nokogiri 1.18.0.rc1-x86_64-linux-gnu

data/ext/nokogiri/html4_document.c

@@ -0,0 +1,171 @@
+#include <nokogiri.h>
+
+VALUE cNokogiriHtml4Document ;
+
+static ID id_encoding_found;
+static ID id_to_s;
+
+/*
+ * call-seq:
+ *  new(uri=nil, external_id=nil) → HTML4::Document
+ *
+ * Create a new empty document with base URI +uri+ and external ID +external_id+.
+ */
+static VALUE
+rb_html_document_s_new(int argc, VALUE *argv, VALUE klass)
+{
+  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 = noko_xml_document_wrap_with_init_args(klass, 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::HTML4.parse
+ */
+static VALUE
+rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
+{
+  VALUE rb_doc;
+  VALUE rb_error_list = rb_ary_new();
+  htmlDocPtr c_doc;
+  const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
+  const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
+  int options = NUM2INT(rb_options);
+
+  xmlSetStructuredErrorFunc((void *)rb_error_list, noko__error_array_pusher);
+
+  c_doc = htmlReadIO(noko_io_read, noko_io_close, (void *)rb_io, c_url, c_encoding, options);
+
+  xmlSetStructuredErrorFunc(NULL, NULL);
+
+  /*
+   * If EncodingFound has occurred in EncodingReader, make sure to do
+   * a cleanup and propagate the error.
+   */
+  if (rb_respond_to(rb_io, id_encoding_found)) {
+    VALUE encoding_found = rb_funcall(rb_io, id_encoding_found, 0);
+    if (!NIL_P(encoding_found)) {
+      xmlFreeDoc(c_doc);
+      rb_exc_raise(encoding_found);
+    }
+  }
+
+  if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
+    VALUE rb_error ;
+
+    xmlFreeDoc(c_doc);
+
+    rb_error = rb_ary_entry(rb_error_list, 0);
+    if (rb_error == Qnil) {
+      rb_raise(rb_eRuntimeError, "Could not parse document");
+    } else {
+      VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
+      exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
+                                        exception_message);
+      rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
+    }
+
+    return Qnil;
+  }
+
+  rb_doc = noko_xml_document_wrap(klass, c_doc);
+  rb_iv_set(rb_doc, "@errors", rb_error_list);
+  return rb_doc;
+}
+
+/*
+ * call-seq:
+ *  read_memory(string, url, encoding, options)
+ *
+ * Read the HTML document contained in +string+ with given +url+, +encoding+,
+ * and +options+.  See Nokogiri::HTML4.parse
+ */
+static VALUE
+rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
+{
+  VALUE rb_doc;
+  VALUE rb_error_list = rb_ary_new();
+  htmlDocPtr c_doc;
+  const char *c_buffer = StringValuePtr(rb_html);
+  const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
+  const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
+  int html_len = (int)RSTRING_LEN(rb_html);
+  int options = NUM2INT(rb_options);
+
+  xmlSetStructuredErrorFunc((void *)rb_error_list, noko__error_array_pusher);
+
+  c_doc = htmlReadMemory(c_buffer, html_len, c_url, c_encoding, options);
+
+  xmlSetStructuredErrorFunc(NULL, NULL);
+
+  if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
+    VALUE rb_error ;
+
+    xmlFreeDoc(c_doc);
+
+    rb_error = rb_ary_entry(rb_error_list, 0);
+    if (rb_error == Qnil) {
+      rb_raise(rb_eRuntimeError, "Could not parse document");
+    } else {
+      VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
+      exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
+                                        exception_message);
+      rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
+    }
+
+    return Qnil;
+  }
+
+  rb_doc = noko_xml_document_wrap(klass, c_doc);
+  rb_iv_set(rb_doc, "@errors", rb_error_list);
+  return rb_doc;
+}
+
+/*
+ * call-seq:
+ *  type
+ *
+ * The type for this document
+ */
+static VALUE
+rb_html_document_type(VALUE self)
+{
+  htmlDocPtr doc = noko_xml_document_unwrap(self);
+  return INT2NUM(doc->type);
+}
+
+void
+noko_init_html_document(void)
+{
+  /* this is here so that rdoc doesn't ignore this file. */
+  /*
+    mNokogiri         = rb_define_module("Nokogiri");
+    mNokogiriHtml4    = rb_define_module_under(mNokogiri, "HTML4");
+  */
+
+  assert(cNokogiriXmlDocument);
+  cNokogiriHtml4Document = rb_define_class_under(mNokogiriHtml4, "Document", cNokogiriXmlDocument);
+
+  rb_define_singleton_method(cNokogiriHtml4Document, "read_memory", rb_html_document_s_read_memory, 4);
+  rb_define_singleton_method(cNokogiriHtml4Document, "read_io", rb_html_document_s_read_io, 4);
+  rb_define_singleton_method(cNokogiriHtml4Document, "new", rb_html_document_s_new, -1);
+
+  rb_define_method(cNokogiriHtml4Document, "type", rb_html_document_type, 0);
+
+  id_encoding_found = rb_intern("encoding_found");
+  id_to_s = rb_intern("to_s");
+}

data/ext/nokogiri/html4_element_description.c

@@ -0,0 +1,299 @@
+#include <nokogiri.h>
+
+static const rb_data_type_t html_elem_desc_type = {
+  .wrap_struct_name = "htmlElemDesc",
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
+VALUE cNokogiriHtml4ElementDescription ;
+
+/*
+ * call-seq:
+ *  required_attributes
+ *
+ * A list of required attributes for this element
+ */
+static VALUE
+required_attributes(VALUE self)
+{
+  const htmlElemDesc *description;
+  VALUE list;
+  int i;
+
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, 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)
+{
+  const htmlElemDesc *description;
+  VALUE list;
+  int i;
+
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, 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)
+{
+  const htmlElemDesc *description;
+  VALUE list;
+  int i;
+
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, 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)
+{
+  const htmlElemDesc *description;
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, 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)
+{
+  const htmlElemDesc *description;
+  VALUE list;
+  int i;
+
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, 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)
+{
+  const htmlElemDesc *description;
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, description);
+
+  return NOKOGIRI_STR_NEW2(description->desc);
+}
+
+/*
+ * call-seq:
+ *  inline?
+ *
+ * Is this element an inline element?
+ */
+static VALUE
+inline_eh(VALUE self)
+{
+  const htmlElemDesc *description;
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, description);
+
+  if (description->isinline) { return Qtrue; }
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  deprecated?
+ *
+ * Is this element deprecated?
+ */
+static VALUE
+deprecated_eh(VALUE self)
+{
+  const htmlElemDesc *description;
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, description);
+
+  if (description->depr) { return Qtrue; }
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  empty?
+ *
+ * Is this an empty element?
+ */
+static VALUE
+empty_eh(VALUE self)
+{
+  const htmlElemDesc *description;
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, 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)
+{
+  const htmlElemDesc *description;
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, 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)
+{
+  const htmlElemDesc *description;
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, 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)
+{
+  const htmlElemDesc *description;
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, description);
+
+  if (description->startTag) { return Qtrue; }
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  name
+ *
+ * Get the tag name for this ElementDescription
+ */
+static VALUE
+name(VALUE self)
+{
+  const htmlElemDesc *description;
+  TypedData_Get_Struct(self, htmlElemDesc, &html_elem_desc_type, description);
+
+  if (NULL == description->name) { return Qnil; }
+  return NOKOGIRI_STR_NEW2(description->name);
+}
+
+/*
+ * call-seq:
+ *  [](tag_name)
+ *
+ * Get ElementDescription 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 TypedData_Wrap_Struct(klass, &html_elem_desc_type, DISCARD_CONST_QUAL(void *, description));
+}
+
+void
+noko_init_html_element_description(void)
+{
+  cNokogiriHtml4ElementDescription = rb_define_class_under(mNokogiriHtml4, "ElementDescription", rb_cObject);
+
+  rb_undef_alloc_func(cNokogiriHtml4ElementDescription);
+
+  rb_define_singleton_method(cNokogiriHtml4ElementDescription, "[]", get_description, 1);
+
+  rb_define_method(cNokogiriHtml4ElementDescription, "name", name, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "implied_start_tag?", implied_start_tag_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "implied_end_tag?", implied_end_tag_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "save_end_tag?", save_end_tag_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "empty?", empty_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "deprecated?", deprecated_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "inline?", inline_eh, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "description", description, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "sub_elements", sub_elements, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "default_sub_element", default_sub_element, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "optional_attributes", optional_attributes, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "deprecated_attributes", deprecated_attributes, 0);
+  rb_define_method(cNokogiriHtml4ElementDescription, "required_attributes", required_attributes, 0);
+}

data/ext/nokogiri/html4_entity_lookup.c

@@ -0,0 +1,37 @@
+#include <nokogiri.h>
+
+static VALUE cNokogiriHtml4EntityLookup;
+
+/*
+ * call-seq:
+ *  get(key)
+ *
+ * Get the HTML4::EntityDescription for +key+
+ */
+static VALUE
+get(VALUE _, VALUE rb_entity_name)
+{
+  VALUE cNokogiriHtml4EntityDescription;
+  const htmlEntityDesc *c_entity_desc;
+  VALUE rb_constructor_args[3];
+
+  c_entity_desc = htmlEntityLookup((const xmlChar *)StringValueCStr(rb_entity_name));
+  if (NULL == c_entity_desc) {
+    return Qnil;
+  }
+
+  rb_constructor_args[0] = UINT2NUM(c_entity_desc->value);
+  rb_constructor_args[1] = NOKOGIRI_STR_NEW2(c_entity_desc->name);
+  rb_constructor_args[2] = NOKOGIRI_STR_NEW2(c_entity_desc->desc);
+
+  cNokogiriHtml4EntityDescription = rb_const_get_at(mNokogiriHtml4, rb_intern("EntityDescription"));
+  return rb_class_new_instance(3, rb_constructor_args, cNokogiriHtml4EntityDescription);
+}
+
+void
+noko_init_html_entity_lookup(void)
+{
+  cNokogiriHtml4EntityLookup = rb_define_class_under(mNokogiriHtml4, "EntityLookup", rb_cObject);
+
+  rb_define_method(cNokogiriHtml4EntityLookup, "get", get, 1);
+}

data/ext/nokogiri/html4_sax_parser.c

@@ -0,0 +1,40 @@
+#include <nokogiri.h>
+
+VALUE cNokogiriHtml4SaxParser;
+
+static ID id_start_document;
+
+static void
+noko_html4_sax_parser_start_document(void *ctx)
+{
+  xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr)ctx;
+  VALUE self = (VALUE)ctxt->_private;
+  VALUE doc = rb_iv_get(self, "@document");
+
+  xmlSAX2StartDocument(ctx);
+
+  rb_funcall(doc, id_start_document, 0);
+}
+
+static VALUE
+noko_html4_sax_parser_initialize(VALUE self)
+{
+  xmlSAXHandlerPtr handler = noko_xml_sax_parser_unwrap(self);
+
+  rb_call_super(0, NULL);
+
+  handler->startDocument = noko_html4_sax_parser_start_document;
+
+  return self;
+}
+
+void
+noko_init_html4_sax_parser(void)
+{
+  cNokogiriHtml4SaxParser = rb_define_class_under(mNokogiriHtml4Sax, "Parser", cNokogiriXmlSaxParser);
+
+  rb_define_private_method(cNokogiriHtml4SaxParser, "initialize_native",
+                           noko_html4_sax_parser_initialize, 0);
+
+  id_start_document = rb_intern("start_document");
+}

data/ext/nokogiri/html4_sax_parser_context.c

@@ -0,0 +1,98 @@
+#include <nokogiri.h>
+
+VALUE cNokogiriHtml4SaxParserContext ;
+
+/* :nodoc: */
+static VALUE
+noko_html4_sax_parser_context_s_native_memory(VALUE rb_class, VALUE rb_input, VALUE rb_encoding)
+{
+  Check_Type(rb_input, T_STRING);
+  if (!(int)RSTRING_LEN(rb_input)) {
+    rb_raise(rb_eRuntimeError, "input string cannot be empty");
+  }
+
+  if (!NIL_P(rb_encoding) && !rb_obj_is_kind_of(rb_encoding, rb_cEncoding)) {
+    rb_raise(rb_eTypeError, "argument must be an Encoding object");
+  }
+
+  htmlParserCtxtPtr c_context =
+    htmlCreateMemoryParserCtxt(StringValuePtr(rb_input), (int)RSTRING_LEN(rb_input));
+  if (!c_context) {
+    rb_raise(rb_eRuntimeError, "failed to create xml sax parser context");
+  }
+
+  noko_xml_sax_parser_context_set_encoding(c_context, rb_encoding);
+
+  if (c_context->sax) {
+    xmlFree(c_context->sax);
+    c_context->sax = NULL;
+  }
+
+  return noko_xml_sax_parser_context_wrap(rb_class, c_context);
+}
+
+/* :nodoc: */
+static VALUE
+noko_html4_sax_parser_context_s_native_file(VALUE rb_class, VALUE rb_filename, VALUE rb_encoding)
+{
+  if (!NIL_P(rb_encoding) && !rb_obj_is_kind_of(rb_encoding, rb_cEncoding)) {
+    rb_raise(rb_eTypeError, "argument must be an Encoding object");
+  }
+
+  htmlParserCtxtPtr c_context = htmlCreateFileParserCtxt(StringValueCStr(rb_filename), NULL);
+  if (!c_context) {
+    rb_raise(rb_eRuntimeError, "failed to create xml sax parser context");
+  }
+
+  noko_xml_sax_parser_context_set_encoding(c_context, rb_encoding);
+
+  if (c_context->sax) {
+    xmlFree(c_context->sax);
+    c_context->sax = NULL;
+  }
+
+  return noko_xml_sax_parser_context_wrap(rb_class, c_context);
+}
+
+static VALUE
+noko_html4_sax_parser_context__parse_with(VALUE rb_context, VALUE rb_sax_parser)
+{
+  htmlParserCtxtPtr ctxt;
+  htmlSAXHandlerPtr sax;
+
+  if (!rb_obj_is_kind_of(rb_sax_parser, cNokogiriXmlSaxParser)) {
+    rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
+  }
+
+  ctxt = noko_xml_sax_parser_context_unwrap(rb_context);
+  sax = noko_xml_sax_parser_unwrap(rb_sax_parser);
+
+  ctxt->sax = sax;
+  ctxt->userData = ctxt; /* so we can use libxml2/SAX2.c handlers if we want to */
+  ctxt->_private = (void *)rb_sax_parser;
+
+  xmlSetStructuredErrorFunc(NULL, NULL);
+
+  /* although we're calling back into Ruby here, we don't need to worry about exceptions, because we
+   * don't have any cleanup to do. The only memory we need to free is handled by
+   * xml_sax_parser_context_type_free */
+  htmlParseDocument(ctxt);
+
+  return Qnil;
+}
+
+void
+noko_init_html_sax_parser_context(void)
+{
+  assert(cNokogiriXmlSaxParserContext);
+  cNokogiriHtml4SaxParserContext = rb_define_class_under(mNokogiriHtml4Sax, "ParserContext",
+                                   cNokogiriXmlSaxParserContext);
+
+  rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "native_memory",
+                             noko_html4_sax_parser_context_s_native_memory, 2);
+  rb_define_singleton_method(cNokogiriHtml4SaxParserContext, "native_file",
+                             noko_html4_sax_parser_context_s_native_file, 2);
+
+  rb_define_method(cNokogiriHtml4SaxParserContext, "parse_with",
+                   noko_html4_sax_parser_context__parse_with, 1);
+}