nokogiri 1.10.9-java → 1.11.0.rc4-java

data/ext/nokogiri/xml_node.c

@@ -301,7 +301,7 @@ ok:
        *  issue #391, where new node's prefix may become the string "default"
        *  see libxml2 tree.c xmlNewReconciliedNs which implements this behavior.
        */
-      xmlFree(reparentee->ns->prefix);
+      xmlFree((xmlChar*)reparentee->ns->prefix);
       reparentee->ns->prefix = NULL;
     }
   }
@@ -1332,6 +1332,25 @@ static VALUE line(VALUE self)
   return INT2NUM(xmlGetLineNo(node));
 }
 
+/*
+ * call-seq:
+ *  line=(num)
+ *
+ * Sets the line for this Node. num must be less than 65535.
+ */
+static VALUE set_line(VALUE self, VALUE num)
+{
+  xmlNodePtr node;
+  Data_Get_Struct(self, xmlNode, node);
+
+  int value = NUM2INT(num);
+  if (value < 65535) {
+    node->line = value;
+  }
+
+  return num;
+}
+
 /*
  * call-seq:
  *  add_namespace_definition(prefix, href)
@@ -1728,6 +1747,7 @@ void init_xml_node()
   rb_define_method(klass, "create_external_subset", create_external_subset, 3);
   rb_define_method(klass, "pointer_id", pointer_id, 0);
   rb_define_method(klass, "line", line, 0);
+  rb_define_method(klass, "line=", set_line, 1);
   rb_define_method(klass, "content", get_native_content, 0);
   rb_define_method(klass, "native_content=", set_native_content, 1);
   rb_define_method(klass, "lang", get_lang, 0);

data/ext/nokogiri/xml_node_set.c

@@ -319,7 +319,7 @@ static VALUE slice(int argc, VALUE *argv, VALUE self)
  *
  * Return this list as an Array
  */
-static VALUE to_array(VALUE self, VALUE rb_node)
+static VALUE to_array(VALUE self)
 {
   xmlNodeSetPtr node_set ;
   VALUE list;

data/ext/nokogiri/xml_reader.c

@@ -28,35 +28,24 @@ static int has_attributes(xmlTextReaderPtr reader)
 static void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash)
 {
   xmlNsPtr ns;
-  static char buffer[XMLNS_BUFFER_LEN] ;
-  char *key ;
-  size_t keylen ;
+  VALUE key;
 
   if (node->type != XML_ELEMENT_NODE) return ;
 
   ns = node->nsDef;
   while (ns != NULL) {
 
-    keylen = XMLNS_PREFIX_LEN + (ns->prefix ? (strlen((const char*)ns->prefix) + 1) : 0) ;
-    if (keylen > XMLNS_BUFFER_LEN) {
-      key = (char*)malloc(keylen) ;
-    } else {
-      key = buffer ;
-    }
-
+    key = rb_enc_str_new_cstr(XMLNS_PREFIX, rb_utf8_encoding());
     if (ns->prefix) {
-      sprintf(key, "%s:%s", XMLNS_PREFIX, ns->prefix);
-    } else {
-      sprintf(key, "%s", XMLNS_PREFIX);
+      rb_str_cat_cstr(key, ":");
+      rb_str_cat_cstr(key, (const char*)ns->prefix);
     }
 
+    key = rb_str_conv_enc(key, rb_utf8_encoding(), rb_default_internal_encoding());
     rb_hash_aset(attr_hash,
-        NOKOGIRI_STR_NEW2(key),
+        key,
         (ns->href ? NOKOGIRI_STR_NEW2(ns->href) : Qnil)
     );
-    if (key != buffer) {
-      free(key);
-    }
     ns = ns->next ;
   }
 }

data/ext/nokogiri/xml_relax_ng.c

@@ -53,16 +53,24 @@ static VALUE validate_document(VALUE self, VALUE document)
  *
  * Create a new RelaxNG from the contents of +string+
  */
-static VALUE read_memory(VALUE klass, VALUE content)
+static VALUE read_memory(int argc, VALUE *argv, VALUE klass)
 {
-  xmlRelaxNGParserCtxtPtr ctx = xmlRelaxNGNewMemParserCtxt(
-      (const char *)StringValuePtr(content),
-      (int)RSTRING_LEN(content)
-  );
+  VALUE content;
+  VALUE parse_options;
+  xmlRelaxNGParserCtxtPtr ctx;
   xmlRelaxNGPtr schema;
-  VALUE errors = rb_ary_new();
+  VALUE errors;
   VALUE rb_schema;
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &content, &parse_options);
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
 
+  ctx = xmlRelaxNGNewMemParserCtxt((const char *)StringValuePtr(content), (int)RSTRING_LEN(content));
+
+  errors = rb_ary_new();
   xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
 
 #ifdef HAVE_XMLRELAXNGSETPARSERSTRUCTUREDERRORS
@@ -90,6 +98,7 @@ static VALUE read_memory(VALUE klass, VALUE content)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 }
@@ -100,18 +109,25 @@ static VALUE read_memory(VALUE klass, VALUE content)
  *
  * Create a new RelaxNG schema from the Nokogiri::XML::Document +doc+
  */
-static VALUE from_document(VALUE klass, VALUE document)
+static VALUE from_document(int argc, VALUE *argv, VALUE klass)
 {
+  VALUE document;
+  VALUE parse_options;
   xmlDocPtr doc;
   xmlRelaxNGParserCtxtPtr ctx;
   xmlRelaxNGPtr schema;
   VALUE errors;
   VALUE rb_schema;
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &document, &parse_options);
 
   Data_Get_Struct(document, xmlDoc, doc);
+  doc = doc->doc; /* In case someone passes us a node. ugh. */
 
-  /* In case someone passes us a node. ugh. */
-  doc = doc->doc;
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
 
   ctx = xmlRelaxNGNewDocParserCtxt(doc);
 
@@ -129,6 +145,7 @@ static VALUE from_document(VALUE klass, VALUE document)
   schema = xmlRelaxNGParse(ctx);
 
   xmlSetStructuredErrorFunc(NULL, NULL);
+  xmlRelaxNGFreeParserCtxt(ctx);
 
   if(NULL == schema) {
     xmlErrorPtr error = xmlGetLastError();
@@ -142,6 +159,7 @@ static VALUE from_document(VALUE klass, VALUE document)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 }
@@ -155,7 +173,7 @@ void init_xml_relax_ng()
 
   cNokogiriXmlRelaxNG = klass;
 
-  rb_define_singleton_method(klass, "read_memory", read_memory, 1);
-  rb_define_singleton_method(klass, "from_document", from_document, 1);
+  rb_define_singleton_method(klass, "read_memory", read_memory, -1);
+  rb_define_singleton_method(klass, "from_document", from_document, -1);
   rb_define_private_method(klass, "validate_document", validate_document, 1);
 }

data/ext/nokogiri/xml_sax_parser.c

@@ -1,8 +1,5 @@
 #include <xml_sax_parser.h>
 
-int vasprintf (char **strp, const char *fmt, va_list ap);
-void vasprintf_free (void *p);
-
 static ID id_start_document, id_end_document, id_start_element, id_end_element;
 static ID id_start_element_namespace, id_end_element_namespace;
 static ID id_comment, id_characters, id_xmldecl, id_error, id_warning;
@@ -206,7 +203,7 @@ static void warning_func(void * ctx, const char *msg, ...)
   va_end(args);
 
   ruby_message = NOKOGIRI_STR_NEW2(message);
-  vasprintf_free(message);
+  free(message);
   rb_funcall(doc, id_warning, 1, ruby_message);
 }
 
@@ -223,7 +220,7 @@ static void error_func(void * ctx, const char *msg, ...)
   va_end(args);
 
   ruby_message = NOKOGIRI_STR_NEW2(message);
-  vasprintf_free(message);
+  free(message);
   rb_funcall(doc, id_error, 1, ruby_message);
 }
 
@@ -262,8 +259,6 @@ static VALUE allocate(VALUE klass)
 {
   xmlSAXHandlerPtr handler = calloc((size_t)1, sizeof(xmlSAXHandler));
 
-  xmlSetStructuredErrorFunc(NULL, NULL);
-
   handler->startDocument = start_document;
   handler->endDocument = end_document;
   handler->startElement = start_element;

data/ext/nokogiri/xml_sax_parser_context.c

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

data/ext/nokogiri/xml_schema.c

@@ -93,15 +93,27 @@ static VALUE validate_file(VALUE self, VALUE rb_filename)
  *
  * Create a new Schema from the contents of +string+
  */
-static VALUE read_memory(VALUE klass, VALUE content)
+static VALUE read_memory(int argc, VALUE *argv, VALUE klass)
 {
+  VALUE content;
+  VALUE parse_options;
+  int parse_options_int;
+  xmlSchemaParserCtxtPtr ctx;
   xmlSchemaPtr schema;
-  xmlSchemaParserCtxtPtr ctx = xmlSchemaNewMemParserCtxt(
-      (const char *)StringValuePtr(content),
-      (int)RSTRING_LEN(content)
-  );
+  VALUE errors;
   VALUE rb_schema;
-  VALUE errors = rb_ary_new();
+  int scanned_args = 0;
+  xmlExternalEntityLoader old_loader = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &content, &parse_options);
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
+  parse_options_int = (int)NUM2INT(rb_funcall(parse_options, rb_intern("to_i"), 0));
+
+  ctx = xmlSchemaNewMemParserCtxt((const char *)StringValuePtr(content), (int)RSTRING_LEN(content));
+
+  errors = rb_ary_new();
   xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
 
 #ifdef HAVE_XMLSCHEMASETPARSERSTRUCTUREDERRORS
@@ -109,10 +121,19 @@ static VALUE read_memory(VALUE klass, VALUE content)
     ctx,
     Nokogiri_error_array_pusher,
     (void *)errors
-  );
+    );
 #endif
 
-   schema = xmlSchemaParse(ctx);
+  if (parse_options_int & XML_PARSE_NONET) {
+    old_loader = xmlGetExternalEntityLoader();
+    xmlSetExternalEntityLoader(xmlNoNetExternalEntityLoader);
+  }
+
+  schema = xmlSchemaParse(ctx);
+
+  if (old_loader) {
+    xmlSetExternalEntityLoader(old_loader);
+  }
 
   xmlSetStructuredErrorFunc(NULL, NULL);
   xmlSchemaFreeParserCtxt(ctx);
@@ -129,6 +150,7 @@ static VALUE read_memory(VALUE klass, VALUE content)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 }
@@ -164,18 +186,28 @@ static int has_blank_nodes_p(VALUE cache)
  *
  * Create a new Schema from the Nokogiri::XML::Document +doc+
  */
-static VALUE from_document(VALUE klass, VALUE document)
+static VALUE from_document(int argc, VALUE *argv, VALUE klass)
 {
+  VALUE document;
+  VALUE parse_options;
+  int parse_options_int;
   xmlDocPtr doc;
   xmlSchemaParserCtxtPtr ctx;
   xmlSchemaPtr schema;
   VALUE errors;
   VALUE rb_schema;
+  int scanned_args = 0;
+  xmlExternalEntityLoader old_loader = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &document, &parse_options);
 
   Data_Get_Struct(document, xmlDoc, doc);
+  doc = doc->doc; /* In case someone passes us a node. ugh. */
 
-  /* In case someone passes us a node. ugh. */
-  doc = doc->doc;
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
+  parse_options_int = (int)NUM2INT(rb_funcall(parse_options, rb_intern("to_i"), 0));
 
   if (has_blank_nodes_p(DOC_NODE_CACHE(doc))) {
     rb_raise(rb_eArgError, "Creating a schema from a document that has blank nodes exposed to Ruby is dangerous");
@@ -194,8 +226,17 @@ static VALUE from_document(VALUE klass, VALUE document)
   );
 #endif
 
+  if (parse_options_int & XML_PARSE_NONET) {
+    old_loader = xmlGetExternalEntityLoader();
+    xmlSetExternalEntityLoader(xmlNoNetExternalEntityLoader);
+  }
+
   schema = xmlSchemaParse(ctx);
 
+  if (old_loader) {
+    xmlSetExternalEntityLoader(old_loader);
+  }
+
   xmlSetStructuredErrorFunc(NULL, NULL);
   xmlSchemaFreeParserCtxt(ctx);
 
@@ -211,6 +252,7 @@ static VALUE from_document(VALUE klass, VALUE document)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 
@@ -226,8 +268,8 @@ void init_xml_schema()
 
   cNokogiriXmlSchema = klass;
 
-  rb_define_singleton_method(klass, "read_memory", read_memory, 1);
-  rb_define_singleton_method(klass, "from_document", from_document, 1);
+  rb_define_singleton_method(klass, "read_memory", read_memory, -1);
+  rb_define_singleton_method(klass, "from_document", from_document, -1);
 
   rb_define_private_method(klass, "validate_document", validate_document, 1);
   rb_define_private_method(klass, "validate_file",     validate_file, 1);

data/ext/nokogiri/xml_xpath_context.c

@@ -1,6 +1,11 @@
 #include <xml_xpath_context.h>
 
-int vasprintf (char **strp, const char *fmt, va_list ap);
+/*
+ * these constants have matching declarations in
+ * ext/java/nokogiri/internals/NokogiriNamespaceContext.java
+ */
+static const xmlChar *NOKOGIRI_BUILTIN_PREFIX = (const xmlChar *)"nokogiri-builtin";
+static const xmlChar *NOKOGIRI_BUILTIN_URI = (const xmlChar *)"https://www.nokogiri.org/default_ns/ruby/builtins";
 
 static void deallocate(xmlXPathContextPtr ctx)
 {
@@ -9,6 +14,73 @@ static void deallocate(xmlXPathContextPtr ctx)
   NOKOGIRI_DEBUG_END(ctx);
 }
 
+/* find a CSS class in an HTML element's `class` attribute */
+const xmlChar* builtin_css_class(const xmlChar* str, const xmlChar *val)
+{
+  int val_len;
+
+  if (str == NULL) { return(NULL); }
+  if (val == NULL) { return(NULL); }
+
+  val_len = xmlStrlen(val);
+  if (val_len == 0) { return(str); }
+
+  while (*str != 0) {
+    if ((*str == *val) && !xmlStrncmp(str, val, val_len)) {
+      const xmlChar* next_byte = str + val_len;
+
+      /* only match if the next byte is whitespace or end of string */
+      if ((*next_byte == 0) || (IS_BLANK_CH(*next_byte))) {
+        return((const xmlChar*)str);
+      }
+    }
+
+    /* advance str to whitespace */
+    while ((*str != 0) && !IS_BLANK_CH(*str)) {
+      str++;
+    }
+
+    /* advance str to start of next word or end of string */
+    while ((*str != 0) && IS_BLANK_CH(*str)) {
+      str++;
+    }
+  }
+
+  return(NULL);
+}
+
+/* xmlXPathFunction to wrap builtin_css_class() */
+static void xpath_builtin_css_class(xmlXPathParserContextPtr ctxt, int nargs)
+{
+  xmlXPathObjectPtr hay, needle;
+
+  CHECK_ARITY(2);
+
+  CAST_TO_STRING;
+  needle = valuePop(ctxt);
+  if ((needle == NULL) || (needle->type != XPATH_STRING)) {
+    xmlXPathFreeObject(needle);
+    XP_ERROR(XPATH_INVALID_TYPE);
+  }
+
+  CAST_TO_STRING;
+  hay = valuePop(ctxt);
+  if ((hay == NULL) || (hay->type != XPATH_STRING)) {
+    xmlXPathFreeObject(hay);
+    xmlXPathFreeObject(needle);
+    XP_ERROR(XPATH_INVALID_TYPE);
+  }
+
+  if (builtin_css_class(hay->stringval, needle->stringval)) {
+    valuePush(ctxt, xmlXPathNewBoolean(1));
+  } else {
+    valuePush(ctxt, xmlXPathNewBoolean(0));
+  }
+
+  xmlXPathFreeObject(hay);
+  xmlXPathFreeObject(needle);
+}
+
 /*
  * call-seq:
  *  register_ns(prefix, uri)
@@ -263,14 +335,18 @@ static VALUE new(VALUE klass, VALUE nodeobj)
   xmlXPathContextPtr ctx;
   VALUE self;
 
-  xmlXPathInit();
-
   Data_Get_Struct(nodeobj, xmlNode, node);
 
+  xmlXPathInit();
+
   ctx = xmlXPathNewContext(node->doc);
   ctx->node = node;
+
+  xmlXPathRegisterNs(ctx, NOKOGIRI_BUILTIN_PREFIX, NOKOGIRI_BUILTIN_URI);
+  xmlXPathRegisterFuncNS(ctx, (const xmlChar *)"css-class", NOKOGIRI_BUILTIN_URI,
+                         xpath_builtin_css_class);
+
   self = Data_Wrap_Struct(klass, 0, deallocate, ctx);
-  /*rb_iv_set(self, "@xpath_handler", Qnil); */
   return self;
 }