nokogiri-maven 1.5.0-java

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 *)StringValuePtr(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 +1,94 @@
+#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(RTEST(encoding)) {
+    xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValuePtr(encoding));
+    if(enc != NULL) {
+      xmlSwitchToEncoding(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)
+{
+  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);
+
+  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

@@ -0,0 +1,115 @@
+#include <nokogiri.h>
+
+VALUE mNokogiri ;
+VALUE mNokogiriXml ;
+VALUE mNokogiriHtml ;
+VALUE mNokogiriXslt ;
+VALUE mNokogiriXmlSax ;
+VALUE mNokogiriHtmlSax ;
+
+#ifdef USE_INCLUDED_VASPRINTF
+/*
+ * I srsly hate windows.  it doesn't have vasprintf.
+ * Thank you Geoffroy Couprie for this implementation of vasprintf!
+ */
+int vasprintf (char **strp, const char *fmt, va_list ap)
+{
+  int len = vsnprintf (NULL, 0, fmt, ap) + 1;
+  char *res = (char *)malloc((unsigned int)len);
+  if (res == NULL)
+      return -1;
+  *strp = res;
+  return vsnprintf(res, (unsigned int)len, fmt, ap);
+}
+#endif
+
+#ifdef USING_SYSTEM_ALLOCATOR_LIBRARY /* Ruby Enterprise Edition with tcmalloc */
+void vasprintf_free (void *p)
+{
+  system_free(p);
+}
+#else
+void vasprintf_free (void *p)
+{
+  free(p);
+}
+#endif
+
+#ifndef __MACRUBY__
+/* Allocate strdupped strings with the same memory allocator Ruby uses. */
+static char *ruby_strdup(const char *s)
+{
+  size_t len = strlen(s);
+  char *result = ruby_xmalloc((ssize_t) (len + 1));
+  memcpy(result, s, len + 1);
+  return result;
+}
+#endif
+
+void Init_nokogiri()
+{
+#ifndef __MACRUBY__
+  xmlMemSetup(
+      (xmlFreeFunc)ruby_xfree,
+      (xmlMallocFunc)ruby_xmalloc,
+      (xmlReallocFunc)ruby_xrealloc,
+      ruby_strdup
+  );
+#endif
+
+  mNokogiri         = rb_define_module("Nokogiri");
+  mNokogiriXml      = rb_define_module_under(mNokogiri, "XML");
+  mNokogiriHtml     = rb_define_module_under(mNokogiri, "HTML");
+  mNokogiriXslt     = rb_define_module_under(mNokogiri, "XSLT");
+  mNokogiriXmlSax   = rb_define_module_under(mNokogiriXml, "SAX");
+  mNokogiriHtmlSax  = rb_define_module_under(mNokogiriHtml, "SAX");
+
+  rb_const_set( mNokogiri,
+                rb_intern("LIBXML_VERSION"),
+                NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION)
+              );
+  rb_const_set( mNokogiri,
+                rb_intern("LIBXML_PARSER_VERSION"),
+                NOKOGIRI_STR_NEW2(xmlParserVersion)
+              );
+
+#ifdef LIBXML_ICONV_ENABLED
+  rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qtrue);
+#else
+  rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qfalse);
+#endif
+
+  xmlInitParser();
+
+  init_xml_document();
+  init_html_document();
+  init_xml_node();
+  init_xml_document_fragment();
+  init_xml_text();
+  init_xml_cdata();
+  init_xml_processing_instruction();
+  init_xml_attr();
+  init_xml_entity_reference();
+  init_xml_comment();
+  init_xml_node_set();
+  init_xml_xpath_context();
+  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_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();
+  init_xml_encoding_handler();
+}

data/ext/nokogiri/nokogiri.h

@@ -0,0 +1,160 @@
+#ifndef NOKOGIRI_NATIVE
+#define NOKOGIRI_NATIVE
+
+#include <stdlib.h>
+#include <string.h>
+#include <assert.h>
+#include <stdarg.h>
+
+#ifdef USE_INCLUDED_VASPRINTF
+int vasprintf (char **strp, const char *fmt, va_list ap);
+#else
+
+#define _GNU_SOURCE
+#  include <stdio.h>
+#undef _GNU_SOURCE
+
+#endif
+
+#include <libxml/parser.h>
+#include <libxml/entities.h>
+#include <libxml/parserInternals.h>
+#include <libxml/xpath.h>
+#include <libxml/xpathInternals.h>
+#include <libxml/xmlreader.h>
+#include <libxml/xmlsave.h>
+#include <libxml/xmlschemas.h>
+#include <libxml/HTMLparser.h>
+#include <libxml/HTMLtree.h>
+#include <libxml/relaxng.h>
+#include <libxslt/extensions.h>
+#include <ruby.h>
+
+#ifdef HAVE_RUBY_ENCODING_H
+#include <ruby/st.h>
+#else
+#include <st.h>
+#endif
+
+#ifndef UNUSED
+# if defined(__GNUC__)
+#  define MAYBE_UNUSED(name) name __attribute__((unused))
+#  define UNUSED(name) MAYBE_UNUSED(UNUSED_ ## name)
+# else
+#  define MAYBE_UNUSED(name) name
+#  define UNUSED(name) name
+# endif
+#endif
+
+#ifndef NORETURN
+# if defined(__GNUC__)
+#  define NORETURN(name) __attribute__((noreturn)) name
+# else
+#  define NORETURN(name) name
+# endif
+#endif
+
+#ifdef HAVE_RUBY_ENCODING_H
+
+#include <ruby/encoding.h>
+
+#define NOKOGIRI_STR_NEW2(str) \
+  NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
+
+#define NOKOGIRI_STR_NEW(str, len) \
+  rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
+
+#else
+
+#define NOKOGIRI_STR_NEW2(str) \
+  rb_str_new2((const char *)(str))
+
+#define NOKOGIRI_STR_NEW(str, len) \
+  rb_str_new((const char *)(str), (long)(len))
+#endif
+
+#define RBSTR_OR_QNIL(_str) \
+  (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
+
+#include <xml_libxml2_hacks.h>
+
+#include <xml_io.h>
+#include <xml_document.h>
+#include <html_entity_lookup.h>
+#include <html_document.h>
+#include <xml_node.h>
+#include <xml_text.h>
+#include <xml_cdata.h>
+#include <xml_attr.h>
+#include <xml_processing_instruction.h>
+#include <xml_entity_reference.h>
+#include <xml_document_fragment.h>
+#include <xml_comment.h>
+#include <xml_node_set.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_context.h>
+#include <xslt_stylesheet.h>
+#include <xml_syntax_error.h>
+#include <xml_schema.h>
+#include <xml_relax_ng.h>
+#include <html_element_description.h>
+#include <xml_namespace.h>
+#include <xml_encoding_handler.h>
+
+extern VALUE mNokogiri ;
+extern VALUE mNokogiriXml ;
+extern VALUE mNokogiriXmlSax ;
+extern VALUE mNokogiriHtml ;
+extern VALUE mNokogiriHtmlSax ;
+extern VALUE mNokogiriXslt ;
+
+#define NOKOGIRI_ROOT_NODE(_node) \
+  st_insert(((nokogiriTuplePtr)(_node)->doc->_private)->unlinkedNodes, (st_data_t)(_node), (st_data_t)(_node))
+
+#define NOKOGIRI_ROOT_NSDEF(_nsDef, _doc)     \
+  st_insert(((nokogiriTuplePtr)(_doc)->_private)->unlinkedNodes, (st_data_t)(_nsDef), (st_data_t)(_nsDef))
+
+#ifdef DEBUG
+
+#define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
+#define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
+
+#else
+
+#define NOKOGIRI_DEBUG_START(p)
+#define NOKOGIRI_DEBUG_END(p)
+
+#endif
+
+#ifndef RSTRING_PTR
+#define RSTRING_PTR(s) (RSTRING(s)->ptr)
+#endif
+
+#ifndef RSTRING_LEN
+#define RSTRING_LEN(s) (RSTRING(s)->len)
+#endif
+
+#ifndef RARRAY_PTR
+#define RARRAY_PTR(a) RARRAY(a)->ptr
+#endif
+
+#ifndef RARRAY_LEN
+#define RARRAY_LEN(a) RARRAY(a)->len
+#endif
+
+#ifndef __builtin_expect
+# if defined(__GNUC__)
+#  define __builtin_expect(expr, c) __builtin_expect((long)(expr), (long)(c))
+# endif
+#endif
+
+#endif

data/ext/nokogiri/xml_attr.c

@@ -0,0 +1,94 @@
+#include <xml_attr.h>
+
+/*
+ * call-seq:
+ *  value=(content)
+ *
+ * Set the value for this Attr to +content+
+ */
+static VALUE set_value(VALUE self, VALUE content)
+{
+  xmlAttrPtr attr;
+  Data_Get_Struct(self, xmlAttr, attr);
+
+  if(attr->children) xmlFreeNodeList(attr->children);
+
+  attr->children = attr->last = NULL;
+
+  if(content) {
+    xmlChar *buffer;
+    xmlNode *tmp;
+
+    /* Encode our content */
+    buffer = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValuePtr(content));
+
+    attr->children = xmlStringGetNodeList(attr->doc, buffer);
+    attr->last = NULL;
+    tmp = attr->children;
+
+    /* Loop through the children */
+    for(tmp = attr->children; tmp; tmp = tmp->next) {
+      tmp->parent = (xmlNode *)attr;
+      tmp->doc = attr->doc;
+      if(tmp->next == NULL) attr->last = tmp;
+    }
+
+    /* Free up memory */
+    xmlFree(buffer);
+  }
+
+  return content;
+}
+
+/*
+ * call-seq:
+ *  new(document, name)
+ *
+ * Create a new Attr element on the +document+ with +name+
+ */
+static VALUE new(int argc, VALUE *argv, VALUE klass)
+{
+  xmlDocPtr xml_doc;
+  VALUE document;
+  VALUE name;
+  VALUE rest;
+  xmlAttrPtr node;
+  VALUE rb_node;
+
+  rb_scan_args(argc, argv, "2*", &document, &name, &rest);
+
+  Data_Get_Struct(document, xmlDoc, xml_doc);
+
+  node = xmlNewDocProp(
+      xml_doc,
+      (const xmlChar *)StringValuePtr(name),
+      NULL
+  );
+
+  NOKOGIRI_ROOT_NODE((xmlNodePtr)node);
+
+  rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
+  rb_obj_call_init(rb_node, argc, argv);
+
+  if(rb_block_given_p()) rb_yield(rb_node);
+
+  return rb_node;
+}
+
+VALUE cNokogiriXmlAttr;
+void init_xml_attr()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE xml = rb_define_module_under(nokogiri, "XML");
+  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
+
+  /*
+   * Attr represents a Attr node in an xml document.
+   */
+  VALUE klass = rb_define_class_under(xml, "Attr", node);
+
+  cNokogiriXmlAttr = klass;
+
+  rb_define_singleton_method(klass, "new", new, -1);
+  rb_define_method(klass, "value=", set_value, 1);
+}