTP_AUTHORIZATION);
+  rb_s_HTTP_AUTHORIZATION = rb_interned_str_cstr("HTTP_AUTHORIZATION");
+  rb_gc_register_address(&rb_s_CONTENT_LENGTH);
+  rb_s_CONTENT_LENGTH = rb_interned_str_cstr("CONTENT_LENGTH");
+  rb_gc_register_address(&rb_s_HTTP_ACCEPT_ENCODING);
+  rb_s_HTTP_ACCEPT_ENCODING = rb_interned_str_cstr("HTTP_ACCEPT_ENCODING");
+  rb_gc_register_address(&rb_s_HTTP_ACCEPT_LANGUAGE);
+  rb_s_HTTP_ACCEPT_LANGUAGE = rb_interned_str_cstr("HTTP_ACCEPT_LANGUAGE");
+  rb_gc_register_address(&rb_s_GET);
+  rb_s_GET = rb_interned_str_cstr("GET");
+  rb_gc_register_address(&rb_s_PUT);
+  rb_s_PUT = rb_interned_str_cstr("PUT");
+  rb_gc_register_address(&rb_s_POST);
+  rb_s_POST = rb_interned_str_cstr("POST");
+  rb_gc_register_address(&rb_s_HEAD);
+  rb_s_HEAD = rb_interned_str_cstr("HEAD");
+  rb_gc_register_address(&rb_s_PATCH);
+  rb_s_PATCH = rb_interned_str_cstr("PATCH");
+  rb_gc_register_address(&rb_s_DELETE);
+  rb_s_DELETE = rb_interned_str_cstr("DELETE");
+  rb_gc_register_address(&rb_s_OPTIONS);
+  rb_s_OPTIONS = rb_interned_str_cstr("OPTIONS");
+}
+

data/lib/picohttp/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Picohttp
-  VERSION = "0.1.1"
+  VERSION = "0.2.0"
 end

data/tool/generate_lookup.rb

@@ -0,0 +1,98 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+# Generates fast C lookup code for known string sets
+
+puts "/* This file is auto-generated by #{__FILE__} */"
+puts
+
+class StringLookup
+  def initialize
+    @string_table = Hash.new do |h, k|
+      h[k] = name_for_string(k)
+    end
+    @functions = []
+  end
+
+  def generate
+    @string_table.each do |v, name|
+      puts "static VALUE #{name}; /* #{v.inspect} */"
+    end
+    puts
+
+    @functions.each do |code|
+      puts code
+      puts
+    end
+
+    puts "static void init_string_lookup(void) {"
+    @string_table.each do |v, name|
+      puts "  rb_gc_register_address(&#{name});"
+      puts "  #{name} = rb_interned_str_cstr(#{v.dump});"
+    end
+    puts "}"
+    puts
+  end
+
+  def name_for_string(string)
+    "rb_s_#{string.gsub(/[^a-z0-9]/i, "_")}"
+  end
+
+  def generate_comparison(var, string, ignore_case: false)
+    0.upto(string.length - 1).map do |i|
+      if ignore_case
+        "(#{var}[#{i}] | 32) == '#{string[i].downcase}'"
+      else
+        "#{var}[#{i}] == '#{string[i]}'"
+      end
+    end.join(" && ")
+  end
+
+  def add_function(function, mapping, ignore_case: false)
+    if mapping.is_a?(Array)
+      mapping = mapping.map do |str|
+        [str, str]
+      end.to_h
+    end
+    by_length = mapping.group_by do |(k,v)|
+      k.length
+    end
+
+    code = []
+    code << "static VALUE lookup_#{function}(const char *s, size_t len) {"
+    code << "  switch (len) {"
+
+    by_length.sort.each do |len, strs|
+      code << "  case #{len}:"
+      strs.each do |match, target|
+        name = @string_table[target]
+        code << "    if (#{generate_comparison("s", match, ignore_case:)}) return #{name};"
+      end
+      code << "    break;"
+    end
+
+    code << "  }"
+    code << "  return Qnil;"
+    code << "}"
+
+    @functions << code.join("\n")
+  end
+end
+
+HEADERS = %w[Content-Type Content-Length Host User-Agent Accept Accept-Encoding
+             Accept-Language Connection Cache-Control Cookie Authorization Referer]
+
+METHODS = %w[GET POST PUT DELETE HEAD OPTIONS PATCH]
+
+lookup = StringLookup.new
+header_mapping = HEADERS.map do |header|
+  header = header.downcase
+  rack_key = header.upcase.gsub("-", "_")
+  unless header == "content-type" || header == "content-length"
+    rack_key = "HTTP_#{rack_key}"
+  end
+  [header, rack_key]
+end.to_h
+lookup.add_function("header", header_mapping, ignore_case: true)
+lookup.add_function("method", METHODS)
+lookup.generate

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: picohttp
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.2.0
 platform: ruby
 authors:
 - John Hawthorn
@@ -27,8 +27,10 @@ files:
 - ext/picohttp/picohttp.h
 - ext/picohttp/picohttpparser.c
 - ext/picohttp/picohttpparser.h
+- ext/picohttp/string_lookup.inc
 - lib/picohttp.rb
 - lib/picohttp/version.rb
+- tool/generate_lookup.rb
 homepage: https://github.com/jhawthorn/picohttp
 licenses:
 - MIT
@@ -52,7 +54,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.9
+rubygems_version: 4.0.2
 specification_version: 4
 summary: Fast HTTP request parser using picohttpparser
 test_files: []