nokolexbor 0.2.0

data/ext/nokolexbor/libxml/xpointer.h

@@ -0,0 +1,137 @@
+/*
+ * Summary: API to handle XML Pointers
+ * Description: API to handle XML Pointers
+ * Base implementation was made accordingly to
+ * W3C Candidate Recommendation 7 June 2000
+ * http://www.w3.org/TR/2000/CR-xptr-20000607
+ *
+ * Added support for the element() scheme described in:
+ * W3C Proposed Recommendation 13 November 2002
+ * http://www.w3.org/TR/2002/PR-xptr-element-20021113/
+ *
+ * Copy: See Copyright for the status of this software.
+ *
+ * Author: Daniel Veillard
+ */
+
+#ifndef __XML_XPTR_H__
+#define __XML_XPTR_H__
+
+#include "xmlversion.h"
+
+#ifdef LIBXML_XPTR_ENABLED
+
+#include "tree.h"
+#include "xpath.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifdef LIBXML_XPTR_LOCS_ENABLED
+/*
+ * A Location Set
+ */
+typedef struct _xmlLocationSet xmlLocationSet;
+typedef xmlLocationSet *xmlLocationSetPtr;
+struct _xmlLocationSet {
+    int locNr;		      /* number of locations in the set */
+    int locMax;		      /* size of the array as allocated */
+    xmlXPathObjectPtr *locTab;/* array of locations */
+};
+
+/*
+ * Handling of location sets.
+ */
+
+XML_DEPRECATED
+XMLPUBFUN xmlLocationSetPtr XMLCALL
+		    xmlXPtrLocationSetCreate	(xmlXPathObjectPtr val);
+XML_DEPRECATED
+XMLPUBFUN void XMLCALL
+		    xmlXPtrFreeLocationSet	(xmlLocationSetPtr obj);
+XML_DEPRECATED
+XMLPUBFUN xmlLocationSetPtr XMLCALL
+		    xmlXPtrLocationSetMerge	(xmlLocationSetPtr val1,
+						 xmlLocationSetPtr val2);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrNewRange		(lxb_dom_node_t_ptr start,
+						 int startindex,
+						 lxb_dom_node_t_ptr end,
+						 int endindex);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrNewRangePoints	(xmlXPathObjectPtr start,
+						 xmlXPathObjectPtr end);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrNewRangeNodePoint	(lxb_dom_node_t_ptr start,
+						 xmlXPathObjectPtr end);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrNewRangePointNode	(xmlXPathObjectPtr start,
+						 lxb_dom_node_t_ptr end);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrNewRangeNodes	(lxb_dom_node_t_ptr start,
+						 lxb_dom_node_t_ptr end);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrNewLocationSetNodes	(lxb_dom_node_t_ptr start,
+						 lxb_dom_node_t_ptr end);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrNewLocationSetNodeSet(xmlNodeSetPtr set);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrNewRangeNodeObject	(lxb_dom_node_t_ptr start,
+						 xmlXPathObjectPtr end);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrNewCollapsedRange	(lxb_dom_node_t_ptr start);
+XML_DEPRECATED
+XMLPUBFUN void XMLCALL
+		    xmlXPtrLocationSetAdd	(xmlLocationSetPtr cur,
+						 xmlXPathObjectPtr val);
+XML_DEPRECATED
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrWrapLocationSet	(xmlLocationSetPtr val);
+XML_DEPRECATED
+XMLPUBFUN void XMLCALL
+		    xmlXPtrLocationSetDel	(xmlLocationSetPtr cur,
+						 xmlXPathObjectPtr val);
+XML_DEPRECATED
+XMLPUBFUN void XMLCALL
+		    xmlXPtrLocationSetRemove	(xmlLocationSetPtr cur,
+						 int val);
+#endif /* LIBXML_XPTR_LOCS_ENABLED */
+
+/*
+ * Functions.
+ */
+XMLPUBFUN xmlXPathContextPtr XMLCALL
+		    xmlXPtrNewContext		(lxb_dom_document_t_ptr doc,
+						 lxb_dom_node_t_ptr here,
+						 lxb_dom_node_t_ptr origin);
+XMLPUBFUN xmlXPathObjectPtr XMLCALL
+		    xmlXPtrEval			(const xmlChar *str,
+						 xmlXPathContextPtr ctx);
+#ifdef LIBXML_XPTR_LOCS_ENABLED
+XML_DEPRECATED
+XMLPUBFUN void XMLCALL
+		    xmlXPtrRangeToFunction	(xmlXPathParserContextPtr ctxt,
+						 int nargs);
+XML_DEPRECATED
+XMLPUBFUN lxb_dom_node_t_ptr XMLCALL
+		    xmlXPtrBuildNodeList	(xmlXPathObjectPtr obj);
+XML_DEPRECATED
+XMLPUBFUN void XMLCALL
+		    xmlXPtrEvalRangePredicate	(xmlXPathParserContextPtr ctxt);
+#endif /* LIBXML_XPTR_LOCS_ENABLED */
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* LIBXML_XPTR_ENABLED */
+#endif /* __XML_XPTR_H__ */

data/ext/nokolexbor/libxml.h

@@ -0,0 +1,76 @@
+/*
+ * libxml.h: internal header only used during the compilation of libxml
+ *
+ * See COPYRIGHT for the status of this software
+ *
+ * Author: breese@users.sourceforge.net
+ */
+
+#ifndef __XML_LIBXML_H__
+#define __XML_LIBXML_H__
+
+/*
+ * These macros must be defined before including system headers.
+ * Do not add any #include directives above this block.
+ */
+#ifndef NO_LARGEFILE_SOURCE
+#ifndef _LARGEFILE_SOURCE
+#define _LARGEFILE_SOURCE
+#endif
+#ifndef _FILE_OFFSET_BITS
+#define _FILE_OFFSET_BITS 64
+#endif
+#endif
+
+/*
+ * Currently supported platforms use either autoconf or
+ * copy to config.h own "preset" configuration file.
+ * As result ifdef HAVE_CONFIG_H is omitted here.
+ */
+#include "config.h"
+#include "libxml/xmlversion.h"
+
+#ifndef SYSCONFDIR
+  #define SYSCONFDIR "/etc"
+#endif
+
+#if defined(__Lynx__)
+#include <stdio.h> /* pull definition of size_t */
+#include <varargs.h>
+int snprintf(char *, size_t, const char *, ...);
+int vfprintf(FILE *, const char *, va_list);
+#endif
+
+#ifndef WITH_TRIO
+#include <stdio.h>
+#else
+/**
+ * TRIO_REPLACE_STDIO:
+ *
+ * This macro is defined if the trio string formatting functions are to
+ * be used instead of the default stdio ones.
+ */
+#define TRIO_REPLACE_STDIO
+#include "trio.h"
+#endif
+
+#if !defined(_WIN32) && \
+    !defined(__CYGWIN__) && \
+    (defined(__clang__) || \
+     (defined(__GNUC__) && (__GNUC__ >= 4)))
+#define XML_HIDDEN __attribute__((visibility("hidden")))
+#else
+#define XML_HIDDEN
+#endif
+
+#if defined(__clang__) || \
+    (defined(__GNUC__) && (__GNUC__ >= 8))
+#define ATTRIBUTE_NO_SANITIZE(arg) __attribute__((no_sanitize(arg)))
+#else
+#define ATTRIBUTE_NO_SANITIZE(arg)
+#endif
+
+#if !defined(PIC) && !defined(NOLIBTOOL) && !defined(LIBXML_STATIC)
+#  define LIBXML_STATIC
+#endif
+#endif /* ! __XML_LIBXML_H__ */

data/ext/nokolexbor/memory.c

@@ -0,0 +1,39 @@
+/*
+ * This file overrides the default `lexbor_malloc` and other
+ * related functions to reduce memory bloat and allow Ruby to
+ * GC more often.
+ * By calling `ruby_xmalloc` instead of `malloc`, Ruby can
+ * count the memory usage of the C extension and run GC
+ * whenever `malloc_increase_byte` exceeds the limit. Similar
+ * to Nokogiri's solution by calling `xmlMemSetup`.
+ * The downside will be the downgrade of performance because
+ * of more frequent GC.
+ */
+
+#include <ruby.h>
+#include "lexbor/core/base.h"
+
+void *
+lexbor_malloc(size_t size)
+{
+    return ruby_xmalloc(size);
+}
+
+void *
+lexbor_realloc(void *dst, size_t size)
+{
+    return ruby_xrealloc(dst, size);
+}
+
+void *
+lexbor_calloc(size_t num, size_t size)
+{
+    return ruby_xcalloc(num, size);
+}
+
+void *
+lexbor_free(void *dst)
+{
+    ruby_xfree(dst);
+    return NULL;
+}

data/ext/nokolexbor/nl_document.c

@@ -0,0 +1,51 @@
+#include "nokolexbor.h"
+
+extern VALUE mNokolexbor;
+extern VALUE cNokolexborNode;
+VALUE cNokolexborDocument;
+
+static void
+free_nl_document(lxb_dom_document_t *document)
+{
+  lxb_html_document_destroy(document);
+}
+
+const rb_data_type_t nl_document_type = {
+    "Nokolexbor::Document",
+    {
+        0,
+        free_nl_document,
+    },
+    0,
+    0,
+    RUBY_TYPED_FREE_IMMEDIATELY,
+};
+
+static VALUE
+nl_document_parse(VALUE self, VALUE rb_html)
+{
+  const char *html_c = StringValuePtr(rb_html);
+  int html_len = RSTRING_LEN(rb_html);
+
+  lxb_html_document_t *document;
+
+  document = lxb_html_document_create();
+  if (document == NULL)
+  {
+    rb_raise(rb_eRuntimeError, "Error creating document");
+  }
+
+  lxb_status_t status = lxb_html_document_parse(document, html_c, html_len);
+  if (status != LXB_STATUS_OK)
+  {
+    nl_raise_lexbor_error(status);
+  }
+
+  return TypedData_Wrap_Struct(cNokolexborDocument, &nl_document_type, document);
+}
+
+void Init_nl_document(void)
+{
+  cNokolexborDocument = rb_define_class_under(mNokolexbor, "Document", cNokolexborNode);
+  rb_define_singleton_method(cNokolexborDocument, "parse", nl_document_parse, 1);
+}