nokogiri 1.10.10-java → 1.11.0-java

data/ext/nokogiri/html_document.c

@@ -1,6 +1,7 @@
 #include <html_document.h>
 
 static ID id_encoding_found;
+static ID id_to_s;
 
 /*
  * call-seq:
@@ -8,19 +9,20 @@ 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
-  );
+          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 ;
@@ -33,60 +35,55 @@ 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(io_read_callback, io_close_callback, (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 = Nokogiri_wrap_xml_document(klass, c_doc);
+  rb_iv_set(rb_doc, "@errors", rb_error_list);
+  return rb_doc;
 }
 
 /*
@@ -96,43 +93,45 @@ 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);
 
-  xmlResetLastError();
-  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
+  xmlSetStructuredErrorFunc((void *)rb_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 = Nokogiri_wrap_xml_document(klass, c_doc);
+  rb_iv_set(rb_doc, "@errors", rb_error_list);
+  return rb_doc;
 }
 
 /*
@@ -141,7 +140,8 @@ 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);
@@ -149,22 +149,23 @@ static VALUE type(VALUE self)
 }
 
 VALUE cNokogiriHtmlDocument ;
-void 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;
+void
+init_html_document()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE nokogiri_xml = rb_define_module_under(nokogiri, "XML");
+  VALUE nokogiri_xml_node = rb_define_class_under(nokogiri_xml, "Node", rb_cObject);
+  VALUE nokogiri_xml_document = rb_define_class_under(nokogiri_xml, "Document", nokogiri_xml_node);
+  VALUE nokogiri_html = rb_define_module_under(nokogiri, "HTML");
+  cNokogiriHtmlDocument = rb_define_class_under(nokogiri_html, "Document", nokogiri_xml_document);
 
-  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_sax_parser_context.c

@@ -4,13 +4,13 @@ VALUE cNokogiriHtmlSaxParserContext ;
 
 static void deallocate(xmlParserCtxtPtr ctxt)
 {
-  NOKOGIRI_DEBUG_START(handler);
+  NOKOGIRI_DEBUG_START(ctxt);
 
   ctxt->sax = NULL;
 
   htmlFreeParserCtxt(ctxt);
 
-  NOKOGIRI_DEBUG_END(handler);
+  NOKOGIRI_DEBUG_END(ctxt);
 }
 
 static VALUE

data/ext/nokogiri/nokogiri.c

@@ -7,12 +7,11 @@ VALUE mNokogiriXslt ;
 VALUE mNokogiriXmlSax ;
 VALUE mNokogiriHtmlSax ;
 
-#ifdef USE_INCLUDED_VASPRINTF
+#ifndef HAVE_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 vasprintf(char **strp, const char *fmt, va_list ap)
 {
   /* Mingw32/64 have a broken vsnprintf implementation that fails when
    * using a zero-byte limit in order to retrieve the required size for malloc.
@@ -21,23 +20,15 @@ int vasprintf (char **strp, const char *fmt, va_list ap)
   char tmp[1];
   int len = vsnprintf (tmp, 1, fmt, ap) + 1;
   char *res = (char *)malloc((unsigned int)len);
-  if (res == NULL)
-      return -1;
+  if (res == NULL) {
+    return -1;
+  }
   *strp = res;
   return vsnprintf(res, (unsigned int)len, fmt, ap);
 }
 #endif
 
-void vasprintf_free (void *p)
-{
-  free(p);
-}
-
-#ifdef HAVE_RUBY_UTIL_H
 #include "ruby/util.h"
-#else
-#include "util.h"
-#endif
 
 void nokogiri_root_node(xmlNodePtr node)
 {
@@ -45,7 +36,7 @@ void nokogiri_root_node(xmlNodePtr node)
   nokogiriTuplePtr tuple;
 
   doc = node->doc;
-  if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
+  if (doc->type == XML_DOCUMENT_FRAG_NODE) { doc = doc->doc; }
   tuple = (nokogiriTuplePtr)doc->_private;
   st_insert(tuple->unlinkedNodes, (st_data_t)node, (st_data_t)node);
 }
@@ -54,7 +45,7 @@ void nokogiri_root_nsdef(xmlNsPtr ns, xmlDocPtr doc)
 {
   nokogiriTuplePtr tuple;
 
-  if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
+  if (doc->type == XML_DOCUMENT_FRAG_NODE) { doc = doc->doc; }
   tuple = (nokogiriTuplePtr)doc->_private;
   st_insert(tuple->unlinkedNodes, (st_data_t)ns, (st_data_t)ns);
 }
@@ -62,10 +53,10 @@ void nokogiri_root_nsdef(xmlNsPtr ns, xmlDocPtr doc)
 void Init_nokogiri()
 {
   xmlMemSetup(
-      (xmlFreeFunc)ruby_xfree,
-      (xmlMallocFunc)ruby_xmalloc,
-      (xmlReallocFunc)ruby_xrealloc,
-      ruby_strdup
+    (xmlFreeFunc)ruby_xfree,
+    (xmlMallocFunc)ruby_xmalloc,
+    (xmlReallocFunc)ruby_xrealloc,
+    ruby_strdup
   );
 
   mNokogiri         = rb_define_module("Nokogiri");
@@ -75,27 +66,26 @@ void Init_nokogiri()
   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 NOKOGIRI_USE_PACKAGED_LIBRARIES
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_USE_PACKAGED_LIBRARIES"), Qtrue);
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATH"), NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATH));
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATH"), NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATH));
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATCHES), " "));
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATCHES), " "));
+  rb_const_set(mNokogiri, rb_intern("LIBXML_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION));
+  rb_const_set(mNokogiri, rb_intern("LIBXML_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xmlParserVersion));
+
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXSLT_DOTTED_VERSION));
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xsltEngineVersion));
+
+#ifdef NOKOGIRI_PACKAGED_LIBRARIES
+  rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qtrue);
+#ifdef NOKOGIRI_PRECOMPILED_LIBRARIES
+  rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qtrue);
+#else
+  rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
+#endif
+  rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATCHES), " "));
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATCHES), " "));
 #else
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_USE_PACKAGED_LIBRARIES"), Qfalse);
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATH"), Qnil);
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATH"), Qnil);
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATCHES"), Qnil);
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATCHES"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qfalse);
+  rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
+  rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), Qnil);
 #endif
 
 #ifdef LIBXML_ICONV_ENABLED
@@ -104,6 +94,10 @@ void Init_nokogiri()
   rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qfalse);
 #endif
 
+#ifdef NOKOGIRI_OTHER_LIBRARY_VERSIONS
+  rb_const_set(mNokogiri, rb_intern("OTHER_LIBRARY_VERSIONS"), NOKOGIRI_STR_NEW2(NOKOGIRI_OTHER_LIBRARY_VERSIONS));
+#endif
+
   xmlInitParser();
 
   init_xml_document();

data/ext/nokogiri/xml_document.c

@@ -1,6 +1,6 @@
 #include <xml_document.h>
 
-static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
+static int dealloc_node_i2(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
 {
   switch(node->type) {
   case XML_ATTRIBUTE_NODE:
@@ -20,6 +20,11 @@ static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
   return ST_CONTINUE;
 }
 
+static int dealloc_node_i(st_data_t key, st_data_t node, st_data_t doc)
+{
+  return dealloc_node_i2((xmlNodePtr)key, (xmlNodePtr)node, (xmlDocPtr)doc);
+}
+
 static void remove_private(xmlNodePtr node)
 {
   xmlNodePtr child;
@@ -38,6 +43,15 @@ static void remove_private(xmlNodePtr node)
   node->_private = NULL;
 }
 
+static void mark(xmlDocPtr doc)
+{
+  nokogiriTuplePtr tuple = (nokogiriTuplePtr)doc->_private;
+  if(tuple) {
+      rb_gc_mark(tuple->doc);
+      rb_gc_mark(tuple->node_cache);
+  }
+}
+
 static void dealloc(xmlDocPtr doc)
 {
   st_table *node_hash;
@@ -492,7 +506,7 @@ static int block_caller(void * ctx, xmlNodePtr _node, xmlNodePtr _parent)
  * The block must return a non-nil, non-false value if the +obj+ passed in
  * should be included in the canonicalized document.
  */
-static VALUE canonicalize(int argc, VALUE* argv, VALUE self)
+static VALUE nokogiri_xml_document_canonicalize(int argc, VALUE* argv, VALUE self)
 {
   VALUE mode;
   VALUE incl_ns;
@@ -573,7 +587,7 @@ void init_xml_document()
   rb_define_method(klass, "encoding", encoding, 0);
   rb_define_method(klass, "encoding=", set_encoding, 1);
   rb_define_method(klass, "version", version, 0);
-  rb_define_method(klass, "canonicalize", canonicalize, -1);
+  rb_define_method(klass, "canonicalize", nokogiri_xml_document_canonicalize, -1);
   rb_define_method(klass, "dup", duplicate_document, -1);
   rb_define_method(klass, "url", url, 0);
   rb_define_method(klass, "create_entity", create_entity, -1);
@@ -588,7 +602,7 @@ VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
 
   VALUE rb_doc = Data_Wrap_Struct(
       klass ? klass : cNokogiriXmlDocument,
-      0,
+      mark,
       dealloc,
       doc
   );

data/ext/nokogiri/xml_io.c

@@ -2,12 +2,13 @@
 
 static ID id_read, id_write;
 
-VALUE read_check(VALUE *args) {
+VALUE read_check(VALUE val) {
+  VALUE *args = (VALUE *)val;
   return rb_funcall(args[0], id_read, 1, args[1]);
 }
 
-VALUE read_failed(void) {
-	return Qundef;
+VALUE read_failed(VALUE arg, VALUE exc) {
+  return Qundef;
 }
 
 int io_read_callback(void * ctx, char * buffer, int len) {
@@ -30,12 +31,13 @@ int io_read_callback(void * ctx, char * buffer, int len) {
   return (int)safe_len;
 }
 
-VALUE write_check(VALUE *args) {
+VALUE write_check(VALUE val) {
+  VALUE *args = (VALUE *)val;
   return rb_funcall(args[0], id_write, 1, args[1]);
 }
 
-VALUE write_failed(void) {
-	return Qundef;
+VALUE write_failed(VALUE arg, VALUE exc) {
+  return Qundef;
 }
 
 int io_write_callback(void * ctx, char * buffer, int len) {