nokogiri 1.3.3 → 1.4.0

data/bin/nokogiri

@@ -6,6 +6,8 @@ require 'uri'
 require 'rubygems'
 require 'nokogiri'
 
+parse_class = Nokogiri
+
 opts = OptionParser.new do |opts|
   opts.banner = "Nokogiri: an HTML, XML, SAX, and Reader parser"
   opts.define_head "Usage: nokogiri <uri|path> [options]"
@@ -16,6 +18,10 @@ opts = OptionParser.new do |opts|
   opts.separator ""
   opts.separator "Options:"
 
+  opts.on("--type [TYPE]", [:xml, :html]) do |v|
+    parse_class = {:xml => Nokogiri::XML, :html => Nokogiri::HTML}[v]
+  end
+
   opts.on_tail("-?", "--help", "Show this message") do
     puts opts
     exit
@@ -36,11 +42,7 @@ if uri.to_s.strip.empty?
   exit 1
 end
 
-if URI.parse(uri).scheme
-  @doc = Nokogiri(open(uri).read)
-else
-  @doc = Nokogiri(File.read(uri))
-end
+@doc = parse_class.parse(open(uri).read)
 
 puts "Your document is stored in @doc..."
 IRB.start

data/ext/nokogiri/extconf.rb

@@ -129,27 +129,11 @@ unless find_library('exslt', 'exsltFuncRegister', *LIB_DIRS)
   abort "libxslt is missing.  try 'port install libxslt' or 'yum install libxslt-devel'"
 end
 
-def nokogiri_link_command ldflags, opt='', libpath=$LIBPATH
-  old_link_command ldflags, opt, libpath
-end
-
-def with_custom_link
-  alias :old_link_command :link_command
-  alias :link_command :nokogiri_link_command
-  yield
-ensure
-  alias :link_command :old_link_command
-end
-
-with_custom_link do
-  with_cppflags $INCFLAGS do
-    have_func('xmlRelaxNGSetParserStructuredErrors')
-    have_func('xmlRelaxNGSetParserStructuredErrors')
-    have_func('xmlRelaxNGSetValidStructuredErrors')
-    have_func('xmlSchemaSetValidStructuredErrors')
-    have_func('xmlSchemaSetParserStructuredErrors')
-  end
-end
+have_func('xmlRelaxNGSetParserStructuredErrors')
+have_func('xmlRelaxNGSetParserStructuredErrors')
+have_func('xmlRelaxNGSetValidStructuredErrors')
+have_func('xmlSchemaSetValidStructuredErrors')
+have_func('xmlSchemaSetParserStructuredErrors')
 
 if ENV['CPUPROFILE']
   unless find_library('profiler', 'ProfilerEnable', *LIB_DIRS)

data/ext/nokogiri/html_document.c

@@ -11,15 +11,15 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   VALUE uri, external_id, rest, rb_doc;
 
   rb_scan_args(argc, argv, "0*", &rest);
-  uri         = rb_ary_entry(rest, 0);
-  external_id = rb_ary_entry(rest, 1);
+  uri         = rb_ary_entry(rest, (long)0);
+  external_id = rb_ary_entry(rest, (long)1);
 
   htmlDocPtr doc = htmlNewDoc(
       RTEST(uri) ? (const xmlChar *)StringValuePtr(uri) : NULL,
       RTEST(external_id) ? (const xmlChar *)StringValuePtr(external_id) : NULL
   );
   rb_doc = Nokogiri_wrap_xml_document(klass, doc);
-  rb_funcall2(rb_doc, rb_intern("initialize"), argc, argv);
+  rb_obj_call_init(rb_doc, argc, argv);
   return rb_doc ;
 }
 
@@ -36,8 +36,8 @@ static VALUE read_io( VALUE klass,
                       VALUE encoding,
                       VALUE options )
 {
-  const char * c_url    = (url == Qnil) ? NULL : StringValuePtr(url);
-  const char * c_enc    = (encoding == Qnil) ? NULL : StringValuePtr(encoding);
+  const char * c_url    = NIL_P(url)      ? NULL : StringValuePtr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
   VALUE error_list      = rb_ary_new();
 
   xmlResetLastError();
@@ -49,7 +49,7 @@ static VALUE read_io( VALUE klass,
       (void *)io,
       c_url,
       c_enc,
-      NUM2INT(options)
+      (int)NUM2INT(options)
   );
   xmlSetStructuredErrorFunc(NULL, NULL);
 
@@ -58,9 +58,7 @@ static VALUE read_io( VALUE klass,
 
     xmlErrorPtr error = xmlGetLastError();
     if(error)
-      rb_funcall(rb_mKernel, rb_intern("raise"), 1,
-          Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error)
-      );
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
     else
       rb_raise(rb_eRuntimeError, "Could not parse document");
 
@@ -68,7 +66,7 @@ static VALUE read_io( VALUE klass,
   }
 
   VALUE document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_funcall(document, rb_intern("errors="), 1, error_list);
+  rb_iv_set(document, "@errors", error_list);
   return document;
 }
 
@@ -86,15 +84,15 @@ static VALUE read_memory( VALUE klass,
                           VALUE options )
 {
   const char * c_buffer = StringValuePtr(string);
-  const char * c_url    = (url == Qnil) ? NULL : StringValuePtr(url);
-  const char * c_enc    = (encoding == Qnil) ? NULL : StringValuePtr(encoding);
+  const char * c_url    = NIL_P(url)      ? NULL : StringValuePtr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
   int len               = RSTRING_LEN(string);
   VALUE error_list      = rb_ary_new();
 
   xmlResetLastError();
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 
-  htmlDocPtr doc = htmlReadMemory(c_buffer, len, c_url, c_enc, NUM2INT(options));
+  htmlDocPtr doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
   xmlSetStructuredErrorFunc(NULL, NULL);
 
   if(doc == NULL) {
@@ -102,9 +100,7 @@ static VALUE read_memory( VALUE klass,
 
     xmlErrorPtr error = xmlGetLastError();
     if(error)
-      rb_funcall(rb_mKernel, rb_intern("raise"), 1,
-          Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error)
-      );
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
     else
       rb_raise(rb_eRuntimeError, "Could not parse document");
 
@@ -112,7 +108,7 @@ static VALUE read_memory( VALUE klass,
   }
 
   VALUE document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_funcall(document, rb_intern("errors="), 1, error_list);
+  rb_iv_set(document, "@errors", error_list);
   return document;
 }
 
@@ -126,37 +122,7 @@ static VALUE type(VALUE self)
 {
   htmlDocPtr doc;
   Data_Get_Struct(self, xmlDoc, doc);
-  return INT2NUM((int)doc->type);
-}
-
-/*
- * call-seq:
- *  meta_encoding=
- *
- * Set the meta tag encoding for this document.
- */
-static VALUE set_meta_encoding(VALUE self, VALUE encoding)
-{
-  htmlDocPtr doc;
-  Data_Get_Struct(self, xmlDoc, doc);
-
-  htmlSetMetaEncoding(doc, (const xmlChar *)StringValuePtr(encoding));
-
-  return encoding;
-}
-
-/*
- * call-seq:
- *  meta_encoding
- *
- * Get the meta tag encoding for this document.
- */
-static VALUE meta_encoding(VALUE self)
-{
-  htmlDocPtr doc;
-  Data_Get_Struct(self, xmlDoc, doc);
-
-  return NOKOGIRI_STR_NEW2(htmlGetMetaEncoding(doc), doc->encoding);
+  return INT2NUM((long)doc->type);
 }
 
 VALUE cNokogiriHtmlDocument ;
@@ -176,6 +142,4 @@ void init_html_document()
   rb_define_singleton_method(klass, "new", new, -1);
 
   rb_define_method(klass, "type", type, 0);
-  rb_define_method(klass, "meta_encoding", meta_encoding, 0);
-  rb_define_method(klass, "meta_encoding=", set_meta_encoding, 1);
 }

data/ext/nokogiri/html_element_description.c

@@ -17,7 +17,7 @@ static VALUE required_attributes(VALUE self)
 
   int i = 0;
   while(description->attrs_req[i]) {
-    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i], "UTF-8"));
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
     i++;
   }
 
@@ -41,7 +41,7 @@ static VALUE deprecated_attributes(VALUE self)
 
   int i = 0;
   while(description->attrs_depr[i]) {
-    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i], "UTF-8"));
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
     i++;
   }
 
@@ -65,7 +65,7 @@ static VALUE optional_attributes(VALUE self)
 
   int i = 0;
   while(description->attrs_opt[i]) {
-    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i], "UTF-8"));
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
     i++;
   }
 
@@ -83,7 +83,7 @@ static VALUE default_sub_element(VALUE self)
   htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  return NOKOGIRI_STR_NEW2(description->defaultsubelt, "UTF-8");
+  return NOKOGIRI_STR_NEW2(description->defaultsubelt);
 }
 
 /*
@@ -103,7 +103,7 @@ static VALUE sub_elements(VALUE self)
 
   int i = 0;
   while(description->subelts[i]) {
-    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i], "UTF-8"));
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
     i++;
   }
 
@@ -121,7 +121,7 @@ static VALUE description(VALUE self)
   htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  return NOKOGIRI_STR_NEW2(description->desc, "UTF-8");
+  return NOKOGIRI_STR_NEW2(description->desc);
 }
 
 /*
@@ -226,7 +226,7 @@ static VALUE name(VALUE self)
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(NULL == description->name) return Qnil;
-  return NOKOGIRI_STR_NEW2(description->name, "UTF-8");
+  return NOKOGIRI_STR_NEW2(description->name);
 }
 
 /*

data/ext/nokogiri/html_entity_lookup.c

@@ -14,10 +14,12 @@ static VALUE get(VALUE self, VALUE key)
   if(NULL == desc) return Qnil;
   VALUE klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
 
-  return rb_funcall(klass, rb_intern("new"), 3,
-      INT2NUM((int)desc->value),
-      NOKOGIRI_STR_NEW2(desc->name, "UTF-8"),
-      NOKOGIRI_STR_NEW2(desc->desc, "UTF-8"));
+  VALUE args[3];
+  args[0] = INT2NUM((long)desc->value);
+  args[1] = NOKOGIRI_STR_NEW2(desc->name);
+  args[2] = NOKOGIRI_STR_NEW2(desc->desc);
+
+  return rb_class_new_instance(3, args, klass);
 }
 
 void init_html_entity_lookup()

data/ext/nokogiri/html_sax_parser_context.c

@@ -0,0 +1,92 @@
+#include <html_sax_parser_context.h>
+
+VALUE cNokogiriHtmlSaxParserContext ;
+
+static void deallocate(xmlParserCtxtPtr ctxt)
+{
+  NOKOGIRI_DEBUG_START(handler);
+
+  ctxt->sax = NULL;
+
+  htmlFreeParserCtxt(ctxt);
+
+  NOKOGIRI_DEBUG_END(handler);
+}
+
+static VALUE parse_memory(VALUE klass, VALUE data, VALUE encoding)
+{
+  if(NIL_P(data)) rb_raise(rb_eArgError, "data cannot be nil");
+  if(!(int)RSTRING_LEN(data))
+    rb_raise(rb_eRuntimeError, "data cannot be empty");
+
+  htmlParserCtxtPtr ctxt = htmlCreateMemoryParserCtxt(
+      StringValuePtr(data),
+      (int)RSTRING_LEN(data)
+  );
+
+  if(RTEST(encoding)) {
+    xmlCharEncoding enc = xmlParseCharEncoding(StringValuePtr(encoding));
+    if(enc != XML_CHAR_ENCODING_ERROR) {
+      xmlSwitchEncoding(ctxt, enc);
+      if(ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
+        rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
+            StringValuePtr(encoding));
+      }
+    }
+  }
+
+  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+}
+
+static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
+{
+  htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
+      StringValuePtr(filename),
+      StringValuePtr(encoding)
+  );
+  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+}
+
+static VALUE parse_with(VALUE self, VALUE sax_handler)
+{
+  if(!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
+    rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
+
+  htmlParserCtxtPtr ctxt;
+  Data_Get_Struct(self, htmlParserCtxt, ctxt);
+
+  htmlSAXHandlerPtr sax;
+  Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
+
+  // Free the sax handler since we'll assign our own
+  if(ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
+    xmlFree(ctxt->sax);
+
+  ctxt->sax = sax;
+  ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
+
+  htmlParseDocument(ctxt);
+
+  if(NULL != ctxt->myDoc) xmlFreeDoc(ctxt->myDoc);
+
+  NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
+  return self;
+}
+
+void init_html_sax_parser_context()
+{
+  VALUE nokogiri  = rb_define_module("Nokogiri");
+  VALUE xml       = rb_define_module_under(nokogiri, "XML");
+  VALUE html      = rb_define_module_under(nokogiri, "HTML");
+  VALUE sax       = rb_define_module_under(xml, "SAX");
+  VALUE hsax      = rb_define_module_under(html, "SAX");
+  VALUE pc        = rb_define_class_under(sax, "ParserContext", rb_cObject);
+  VALUE klass     = rb_define_class_under(hsax, "ParserContext", pc);
+
+  cNokogiriHtmlSaxParserContext = klass;
+
+  rb_define_singleton_method(klass, "memory", parse_memory, 2);
+  rb_define_singleton_method(klass, "file", parse_file, 2);
+
+  rb_define_method(klass, "parse_with", parse_with, 1);
+}

data/ext/nokogiri/html_sax_parser_context.h

@@ -0,0 +1,11 @@
+#ifndef NOKOGIRI_HTML_SAX_PARSER_CONTEXT
+#define NOKOGIRI_HTML_SAX_PARSER_CONTEXT
+
+#include <nokogiri.h>
+
+extern VALUE cNokogiriHtmlSaxParserContext;
+
+void init_html_sax_parser_context();
+
+#endif
+

data/ext/nokogiri/nokogiri.c

@@ -46,11 +46,11 @@ void Init_nokogiri()
 
   rb_const_set( mNokogiri,
                 rb_intern("LIBXML_VERSION"),
-                NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION, "UTF-8")
+                NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION)
               );
   rb_const_set( mNokogiri,
                 rb_intern("LIBXML_PARSER_VERSION"),
-                NOKOGIRI_STR_NEW2(xmlParserVersion, "UTF-8")
+                NOKOGIRI_STR_NEW2(xmlParserVersion)
               );
 
   xmlInitParser();
@@ -68,16 +68,22 @@ void Init_nokogiri()
   init_xml_node_set();
   init_xml_xpath_context();
   init_xml_xpath();
+  init_xml_sax_parser_context();
   init_xml_sax_parser();
   init_xml_sax_push_parser();
   init_xml_reader();
   init_xml_dtd();
+  init_xml_element_content();
+  init_xml_attribute_decl();
+  init_xml_element_decl();
+  init_xml_entity_decl();
   init_xml_namespace();
-  init_html_sax_parser();
+  init_html_sax_parser_context();
   init_xslt_stylesheet();
   init_xml_syntax_error();
   init_html_entity_lookup();
   init_html_element_description();
   init_xml_schema();
   init_xml_relax_ng();
+  init_nokogiri_io();
 }

data/ext/nokogiri/nokogiri.h

@@ -3,8 +3,8 @@
 
 #include <stdlib.h>
 #include <assert.h>
-#include <ruby.h>
 #include <libxml/parser.h>
+#include <libxml/parserInternals.h>
 #include <libxml/xpath.h>
 #include <libxml/xpathInternals.h>
 #include <libxml/xmlreader.h>
@@ -13,6 +13,7 @@
 #include <libxml/HTMLparser.h>
 #include <libxml/HTMLtree.h>
 #include <libxml/relaxng.h>
+#include <ruby.h>
 
 #ifdef USE_INCLUDED_VASPRINTF
 int vasprintf (char **strp, const char *fmt, va_list ap);
@@ -42,38 +43,28 @@ int is_2_6_16(void) ;
 
 #include <ruby/encoding.h>
 
-#define NOKOGIRI_STR_NEW2(str, encoding) \
+#define NOKOGIRI_STR_NEW2(str) \
   ({ \
     VALUE _string = rb_str_new2((const char *)str); \
-    if(NULL != encoding) { \
-      int _enc = rb_enc_find_index("UTF-8"); \
-      if(_enc == -1) \
-        rb_enc_associate_index(_string, rb_enc_find_index("ASCII")); \
-      else \
-        rb_enc_associate_index(_string, _enc); \
-    } \
+    int _enc = rb_enc_find_index("UTF-8"); \
+    rb_enc_associate_index(_string, _enc); \
     _string; \
   })
 
-#define NOKOGIRI_STR_NEW(str, len, encoding) \
+#define NOKOGIRI_STR_NEW(str, len) \
   ({ \
     VALUE _string = rb_str_new((const char *)str, (long)len); \
-    if(NULL != encoding) { \
-      int _enc = rb_enc_find_index("UTF-8"); \
-      if(_enc == -1) \
-        rb_enc_associate_index(_string, rb_enc_find_index("ASCII")); \
-      else \
-        rb_enc_associate_index(_string, _enc); \
-    } \
+    int _enc = rb_enc_find_index("UTF-8"); \
+    rb_enc_associate_index(_string, _enc); \
     _string; \
   })
 
 #else
 
-#define NOKOGIRI_STR_NEW2(str, doc) \
+#define NOKOGIRI_STR_NEW2(str) \
   rb_str_new2((const char *)str)
 
-#define NOKOGIRI_STR_NEW(str, len, doc) \
+#define NOKOGIRI_STR_NEW(str, len) \
   rb_str_new((const char *)str, (long)len)
 #endif
 
@@ -92,11 +83,16 @@ int is_2_6_16(void) ;
 #include <xml_node_set.h>
 #include <xml_xpath.h>
 #include <xml_dtd.h>
+#include <xml_attribute_decl.h>
+#include <xml_element_decl.h>
+#include <xml_entity_decl.h>
 #include <xml_xpath_context.h>
+#include <xml_element_content.h>
+#include <xml_sax_parser_context.h>
 #include <xml_sax_parser.h>
 #include <xml_sax_push_parser.h>
 #include <xml_reader.h>
-#include <html_sax_parser.h>
+#include <html_sax_parser_context.h>
 #include <xslt_stylesheet.h>
 #include <xml_syntax_error.h>
 #include <xml_schema.h>