nokogiri 1.11.0.rc1-x86-linux → 1.11.1-x86-linux

data/ext/nokogiri/html_document.c

@@ -1,6 +1,7 @@
 #include <html_document.h>
 
 static ID id_encoding_found;
+static ID id_to_s;
 
 /*
  * call-seq:
@@ -8,19 +9,20 @@ static ID id_encoding_found;
  *
  * Create a new document
  */
-static VALUE new(int argc, VALUE *argv, VALUE klass)
+static VALUE
+rb_html_document_s_new(int argc, VALUE *argv, VALUE klass)
 {
   VALUE uri, external_id, rest, rb_doc;
   htmlDocPtr doc;
 
   rb_scan_args(argc, argv, "0*", &rest);
-  uri         = rb_ary_entry(rest, (long)0);
+  uri = rb_ary_entry(rest, (long)0);
   external_id = rb_ary_entry(rest, (long)1);
 
   doc = htmlNewDoc(
-      RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
-      RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
-  );
+          RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
+          RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
+        );
   rb_doc = Nokogiri_wrap_xml_document(klass, doc);
   rb_obj_call_init(rb_doc, argc, argv);
   return rb_doc ;
@@ -33,60 +35,55 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
  * Read the HTML document from +io+ with given +url+, +encoding+,
  * and +options+.  See Nokogiri::HTML.parse
  */
-static VALUE read_io( VALUE klass,
-                      VALUE io,
-                      VALUE url,
-                      VALUE encoding,
-                      VALUE options )
+static VALUE
+rb_html_document_s_read_io(VALUE klass, VALUE rb_io, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
 {
-  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
-  VALUE error_list      = rb_ary_new();
-  VALUE document;
-  htmlDocPtr doc;
+  VALUE rb_doc;
+  VALUE rb_error_list = rb_ary_new();
+  htmlDocPtr c_doc;
+  const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
+  const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
+  int options = NUM2INT(rb_options);
+
+  xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
+
+  c_doc = htmlReadIO(io_read_callback, io_close_callback, (void *)rb_io, c_url, c_encoding, options);
 
-  xmlResetLastError();
-  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
-
-  doc = htmlReadIO(
-      io_read_callback,
-      io_close_callback,
-      (void *)io,
-      c_url,
-      c_enc,
-      (int)NUM2INT(options)
-  );
   xmlSetStructuredErrorFunc(NULL, NULL);
 
   /*
    * If EncodingFound has occurred in EncodingReader, make sure to do
    * a cleanup and propagate the error.
    */
-  if (rb_respond_to(io, id_encoding_found)) {
-    VALUE encoding_found = rb_funcall(io, id_encoding_found, 0);
+  if (rb_respond_to(rb_io, id_encoding_found)) {
+    VALUE encoding_found = rb_funcall(rb_io, id_encoding_found, 0);
     if (!NIL_P(encoding_found)) {
-      xmlFreeDoc(doc);
+      xmlFreeDoc(c_doc);
       rb_exc_raise(encoding_found);
     }
   }
 
-  if(doc == NULL) {
-    xmlErrorPtr error;
+  if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
+    VALUE rb_error ;
 
-    xmlFreeDoc(doc);
+    xmlFreeDoc(c_doc);
 
-    error = xmlGetLastError();
-    if(error)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+    rb_error = rb_ary_entry(rb_error_list, 0);
+    if (rb_error == Qnil) {
       rb_raise(rb_eRuntimeError, "Could not parse document");
+    } else {
+      VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
+      exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
+                                        exception_message);
+      rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
+    }
 
     return Qnil;
   }
 
-  document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_iv_set(document, "@errors", error_list);
-  return document;
+  rb_doc = Nokogiri_wrap_xml_document(klass, c_doc);
+  rb_iv_set(rb_doc, "@errors", rb_error_list);
+  return rb_doc;
 }
 
 /*
@@ -96,43 +93,45 @@ static VALUE read_io( VALUE klass,
  * Read the HTML document contained in +string+ with given +url+, +encoding+,
  * and +options+.  See Nokogiri::HTML.parse
  */
-static VALUE read_memory( VALUE klass,
-                          VALUE string,
-                          VALUE url,
-                          VALUE encoding,
-                          VALUE options )
+static VALUE
+rb_html_document_s_read_memory(VALUE klass, VALUE rb_html, VALUE rb_url, VALUE rb_encoding, VALUE rb_options)
 {
-  const char * c_buffer = StringValuePtr(string);
-  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
-  int len               = (int)RSTRING_LEN(string);
-  VALUE error_list      = rb_ary_new();
-  VALUE document;
-  htmlDocPtr doc;
+  VALUE rb_doc;
+  VALUE rb_error_list = rb_ary_new();
+  htmlDocPtr c_doc;
+  const char *c_buffer = StringValuePtr(rb_html);
+  const char *c_url = NIL_P(rb_url) ? NULL : StringValueCStr(rb_url);
+  const char *c_encoding = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
+  int html_len = (int)RSTRING_LEN(rb_html);
+  int options = NUM2INT(rb_options);
 
-  xmlResetLastError();
-  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
+  xmlSetStructuredErrorFunc((void *)rb_error_list, Nokogiri_error_array_pusher);
+
+  c_doc = htmlReadMemory(c_buffer, html_len, c_url, c_encoding, options);
 
-  doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
   xmlSetStructuredErrorFunc(NULL, NULL);
 
-  if(doc == NULL) {
-    xmlErrorPtr error;
+  if ((c_doc == NULL) || (!(options & XML_PARSE_RECOVER) && (RARRAY_LEN(rb_error_list) > 0))) {
+    VALUE rb_error ;
 
-    xmlFreeDoc(doc);
+    xmlFreeDoc(c_doc);
 
-    error = xmlGetLastError();
-    if(error)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+    rb_error = rb_ary_entry(rb_error_list, 0);
+    if (rb_error == Qnil) {
       rb_raise(rb_eRuntimeError, "Could not parse document");
+    } else {
+      VALUE exception_message = rb_funcall(rb_error, id_to_s, 0);
+      exception_message = rb_str_concat(rb_str_new2("Parser without recover option encountered error or warning: "),
+                                        exception_message);
+      rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
+    }
 
     return Qnil;
   }
 
-  document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_iv_set(document, "@errors", error_list);
-  return document;
+  rb_doc = Nokogiri_wrap_xml_document(klass, c_doc);
+  rb_iv_set(rb_doc, "@errors", rb_error_list);
+  return rb_doc;
 }
 
 /*
@@ -141,7 +140,8 @@ static VALUE read_memory( VALUE klass,
  *
  * The type for this document
  */
-static VALUE type(VALUE self)
+static VALUE
+rb_html_document_type(VALUE self)
 {
   htmlDocPtr doc;
   Data_Get_Struct(self, xmlDoc, doc);
@@ -149,22 +149,23 @@ static VALUE type(VALUE self)
 }
 
 VALUE cNokogiriHtmlDocument ;
-void init_html_document()
-{
-  VALUE nokogiri  = rb_define_module("Nokogiri");
-  VALUE html      = rb_define_module_under(nokogiri, "HTML");
-  VALUE xml       = rb_define_module_under(nokogiri, "XML");
-  VALUE node      = rb_define_class_under(xml, "Node", rb_cObject);
-  VALUE xml_doc   = rb_define_class_under(xml, "Document", node);
-  VALUE klass     = rb_define_class_under(html, "Document", xml_doc);
 
-  cNokogiriHtmlDocument = klass;
+void
+init_html_document()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE nokogiri_xml = rb_define_module_under(nokogiri, "XML");
+  VALUE nokogiri_xml_node = rb_define_class_under(nokogiri_xml, "Node", rb_cObject);
+  VALUE nokogiri_xml_document = rb_define_class_under(nokogiri_xml, "Document", nokogiri_xml_node);
+  VALUE nokogiri_html = rb_define_module_under(nokogiri, "HTML");
+  cNokogiriHtmlDocument = rb_define_class_under(nokogiri_html, "Document", nokogiri_xml_document);
 
-  rb_define_singleton_method(klass, "read_memory", read_memory, 4);
-  rb_define_singleton_method(klass, "read_io", read_io, 4);
-  rb_define_singleton_method(klass, "new", new, -1);
+  rb_define_singleton_method(cNokogiriHtmlDocument, "read_memory", rb_html_document_s_read_memory, 4);
+  rb_define_singleton_method(cNokogiriHtmlDocument, "read_io", rb_html_document_s_read_io, 4);
+  rb_define_singleton_method(cNokogiriHtmlDocument, "new", rb_html_document_s_new, -1);
 
-  rb_define_method(klass, "type", type, 0);
+  rb_define_method(cNokogiriHtmlDocument, "type", rb_html_document_type, 0);
 
   id_encoding_found = rb_intern("encoding_found");
+  id_to_s = rb_intern("to_s");
 }

data/ext/nokogiri/html_sax_parser_context.c

@@ -4,13 +4,13 @@ VALUE cNokogiriHtmlSaxParserContext ;
 
 static void deallocate(xmlParserCtxtPtr ctxt)
 {
-  NOKOGIRI_DEBUG_START(handler);
+  NOKOGIRI_DEBUG_START(ctxt);
 
   ctxt->sax = NULL;
 
   htmlFreeParserCtxt(ctxt);
 
-  NOKOGIRI_DEBUG_END(handler);
+  NOKOGIRI_DEBUG_END(ctxt);
 }
 
 static VALUE
@@ -92,6 +92,8 @@ parse_with(VALUE self, VALUE 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);
 
     return self;

data/ext/nokogiri/html_sax_push_parser.c

@@ -9,9 +9,10 @@
 static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
 {
   xmlParserCtxtPtr ctx;
-  const char * chunk  = NULL;
-  int size            = 0;
-
+  const char * chunk = NULL;
+  int size = 0;
+  int status = 0;
+  libxmlStructuredErrorHandlerState handler_state;
 
   Data_Get_Struct(self, xmlParserCtxt, ctx);
 
@@ -20,11 +21,16 @@ static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
     size = (int)RSTRING_LEN(_chunk);
   }
 
-  if(htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
-    if (!(ctx->options & XML_PARSE_RECOVER)) {
-      xmlErrorPtr e = xmlCtxtGetLastError(ctx);
-      Nokogiri_error_raise(NULL, e);
-    }
+  Nokogiri_structured_error_func_save_and_set(&handler_state, NULL, NULL);
+
+  status = htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0);
+
+  Nokogiri_structured_error_func_restore(&handler_state);
+
+  if ((status != 0) && !(ctx->options & XML_PARSE_RECOVER)) {
+    // TODO: there appear to be no tests for this block
+    xmlErrorPtr e = xmlCtxtGetLastError(ctx);
+    Nokogiri_error_raise(NULL, e);
   }
 
   return self;

data/ext/nokogiri/include/libexslt/exslt.h

@@ -0,0 +1,102 @@
+
+#ifndef __EXSLT_H__
+#define __EXSLT_H__
+
+#include <libxml/tree.h>
+#include <libxml/xpath.h>
+#include "exsltexports.h"
+#include <libexslt/exsltconfig.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+EXSLTPUBVAR const char *exsltLibraryVersion;
+EXSLTPUBVAR const int exsltLibexsltVersion;
+EXSLTPUBVAR const int exsltLibxsltVersion;
+EXSLTPUBVAR const int exsltLibxmlVersion;
+
+/**
+ * EXSLT_COMMON_NAMESPACE:
+ *
+ * Namespace for EXSLT common functions
+ */
+#define EXSLT_COMMON_NAMESPACE ((const xmlChar *) "http://exslt.org/common")
+/**
+ * EXSLT_CRYPTO_NAMESPACE:
+ *
+ * Namespace for EXSLT crypto functions
+ */
+#define EXSLT_CRYPTO_NAMESPACE ((const xmlChar *) "http://exslt.org/crypto")
+/**
+ * EXSLT_MATH_NAMESPACE:
+ *
+ * Namespace for EXSLT math functions
+ */
+#define EXSLT_MATH_NAMESPACE ((const xmlChar *) "http://exslt.org/math")
+/**
+ * EXSLT_SETS_NAMESPACE:
+ *
+ * Namespace for EXSLT set functions
+ */
+#define EXSLT_SETS_NAMESPACE ((const xmlChar *) "http://exslt.org/sets")
+/**
+ * EXSLT_FUNCTIONS_NAMESPACE:
+ *
+ * Namespace for EXSLT functions extension functions
+ */
+#define EXSLT_FUNCTIONS_NAMESPACE ((const xmlChar *) "http://exslt.org/functions")
+/**
+ * EXSLT_STRINGS_NAMESPACE:
+ *
+ * Namespace for EXSLT strings functions
+ */
+#define EXSLT_STRINGS_NAMESPACE ((const xmlChar *) "http://exslt.org/strings")
+/**
+ * EXSLT_DATE_NAMESPACE:
+ *
+ * Namespace for EXSLT date functions
+ */
+#define EXSLT_DATE_NAMESPACE ((const xmlChar *) "http://exslt.org/dates-and-times")
+/**
+ * EXSLT_DYNAMIC_NAMESPACE:
+ *
+ * Namespace for EXSLT dynamic functions
+ */
+#define EXSLT_DYNAMIC_NAMESPACE ((const xmlChar *) "http://exslt.org/dynamic")
+
+/**
+ * SAXON_NAMESPACE:
+ *
+ * Namespace for SAXON extensions functions
+ */
+#define SAXON_NAMESPACE ((const xmlChar *) "http://icl.com/saxon")
+
+EXSLTPUBFUN void EXSLTCALL exsltCommonRegister (void);
+#ifdef EXSLT_CRYPTO_ENABLED
+EXSLTPUBFUN void EXSLTCALL exsltCryptoRegister (void);
+#endif
+EXSLTPUBFUN void EXSLTCALL exsltMathRegister (void);
+EXSLTPUBFUN void EXSLTCALL exsltSetsRegister (void);
+EXSLTPUBFUN void EXSLTCALL exsltFuncRegister (void);
+EXSLTPUBFUN void EXSLTCALL exsltStrRegister (void);
+EXSLTPUBFUN void EXSLTCALL exsltDateRegister (void);
+EXSLTPUBFUN void EXSLTCALL exsltSaxonRegister (void);
+EXSLTPUBFUN void EXSLTCALL exsltDynRegister(void);
+
+EXSLTPUBFUN void EXSLTCALL exsltRegisterAll (void);
+
+EXSLTPUBFUN int EXSLTCALL exsltDateXpathCtxtRegister (xmlXPathContextPtr ctxt,
+                                                      const xmlChar *prefix);
+EXSLTPUBFUN int EXSLTCALL exsltMathXpathCtxtRegister (xmlXPathContextPtr ctxt,
+                                                      const xmlChar *prefix);
+EXSLTPUBFUN int EXSLTCALL exsltSetsXpathCtxtRegister (xmlXPathContextPtr ctxt,
+                                                      const xmlChar *prefix);
+EXSLTPUBFUN int EXSLTCALL exsltStrXpathCtxtRegister (xmlXPathContextPtr ctxt,
+                                                     const xmlChar *prefix);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* __EXSLT_H__ */
+

data/ext/nokogiri/include/libexslt/exsltconfig.h

@@ -0,0 +1,70 @@
+/*
+ * exsltconfig.h: compile-time version information for the EXSLT library
+ *
+ * See Copyright for the status of this software.
+ *
+ * daniel@veillard.com
+ */
+
+#ifndef __XML_EXSLTCONFIG_H__
+#define __XML_EXSLTCONFIG_H__
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/**
+ * LIBEXSLT_DOTTED_VERSION:
+ *
+ * the version string like "1.2.3"
+ */
+#define LIBEXSLT_DOTTED_VERSION "1.1.34"
+
+/**
+ * LIBEXSLT_VERSION:
+ *
+ * the version number: 1.2.3 value is 10203
+ */
+#define LIBEXSLT_VERSION 820
+
+/**
+ * LIBEXSLT_VERSION_STRING:
+ *
+ * the version number string, 1.2.3 value is "10203"
+ */
+#define LIBEXSLT_VERSION_STRING "820"
+
+/**
+ * LIBEXSLT_VERSION_EXTRA:
+ *
+ * extra version information, used to show a CVS compilation
+ */
+#define	LIBEXSLT_VERSION_EXTRA ""
+
+/**
+ * WITH_CRYPTO:
+ *
+ * Whether crypto support is configured into exslt
+ */
+#if 0
+#define EXSLT_CRYPTO_ENABLED
+#endif
+
+/**
+ * ATTRIBUTE_UNUSED:
+ *
+ * This macro is used to flag unused function parameters to GCC
+ */
+#ifdef __GNUC__
+#ifndef ATTRIBUTE_UNUSED
+#define ATTRIBUTE_UNUSED __attribute__((unused))
+#endif
+#else
+#define ATTRIBUTE_UNUSED
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* __XML_EXSLTCONFIG_H__ */