nokogiri 1.9.1 → 1.15.3

data/ext/nokogiri/xml_sax_parser_context.c

@@ -1,18 +1,44 @@
-#include <xml_sax_parser_context.h>
+#include <nokogiri.h>
 
 VALUE cNokogiriXmlSaxParserContext ;
 
-static void deallocate(xmlParserCtxtPtr ctxt)
-{
-  NOKOGIRI_DEBUG_START(handler);
+static ID id_read;
 
+static void
+xml_sax_parser_context_free(void *data)
+{
+  xmlParserCtxtPtr ctxt = data;
   ctxt->sax = NULL;
-
   xmlFreeParserCtxt(ctxt);
+}
+
+/*
+ *  note that htmlParserCtxtPtr == xmlParserCtxtPtr and xmlFreeParserCtxt() == htmlFreeParserCtxt()
+ *  so we use this type for both XML::SAX::ParserContext and HTML::SAX::ParserContext
+ */
+static const rb_data_type_t xml_sax_parser_context_type = {
+  .wrap_struct_name = "Nokogiri::XML::SAX::ParserContext",
+  .function = {
+    .dfree = xml_sax_parser_context_free,
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
+xmlParserCtxtPtr
+noko_xml_sax_parser_context_unwrap(VALUE rb_context)
+{
+  xmlParserCtxtPtr c_context;
+  TypedData_Get_Struct(rb_context, xmlParserCtxt, &xml_sax_parser_context_type, c_context);
+  return c_context;
+}
 
-  NOKOGIRI_DEBUG_END(handler);
+VALUE
+noko_xml_sax_parser_context_wrap(VALUE klass, xmlParserCtxtPtr c_context)
+{
+  return TypedData_Wrap_Struct(klass, &xml_sax_parser_context_type, c_context);
 }
 
+
 /*
  * call-seq:
  *  parse_io(io, encoding)
@@ -22,19 +48,23 @@ static void deallocate(xmlParserCtxtPtr ctxt)
 static VALUE
 parse_io(VALUE klass, VALUE io, VALUE encoding)
 {
-    xmlParserCtxtPtr ctxt;
-    xmlCharEncoding enc = (xmlCharEncoding)NUM2INT(encoding);
-
-    ctxt = xmlCreateIOParserCtxt(NULL, NULL,
-				 (xmlInputReadCallback)io_read_callback,
-				 (xmlInputCloseCallback)io_close_callback,
-				 (void *)io, enc);
-    if (ctxt->sax) {
-	xmlFree(ctxt->sax);
-	ctxt->sax = NULL;
-    }
-
-    return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+  xmlParserCtxtPtr ctxt;
+  xmlCharEncoding enc = (xmlCharEncoding)NUM2INT(encoding);
+
+  if (!rb_respond_to(io, id_read)) {
+    rb_raise(rb_eTypeError, "argument expected to respond to :read");
+  }
+
+  ctxt = xmlCreateIOParserCtxt(NULL, NULL,
+                               (xmlInputReadCallback)noko_io_read,
+                               (xmlInputCloseCallback)noko_io_close,
+                               (void *)io, enc);
+  if (ctxt->sax) {
+    xmlFree(ctxt->sax);
+    ctxt->sax = NULL;
+  }
+
+  return noko_xml_sax_parser_context_wrap(klass, ctxt);
 }
 
 /*
@@ -43,10 +73,17 @@ parse_io(VALUE klass, VALUE io, VALUE encoding)
  *
  * Parse file given +filename+
  */
-static VALUE parse_file(VALUE klass, VALUE filename)
+static VALUE
+parse_file(VALUE klass, VALUE filename)
 {
   xmlParserCtxtPtr ctxt = xmlCreateFileParserCtxt(StringValueCStr(filename));
-  return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+
+  if (ctxt->sax) {
+    xmlFree(ctxt->sax);
+    ctxt->sax = NULL;
+  }
+
+  return noko_xml_sax_parser_context_wrap(klass, ctxt);
 }
 
 /*
@@ -58,41 +95,43 @@ static VALUE parse_file(VALUE klass, VALUE filename)
 static VALUE
 parse_memory(VALUE klass, VALUE data)
 {
-    xmlParserCtxtPtr ctxt;
+  xmlParserCtxtPtr ctxt;
+
+  Check_Type(data, T_STRING);
 
-    if (NIL_P(data))
-	rb_raise(rb_eArgError, "data cannot be nil");
-    if (!(int)RSTRING_LEN(data))
-	rb_raise(rb_eRuntimeError, "data cannot be empty");
+  if (!(int)RSTRING_LEN(data)) {
+    rb_raise(rb_eRuntimeError, "data cannot be empty");
+  }
 
-    ctxt = xmlCreateMemoryParserCtxt(StringValuePtr(data),
-				     (int)RSTRING_LEN(data));
-    if (ctxt->sax) {
-	xmlFree(ctxt->sax);
-	ctxt->sax = NULL;
-    }
+  ctxt = xmlCreateMemoryParserCtxt(StringValuePtr(data),
+                                   (int)RSTRING_LEN(data));
+  if (ctxt->sax) {
+    xmlFree(ctxt->sax);
+    ctxt->sax = NULL;
+  }
 
-    return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
+  return noko_xml_sax_parser_context_wrap(klass, ctxt);
 }
 
 static VALUE
 parse_doc(VALUE ctxt_val)
 {
-    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr)ctxt_val;
-    xmlParseDocument(ctxt);
-    return Qnil;
+  xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr)ctxt_val;
+  xmlParseDocument(ctxt);
+  return Qnil;
 }
 
 static VALUE
 parse_doc_finalize(VALUE ctxt_val)
 {
-    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr)ctxt_val;
+  xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr)ctxt_val;
 
-    if (NULL != ctxt->myDoc)
-	xmlFreeDoc(ctxt->myDoc);
+  if (NULL != ctxt->myDoc) {
+    xmlFreeDoc(ctxt->myDoc);
+  }
 
-    NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
-    return Qnil;
+  NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
+  return Qnil;
 }
 
 /*
@@ -104,25 +143,24 @@ parse_doc_finalize(VALUE ctxt_val)
 static VALUE
 parse_with(VALUE self, VALUE sax_handler)
 {
-    xmlParserCtxtPtr ctxt;
-    xmlSAXHandlerPtr sax;
+  xmlParserCtxtPtr ctxt;
+  xmlSAXHandlerPtr sax;
 
-    if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
-	rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
+  if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser)) {
+    rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
+  }
 
-    Data_Get_Struct(self, xmlParserCtxt, ctxt);
-    Data_Get_Struct(sax_handler, xmlSAXHandler, sax);
+  ctxt = noko_xml_sax_parser_context_unwrap(self);
+  sax = noko_sax_handler_unwrap(sax_handler);
 
-    /* Free the sax handler since we'll assign our own */
-    if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
-	xmlFree(ctxt->sax);
+  ctxt->sax = sax;
+  ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
 
-    ctxt->sax = sax;
-    ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
+  xmlSetStructuredErrorFunc(NULL, NULL);
 
-    rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
+  rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
 
-    return Qnil;
+  return Qnil;
 }
 
 /*
@@ -132,15 +170,16 @@ parse_with(VALUE self, VALUE sax_handler)
  * Should this parser replace entities?  &amp; will get converted to '&' if
  * set to true
  */
-static VALUE set_replace_entities(VALUE self, VALUE value)
+static VALUE
+set_replace_entities(VALUE self, VALUE value)
 {
-  xmlParserCtxtPtr ctxt;
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
-  if(Qfalse == value)
+  if (Qfalse == value) {
     ctxt->replaceEntities = 0;
-  else
+  } else {
     ctxt->replaceEntities = 1;
+  }
 
   return value;
 }
@@ -152,15 +191,16 @@ static VALUE set_replace_entities(VALUE self, VALUE value)
  * Should this parser replace entities?  &amp; will get converted to '&' if
  * set to true
  */
-static VALUE get_replace_entities(VALUE self)
+static VALUE
+get_replace_entities(VALUE self)
 {
-  xmlParserCtxtPtr ctxt;
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
-  if(0 == ctxt->replaceEntities)
+  if (0 == ctxt->replaceEntities) {
     return Qfalse;
-  else
+  } else {
     return Qtrue;
+  }
 }
 
 /*
@@ -168,16 +208,16 @@ static VALUE get_replace_entities(VALUE self)
  *
  * Get the current line the parser context is processing.
  */
-static VALUE line(VALUE self)
+static VALUE
+line(VALUE self)
 {
-  xmlParserCtxtPtr ctxt;
   xmlParserInputPtr io;
-
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
   io = ctxt->input;
-  if(io)
+  if (io) {
     return INT2NUM(io->line);
+  }
 
   return Qnil;
 }
@@ -187,16 +227,16 @@ static VALUE line(VALUE self)
  *
  * Get the current column the parser context is processing.
  */
-static VALUE column(VALUE self)
+static VALUE
+column(VALUE self)
 {
-  xmlParserCtxtPtr ctxt;
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
   xmlParserInputPtr io;
 
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
-
   io = ctxt->input;
-  if(io)
+  if (io) {
     return INT2NUM(io->col);
+  }
 
   return Qnil;
 }
@@ -208,15 +248,16 @@ static VALUE column(VALUE self)
  * Should this parser recover from structural errors? It will not stop processing
  * file on structural errors if set to true
  */
-static VALUE set_recovery(VALUE self, VALUE value)
+static VALUE
+set_recovery(VALUE self, VALUE value)
 {
-  xmlParserCtxtPtr ctxt;
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
-  if(value == Qfalse)
+  if (value == Qfalse) {
     ctxt->recovery = 0;
-  else
+  } else {
     ctxt->recovery = 1;
+  }
 
   return value;
 }
@@ -228,35 +269,36 @@ static VALUE set_recovery(VALUE self, VALUE value)
  * Should this parser recover from structural errors? It will not stop processing
  * file on structural errors if set to true
  */
-static VALUE get_recovery(VALUE self)
+static VALUE
+get_recovery(VALUE self)
 {
-  xmlParserCtxtPtr ctxt;
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  xmlParserCtxtPtr ctxt = noko_xml_sax_parser_context_unwrap(self);
 
-  if(ctxt->recovery == 0)
+  if (ctxt->recovery == 0) {
     return Qfalse;
-  else
+  } else {
     return Qtrue;
+  }
 }
 
-void init_xml_sax_parser_context()
+void
+noko_init_xml_sax_parser_context(void)
 {
-  VALUE nokogiri  = rb_define_module("Nokogiri");
-  VALUE xml       = rb_define_module_under(nokogiri, "XML");
-  VALUE sax       = rb_define_module_under(xml, "SAX");
-  VALUE klass     = rb_define_class_under(sax, "ParserContext", rb_cObject);
-
-  cNokogiriXmlSaxParserContext = klass;
-
-  rb_define_singleton_method(klass, "io", parse_io, 2);
-  rb_define_singleton_method(klass, "memory", parse_memory, 1);
-  rb_define_singleton_method(klass, "file", parse_file, 1);
-
-  rb_define_method(klass, "parse_with", parse_with, 1);
-  rb_define_method(klass, "replace_entities=", set_replace_entities, 1);
-  rb_define_method(klass, "replace_entities", get_replace_entities, 0);
-  rb_define_method(klass, "recovery=", set_recovery, 1);
-  rb_define_method(klass, "recovery", get_recovery, 0);
-  rb_define_method(klass, "line", line, 0);
-  rb_define_method(klass, "column", column, 0);
+  cNokogiriXmlSaxParserContext = rb_define_class_under(mNokogiriXmlSax, "ParserContext", rb_cObject);
+
+  rb_undef_alloc_func(cNokogiriXmlSaxParserContext);
+
+  rb_define_singleton_method(cNokogiriXmlSaxParserContext, "io", parse_io, 2);
+  rb_define_singleton_method(cNokogiriXmlSaxParserContext, "memory", parse_memory, 1);
+  rb_define_singleton_method(cNokogiriXmlSaxParserContext, "file", parse_file, 1);
+
+  rb_define_method(cNokogiriXmlSaxParserContext, "parse_with", parse_with, 1);
+  rb_define_method(cNokogiriXmlSaxParserContext, "replace_entities=", set_replace_entities, 1);
+  rb_define_method(cNokogiriXmlSaxParserContext, "replace_entities", get_replace_entities, 0);
+  rb_define_method(cNokogiriXmlSaxParserContext, "recovery=", set_recovery, 1);
+  rb_define_method(cNokogiriXmlSaxParserContext, "recovery", get_recovery, 0);
+  rb_define_method(cNokogiriXmlSaxParserContext, "line", line, 0);
+  rb_define_method(cNokogiriXmlSaxParserContext, "column", column, 0);
+
+  id_read = rb_intern("read");
 }

data/ext/nokogiri/xml_sax_push_parser.c

@@ -1,18 +1,37 @@
-#include <xml_sax_push_parser.h>
+#include <nokogiri.h>
 
-static void deallocate(xmlParserCtxtPtr ctx)
+VALUE cNokogiriXmlSaxPushParser ;
+
+static void
+xml_sax_push_parser_free(void *data)
 {
-  NOKOGIRI_DEBUG_START(ctx);
+  xmlParserCtxtPtr ctx = data;
   if (ctx != NULL) {
     NOKOGIRI_SAX_TUPLE_DESTROY(ctx->userData);
     xmlFreeParserCtxt(ctx);
   }
-  NOKOGIRI_DEBUG_END(ctx);
 }
 
-static VALUE allocate(VALUE klass)
+static const rb_data_type_t xml_sax_push_parser_type = {
+  .wrap_struct_name = "Nokogiri::XML::SAX::PushParser",
+  .function = {
+    .dfree = xml_sax_push_parser_free,
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
+static VALUE
+allocate(VALUE klass)
+{
+  return TypedData_Wrap_Struct(klass, &xml_sax_push_parser_type, NULL);
+}
+
+xmlParserCtxtPtr
+noko_xml_sax_push_parser_unwrap(VALUE rb_parser)
 {
-  return Data_Wrap_Struct(klass, NULL, deallocate, NULL);
+  xmlParserCtxtPtr c_parser;
+  TypedData_Get_Struct(rb_parser, xmlParserCtxt, &xml_sax_push_parser_type, c_parser);
+  return c_parser;
 }
 
 /*
@@ -21,20 +40,23 @@ static VALUE allocate(VALUE klass)
  *
  * Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
  */
-static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
+static VALUE
+native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
 {
   xmlParserCtxtPtr ctx;
-  const char * chunk  = NULL;
+  const char *chunk  = NULL;
   int size            = 0;
 
 
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   if (Qnil != _chunk) {
     chunk = StringValuePtr(_chunk);
     size = (int)RSTRING_LEN(_chunk);
   }
 
+  xmlSetStructuredErrorFunc(NULL, NULL);
+
   if (xmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
     if (!(ctx->options & XML_PARSE_RECOVER)) {
       xmlErrorPtr e = xmlCtxtGetLastError(ctx);
@@ -51,13 +73,14 @@ static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
  *
  * Initialize the push parser with +xml_sax+ using +filename+
  */
-static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
+static VALUE
+initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
 {
   xmlSAXHandlerPtr sax;
-  const char * filename = NULL;
+  const char *filename = NULL;
   xmlParserCtxtPtr ctx;
 
-  Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
+  sax = noko_sax_handler_unwrap(_xml_sax);
 
   if (_filename != Qnil) { filename = StringValueCStr(_filename); }
 
@@ -79,18 +102,22 @@ static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
   return self;
 }
 
-static VALUE get_options(VALUE self)
+static VALUE
+get_options(VALUE self)
 {
   xmlParserCtxtPtr ctx;
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   return INT2NUM(ctx->options);
 }
 
-static VALUE set_options(VALUE self, VALUE options)
+static VALUE
+set_options(VALUE self, VALUE options)
 {
   xmlParserCtxtPtr ctx;
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   if (xmlCtxtUseOptions(ctx, (int)NUM2INT(options)) != 0) {
     rb_raise(rb_eRuntimeError, "Cannot set XML parser context options");
@@ -106,10 +133,12 @@ static VALUE set_options(VALUE self, VALUE options)
  * Should this parser replace entities?  &amp; will get converted to '&' if
  * set to true
  */
-static VALUE get_replace_entities(VALUE self)
+static VALUE
+get_replace_entities(VALUE self)
 {
   xmlParserCtxtPtr ctx;
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   if (0 == ctx->replaceEntities) {
     return Qfalse;
@@ -125,10 +154,12 @@ static VALUE get_replace_entities(VALUE self)
  * Should this parser replace entities?  &amp; will get converted to '&' if
  * set to true
  */
-static VALUE set_replace_entities(VALUE self, VALUE value)
+static VALUE
+set_replace_entities(VALUE self, VALUE value)
 {
   xmlParserCtxtPtr ctx;
-  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  ctx = noko_xml_sax_push_parser_unwrap(self);
 
   if (Qfalse == value) {
     ctx->replaceEntities = 0;
@@ -139,21 +170,18 @@ static VALUE set_replace_entities(VALUE self, VALUE value)
   return value;
 }
 
-VALUE cNokogiriXmlSaxPushParser ;
-void init_xml_sax_push_parser()
+void
+noko_init_xml_sax_push_parser(void)
 {
-  VALUE nokogiri = rb_define_module("Nokogiri");
-  VALUE xml = rb_define_module_under(nokogiri, "XML");
-  VALUE sax = rb_define_module_under(xml, "SAX");
-  VALUE klass = rb_define_class_under(sax, "PushParser", rb_cObject);
-
-  cNokogiriXmlSaxPushParser = klass;
-
-  rb_define_alloc_func(klass, allocate);
-  rb_define_private_method(klass, "initialize_native", initialize_native, 2);
-  rb_define_private_method(klass, "native_write", native_write, 2);
-  rb_define_method(klass, "options", get_options, 0);
-  rb_define_method(klass, "options=", set_options, 1);
-  rb_define_method(klass, "replace_entities", get_replace_entities, 0);
-  rb_define_method(klass, "replace_entities=", set_replace_entities, 1);
+  cNokogiriXmlSaxPushParser = rb_define_class_under(mNokogiriXmlSax, "PushParser", rb_cObject);
+
+  rb_define_alloc_func(cNokogiriXmlSaxPushParser, allocate);
+
+  rb_define_method(cNokogiriXmlSaxPushParser, "options", get_options, 0);
+  rb_define_method(cNokogiriXmlSaxPushParser, "options=", set_options, 1);
+  rb_define_method(cNokogiriXmlSaxPushParser, "replace_entities", get_replace_entities, 0);
+  rb_define_method(cNokogiriXmlSaxPushParser, "replace_entities=", set_replace_entities, 1);
+
+  rb_define_private_method(cNokogiriXmlSaxPushParser, "initialize_native", initialize_native, 2);
+  rb_define_private_method(cNokogiriXmlSaxPushParser, "native_write", native_write, 2);
 }