hpricot 0.6-mswin32 → 0.6.164-mswin32

data/ext/hpricot_scan/hpricot_scan.java.rl

@@ -6,7 +6,9 @@ import org.jruby.RubyClass;
 import org.jruby.RubyHash;
 import org.jruby.RubyModule;
 import org.jruby.RubyNumeric;
+import org.jruby.RubyObjectAdapter;
 import org.jruby.RubyString;
+import org.jruby.javasupport.JavaEmbedUtils;
 import org.jruby.runtime.Block;
 import org.jruby.runtime.CallbackFactory;
 import org.jruby.runtime.builtin.IRubyObject;
@@ -15,13 +17,14 @@ import org.jruby.runtime.load.BasicLibraryService;
 
 public class HpricotScanService implements BasicLibraryService {
        public static String NO_WAY_SERIOUSLY="*** This should not happen, please send a bug report with the HTML you're parsing to why@whytheluckystiff.net.  So sorry!";
+       private static RubyObjectAdapter rubyApi;
 
        public void ELE(IRubyObject N) {
-         if (tokend > tokstart || text) {
+         if (te > ts || text) {
            IRubyObject raw_string = runtime.getNil();
            ele_open = false; text = false;
-           if (tokstart != -1 && N != cdata && N != sym_text && N != procins && N != comment) { 
-             raw_string = runtime.newString(new String(buf,tokstart,tokend-tokstart));
+           if (ts != -1 && N != cdata && N != sym_text && N != procins && N != comment) { 
+             raw_string = runtime.newString(new String(buf,ts,te-ts));
            } 
            rb_yield_tokens(N, tag[0], attr, raw_string, taint);
          }
@@ -75,13 +78,13 @@ public class HpricotScanService implements BasicLibraryService {
            } else if(N == aval) {
              mark = mark_aval;
            }
-           if(mark > tokstart) {
+           if(mark > ts) {
              if(N == tag) {
-               mark_tag  -= tokstart;
+               mark_tag  -= ts;
              } else if(N == akey) {
-               mark_akey -= tokstart;
+               mark_akey -= ts;
              } else if(N == aval) {
-               mark_aval -= tokstart;
+               mark_aval -= ts;
              }
            }
        }
@@ -91,7 +94,8 @@ public class HpricotScanService implements BasicLibraryService {
            if(attr.isNil()) {
              attr = RubyHash.newHash(runtime);
            }
-           ((RubyHash)attr).aset(K,V);
+           ((RubyHash)attr).op_aset(runtime.getCurrentContext(),K,V);
+           // ((RubyHash)attr).aset(K,V);
          }
        }
 
@@ -111,8 +115,8 @@ public class HpricotScanService implements BasicLibraryService {
          if(!text) { 
            if(ele_open) { 
              ele_open = false; 
-             if(tokstart > -1) { 
-               mark_tag = tokstart; 
+             if(ts > -1) { 
+               mark_tag = ts; 
              } 
            } else {
              mark_tag = p; 
@@ -180,7 +184,7 @@ public class HpricotScanService implements BasicLibraryService {
     ATTR(akey, aval);
   }
 
-  include hpricot_common "ext/hpricot_scan/hpricot_common.rl";
+  include hpricot_common "hpricot_common.rl";
 
 }%%
 
@@ -206,7 +210,8 @@ private void rb_yield_tokens(IRubyObject sym, IRubyObject tag, IRubyObject attr,
 
 int cs, act, have = 0, nread = 0, curline = 1, p=-1;
 boolean text = false;
-int tokstart=-1, tokend;
+int ts=-1, te;
+int eof=-1;
 char[] buf;
 Ruby runtime;
 IRubyObject attr, bufsize;
@@ -239,8 +244,8 @@ IRubyObject hpricot_scan(IRubyObject recv, IRubyObject port) {
   }
 
   buffer_size = BUFSIZE;
-  if (recv.getInstanceVariable("@buffer_size") != null) {
-    bufsize = recv.getInstanceVariable("@buffer_size");
+  if (rubyApi.getInstanceVariable(recv, "@buffer_size") != null) {
+    bufsize = rubyApi.getInstanceVariable(recv, "@buffer_size");
     if (!bufsize.isNil()) {
       buffer_size = RubyNumeric.fix2int(bufsize);
     }
@@ -296,16 +301,16 @@ IRubyObject hpricot_scan(IRubyObject recv, IRubyObject port) {
     
     if ( done && ele_open ) {
       ele_open = false;
-      if(tokstart > -1) {
-        mark_tag = tokstart;
-        tokstart = -1;
+      if(ts > -1) {
+        mark_tag = ts;
+        ts = -1;
         text = true;
       }
     }
 
-    if(tokstart == -1) {
+    if(ts == -1) {
       have = 0;
-      /* text nodes have no tokstart because each byte is parsed alone */
+      /* text nodes have no ts because each byte is parsed alone */
       if(mark_tag != -1 && text) {
         if (done) {
           if(mark_tag < p-1) {
@@ -318,13 +323,13 @@ IRubyObject hpricot_scan(IRubyObject recv, IRubyObject port) {
       }
       mark_tag = 0;
     } else {
-      have = pe - tokstart;
-      System.arraycopy(buf,tokstart,buf,0,have);
+      have = pe - ts;
+      System.arraycopy(buf,ts,buf,0,have);
       SLIDE(tag);
       SLIDE(akey);
       SLIDE(aval);
-      tokend = (tokend - tokstart);
-      tokstart = 0;
+      te = (te - ts);
+      ts = 0;
     }
   }
   return runtime.getNil();
@@ -355,9 +360,10 @@ public boolean basicLoad(final Ruby runtime) throws IOException {
 
 public static void Init_hpricot_scan(Ruby runtime) {
   RubyModule mHpricot = runtime.defineModule("Hpricot");
-  mHpricot.getMetaClass().attr_accessor(new IRubyObject[]{runtime.newSymbol("buffer_size")});
+  mHpricot.getMetaClass().attr_accessor(runtime.getCurrentContext(),new IRubyObject[]{runtime.newSymbol("buffer_size")});
   CallbackFactory fact = runtime.callbackFactory(HpricotScanService.class);
   mHpricot.getMetaClass().defineMethod("scan",fact.getSingletonMethod("__hpricot_scan",IRubyObject.class));
-  mHpricot.defineClassUnder("ParseError",runtime.getClass("Exception"),runtime.getClass("Exception").getAllocator());
+  mHpricot.defineClassUnder("ParseError",runtime.getClass("StandardError"),runtime.getClass("StandardError").getAllocator());
+  rubyApi = JavaEmbedUtils.newObjectAdapter();
 }
 }

data/ext/hpricot_scan/hpricot_scan.rl

@@ -8,6 +8,12 @@
  */
 #include <ruby.h>
 
+#ifndef RARRAY_LEN
+#define RARRAY_LEN(arr)  RARRAY(arr)->len
+#define RSTRING_LEN(str) RSTRING(str)->len
+#define RSTRING_PTR(str) RSTRING(str)->ptr
+#endif
+
 #define NO_WAY_SERIOUSLY "*** This should not happen, please send a bug report with the HTML you're parsing to why@whytheluckystiff.net.  So sorry!"
 
 static VALUE sym_xmldecl, sym_doctype, sym_procins, sym_stag, sym_etag, sym_emptytag, sym_comment,
@@ -16,11 +22,11 @@ static VALUE rb_eHpricotParseError;
 static ID s_read, s_to_str;
 
 #define ELE(N) \
-  if (tokend > tokstart || text == 1) { \
+  if (te > ts || text == 1) { \
     VALUE raw_string = Qnil; \
     ele_open = 0; text = 0; \
-    if (tokstart != 0 && sym_##N != sym_cdata && sym_##N != sym_text && sym_##N != sym_procins && sym_##N != sym_comment) { \
-      raw_string = rb_str_new(tokstart, tokend-tokstart); \
+    if (ts != 0 && sym_##N != sym_cdata && sym_##N != sym_text && sym_##N != sym_procins && sym_##N != sym_comment) { \
+      raw_string = rb_str_new(ts, te-ts); \
     } \
     rb_yield_tokens(sym_##N, tag, attr, raw_string, taint); \
   }
@@ -33,7 +39,7 @@ static ID s_read, s_to_str;
 
 #define CAT(N, E) if (NIL_P(N)) { SET(N, E); } else { rb_str_cat(N, mark_##N, E - mark_##N); }
 
-#define SLIDE(N) if ( mark_##N > tokstart ) mark_##N = buf + (mark_##N - tokstart);
+#define SLIDE(N) if ( mark_##N > ts ) mark_##N = buf + (mark_##N - ts);
 
 #define ATTR(K, V) \
     if (!NIL_P(K)) { \
@@ -46,8 +52,8 @@ static ID s_read, s_to_str;
     { \
       if (ele_open == 1) { \
         ele_open = 0; \
-        if (tokstart > 0) { \
-          mark_tag = tokstart; \
+        if (ts > 0) { \
+          mark_tag = ts; \
         } \
       } else { \
         mark_tag = p; \
@@ -102,7 +108,7 @@ static ID s_read, s_to_str;
     ATTR(akey, aval);
   }
 
-  include hpricot_common "ext/hpricot_scan/hpricot_common.rl";
+  include hpricot_common "hpricot_common.rl";
 
 }%%
 
@@ -129,7 +135,7 @@ void rb_yield_tokens(VALUE sym, VALUE tag, VALUE attr, VALUE raw, int taint)
 VALUE hpricot_scan(VALUE self, VALUE port)
 {
   int cs, act, have = 0, nread = 0, curline = 1, text = 0;
-  char *tokstart = 0, *tokend = 0, *buf = NULL;
+  char *ts = 0, *te = 0, *buf = NULL, *eof = NULL;
 
   VALUE attr = Qnil, tag = Qnil, akey = Qnil, aval = Qnil, bufsize = Qnil;
   char *mark_tag = 0, *mark_akey = 0, *mark_aval = 0;
@@ -169,7 +175,7 @@ VALUE hpricot_scan(VALUE self, VALUE port)
       /* We've used up the entire buffer storing an already-parsed token
        * prefix that must be preserved.  Likely caused by super-long attributes.
        * See ticket #13. */
-      rb_raise(rb_eHpricotParseError, "ran out of buffer space on element <%s>, starting on line %d.", RSTRING(tag)->ptr, curline);
+      rb_raise(rb_eHpricotParseError, "ran out of buffer space on element <%s>, starting on line %d.", RSTRING_PTR(tag), curline);
     }
 
     if ( rb_respond_to( port, s_read ) )
@@ -182,8 +188,8 @@ VALUE hpricot_scan(VALUE self, VALUE port)
     }
 
     StringValue(str);
-    memcpy( p, RSTRING(str)->ptr, RSTRING(str)->len );
-    len = RSTRING(str)->len;
+    memcpy( p, RSTRING_PTR(str), RSTRING_LEN(str) );
+    len = RSTRING_LEN(str);
     nread += len;
 
     /* If this is the last buffer, tack on an EOF. */
@@ -199,7 +205,7 @@ VALUE hpricot_scan(VALUE self, VALUE port)
       free(buf);
       if ( !NIL_P(tag) )
       {
-        rb_raise(rb_eHpricotParseError, "parse error on element <%s>, starting on line %d.\n" NO_WAY_SERIOUSLY, RSTRING(tag)->ptr, curline);
+        rb_raise(rb_eHpricotParseError, "parse error on element <%s>, starting on line %d.\n" NO_WAY_SERIOUSLY, RSTRING_PTR(tag), curline);
       }
       else
       {
@@ -210,17 +216,17 @@ VALUE hpricot_scan(VALUE self, VALUE port)
     if ( done && ele_open )
     {
       ele_open = 0;
-      if (tokstart > 0) {
-        mark_tag = tokstart;
-        tokstart = 0;
+      if (ts > 0) {
+        mark_tag = ts;
+        ts = 0;
         text = 1;
       }
     }
 
-    if ( tokstart == 0 )
+    if ( ts == 0 )
     {
       have = 0;
-      /* text nodes have no tokstart because each byte is parsed alone */
+      /* text nodes have no ts because each byte is parsed alone */
       if ( mark_tag != NULL && text == 1 )
       {
         if (done)
@@ -240,13 +246,13 @@ VALUE hpricot_scan(VALUE self, VALUE port)
     }
     else
     {
-      have = pe - tokstart;
-      memmove( buf, tokstart, have );
+      have = pe - ts;
+      memmove( buf, ts, have );
       SLIDE(tag);
       SLIDE(akey);
       SLIDE(aval);
-      tokend = buf + (tokend - tokstart);
-      tokstart = buf;
+      te = buf + (te - ts);
+      ts = buf;
     }
   }
   free(buf);
@@ -257,7 +263,7 @@ void Init_hpricot_scan()
   VALUE mHpricot = rb_define_module("Hpricot");
   rb_define_attr(rb_singleton_class(mHpricot), "buffer_size", 1, 1);
   rb_define_singleton_method(mHpricot, "scan", hpricot_scan, 1);
-  rb_eHpricotParseError = rb_define_class_under(mHpricot, "ParseError", rb_eException);
+  rb_eHpricotParseError = rb_define_class_under(mHpricot, "ParseError", rb_eStandardError);
 
   s_read = rb_intern("read");
   s_to_str = rb_intern("to_str");

data/ext/hpricot_scan/test.rb

@@ -0,0 +1,5 @@
+require './hpricot_scan.so'
+
+doc = "<doc><person><test>YESSS</test></person><train>SET</train></doc>"
+Hpricot.scan(doc) { |x| p x }
+p Hpricot.lemon(doc)

data/lib/hpricot/builder.rb

@@ -1,8 +1,23 @@
 require 'hpricot/tags'
-require 'hpricot/xchar'
+require 'fast_xs'
 require 'hpricot/blankslate'
 
 module Hpricot
+  PREDEFINED = {
+    34 => '"', # quotation mark
+    38 => '&',  # ampersand
+    60 => '<',   # left angle bracket
+    62 => '>'    # right angle bracket
+  }
+  PREDEFINED_U = PREDEFINED.inject({}) { |hsh, (k, v)| hsh[v] = k; hsh }
+
+  # XML unescape
+  def self.uxs(str)
+    str.to_s.
+        gsub(/\&\w+;/) { |x| (PREDEFINED_U[x] || ??).chr }.
+        gsub(/\&\#(\d+);/) { [$1.to_i].pack("U*") }
+  end
+
   def self.build(ele = Doc.new, assigns = {}, &blk)
     ele.extend Builder
     assigns.each do |k, v|
@@ -32,7 +47,7 @@ module Hpricot
 
     # Write a +string+ to the HTML stream, making sure to escape it.
     def text!(string)
-      @children << Text.new(Hpricot.xs(string))
+      @children << Text.new(string.fast_xs)
     end
 
     # Write a +string+ to the HTML stream without escaping it.
@@ -75,16 +90,16 @@ module Hpricot
       # turn arguments into children or attributes
       childs = []
       attrs = args.grep(Hash)
-      childs.concat((args - attrs).map do |x|
+      childs.concat((args - attrs).flatten.map do |x|
         if x.respond_to? :to_html
           Hpricot.make(x.to_html)
         elsif x
-          Text.new(Hpricot.xs(x))
+          Text.new(x.fast_xs)
         end
       end.flatten)
       attrs = attrs.inject({}) do |hsh, ath|
         ath.each do |k, v|
-          hsh[k] = Hpricot.xs(v.to_s) if v
+          hsh[k] = v.to_s.fast_xs if v
         end
         hsh
       end

data/lib/hpricot/elements.rb

@@ -130,25 +130,25 @@ module Hpricot
     # Add to the end of the contents inside each element in this list.
     # Pass in an HTML +str+, which is turned into Hpricot elements.
     def append(str = nil, &blk)
-      each { |x| x.html(x.children + Hpricot.make(str, &blk)) }
+      each { |x| x.html(x.children + x.make(str, &blk)) }
     end
 
     # Add to the start of the contents inside each element in this list.
     # Pass in an HTML +str+, which is turned into Hpricot elements.
     def prepend(str = nil, &blk)
-      each { |x| x.html(Hpricot.make(str, &blk) + x.children) }
+      each { |x| x.html(x.make(str, &blk) + x.children) }
     end
  
     # Add some HTML just previous to each element in this list.
     # Pass in an HTML +str+, which is turned into Hpricot elements.
     def before(str = nil, &blk)
-      each { |x| x.parent.insert_before Hpricot.make(str, &blk), x }
+      each { |x| x.parent.insert_before x.make(str, &blk), x }
     end
 
     # Just after each element in this list, add some HTML.
     # Pass in an HTML +str+, which is turned into Hpricot elements.
     def after(str = nil, &blk)
-      each { |x| x.parent.insert_after Hpricot.make(str, &blk), x }
+      each { |x| x.parent.insert_after x.make(str, &blk), x }
     end
 
     # Wraps each element in the list inside the element created by HTML +str+. 
@@ -161,10 +161,10 @@ module Hpricot
     # This code wraps every link on the page inside a +div.link+ and a +div.link_inner+ nest.
     def wrap(str = nil, &blk)
       each do |x|
-        wrap = Hpricot.make(str, &blk)
+        wrap = x.make(str, &blk)
         nest = wrap.detect { |w| w.respond_to? :children }
         unless nest
-          raise Exception, "No wrapping element found."
+          raise "No wrapping element found."
         end
         x.parent.replace_child(x, wrap)
         nest = nest.children.first until nest.empty?
@@ -261,7 +261,7 @@ module Hpricot
       self      
     end
 
-    ATTR_RE = %r!\[ *(?:(@)([\w\(\)-]+)|([\w\(\)-]+\(\))) *([~\!\|\*$\^=]*) *'?"?([^\]'"]*)'?"? *\]!i
+    ATTR_RE = %r!\[ *(?:(@)([\w\(\)-]+)|([\w\(\)-]+\(\))) *([~\!\|\*$\^=]*) *'?"?([^'"]*)'?"? *\]!i
     BRACK_RE = %r!(\[) *([^\]]*) *\]+!i
     FUNC_RE = %r!(:)?([a-zA-Z0-9\*_-]*)\( *[\"']?([^ \)]*?)['\"]? *\)!
     CUST_RE = %r!(:)([a-zA-Z0-9\*_-]*)()!

data/lib/hpricot/parse.rb

@@ -12,13 +12,14 @@ module Hpricot
   # Hpricot.parse parses <i>input</i> and return a document tree.
   # represented by Hpricot::Doc.
   def Hpricot.parse(input = nil, opts = {}, &blk)
-    Doc.new(make(input, opts, &blk))
+    Doc.new(make(input, opts, &blk), opts)
   end
 
   # Hpricot::XML parses <i>input</i>, disregarding all the HTML rules
   # and returning a document tree.
-  def Hpricot.XML(input, opts = {})
-    Doc.new(make(input, opts.merge(:xml => true)))
+  def Hpricot.XML(input = nil, opts = {}, &blk)
+    opts.merge! :xml => true
+    Doc.new(make(input, opts, &blk), opts)
   end
 
   # :stopdoc:
@@ -137,7 +138,7 @@ module Hpricot
               matched_elem = stack[i]
               stack[i][1] += token
               eles = stack.slice!((i+1)..-1)
-              stack.last[2] += eles
+              stack.last[2] += eles if eles
               break
             end
           end
@@ -208,7 +209,7 @@ module Hpricot
     when :cdata
       Text.parse_cdata_section(structure[1])
     else
-      raise Exception, "[bug] unknown structure: #{structure.inspect}"
+      raise "[bug] unknown structure: #{structure.inspect}"
     end
   end
 

data/lib/hpricot/tag.rb

@@ -3,8 +3,9 @@ module Hpricot
 
   class Doc
     attr_accessor :children
-    def initialize(children = [])
+    def initialize(children = [], options = {})
       @children = children ? children.each { |c| c.parent = self }  : []
+      @options = options
     end
     def output(out, opts = {})
       @children.each do |n|
@@ -12,6 +13,9 @@ module Hpricot
       end
       out
     end
+    def make(input = nil, &blk)
+      Hpricot.make(input, @options, &blk)
+    end
     def altered!; end
   end
 
@@ -100,7 +104,7 @@ module Hpricot
       if @raw_attributes
         @raw_attributes.map do |aname, aval|
           " #{aname}" +
-            (aval ? "=\"#{aval}\"" : "")
+            (aval ? "=#{html_quote aval}" : "")
         end.join
       end
     end