nokogiri 1.10.4 → 1.11.2

data/ext/nokogiri/html_document.c

@@ -1,6 +1,9 @@
-#include <html_document.h>
+#include <nokogiri.h>
+
+VALUE cNokogiriHtmlDocument ;
 
 static ID id_encoding_found;
+static ID id_to_s;
 
 /*
  * call-seq:
@@ -8,21 +11,21 @@ static ID id_encoding_found;
  *
  * Create a new document
  */
-static VALUE new(int argc, VALUE *argv, VALUE klass)
+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);
+  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);
+          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 ;
 }
 
@@ -33,60 +36,55 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
  * 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 )
+static VALUE
+rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_encoding, VALUE rb_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;
+  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, Nokogiri_error_array_pusher);
+
+  c_doc = htmlReadIO(noko_io_read, noko_io_close, (void *)rb_io, c_url, c_encoding, options);
 
-  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);
 
   /*
    * 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 (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(doc);
+      xmlFreeDoc(c_doc);
       rb_exc_raise(encoding_found);
     }
   }
 
-  if(doc == NULL) {
-    xmlErrorPtr error;
+  if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
+    VALUE rb_error ;
 
-    xmlFreeDoc(doc);
+    xmlFreeDoc(c_doc);
 
-    error = xmlGetLastError();
-    if(error)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+    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;
   }
 
-  document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_iv_set(document, "@errors", error_list);
-  return document;
+  rb_doc = noko_xml_document_wrap(klass, c_doc);
+  rb_iv_set(rb_doc, "@errors", rb_error_list);
+  return rb_doc;
 }
 
 /*
@@ -96,43 +94,45 @@ static VALUE read_io( VALUE klass,
  * Read the HTML document contained in +string+ with given +url+, +encoding+,
  * and +options+.  See Nokogiri::HTML.parse
  */
-static VALUE read_memory( VALUE klass,
-                          VALUE string,
-                          VALUE url,
-                          VALUE encoding,
-                          VALUE options )
+static VALUE
+rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
 {
-  const char * c_buffer = StringValuePtr(string);
-  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;
+  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, Nokogiri_error_array_pusher);
 
-  xmlResetLastError();
-  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
+  c_doc = htmlReadMemory(c_buffer, html_len, c_url, c_encoding, options);
 
-  doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
   xmlSetStructuredErrorFunc(NULL, NULL);
 
-  if(doc == NULL) {
-    xmlErrorPtr error;
+  if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
+    VALUE rb_error ;
 
-    xmlFreeDoc(doc);
+    xmlFreeDoc(c_doc);
 
-    error = xmlGetLastError();
-    if(error)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+    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;
   }
 
-  document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_iv_set(document, "@errors", error_list);
-  return document;
+  rb_doc = noko_xml_document_wrap(klass, c_doc);
+  rb_iv_set(rb_doc, "@errors", rb_error_list);
+  return rb_doc;
 }
 
 /*
@@ -141,30 +141,26 @@ static VALUE read_memory( VALUE klass,
  *
  * The type for this document
  */
-static VALUE type(VALUE self)
+static VALUE
+rb_html_document_type(VALUE self)
 {
   htmlDocPtr doc;
   Data_Get_Struct(self, xmlDoc, doc);
   return INT2NUM((long)doc->type);
 }
 
-VALUE cNokogiriHtmlDocument ;
-void init_html_document()
+void
+noko_init_html_document()
 {
-  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);
-
-  cNokogiriHtmlDocument = klass;
+  assert(cNokogiriXmlDocument);
+  cNokogiriHtmlDocument = rb_define_class_under(mNokogiriHtml, "Document", cNokogiriXmlDocument);
 
-  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_singleton_method(cNokogiriHtmlDocument, "read_memory", rb_html_document_s_read_memory, 4);
+  rb_define_singleton_method(cNokogiriHtmlDocument, "read_io", rb_html_document_s_read_io, 4);
+  rb_define_singleton_method(cNokogiriHtmlDocument, "new", rb_html_document_s_new, -1);
 
-  rb_define_method(klass, "type", type, 0);
+  rb_define_method(cNokogiriHtmlDocument, "type", rb_html_document_type, 0);
 
   id_encoding_found = rb_intern("encoding_found");
+  id_to_s = rb_intern("to_s");
 }

data/ext/nokogiri/html_element_description.c

@@ -1,4 +1,6 @@
-#include <html_element_description.h>
+#include <nokogiri.h>
+
+VALUE cNokogiriHtmlElementDescription ;
 
 /*
  * call-seq:
@@ -6,9 +8,10 @@
  *
  * A list of required attributes for this element
  */
-static VALUE required_attributes(VALUE self)
+static VALUE
+required_attributes(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   VALUE list;
   int i;
 
@@ -16,9 +19,9 @@ static VALUE required_attributes(VALUE self)
 
   list = rb_ary_new();
 
-  if(NULL == description->attrs_req) return list;
+  if (NULL == description->attrs_req) { return list; }
 
-  for(i = 0; description->attrs_depr[i]; i++) {
+  for (i = 0; description->attrs_depr[i]; i++) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
   }
 
@@ -31,9 +34,10 @@ static VALUE required_attributes(VALUE self)
  *
  * A list of deprecated attributes for this element
  */
-static VALUE deprecated_attributes(VALUE self)
+static VALUE
+deprecated_attributes(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   VALUE list;
   int i;
 
@@ -41,9 +45,9 @@ static VALUE deprecated_attributes(VALUE self)
 
   list = rb_ary_new();
 
-  if(NULL == description->attrs_depr) return list;
+  if (NULL == description->attrs_depr) { return list; }
 
-  for(i = 0; description->attrs_depr[i]; i++) {
+  for (i = 0; description->attrs_depr[i]; i++) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
   }
 
@@ -56,9 +60,10 @@ static VALUE deprecated_attributes(VALUE self)
  *
  * A list of optional attributes for this element
  */
-static VALUE optional_attributes(VALUE self)
+static VALUE
+optional_attributes(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   VALUE list;
   int i;
 
@@ -66,9 +71,9 @@ static VALUE optional_attributes(VALUE self)
 
   list = rb_ary_new();
 
-  if(NULL == description->attrs_opt) return list;
+  if (NULL == description->attrs_opt) { return list; }
 
-  for(i = 0; description->attrs_opt[i]; i++) {
+  for (i = 0; description->attrs_opt[i]; i++) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
   }
 
@@ -81,13 +86,15 @@ static VALUE optional_attributes(VALUE self)
  *
  * The default sub element for this element
  */
-static VALUE default_sub_element(VALUE self)
+static VALUE
+default_sub_element(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  if (description->defaultsubelt)
+  if (description->defaultsubelt) {
     return NOKOGIRI_STR_NEW2(description->defaultsubelt);
+  }
 
   return Qnil;
 }
@@ -98,9 +105,10 @@ static VALUE default_sub_element(VALUE self)
  *
  * A list of allowed sub elements for this element.
  */
-static VALUE sub_elements(VALUE self)
+static VALUE
+sub_elements(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   VALUE list;
   int i;
 
@@ -108,9 +116,9 @@ static VALUE sub_elements(VALUE self)
 
   list = rb_ary_new();
 
-  if(NULL == description->subelts) return list;
+  if (NULL == description->subelts) { return list; }
 
-  for(i = 0; description->subelts[i]; i++) {
+  for (i = 0; description->subelts[i]; i++) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
   }
 
@@ -123,9 +131,10 @@ static VALUE sub_elements(VALUE self)
  *
  * The description for this element
  */
-static VALUE description(VALUE self)
+static VALUE
+description(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   return NOKOGIRI_STR_NEW2(description->desc);
@@ -137,12 +146,13 @@ static VALUE description(VALUE self)
  *
  * Is this element an inline element?
  */
-static VALUE inline_eh(VALUE self)
+static VALUE
+inline_eh(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  if(description->isinline) return Qtrue;
+  if (description->isinline) { return Qtrue; }
   return Qfalse;
 }
 
@@ -152,12 +162,13 @@ static VALUE inline_eh(VALUE self)
  *
  * Is this element deprecated?
  */
-static VALUE deprecated_eh(VALUE self)
+static VALUE
+deprecated_eh(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  if(description->depr) return Qtrue;
+  if (description->depr) { return Qtrue; }
   return Qfalse;
 }
 
@@ -167,12 +178,13 @@ static VALUE deprecated_eh(VALUE self)
  *
  * Is this an empty element?
  */
-static VALUE empty_eh(VALUE self)
+static VALUE
+empty_eh(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  if(description->empty) return Qtrue;
+  if (description->empty) { return Qtrue; }
   return Qfalse;
 }
 
@@ -182,12 +194,13 @@ static VALUE empty_eh(VALUE self)
  *
  * Should the end tag be saved?
  */
-static VALUE save_end_tag_eh(VALUE self)
+static VALUE
+save_end_tag_eh(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  if(description->saveEndTag) return Qtrue;
+  if (description->saveEndTag) { return Qtrue; }
   return Qfalse;
 }
 
@@ -197,12 +210,13 @@ static VALUE save_end_tag_eh(VALUE self)
  *
  * Can the end tag be implied for this tag?
  */
-static VALUE implied_end_tag_eh(VALUE self)
+static VALUE
+implied_end_tag_eh(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  if(description->endTag) return Qtrue;
+  if (description->endTag) { return Qtrue; }
   return Qfalse;
 }
 
@@ -212,12 +226,13 @@ static VALUE implied_end_tag_eh(VALUE self)
  *
  * Can the start tag be implied for this tag?
  */
-static VALUE implied_start_tag_eh(VALUE self)
+static VALUE
+implied_start_tag_eh(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  if(description->startTag) return Qtrue;
+  if (description->startTag) { return Qtrue; }
   return Qfalse;
 }
 
@@ -227,12 +242,13 @@ static VALUE implied_start_tag_eh(VALUE self)
  *
  * Get the tag name for this ElemementDescription
  */
-static VALUE name(VALUE self)
+static VALUE
+name(VALUE self)
 {
-  const htmlElemDesc * description;
+  const htmlElemDesc *description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  if(NULL == description->name) return Qnil;
+  if (NULL == description->name) { return Qnil; }
   return NOKOGIRI_STR_NEW2(description->name);
 }
 
@@ -242,38 +258,35 @@ static VALUE name(VALUE self)
  *
  * Get ElemementDescription for +tag_name+
  */
-static VALUE get_description(VALUE klass, VALUE tag_name)
+static VALUE
+get_description(VALUE klass, VALUE tag_name)
 {
-  const htmlElemDesc * description = htmlTagLookup(
-      (const xmlChar *)StringValueCStr(tag_name)
-  );
+  const htmlElemDesc *description = htmlTagLookup(
+                                      (const xmlChar *)StringValueCStr(tag_name)
+                                    );
 
-  if(NULL == description) return Qnil;
+  if (NULL == description) { return Qnil; }
   return Data_Wrap_Struct(klass, 0, 0, (void *)(uintptr_t)description);
 }
 
-VALUE cNokogiriHtmlElementDescription ;
-void init_html_element_description()
+void
+noko_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);
+  cNokogiriHtmlElementDescription = rb_define_class_under(mNokogiriHtml, "ElementDescription", rb_cObject);
+
+  rb_define_singleton_method(cNokogiriHtmlElementDescription, "[]", get_description, 1);
+
+  rb_define_method(cNokogiriHtmlElementDescription, "name", name, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "implied_start_tag?", implied_start_tag_eh, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "implied_end_tag?", implied_end_tag_eh, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "save_end_tag?", save_end_tag_eh, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "empty?", empty_eh, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "deprecated?", deprecated_eh, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "inline?", inline_eh, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "description", description, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "sub_elements", sub_elements, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "default_sub_element", default_sub_element, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "optional_attributes", optional_attributes, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "deprecated_attributes", deprecated_attributes, 0);
+  rb_define_method(cNokogiriHtmlElementDescription, "required_attributes", required_attributes, 0);
 }