nokogiri 1.0.0 → 1.6.8.1

data/ext/nokogiri/html_document.c

@@ -1,22 +1,92 @@
 #include <html_document.h>
 
+static ID id_encoding_found;
+
 /*
  * call-seq:
- *  serialize
+ *  new
  *
- * Serialize this document
+ * Create a new document
  */
-static VALUE serialize(VALUE self)
+static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
-  xmlDocPtr doc;
-  xmlChar *buf;
-  int size;
-  Data_Get_Struct(self, xmlDoc, doc);
+  VALUE uri, external_id, rest, rb_doc;
+  htmlDocPtr doc;
+
+  rb_scan_args(argc, argv, "0*", &rest);
+  uri         = rb_ary_entry(rest, (long)0);
+  external_id = rb_ary_entry(rest, (long)1);
+
+  doc = htmlNewDoc(
+      RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
+      RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
+  );
+  rb_doc = Nokogiri_wrap_xml_document(klass, doc);
+  rb_obj_call_init(rb_doc, argc, argv);
+  return rb_doc ;
+}
+
+/*
+ * call-seq:
+ *  read_io(io, url, encoding, options)
+ *
+ * Read the HTML document from +io+ with given +url+, +encoding+,
+ * and +options+.  See Nokogiri::HTML.parse
+ */
+static VALUE read_io( VALUE klass,
+                      VALUE io,
+                      VALUE url,
+                      VALUE encoding,
+                      VALUE options )
+{
+  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
+  VALUE error_list      = rb_ary_new();
+  VALUE document;
+  htmlDocPtr doc;
+
+  xmlResetLastError();
+  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
+
+  doc = htmlReadIO(
+      io_read_callback,
+      io_close_callback,
+      (void *)io,
+      c_url,
+      c_enc,
+      (int)NUM2INT(options)
+  );
+  xmlSetStructuredErrorFunc(NULL, NULL);
 
-  htmlDocDumpMemory(doc, &buf, &size);
-  VALUE rb_str = rb_str_new((char *)buf, (long)size);
-  free(buf);
-  return rb_str;
+  /*
+   * If EncodingFound has occurred in EncodingReader, make sure to do
+   * a cleanup and propagate the error.
+   */
+  if (rb_respond_to(io, id_encoding_found)) {
+    VALUE encoding_found = rb_funcall(io, id_encoding_found, 0);
+    if (!NIL_P(encoding_found)) {
+      xmlFreeDoc(doc);
+      rb_exc_raise(encoding_found);
+    }
+  }
+
+  if(doc == NULL) {
+    xmlErrorPtr error;
+
+    xmlFreeDoc(doc);
+
+    error = xmlGetLastError();
+    if(error)
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
+    else
+      rb_raise(rb_eRuntimeError, "Could not parse document");
+
+    return Qnil;
+  }
+
+  document = Nokogiri_wrap_xml_document(klass, doc);
+  rb_iv_set(document, "@errors", error_list);
+  return document;
 }
 
 /*
@@ -33,16 +103,36 @@ 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);
-  int len               = NUM2INT(rb_funcall(string, rb_intern("length"), 0));
+  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
+  int len               = (int)RSTRING_LEN(string);
+  VALUE error_list      = rb_ary_new();
+  VALUE document;
+  htmlDocPtr doc;
 
-  htmlDocPtr doc = htmlReadMemory(c_buffer, len, c_url, c_enc, NUM2INT(options));
+  xmlResetLastError();
+  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 
-  if(doc == NULL)
-    doc = htmlNewDoc((const xmlChar *)c_url, NULL);
+  doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
+  xmlSetStructuredErrorFunc(NULL, NULL);
 
-  return Nokogiri_wrap_xml_document(klass, doc);
+  if(doc == NULL) {
+    xmlErrorPtr error;
+
+    xmlFreeDoc(doc);
+
+    error = xmlGetLastError();
+    if(error)
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
+    else
+      rb_raise(rb_eRuntimeError, "Could not parse document");
+
+    return Qnil;
+  }
+
+  document = Nokogiri_wrap_xml_document(klass, doc);
+  rb_iv_set(document, "@errors", error_list);
+  return document;
 }
 
 /*
@@ -55,29 +145,26 @@ static VALUE type(VALUE self)
 {
   htmlDocPtr doc;
   Data_Get_Struct(self, xmlDoc, doc);
-  return INT2NUM((int)doc->type);
+  return INT2NUM((long)doc->type);
 }
 
 VALUE cNokogiriHtmlDocument ;
 void init_html_document()
 {
-  /*
-   * HACK.  This is so that rdoc will work with this C file.
-   */
-  /*
-  VALUE nokogiri = rb_define_module("Nokogiri");
-  VALUE html = rb_define_module_under(nokogiri, "HTML");
-  VALUE xml = rb_define_module_under(nokogiri, "XML");
-  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
-  VALUE xml_doc = rb_define_class_under(xml, "Document", node);
-  VALUE klass = rb_define_class_under(html, "Document", xml_doc);
-  */
+  VALUE nokogiri  = rb_define_module("Nokogiri");
+  VALUE html      = rb_define_module_under(nokogiri, "HTML");
+  VALUE xml       = rb_define_module_under(nokogiri, "XML");
+  VALUE node      = rb_define_class_under(xml, "Node", rb_cObject);
+  VALUE xml_doc   = rb_define_class_under(xml, "Document", node);
+  VALUE klass     = rb_define_class_under(html, "Document", xml_doc);
 
-  VALUE klass ;
-  klass = cNokogiriHtmlDocument = rb_const_get(mNokogiriHtml, rb_intern("Document"));
+  cNokogiriHtmlDocument = klass;
 
   rb_define_singleton_method(klass, "read_memory", read_memory, 4);
+  rb_define_singleton_method(klass, "read_io", read_io, 4);
+  rb_define_singleton_method(klass, "new", new, -1);
 
   rb_define_method(klass, "type", type, 0);
-  rb_define_method(klass, "serialize", serialize, 0);
+
+  id_encoding_found = rb_intern("encoding_found");
 }

data/ext/nokogiri/html_document.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_HTML_DOCUMENT
 #define NOKOGIRI_HTML_DOCUMENT
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_html_document();
 

data/ext/nokogiri/html_element_description.c

@@ -0,0 +1,279 @@
+#include <html_element_description.h>
+
+/*
+ * call-seq:
+ *  required_attributes
+ *
+ * A list of required attributes for this element
+ */
+static VALUE required_attributes(VALUE self)
+{
+  htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->attrs_req) return list;
+
+  for(i = 0; description->attrs_depr[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  deprecated_attributes
+ *
+ * A list of deprecated attributes for this element
+ */
+static VALUE deprecated_attributes(VALUE self)
+{
+  htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->attrs_depr) return list;
+
+  for(i = 0; description->attrs_depr[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  optional_attributes
+ *
+ * A list of optional attributes for this element
+ */
+static VALUE optional_attributes(VALUE self)
+{
+  htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->attrs_opt) return list;
+
+  for(i = 0; description->attrs_opt[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  default_sub_element
+ *
+ * The default sub element for this element
+ */
+static VALUE default_sub_element(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if (description->defaultsubelt)
+    return NOKOGIRI_STR_NEW2(description->defaultsubelt);
+
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *  sub_elements
+ *
+ * A list of allowed sub elements for this element.
+ */
+static VALUE sub_elements(VALUE self)
+{
+  htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->subelts) return list;
+
+  for(i = 0; description->subelts[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  description
+ *
+ * The description for this element
+ */
+static VALUE description(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  return NOKOGIRI_STR_NEW2(description->desc);
+}
+
+/*
+ * call-seq:
+ *  inline?
+ *
+ * Is this element an inline element?
+ */
+static VALUE inline_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->isinline) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  deprecated?
+ *
+ * Is this element deprecated?
+ */
+static VALUE deprecated_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->depr) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  empty?
+ *
+ * Is this an empty element?
+ */
+static VALUE empty_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->empty) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  save_end_tag?
+ *
+ * Should the end tag be saved?
+ */
+static VALUE save_end_tag_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->saveEndTag) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  implied_end_tag?
+ *
+ * Can the end tag be implied for this tag?
+ */
+static VALUE implied_end_tag_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->endTag) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  implied_start_tag?
+ *
+ * Can the start tag be implied for this tag?
+ */
+static VALUE implied_start_tag_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->startTag) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  name
+ *
+ * Get the tag name for this ElemementDescription
+ */
+static VALUE name(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(NULL == description->name) return Qnil;
+  return NOKOGIRI_STR_NEW2(description->name);
+}
+
+/*
+ * call-seq:
+ *  [](tag_name)
+ *
+ * Get ElemementDescription for +tag_name+
+ */
+static VALUE get_description(VALUE klass, VALUE tag_name)
+{
+  const htmlElemDesc * description = htmlTagLookup(
+      (const xmlChar *)StringValueCStr(tag_name)
+  );
+
+  if(NULL == description) return Qnil;
+  return Data_Wrap_Struct(klass, 0, 0, (void *)description);
+}
+
+VALUE cNokogiriHtmlElementDescription ;
+void init_html_element_description()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE html     = rb_define_module_under(nokogiri, "HTML");
+  VALUE klass    = rb_define_class_under(html, "ElementDescription",rb_cObject);
+
+  cNokogiriHtmlElementDescription = klass;
+
+  rb_define_singleton_method(klass, "[]", get_description, 1);
+
+  rb_define_method(klass, "name", name, 0);
+  rb_define_method(klass, "implied_start_tag?", implied_start_tag_eh, 0);
+  rb_define_method(klass, "implied_end_tag?", implied_end_tag_eh, 0);
+  rb_define_method(klass, "save_end_tag?", save_end_tag_eh, 0);
+  rb_define_method(klass, "empty?", empty_eh, 0);
+  rb_define_method(klass, "deprecated?", deprecated_eh, 0);
+  rb_define_method(klass, "inline?", inline_eh, 0);
+  rb_define_method(klass, "description", description, 0);
+  rb_define_method(klass, "sub_elements", sub_elements, 0);
+  rb_define_method(klass, "default_sub_element", default_sub_element, 0);
+  rb_define_method(klass, "optional_attributes", optional_attributes, 0);
+  rb_define_method(klass, "deprecated_attributes", deprecated_attributes, 0);
+  rb_define_method(klass, "required_attributes", required_attributes, 0);
+}

data/ext/nokogiri/html_element_description.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_HTML_ELEMENT_DESCRIPTION
+#define NOKOGIRI_HTML_ELEMENT_DESCRIPTION
+
+#include <nokogiri.h>
+
+void init_html_element_description();
+
+extern VALUE cNokogiriHtmlElementDescription ;
+
+#endif

data/ext/nokogiri/html_entity_lookup.c

@@ -0,0 +1,32 @@
+#include <html_entity_lookup.h>
+
+/*
+ * call-seq:
+ *  get(key)
+ *
+ * Get the HTML::EntityDescription for +key+
+ */
+static VALUE get(VALUE self, VALUE key)
+{
+  const htmlEntityDesc * desc =
+    htmlEntityLookup((const xmlChar *)StringValueCStr(key));
+  VALUE klass, args[3];
+
+  if(NULL == desc) return Qnil;
+  klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
+
+  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()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE html = rb_define_module_under(nokogiri, "HTML");
+  VALUE klass = rb_define_class_under(html, "EntityLookup", rb_cObject);
+
+  rb_define_method(klass, "get", get, 1);
+}

data/ext/nokogiri/html_entity_lookup.h

@@ -0,0 +1,8 @@
+#ifndef NOKOGIRI_HTML_ENTITY_LOOKUP
+#define NOKOGIRI_HTML_ENTITY_LOOKUP
+
+#include <nokogiri.h>
+
+void init_html_entity_lookup();
+
+#endif

data/ext/nokogiri/html_sax_parser_context.c

@@ -0,0 +1,116 @@
+#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)
+{
+    htmlParserCtxtPtr ctxt;
+
+    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");
+
+    ctxt = htmlCreateMemoryParserCtxt(StringValuePtr(data),
+				      (int)RSTRING_LEN(data));
+    if (ctxt->sax) {
+	xmlFree(ctxt->sax);
+	ctxt->sax = NULL;
+    }
+
+    if (RTEST(encoding)) {
+	xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValueCStr(encoding));
+	if (enc != NULL) {
+	    xmlSwitchToEncoding(ctxt, enc);
+	    if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
+		rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
+			 StringValueCStr(encoding));
+	    }
+	}
+    }
+
+    return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+}
+
+static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
+{
+  htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
+      StringValueCStr(filename),
+      StringValueCStr(encoding)
+  );
+  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+}
+
+static VALUE
+parse_doc(VALUE ctxt_val)
+{
+    htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
+    htmlParseDocument(ctxt);
+    return Qnil;
+}
+
+static VALUE
+parse_doc_finalize(VALUE ctxt_val)
+{
+    htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
+
+    if (ctxt->myDoc)
+	xmlFreeDoc(ctxt->myDoc);
+
+    NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
+    return Qnil;
+}
+
+static VALUE
+parse_with(VALUE self, VALUE sax_handler)
+{
+    htmlParserCtxtPtr ctxt;
+    htmlSAXHandlerPtr sax;
+
+    if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
+	rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
+
+    Data_Get_Struct(self, htmlParserCtxt, ctxt);
+    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);
+
+    rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
+
+    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
+