nokogiri 1.13.0-x64-mingw-ucrt

data/ext/nokogiri/xml_relax_ng.c

@@ -0,0 +1,185 @@
+#include <nokogiri.h>
+
+VALUE cNokogiriXmlRelaxNG;
+
+static void
+dealloc(xmlRelaxNGPtr schema)
+{
+  NOKOGIRI_DEBUG_START(schema);
+  xmlRelaxNGFree(schema);
+  NOKOGIRI_DEBUG_END(schema);
+}
+
+/*
+ * call-seq:
+ *  validate_document(document)
+ *
+ * Validate a Nokogiri::XML::Document against this RelaxNG schema.
+ */
+static VALUE
+validate_document(VALUE self, VALUE document)
+{
+  xmlDocPtr doc;
+  xmlRelaxNGPtr schema;
+  VALUE errors;
+  xmlRelaxNGValidCtxtPtr valid_ctxt;
+
+  Data_Get_Struct(self, xmlRelaxNG, schema);
+  Data_Get_Struct(document, xmlDoc, doc);
+
+  errors = rb_ary_new();
+
+  valid_ctxt = xmlRelaxNGNewValidCtxt(schema);
+
+  if (NULL == valid_ctxt) {
+    /* we have a problem */
+    rb_raise(rb_eRuntimeError, "Could not create a validation context");
+  }
+
+#ifdef HAVE_XMLRELAXNGSETVALIDSTRUCTUREDERRORS
+  xmlRelaxNGSetValidStructuredErrors(
+    valid_ctxt,
+    Nokogiri_error_array_pusher,
+    (void *)errors
+  );
+#endif
+
+  xmlRelaxNGValidateDoc(valid_ctxt, doc);
+
+  xmlRelaxNGFreeValidCtxt(valid_ctxt);
+
+  return errors;
+}
+
+/*
+ * call-seq:
+ *  read_memory(string)
+ *
+ * Create a new RelaxNG from the contents of +string+
+ */
+static VALUE
+read_memory(int argc, VALUE *argv, VALUE klass)
+{
+  VALUE content;
+  VALUE parse_options;
+  xmlRelaxNGParserCtxtPtr ctx;
+  xmlRelaxNGPtr schema;
+  VALUE errors;
+  VALUE rb_schema;
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &content, &parse_options);
+  if (scanned_args == 1) {
+    parse_options = rb_const_get_at(rb_const_get_at(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
+
+  ctx = xmlRelaxNGNewMemParserCtxt((const char *)StringValuePtr(content), (int)RSTRING_LEN(content));
+
+  errors = rb_ary_new();
+  xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
+
+#ifdef HAVE_XMLRELAXNGSETPARSERSTRUCTUREDERRORS
+  xmlRelaxNGSetParserStructuredErrors(
+    ctx,
+    Nokogiri_error_array_pusher,
+    (void *)errors
+  );
+#endif
+
+  schema = xmlRelaxNGParse(ctx);
+
+  xmlSetStructuredErrorFunc(NULL, NULL);
+  xmlRelaxNGFreeParserCtxt(ctx);
+
+  if (NULL == schema) {
+    xmlErrorPtr error = xmlGetLastError();
+    if (error) {
+      Nokogiri_error_raise(NULL, error);
+    } else {
+      rb_raise(rb_eRuntimeError, "Could not parse document");
+    }
+
+    return Qnil;
+  }
+
+  rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
+  rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
+
+  return rb_schema;
+}
+
+/*
+ * call-seq:
+ *  from_document(doc)
+ *
+ * Create a new RelaxNG schema from the Nokogiri::XML::Document +doc+
+ */
+static VALUE
+from_document(int argc, VALUE *argv, VALUE klass)
+{
+  VALUE document;
+  VALUE parse_options;
+  xmlDocPtr doc;
+  xmlRelaxNGParserCtxtPtr ctx;
+  xmlRelaxNGPtr schema;
+  VALUE errors;
+  VALUE rb_schema;
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &document, &parse_options);
+
+  Data_Get_Struct(document, xmlDoc, doc);
+  doc = doc->doc; /* In case someone passes us a node. ugh. */
+
+  if (scanned_args == 1) {
+    parse_options = rb_const_get_at(rb_const_get_at(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
+
+  ctx = xmlRelaxNGNewDocParserCtxt(doc);
+
+  errors = rb_ary_new();
+  xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
+
+#ifdef HAVE_XMLRELAXNGSETPARSERSTRUCTUREDERRORS
+  xmlRelaxNGSetParserStructuredErrors(
+    ctx,
+    Nokogiri_error_array_pusher,
+    (void *)errors
+  );
+#endif
+
+  schema = xmlRelaxNGParse(ctx);
+
+  xmlSetStructuredErrorFunc(NULL, NULL);
+  xmlRelaxNGFreeParserCtxt(ctx);
+
+  if (NULL == schema) {
+    xmlErrorPtr error = xmlGetLastError();
+    if (error) {
+      Nokogiri_error_raise(NULL, error);
+    } else {
+      rb_raise(rb_eRuntimeError, "Could not parse document");
+    }
+
+    return Qnil;
+  }
+
+  rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
+  rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
+
+  return rb_schema;
+}
+
+void
+noko_init_xml_relax_ng()
+{
+  assert(cNokogiriXmlSchema);
+  cNokogiriXmlRelaxNG = rb_define_class_under(mNokogiriXml, "RelaxNG", cNokogiriXmlSchema);
+
+  rb_define_singleton_method(cNokogiriXmlRelaxNG, "read_memory", read_memory, -1);
+  rb_define_singleton_method(cNokogiriXmlRelaxNG, "from_document", from_document, -1);
+
+  rb_define_private_method(cNokogiriXmlRelaxNG, "validate_document", validate_document, 1);
+}

data/ext/nokogiri/xml_sax_parser.c

@@ -0,0 +1,310 @@
+#include <nokogiri.h>
+
+VALUE cNokogiriXmlSaxParser ;
+
+static ID id_start_document, id_end_document, id_start_element, id_end_element;
+static ID id_start_element_namespace, id_end_element_namespace;
+static ID id_comment, id_characters, id_xmldecl, id_error, id_warning;
+static ID id_cdata_block;
+static ID id_processing_instruction;
+
+static void
+start_document(void *ctx)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+
+  xmlParserCtxtPtr ctxt = NOKOGIRI_SAX_CTXT(ctx);
+
+  if (NULL != ctxt && ctxt->html != 1) {
+    if (ctxt->standalone != -1) { /* -1 means there was no declaration */
+      VALUE encoding = Qnil ;
+      VALUE standalone = Qnil;
+      VALUE version;
+      if (ctxt->encoding) {
+        encoding = NOKOGIRI_STR_NEW2(ctxt->encoding) ;
+      } else if (ctxt->input && ctxt->input->encoding) {
+        encoding = NOKOGIRI_STR_NEW2(ctxt->input->encoding) ;
+      }
+
+      version = ctxt->version ? NOKOGIRI_STR_NEW2(ctxt->version) : Qnil;
+
+      switch (ctxt->standalone) {
+        case 0:
+          standalone = NOKOGIRI_STR_NEW2("no");
+          break;
+        case 1:
+          standalone = NOKOGIRI_STR_NEW2("yes");
+          break;
+      }
+
+      rb_funcall(doc, id_xmldecl, 3, version, encoding, standalone);
+    }
+  }
+
+  rb_funcall(doc, id_start_document, 0);
+}
+
+static void
+end_document(void *ctx)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  rb_funcall(doc, id_end_document, 0);
+}
+
+static void
+start_element(void *ctx, const xmlChar *name, const xmlChar **atts)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  VALUE attributes = rb_ary_new();
+  const xmlChar *attr;
+  int i = 0;
+  if (atts) {
+    while ((attr = atts[i]) != NULL) {
+      const xmlChar *val = atts[i + 1];
+      VALUE value = val != NULL ? NOKOGIRI_STR_NEW2(val) : Qnil;
+      rb_ary_push(attributes, rb_ary_new3(2, NOKOGIRI_STR_NEW2(attr), value));
+      i += 2;
+    }
+  }
+
+  rb_funcall(doc,
+             id_start_element,
+             2,
+             NOKOGIRI_STR_NEW2(name),
+             attributes
+            );
+}
+
+static void
+end_element(void *ctx, const xmlChar *name)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  rb_funcall(doc, id_end_element, 1, NOKOGIRI_STR_NEW2(name));
+}
+
+static VALUE
+attributes_as_array(int attributes_len, const xmlChar **c_attributes)
+{
+  VALUE rb_array = rb_ary_new2((long)attributes_len);
+  VALUE cNokogiriXmlSaxParserAttribute;
+
+  cNokogiriXmlSaxParserAttribute = rb_const_get_at(cNokogiriXmlSaxParser, rb_intern("Attribute"));
+  if (c_attributes) {
+    /* Each attribute is an array of [localname, prefix, URI, value, end] */
+    int i;
+    for (i = 0; i < attributes_len * 5; i += 5) {
+      VALUE rb_constructor_args[4], rb_attribute;
+
+      rb_constructor_args[0] = RBSTR_OR_QNIL(c_attributes[i + 0]); /* localname */
+      rb_constructor_args[1] = RBSTR_OR_QNIL(c_attributes[i + 1]); /* prefix */
+      rb_constructor_args[2] = RBSTR_OR_QNIL(c_attributes[i + 2]); /* URI */
+
+      /* value */
+      rb_constructor_args[3] = NOKOGIRI_STR_NEW((const char *)c_attributes[i + 3],
+                               (c_attributes[i + 4] - c_attributes[i + 3]));
+
+      rb_attribute = rb_class_new_instance(4, rb_constructor_args, cNokogiriXmlSaxParserAttribute);
+      rb_ary_push(rb_array, rb_attribute);
+    }
+  }
+
+  return rb_array;
+}
+
+static void
+start_element_ns(
+  void *ctx,
+  const xmlChar *localname,
+  const xmlChar *prefix,
+  const xmlChar *uri,
+  int nb_namespaces,
+  const xmlChar **namespaces,
+  int nb_attributes,
+  int nb_defaulted,
+  const xmlChar **attributes)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+
+  VALUE attribute_ary = attributes_as_array(nb_attributes, attributes);
+
+  VALUE ns_list = rb_ary_new2((long)nb_namespaces);
+
+  if (namespaces) {
+    int i;
+    for (i = 0; i < nb_namespaces * 2; i += 2) {
+      rb_ary_push(ns_list,
+                  rb_ary_new3((long)2,
+                              RBSTR_OR_QNIL(namespaces[i + 0]),
+                              RBSTR_OR_QNIL(namespaces[i + 1])
+                             )
+                 );
+    }
+  }
+
+  rb_funcall(doc,
+             id_start_element_namespace,
+             5,
+             NOKOGIRI_STR_NEW2(localname),
+             attribute_ary,
+             RBSTR_OR_QNIL(prefix),
+             RBSTR_OR_QNIL(uri),
+             ns_list
+            );
+}
+
+/**
+ * end_element_ns was borrowed heavily from libxml-ruby.
+ */
+static void
+end_element_ns(
+  void *ctx,
+  const xmlChar *localname,
+  const xmlChar *prefix,
+  const xmlChar *uri)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+
+  rb_funcall(doc, id_end_element_namespace, 3,
+             NOKOGIRI_STR_NEW2(localname),
+             RBSTR_OR_QNIL(prefix),
+             RBSTR_OR_QNIL(uri)
+            );
+}
+
+static void
+characters_func(void *ctx, const xmlChar *ch, int len)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  VALUE str = NOKOGIRI_STR_NEW(ch, len);
+  rb_funcall(doc, id_characters, 1, str);
+}
+
+static void
+comment_func(void *ctx, const xmlChar *value)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  VALUE str = NOKOGIRI_STR_NEW2(value);
+  rb_funcall(doc, id_comment, 1, str);
+}
+
+static void
+warning_func(void *ctx, const char *msg, ...)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  char *message;
+  VALUE ruby_message;
+
+  va_list args;
+  va_start(args, msg);
+  vasprintf(&message, msg, args);
+  va_end(args);
+
+  ruby_message = NOKOGIRI_STR_NEW2(message);
+  free(message);
+  rb_funcall(doc, id_warning, 1, ruby_message);
+}
+
+static void
+error_func(void *ctx, const char *msg, ...)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  char *message;
+  VALUE ruby_message;
+
+  va_list args;
+  va_start(args, msg);
+  vasprintf(&message, msg, args);
+  va_end(args);
+
+  ruby_message = NOKOGIRI_STR_NEW2(message);
+  free(message);
+  rb_funcall(doc, id_error, 1, ruby_message);
+}
+
+static void
+cdata_block(void *ctx, const xmlChar *value, int len)
+{
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  VALUE string = NOKOGIRI_STR_NEW(value, len);
+  rb_funcall(doc, id_cdata_block, 1, string);
+}
+
+static void
+processing_instruction(void *ctx, const xmlChar *name, const xmlChar *content)
+{
+  VALUE rb_content;
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+
+  rb_content = content ? NOKOGIRI_STR_NEW2(content) : Qnil;
+
+  rb_funcall(doc,
+             id_processing_instruction,
+             2,
+             NOKOGIRI_STR_NEW2(name),
+             rb_content
+            );
+}
+
+static void
+deallocate(xmlSAXHandlerPtr handler)
+{
+  NOKOGIRI_DEBUG_START(handler);
+  free(handler);
+  NOKOGIRI_DEBUG_END(handler);
+}
+
+static VALUE
+allocate(VALUE klass)
+{
+  xmlSAXHandlerPtr handler = calloc((size_t)1, sizeof(xmlSAXHandler));
+
+  handler->startDocument = start_document;
+  handler->endDocument = end_document;
+  handler->startElement = start_element;
+  handler->endElement = end_element;
+  handler->startElementNs = start_element_ns;
+  handler->endElementNs = end_element_ns;
+  handler->characters = characters_func;
+  handler->comment = comment_func;
+  handler->warning = warning_func;
+  handler->error = error_func;
+  handler->cdataBlock = cdata_block;
+  handler->processingInstruction = processing_instruction;
+  handler->initialized = XML_SAX2_MAGIC;
+
+  return Data_Wrap_Struct(klass, NULL, deallocate, handler);
+}
+
+void
+noko_init_xml_sax_parser()
+{
+  cNokogiriXmlSaxParser = rb_define_class_under(mNokogiriXmlSax, "Parser", rb_cObject);
+
+  rb_define_alloc_func(cNokogiriXmlSaxParser, allocate);
+
+  id_start_document = rb_intern("start_document");
+  id_end_document = rb_intern("end_document");
+  id_start_element = rb_intern("start_element");
+  id_end_element = rb_intern("end_element");
+  id_comment = rb_intern("comment");
+  id_characters = rb_intern("characters");
+  id_xmldecl = rb_intern("xmldecl");
+  id_error = rb_intern("error");
+  id_warning = rb_intern("warning");
+  id_cdata_block = rb_intern("cdata_block");
+  id_start_element_namespace = rb_intern("start_element_namespace");
+  id_end_element_namespace = rb_intern("end_element_namespace");
+  id_processing_instruction = rb_intern("processing_instruction");
+}