nokogiri 1.6.0 → 1.13.2

data/ext/nokogiri/xml_sax_parser_context.c

@@ -1,16 +1,17 @@
-#include <xml_sax_parser_context.h>
+#include <nokogiri.h>
 
 VALUE cNokogiriXmlSaxParserContext ;
 
-static void deallocate(xmlParserCtxtPtr ctxt)
+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);
 }
 
 /*
@@ -22,19 +23,19 @@ 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);
+
+  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 Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
 }
 
 /*
@@ -43,9 +44,10 @@ 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(StringValuePtr(filename));
+  xmlParserCtxtPtr ctxt = xmlCreateFileParserCtxt(StringValueCStr(filename));
   return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
 }
 
@@ -58,41 +60,44 @@ static VALUE parse_file(VALUE klass, VALUE filename)
 static VALUE
 parse_memory(VALUE klass, VALUE data)
 {
-    xmlParserCtxtPtr ctxt;
+  xmlParserCtxtPtr ctxt;
 
-    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 (NIL_P(data)) {
+    rb_raise(rb_eArgError, "data cannot be nil");
+  }
+  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 Data_Wrap_Struct(klass, NULL, deallocate, 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 +109,29 @@ 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);
 
-    Data_Get_Struct(self, xmlParserCtxt, ctxt);
-    Data_Get_Struct(sax_handler, xmlSAXHandler, sax);
+  /* Free the sax handler since we'll assign our own */
+  if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler) {
+    xmlFree(ctxt->sax);
+  }
 
-    /* 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 +141,17 @@ 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);
 
-  if(Qfalse == value)
+  if (Qfalse == value) {
     ctxt->replaceEntities = 0;
-  else
+  } else {
     ctxt->replaceEntities = 1;
+  }
 
   return value;
 }
@@ -152,15 +163,17 @@ 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);
 
-  if(0 == ctxt->replaceEntities)
+  if (0 == ctxt->replaceEntities) {
     return Qfalse;
-  else
+  } else {
     return Qtrue;
+  }
 }
 
 /*
@@ -168,7 +181,8 @@ 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;
@@ -176,8 +190,9 @@ static VALUE line(VALUE self)
   Data_Get_Struct(self, xmlParserCtxt, ctxt);
 
   io = ctxt->input;
-  if(io)
+  if (io) {
     return INT2NUM(io->line);
+  }
 
   return Qnil;
 }
@@ -187,7 +202,8 @@ 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;
   xmlParserInputPtr io;
@@ -195,28 +211,71 @@ static VALUE column(VALUE self)
   Data_Get_Struct(self, xmlParserCtxt, ctxt);
 
   io = ctxt->input;
-  if(io)
+  if (io) {
     return INT2NUM(io->col);
+  }
 
   return Qnil;
 }
 
-void init_xml_sax_parser_context()
+/*
+ * call-seq:
+ *  recovery=(boolean)
+ *
+ * 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)
+{
+  xmlParserCtxtPtr ctxt;
+  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+
+  if (value == Qfalse) {
+    ctxt->recovery = 0;
+  } else {
+    ctxt->recovery = 1;
+  }
+
+  return value;
+}
+
+/*
+ * call-seq:
+ *  recovery
+ *
+ * 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)
 {
-  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, "line", line, 0);
-  rb_define_method(klass, "column", column, 0);
+  xmlParserCtxtPtr ctxt;
+  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+
+  if (ctxt->recovery == 0) {
+    return Qfalse;
+  } else {
+    return Qtrue;
+  }
+}
+
+void
+noko_init_xml_sax_parser_context()
+{
+  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);
 }

data/ext/nokogiri/xml_sax_push_parser.c

@@ -1,16 +1,20 @@
-#include <xml_sax_push_parser.h>
+#include <nokogiri.h>
 
-static void deallocate(xmlParserCtxtPtr ctx)
+VALUE cNokogiriXmlSaxPushParser ;
+
+static void
+deallocate(xmlParserCtxtPtr ctx)
 {
   NOKOGIRI_DEBUG_START(ctx);
-  if(ctx != NULL) {
+  if (ctx != NULL) {
     NOKOGIRI_SAX_TUPLE_DESTROY(ctx->userData);
     xmlFreeParserCtxt(ctx);
   }
   NOKOGIRI_DEBUG_END(ctx);
 }
 
-static VALUE allocate(VALUE klass)
+static VALUE
+allocate(VALUE klass)
 {
   return Data_Wrap_Struct(klass, NULL, deallocate, NULL);
 }
@@ -21,21 +25,24 @@ 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);
 
-  if(Qnil != _chunk) {
+  if (Qnil != _chunk) {
     chunk = StringValuePtr(_chunk);
     size = (int)RSTRING_LEN(_chunk);
   }
 
-  if(xmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
+  xmlSetStructuredErrorFunc(NULL, NULL);
+
+  if (xmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
     if (!(ctx->options & XML_PARSE_RECOVER)) {
       xmlErrorPtr e = xmlCtxtGetLastError(ctx);
       Nokogiri_error_raise(NULL, e);
@@ -51,25 +58,27 @@ 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);
 
-  if(_filename != Qnil) filename = StringValuePtr(_filename);
+  if (_filename != Qnil) { filename = StringValueCStr(_filename); }
 
   ctx = xmlCreatePushParserCtxt(
-      sax,
-      NULL,
-      NULL,
-      0,
-      filename
-  );
-  if(ctx == NULL)
+          sax,
+          NULL,
+          NULL,
+          0,
+          filename
+        );
+  if (ctx == NULL) {
     rb_raise(rb_eRuntimeError, "Could not create a parser context");
+  }
 
   ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
 
@@ -78,7 +87,8 @@ 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);
@@ -86,30 +96,73 @@ static VALUE get_options(VALUE 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);
 
-  if (xmlCtxtUseOptions(ctx, (int)NUM2INT(options)) != 0)
+  if (xmlCtxtUseOptions(ctx, (int)NUM2INT(options)) != 0) {
     rb_raise(rb_eRuntimeError, "Cannot set XML parser context options");
+  }
 
   return Qnil;
 }
 
-VALUE cNokogiriXmlSaxPushParser ;
-void init_xml_sax_push_parser()
+/*
+ * call-seq:
+ *  replace_entities
+ *
+ * Should this parser replace entities?  &amp; will get converted to '&' if
+ * set to true
+ */
+static VALUE
+get_replace_entities(VALUE self)
+{
+  xmlParserCtxtPtr ctx;
+  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  if (0 == ctx->replaceEntities) {
+    return Qfalse;
+  } else {
+    return Qtrue;
+  }
+}
+
+/*
+ * call-seq:
+ *  replace_entities=(boolean)
+ *
+ * Should this parser replace entities?  &amp; will get converted to '&' if
+ * set to true
+ */
+static VALUE
+set_replace_entities(VALUE self, VALUE value)
+{
+  xmlParserCtxtPtr ctx;
+  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  if (Qfalse == value) {
+    ctx->replaceEntities = 0;
+  } else {
+    ctx->replaceEntities = 1;
+  }
+
+  return value;
+}
+
+void
+noko_init_xml_sax_push_parser()
 {
-  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);
+  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);
 }