Changeset 648

Timestamp:

10/11/07 19:11:52 (1 year ago)

Author:

rydahl

Message:

First pass at adding (un)escape and query_string_parse methods. query_string_parse is now known as HttpParser?.form_parse since it parses the body of http requests with type application/x-www-form-urlencoded.
Also removed query_parse/ext/http11/http11_parser.c since it is generated by Ragel.

Files:

• branches/query_parse/Rakefile (modified) (1 diff)
• branches/query_parse/ext/http11/ext_help.h (modified) (1 diff)
• branches/query_parse/ext/http11/form_parser.h (added)
• branches/query_parse/ext/http11/form_parser.rl (added)
• branches/query_parse/ext/http11/http11.c (modified) (6 diffs)
• branches/query_parse/ext/http11/http11_parser.c (deleted)
• branches/query_parse/ext/http11/http11_parser.h (modified) (1 diff)
• branches/query_parse/ext/http11/http11_parser.rl (modified) (2 diffs)
• branches/query_parse/test/test_http11.rb (modified) (2 diffs)

branches/query_parse/Rakefile

@@ -109 +109 @@
-end
-
-
+
+
+
-  sh %{ragel ext/http11/http11_parser.rl | rlgen-cd -G2 -o ext/http11/http11_parser.c}
+end
+
+file('ext/http11/form_parser.c' => 'ext/http11/form_parser.rl') do
+  sh %{ragel ext/http11/form_parser.rl | rlgen-cd -G2 -o ext/http11/form_parser.c}
-end
-

branches/query_parse/ext/http11/ext_help.h

@@ -12 +12 @@
-#endif
-
+#define DEBUG (RTEST(ruby_debug))
+#define INSPECT(n,a) (fprintf(stderr, "*** %s: %s\n", n, RSTRING_PTR(rb_funcall(a, rb_intern("inspect"), 0))))
+
+
-#endif

branches/query_parse/ext/http11/http11.c

@@ -8 +8 @@
-#include <string.h>
-#include "http11_parser.h"
+#include "form_parser.h"
-#include <ctype.h>
-#include "tst.h"
@@ -15 +16 @@
-static VALUE cURIClassifier;
-static VALUE eHttpParserError;
+static VALUE eQSPError;
-
-#define id_handler_map rb_intern("@handler_map")
@@ -499 +501 @@
-{
-  void *handler = NULL;
-
+unsigned 
-  struct tst *tst = NULL;
-  VALUE result;
@@ -533 +535 @@
-  return result;
-}
+
+/***** (Un)escape URL functions *****/
+
+VALUE unescape_helper(const char *at, long len)
+{
+  long out_len = url_unescape_length(at, len);
+  char out[out_len];
+  
+  url_unescape(at, len, out);
+  return rb_str_new(out, out_len);
+}
+
+VALUE escape_helper(const char *at, long len)
+{
+  long out_len = url_escape_length(at, len);
+  char out[out_len];
+  
+  url_escape(at, len, out);
+  return rb_str_new(out, out_len);
+}
+
+VALUE HttpParser_unescape(VALUE self, VALUE input) {
+  Check_Type(input, T_STRING);
+  return unescape_helper(RSTRING_PTR(input), RSTRING_LEN(input));
+}
+
+VALUE HttpParser_escape(VALUE self, VALUE input) {
+  Check_Type(input, T_STRING);
+  return escape_helper(RSTRING_PTR(input), RSTRING_LEN(input));
+  
+}
+
+struct qsp_data {
+  VALUE keys;
+  VALUE value;
+  VALUE hash;
+};
+
+/******** Query Parse Ragel Callbacks *****************************/
+void key_cb(void *data, const char *at, size_t length)
+{
+  struct qsp_data *qsp_data = (struct qsp_data *) data;
+  VALUE new_key;
+  
+  if(DEBUG) INSPECT("keys", qsp_data->keys);
+  
+  if(qsp_data->keys == Qnil) {
+    qsp_data->keys = rb_ary_new();
+    rb_gc_mark(qsp_data->keys);
+  }
+  
+  new_key = unescape_helper(at, (long)length);
+  if(DEBUG) INSPECT("key", new_key);
+  rb_ary_push(qsp_data->keys, new_key);
+}
+
+void value_cb(void *data, const char *at, size_t length)
+{
+  struct qsp_data *qsp_data = (struct qsp_data *) data;
+  
+  if(qsp_data->value != Qnil)
+    rb_raise(eQSPError, "Query pair value already set!");
+  qsp_data->value = unescape_helper(at, (long)length);
+  if(DEBUG) INSPECT("value", qsp_data->value);
+}
+
+void pair_cb(void *data, const char *at, size_t length)
+{
+  struct qsp_data *qsp_data = (struct qsp_data *) data;
+  VALUE last_key, key, a, b;
+  
+  if(DEBUG) INSPECT("pair", Qnil);
+  
+  last_key = rb_ary_pop(qsp_data->keys);
+  
+  a = qsp_data->hash;  
+  while((key = rb_ary_shift(qsp_data->keys))  != Qnil) {
+    b = rb_hash_aref(a, key);
+    if(b == Qnil) {
+      b = rb_hash_new(); 
+      rb_hash_aset(a, key, b);
+    }
+    a = b;
+  }
+  /* a is either an array or hash to put our value into */
+  /* TODO: if last_key is blank, then a should be an array, and we push into
+   * it */
+  rb_hash_aset(a, last_key, qsp_data->value);
+  
+  qsp_data->keys = Qnil;
+  qsp_data->value = Qnil;
+}
+
+void error_cb(void *data)
+{
+  rb_raise(eQSPError, "Parse Error.");
+}
+
+/******** </Query Parse Ragel Callbacks> **************************/
+
+// TODO: make optional 2nd argument to pass in Hash to add to.
+VALUE HttpParser_form_parse(VALUE self, VALUE input) 
+{
+  Check_Type(input, T_STRING);
+  form_parser parser;
+  struct qsp_data qsp_data;
+  size_t nread;
+  TRACE();
+  
+  parser.data = NULL;
+  parser.pair_cb = pair_cb;
+  parser.key_cb = key_cb;
+  parser.value_cb = value_cb;
+  parser.error_cb = error_cb;
+  
+  qsp_data.hash = rb_hash_new(); 
+  rb_gc_mark(qsp_data.hash);
+  qsp_data.keys = Qnil;
+  qsp_data.value = Qnil;
+  parser.data = (void *) &qsp_data;
+  
+  if(DEBUG) INSPECT("parse input", input);
+  
+  nread = form_parser_execute(&parser, 
+                              StringValuePtr(input), 
+                              RSTRING_LEN(input), 
+                              0);
+  if(DEBUG) INSPECT("nread", INT2FIX((int)nread));
+  
+  return qsp_data.hash;
+}
+
-
-
@@ -572 +706 @@
-  rb_define_method(cHttpParser, "finished?", HttpParser_is_finished,0);
-  rb_define_method(cHttpParser, "nread", HttpParser_nread,0);
+  
+  rb_define_singleton_method(cHttpParser, "unescape", HttpParser_unescape, 1);
+  rb_define_singleton_method(cHttpParser, "escape", HttpParser_escape, 1);
+  rb_define_singleton_method(cHttpParser, "form_parse", HttpParser_form_parse, 1);
-
-  cURIClassifier = rb_define_class_under(mMongrel, "URIClassifier", rb_cObject);
@@ -579 +717 @@
-  rb_define_method(cURIClassifier, "unregister", URIClassifier_unregister, 1);
-  rb_define_method(cURIClassifier, "resolve", URIClassifier_resolve, 1);
-
+
+

branches/query_parse/ext/http11/http11_parser.h

@@ -44 +44 @@
-int http_parser_is_finished(http_parser *parser);
-
+
+long url_escape_length(const char *s, long len);
+void url_escape(const char *s, long len, char *out);
+
+long url_unescape_length(const char *s, long len);
+void url_unescape(const char *s, long len, char *out);
+
-#define http_parser_nread(parser) (parser)->nread 
-

branches/query_parse/ext/http11/http11_parser.rl

@@ -13 +13 @@
-#define MARK(M,FPC) (parser->M = (FPC) - buffer)
-#define PTR_TO(F) (buffer + parser->F)
+
+/* For [un]escape */
+#define XDIGIT_TO_NUM(h) ((h) < 'A' ? (h) - '0' : toupper(h) - 'A' + 10)
+#define X2DIGITS_TO_NUM(h1, h2) ((XDIGIT_TO_NUM (h1) << 4) + XDIGIT_TO_NUM (h2))
+#define XNUM_TO_DIGIT(x) ("0123456789ABCDEF"[x] + 0)
+#define SAFE_CHAR(c) ( \
+ ('a' <= c && c <= 'z') || \
+ ('A' <= c && c <= 'Z') || \
+ ('0' <= c && c <= '9') || \
+ c == '$' || \
+ c == '_' || \
+ c == '.' || \
+ c == '-' \
+)
-
-/** machine **/
@@ -191 +205 @@
-  return parser->cs == http_parser_first_final;
-}
+
+
+/* returns the length of a 2 b escaped string. O(n) */
+long url_escape_length(const char *s, long len)
+{
+  long i, escape_count = 0;
+  
+  for(i=0; i<len; i++)
+    if(!SAFE_CHAR(s[i])) escape_count++;
+  return len + 2*escape_count;
+}
+
+/* escapes a string s into a string out.
+ * does not allocate memory for out. you must allocate and free this
+ * string yourself. use url_escape_length() to find out how long the
+ * string needs to be.
+ */
+void url_escape(const char *s, long len, char *out) 
+{
+  long i,j;
+  
+  for(i=0,j=0; i<len; i++) {
+    if(!SAFE_CHAR(s[i])) {
+      out[j++] = '%';
+      out[j++] = XNUM_TO_DIGIT (s[i] >> 4);
+      out[j++] = XNUM_TO_DIGIT (s[i] & 0xf);
+    } else out[j++] = s[i];
+  }
+}
+
+/* returns the length of a 2 b unescaped string. O(n) */
+long url_unescape_length(const char *s, long len)
+{
+  long i, escape_count = 0;
+  
+  for(i=0; i < len-2; i++) {
+    if(s[i] == '%' && isxdigit(s[i+1]) && isxdigit(s[i+2])) {
+      i+=2;
+      escape_count++;
+    }
+  }
+  return len-2*escape_count;
+}
+
+/* unescapes URL. e.g "%21" becomes "!"  */
+void url_unescape(const char *s, long len, char *out) 
+{
+  long i, j;
+  
+  for(i=0,j=0; i<len; i++,j++) {
+    if(s[i] == '+') {
+      out[j] = ' ';
+    } else if(s[i] == '%' && i < len-2 && isxdigit(s[i+1]) && isxdigit(s[i+2])) {
+      out[j] = X2DIGITS_TO_NUM(s[i+1], s[i+2]);
+      i+=2;
+    } else out[j] = s[i];
+  }
+}

branches/query_parse/test/test_http11.rb

@@ -12 +12 @@
-
-include Mongrel
+
+def escape(s)
+  s.to_s.gsub(/([^ a-zA-Z0-9_.-=,]+)/n) {
+    '%'+$1.unpack('H2'*$1.size).join('%').upcase
+  }.tr(' ', '+') 
+end
-
-class HttpParserTest < Test::Unit::TestCase
@@ -121 +127 @@
-
-  def test_query_parse
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-    assert res["zed"], "didn't get the request right"
-    assert res["frank"], "no frank"
-    assert_equal "1", res["zed"], "wrong result"
-Request
+Parser
-
-Request.query
+Parser.form
-    assert res["zed"], "didn't get the request right"
-    assert res["frank"], "no frank"
-
-
+
+
+
+
-  end
-