libxml-ruby 4.1.2 → 6.0.0

data/ext/libxml/ruby_xml_html_parser_context.h

@@ -1,10 +1,11 @@
-/* Please see the LICENSE file for copyright and distribution information */
-
-#ifndef __RXML_HTML_PARSER_CONTEXT__
-#define __RXML_HTML_PARSER_CONTEXT__
-
-extern VALUE cXMLHtmlParserContext;
-
-void rxml_init_html_parser_context(void);
-
-#endif
+/* Please see the LICENSE file for copyright and distribution information */
+
+#ifndef __RXML_HTML_PARSER_CONTEXT__
+#define __RXML_HTML_PARSER_CONTEXT__
+
+extern VALUE cXMLHtmlParserContext;
+extern const rb_data_type_t rxml_html_parser_context_type;
+
+void rxml_init_html_parser_context(void);
+
+#endif

data/ext/libxml/ruby_xml_html_parser_options.c

@@ -1,48 +1,48 @@
-/* Please see the LICENSE file for copyright and distribution information */
-
-#include "ruby_libxml.h"
-
-#include <libxml/HTMLparser.h>
-
-/* Document-class: LibXML::XML::HTMLParser::Options
- *
- * Options to control the operation of the HTMLParser.  The easiest
- * way to set a parser's options is via the methods
- * XML::HTMLParser.file, XML::HTMLParser.io or XML::HTMLParser.string.
- * For additional control, see XML::HTMLParser::Context#options=.
-*/
-
-VALUE mXMLHtmlParserOptions;
-
-void rxml_init_html_parser_options(void)
-{
-  mXMLHtmlParserOptions = rb_define_module_under(cXMLHtmlParser, "Options");
-
-
-#if LIBXML_VERSION >= 20621
-  /* 1: Relax parsing. */
-  rb_define_const(mXMLHtmlParserOptions, "RECOVER", INT2NUM(HTML_PARSE_RECOVER)); 
-#endif
-#if LIBXML_VERSION >= 20708
-  /* 2:  Do not default a doctype if not found */
-  rb_define_const(mXMLHtmlParserOptions, "NODEFDTD", INT2NUM(HTML_PARSE_NODEFDTD));
-#endif
-  /* 32: Suppress error reports. */
-  rb_define_const(mXMLHtmlParserOptions, "NOERROR", INT2NUM(HTML_PARSE_NOERROR)); 
-  /* 64: Suppress warning reports. */
-  rb_define_const(mXMLHtmlParserOptions, "NOWARNING", INT2NUM(HTML_PARSE_NOWARNING));
-  /* 128: Enable pedantic error reporting. */
-  rb_define_const(mXMLHtmlParserOptions, "PEDANTIC", INT2NUM(HTML_PARSE_PEDANTIC)); 
-  /* 256: Remove blank nodes. */
-  rb_define_const(mXMLHtmlParserOptions, "NOBLANKS", INT2NUM(HTML_PARSE_NOBLANKS)); 
-#if LIBXML_VERSION >= 20621
-  /* 2048: Forbid network access. */
-  rb_define_const(mXMLHtmlParserOptions, "NONET", INT2NUM(HTML_PARSE_NONET)); 
-  /* 65536: Compact small text nodes. */
-  rb_define_const(mXMLHtmlParserOptions, "COMPACT", INT2NUM(HTML_PARSE_COMPACT));
-#endif
-#if LIBXML_VERSION >= 20707
-  /* 8192:  Do not add implied html/body... elements */
-  rb_define_const(mXMLHtmlParserOptions, "NOIMPLIED", INT2NUM(HTML_PARSE_NOIMPLIED));
-#endif
-}
+/* Please see the LICENSE file for copyright and distribution information */
+
+#include "ruby_libxml.h"
+
+#include <libxml/HTMLparser.h>
+
+/* Document-class: LibXML::XML::HTMLParser::Options
+ *
+ * Options to control the operation of the HTMLParser.  The easiest
+ * way to set a parser's options is via the methods
+ * XML::HTMLParser.file, XML::HTMLParser.io or XML::HTMLParser.string.
+ * For additional control, see XML::HTMLParser::Context#options=.
+*/
+
+VALUE mXMLHtmlParserOptions;
+
+void rxml_init_html_parser_options(void)
+{
+  mXMLHtmlParserOptions = rb_define_module_under(cXMLHtmlParser, "Options");
+
+
+#if LIBXML_VERSION >= 20621
+  /* 1: Relax parsing. */
+  rb_define_const(mXMLHtmlParserOptions, "RECOVER", INT2NUM(HTML_PARSE_RECOVER)); 
+#endif
+#if LIBXML_VERSION >= 20708
+  /* 2:  Do not default a doctype if not found */
+  rb_define_const(mXMLHtmlParserOptions, "NODEFDTD", INT2NUM(HTML_PARSE_NODEFDTD));
+#endif
+  /* 32: Suppress error reports. */
+  rb_define_const(mXMLHtmlParserOptions, "NOERROR", INT2NUM(HTML_PARSE_NOERROR)); 
+  /* 64: Suppress warning reports. */
+  rb_define_const(mXMLHtmlParserOptions, "NOWARNING", INT2NUM(HTML_PARSE_NOWARNING));
+  /* 128: Enable pedantic error reporting. */
+  rb_define_const(mXMLHtmlParserOptions, "PEDANTIC", INT2NUM(HTML_PARSE_PEDANTIC)); 
+  /* 256: Remove blank nodes. */
+  rb_define_const(mXMLHtmlParserOptions, "NOBLANKS", INT2NUM(HTML_PARSE_NOBLANKS)); 
+#if LIBXML_VERSION >= 20621
+  /* 2048: Forbid network access. */
+  rb_define_const(mXMLHtmlParserOptions, "NONET", INT2NUM(HTML_PARSE_NONET)); 
+  /* 65536: Compact small text nodes. */
+  rb_define_const(mXMLHtmlParserOptions, "COMPACT", INT2NUM(HTML_PARSE_COMPACT));
+#endif
+#if LIBXML_VERSION >= 20707
+  /* 8192:  Do not add implied html/body... elements */
+  rb_define_const(mXMLHtmlParserOptions, "NOIMPLIED", INT2NUM(HTML_PARSE_NOIMPLIED));
+#endif
+}

data/ext/libxml/ruby_xml_html_parser_options.h

@@ -1,10 +1,10 @@
-/* Please see the LICENSE file for copyright and distribution information */
-
-#ifndef __RXML_HTML_PARSER_OPTIONS__
-#define __RXML_HTML_PARSER_OPTIONS__
-
-extern VALUE mXMLHtmlParserOptions;
-
-void rxml_init_html_parser_options(void);
-
-#endif
+/* Please see the LICENSE file for copyright and distribution information */
+
+#ifndef __RXML_HTML_PARSER_OPTIONS__
+#define __RXML_HTML_PARSER_OPTIONS__
+
+extern VALUE mXMLHtmlParserOptions;
+
+void rxml_init_html_parser_options(void);
+
+#endif

data/ext/libxml/ruby_xml_input_cbg.c

@@ -1,188 +1,216 @@
-/* Author: Martin Povolny (xpovolny@fi.muni.cz) */
-
-#include "ruby_libxml.h"
-#include "ruby_xml_input_cbg.h"
-
-/* Document-class: LibXML::XML::InputCallbacks
- *
- * Support for adding custom scheme handlers. */
-
-static ic_scheme *first_scheme = 0;
-
-int ic_match(char const *filename)
-{
-  ic_scheme *scheme;
-
-  //fprintf( stderr, "ic_match: %s\n", filename );
-
-  scheme = first_scheme;
-  while (0 != scheme)
-  {
-    if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST scheme->scheme_name, scheme->name_len))
-    {
-      return 1;
-    }
-    scheme = scheme->next_scheme;
-  }
-  return 0;
-}
-
-void* ic_open(char const *filename)
-{
-  ic_doc_context *ic_doc;
-  ic_scheme *scheme;
-  VALUE res;
-
-  scheme = first_scheme;
-  while (0 != scheme)
-  {
-    if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST scheme->scheme_name, scheme->name_len))
-    {
-      ic_doc = (ic_doc_context*) malloc(sizeof(ic_doc_context));
-
-      res = rb_funcall(scheme->class, rb_intern("document_query"), 1,
-          rb_str_new2(filename));
-
-      ic_doc->buffer = strdup(StringValuePtr(res));
-
-      ic_doc->bpos = ic_doc->buffer;
-      ic_doc->remaining = (int)strlen(ic_doc->buffer);
-      return ic_doc;
-    }
-    scheme = scheme->next_scheme;
-  }
-  return 0;
-}
-
-int ic_read(void *context, char *buffer, int len)
-{
-  ic_doc_context *ic_doc;
-  int ret_len;
-  ic_doc = (ic_doc_context*) context;
-
-  if (len >= ic_doc->remaining)
-  {
-    ret_len = ic_doc->remaining;
-  }
-  else
-  {
-    ret_len = len;
-  }
-  ic_doc->remaining -= ret_len;
-  strncpy(buffer, ic_doc->bpos, ret_len);
-  ic_doc->bpos += ret_len;
-
-  return ret_len;
-}
-
-int ic_close(void *context)
-{
-  ruby_xfree(((ic_doc_context*) context)->buffer);
-  ruby_xfree(context);
-  return 1;
-}
-
-/*
- * call-seq:
- *    register
- *
- * Register a new set of I/O callback for handling parser input.
- */
-static VALUE input_callbacks_register_input_callbacks(VALUE self)
-{
-  xmlRegisterInputCallbacks(ic_match, ic_open, ic_read, ic_close);
-  return (Qtrue);
-}
-
-/*
- * call-seq:
- *    add_scheme
- *
- * No documentation available.
- */
-static VALUE input_callbacks_add_scheme(VALUE self, VALUE scheme_name,
-    VALUE class)
-{
-  ic_scheme *scheme;
-
-  Check_Type(scheme_name, T_STRING);
-
-  scheme = (ic_scheme*) malloc(sizeof(ic_scheme));
-  scheme->next_scheme = 0;
-  scheme->scheme_name = strdup(StringValuePtr(scheme_name)); /* TODO alloc, dealloc */
-  scheme->name_len = (int)strlen(scheme->scheme_name);
-  scheme->class = class; /* TODO alloc, dealloc */
-
-  //fprintf( stderr, "registered: %s, %d, %s\n", scheme->scheme_name, scheme->name_len, scheme->class );
-
-  if (0 == first_scheme)
-    first_scheme = scheme;
-  else
-  {
-    ic_scheme *pos;
-    pos = first_scheme;
-    while (0 != pos->next_scheme)
-      pos = pos->next_scheme;
-    pos->next_scheme = scheme;
-  }
-
-  return (Qtrue);
-}
-
-/*
- * call-seq:
- *    remove_scheme
- *
- * No documentation available.
- */
-static VALUE input_callbacks_remove_scheme(VALUE self, VALUE scheme_name)
-{
-  char *name;
-  ic_scheme *save_scheme, *scheme;
-
-  Check_Type(scheme_name, T_STRING);
-  name = StringValuePtr(scheme_name);
-
-  if (0 == first_scheme)
-    return Qfalse;
-
-  if (!strncmp(name, first_scheme->scheme_name, first_scheme->name_len))
-  {
-    save_scheme = first_scheme->next_scheme;
-
-    ruby_xfree(first_scheme->scheme_name);
-    ruby_xfree(first_scheme);
-
-    first_scheme = save_scheme;
-    return Qtrue;
-  }
-
-  scheme = first_scheme;
-  while (0 != scheme->next_scheme)
-  {
-    if (!strncmp(name, scheme->next_scheme->scheme_name,
-        scheme->next_scheme->name_len))
-    {
-      save_scheme = scheme->next_scheme->next_scheme;
-
-      ruby_xfree(scheme->next_scheme->scheme_name);
-      ruby_xfree(scheme->next_scheme);
-
-      scheme->next_scheme = save_scheme;
-      return Qtrue;
-    }
-    scheme = scheme->next_scheme;
-  }
-  return Qfalse;
-}
-
-void rxml_init_input_callbacks(void)
-{
-  VALUE cInputCallbacks;
-  cInputCallbacks = rb_define_class_under(mXML, "InputCallbacks", rb_cObject);
-
-  /* Class Methods */
-  rb_define_singleton_method(cInputCallbacks, "register", input_callbacks_register_input_callbacks, 0);
-  rb_define_singleton_method(cInputCallbacks, "add_scheme", input_callbacks_add_scheme, 2);
-  rb_define_singleton_method(cInputCallbacks, "remove_scheme", input_callbacks_remove_scheme, 1);
-}
+/* Author: Martin Povolny (xpovolny@fi.muni.cz) */
+
+#include "ruby_libxml.h"
+#include "ruby_xml_input_cbg.h"
+
+/* Document-class: LibXML::XML::InputCallbacks
+ *
+ * Support for adding custom scheme handlers. */
+
+static ic_scheme *first_scheme = 0;
+static int input_callbacks_registered = 0;
+
+static int ic_match(char const *filename);
+static void* ic_open(char const *filename);
+static int ic_read(void *context, char *buffer, int len);
+static int ic_close(void *context);
+
+static void input_callbacks_register_once(void)
+{
+  if (input_callbacks_registered)
+    return;
+
+  xmlRegisterInputCallbacks(ic_match, ic_open, ic_read, ic_close);
+  input_callbacks_registered = 1;
+}
+
+static char *ic_strdup(const char *string)
+{
+  size_t length = strlen(string) + 1;
+  char *copy = ruby_xmalloc(length);
+  memcpy(copy, string, length);
+  return copy;
+}
+
+static void ic_scheme_free(ic_scheme *scheme)
+{
+  rb_gc_unregister_address(&scheme->class);
+  ruby_xfree(scheme->scheme_name);
+  ruby_xfree(scheme);
+}
+
+static int ic_match(char const *filename)
+{
+  ic_scheme *scheme;
+
+  //fprintf( stderr, "ic_match: %s\n", filename );
+
+  scheme = first_scheme;
+  while (0 != scheme)
+  {
+    if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST scheme->scheme_name, scheme->name_len))
+    {
+      return 1;
+    }
+    scheme = scheme->next_scheme;
+  }
+  return 0;
+}
+
+static void* ic_open(char const *filename)
+{
+  ic_doc_context *ic_doc;
+  ic_scheme *scheme;
+  VALUE res;
+
+  scheme = first_scheme;
+  while (0 != scheme)
+  {
+    if (!xmlStrncasecmp(BAD_CAST filename, BAD_CAST scheme->scheme_name, scheme->name_len))
+    {
+      res = rb_funcall(scheme->class, rb_intern("document_query"), 1,
+          rb_str_new2(filename));
+
+      ic_doc = ALLOC(ic_doc_context);
+      ic_doc->buffer = ic_strdup(StringValueCStr(res));
+
+      ic_doc->bpos = ic_doc->buffer;
+      ic_doc->remaining = (int)strlen(ic_doc->buffer);
+      return ic_doc;
+    }
+    scheme = scheme->next_scheme;
+  }
+  return 0;
+}
+
+static int ic_read(void *context, char *buffer, int len)
+{
+  ic_doc_context *ic_doc;
+  int ret_len;
+  ic_doc = (ic_doc_context*) context;
+
+  if (len >= ic_doc->remaining)
+  {
+    ret_len = ic_doc->remaining;
+  }
+  else
+  {
+    ret_len = len;
+  }
+  ic_doc->remaining -= ret_len;
+  strncpy(buffer, ic_doc->bpos, ret_len);
+  ic_doc->bpos += ret_len;
+
+  return ret_len;
+}
+
+static int ic_close(void *context)
+{
+  ruby_xfree(((ic_doc_context*) context)->buffer);
+  ruby_xfree(context);
+  return 1;
+}
+
+/*
+ * call-seq:
+ *    register
+ *
+ * Register a new set of I/O callback for handling parser input.
+ */
+static VALUE input_callbacks_register_input_callbacks(VALUE self)
+{
+  input_callbacks_register_once();
+  return (Qtrue);
+}
+
+/*
+ * call-seq:
+ *    add_scheme
+ *
+ * No documentation available.
+ */
+static VALUE input_callbacks_add_scheme(VALUE self, VALUE scheme_name,
+    VALUE class)
+{
+  ic_scheme *scheme;
+
+  Check_Type(scheme_name, T_STRING);
+
+  input_callbacks_register_once();
+
+  scheme = ALLOC(ic_scheme);
+  scheme->next_scheme = 0;
+  scheme->scheme_name = ic_strdup(StringValueCStr(scheme_name));
+  scheme->name_len = (int)strlen(scheme->scheme_name);
+  scheme->class = class;
+  rb_gc_register_address(&scheme->class);
+
+  //fprintf( stderr, "registered: %s, %d, %s\n", scheme->scheme_name, scheme->name_len, scheme->class );
+
+  if (0 == first_scheme)
+    first_scheme = scheme;
+  else
+  {
+    ic_scheme *pos;
+    pos = first_scheme;
+    while (0 != pos->next_scheme)
+      pos = pos->next_scheme;
+    pos->next_scheme = scheme;
+  }
+
+  return (Qtrue);
+}
+
+/*
+ * call-seq:
+ *    remove_scheme
+ *
+ * No documentation available.
+ */
+static VALUE input_callbacks_remove_scheme(VALUE self, VALUE scheme_name)
+{
+  char *name;
+  ic_scheme *save_scheme, *scheme;
+
+  Check_Type(scheme_name, T_STRING);
+  name = StringValuePtr(scheme_name);
+
+  if (0 == first_scheme)
+    return Qfalse;
+
+  if (!strncmp(name, first_scheme->scheme_name, first_scheme->name_len))
+  {
+    save_scheme = first_scheme->next_scheme;
+
+    ic_scheme_free(first_scheme);
+    first_scheme = save_scheme;
+    return Qtrue;
+  }
+
+  scheme = first_scheme;
+  while (0 != scheme->next_scheme)
+  {
+    if (!strncmp(name, scheme->next_scheme->scheme_name,
+        scheme->next_scheme->name_len))
+    {
+      save_scheme = scheme->next_scheme->next_scheme;
+
+      ic_scheme_free(scheme->next_scheme);
+      scheme->next_scheme = save_scheme;
+      return Qtrue;
+    }
+    scheme = scheme->next_scheme;
+  }
+  return Qfalse;
+}
+
+void rxml_init_input_callbacks(void)
+{
+  VALUE cInputCallbacks;
+  cInputCallbacks = rb_define_class_under(mXML, "InputCallbacks", rb_cObject);
+
+  /* Class Methods */
+  rb_define_singleton_method(cInputCallbacks, "register", input_callbacks_register_input_callbacks, 0);
+  rb_define_singleton_method(cInputCallbacks, "add_scheme", input_callbacks_add_scheme, 2);
+  rb_define_singleton_method(cInputCallbacks, "remove_scheme", input_callbacks_remove_scheme, 1);
+}

data/ext/libxml/ruby_xml_input_cbg.h

@@ -1,20 +1,20 @@
-#ifndef _INPUT_CBG_
-#define _INPUT_CBG_
-
-void rxml_init_input_callbacks(void);
-
-typedef struct ic_doc_context {
-    char *buffer;
-    char *bpos;
-    int remaining;
-} ic_doc_context;
-
-typedef struct ic_scheme {
-    char *scheme_name;
-    VALUE class;
-    int name_len;
-
-    struct ic_scheme *next_scheme;
-} ic_scheme;
-
-#endif
+#ifndef _INPUT_CBG_
+#define _INPUT_CBG_
+
+void rxml_init_input_callbacks(void);
+
+typedef struct ic_doc_context {
+    char *buffer;
+    char *bpos;
+    int remaining;
+} ic_doc_context;
+
+typedef struct ic_scheme {
+    char *scheme_name;
+    VALUE class;
+    int name_len;
+
+    struct ic_scheme *next_scheme;
+} ic_scheme;
+
+#endif