mini-pro-gem 0.0.1

data/nokogiri-1.19.4/ext/nokogiri/html4_sax_push_parser.c

@@ -0,0 +1,96 @@
+#include <nokogiri.h>
+
+VALUE cNokogiriHtml4SaxPushParser;
+
+/*
+ * Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
+ */
+static VALUE
+noko_html4_sax_push_parser__native_write(VALUE self, VALUE rb_chunk, VALUE rb_last_chunk)
+{
+  xmlParserCtxtPtr ctx;
+  const char *chunk = NULL;
+  int size = 0;
+  int status = 0;
+  libxmlStructuredErrorHandlerState handler_state;
+
+  ctx = noko_xml_sax_push_parser_unwrap(self);
+
+  if (Qnil != rb_chunk) {
+    chunk = StringValuePtr(rb_chunk);
+    size = (int)RSTRING_LEN(rb_chunk);
+  }
+
+  noko__structured_error_func_save_and_set(&handler_state, NULL, NULL);
+
+  status = htmlParseChunk(ctx, chunk, size, Qtrue == rb_last_chunk ? 1 : 0);
+
+  noko__structured_error_func_restore(&handler_state);
+
+  if ((status != 0) && !(xmlCtxtGetOptions(ctx) & XML_PARSE_RECOVER)) {
+    // TODO: there appear to be no tests for this block
+    xmlErrorConstPtr e = xmlCtxtGetLastError(ctx);
+    noko__error_raise(NULL, e);
+  }
+
+  return self;
+}
+
+/*
+ * Initialize the push parser with +xml_sax+ using +filename+
+ */
+static VALUE
+noko_html4_sax_push_parser__initialize_native(
+  VALUE self,
+  VALUE rb_xml_sax,
+  VALUE rb_filename,
+  VALUE encoding
+)
+{
+  htmlSAXHandlerPtr sax;
+  const char *filename = NULL;
+  htmlParserCtxtPtr ctx;
+  xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
+
+  sax = noko_xml_sax_parser_unwrap(rb_xml_sax);
+
+  if (rb_filename != Qnil) { filename = StringValueCStr(rb_filename); }
+
+  if (!NIL_P(encoding)) {
+    enc = xmlParseCharEncoding(StringValueCStr(encoding));
+    if (enc == XML_CHAR_ENCODING_ERROR) {
+      rb_raise(rb_eArgError, "Unsupported Encoding");
+    }
+  }
+
+  ctx = htmlCreatePushParserCtxt(
+          sax,
+          NULL,
+          NULL,
+          0,
+          filename,
+          enc
+        );
+  if (ctx == NULL) {
+    rb_raise(rb_eRuntimeError, "Could not create a parser context");
+  }
+
+  ctx->userData = ctx;
+  ctx->_private = (void *)rb_xml_sax;
+
+  DATA_PTR(self) = ctx;
+  return self;
+}
+
+void
+noko_init_html_sax_push_parser(void)
+{
+  assert(cNokogiriXmlSaxPushParser);
+  cNokogiriHtml4SaxPushParser =
+    rb_define_class_under(mNokogiriHtml4Sax, "PushParser", cNokogiriXmlSaxPushParser);
+
+  rb_define_private_method(cNokogiriHtml4SaxPushParser, "initialize_native",
+                           noko_html4_sax_push_parser__initialize_native, 3);
+  rb_define_private_method(cNokogiriHtml4SaxPushParser, "native_write",
+                           noko_html4_sax_push_parser__native_write, 2);
+}

data/nokogiri-1.19.4/ext/nokogiri/libxml2_polyfill.c

@@ -0,0 +1,114 @@
+#include <nokogiri.h>
+
+#ifndef HAVE_XMLCTXTSETOPTIONS
+/* based on libxml2-2.14.0-dev (1d8bd126) parser.c xmlCtxtSetInternalOptions */
+int
+xmlCtxtSetOptions(xmlParserCtxtPtr ctxt, int options)
+{
+  int keepMask = 0;
+  int allMask;
+
+  if (ctxt == NULL) {
+    return (-1);
+  }
+
+  /*
+   * XInclude options aren't handled by the parser.
+   *
+   * XML_PARSE_XINCLUDE
+   * XML_PARSE_NOXINCNODE
+   * XML_PARSE_NOBASEFIX
+   */
+  allMask = XML_PARSE_RECOVER |
+            XML_PARSE_NOENT |
+            XML_PARSE_DTDLOAD |
+            XML_PARSE_DTDATTR |
+            XML_PARSE_DTDVALID |
+            XML_PARSE_NOERROR |
+            XML_PARSE_NOWARNING |
+            XML_PARSE_PEDANTIC |
+            XML_PARSE_NOBLANKS |
+#ifdef LIBXML_SAX1_ENABLED
+            XML_PARSE_SAX1 |
+#endif
+            XML_PARSE_NONET |
+            XML_PARSE_NODICT |
+            XML_PARSE_NSCLEAN |
+            XML_PARSE_NOCDATA |
+            XML_PARSE_COMPACT |
+            XML_PARSE_OLD10 |
+            XML_PARSE_HUGE |
+            XML_PARSE_OLDSAX |
+            XML_PARSE_IGNORE_ENC |
+            XML_PARSE_BIG_LINES;
+
+  ctxt->options = (ctxt->options & keepMask) | (options & allMask);
+
+  /*
+   * For some options, struct members are historically the source
+   * of truth. The values are initalized from global variables and
+   * old code could also modify them directly. Several older API
+   * functions that don't take an options argument rely on these
+   * deprecated mechanisms.
+   *
+   * Once public access to struct members and the globals are
+   * disabled, we can use the options bitmask as source of
+   * truth, making all these struct members obsolete.
+   *
+   * The XML_DETECT_IDS flags is misnamed. It simply enables
+   * loading of the external subset.
+   */
+  ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
+  ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
+  ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
+  ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
+  ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
+  ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
+  ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
+  ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
+
+  /*
+   * Changing SAX callbacks is a bad idea. This should be fixed.
+   */
+  if (options & XML_PARSE_NOBLANKS) {
+    ctxt->sax->ignorableWhitespace = xmlSAX2IgnorableWhitespace;
+  }
+  if (options & XML_PARSE_NOCDATA) {
+    ctxt->sax->cdataBlock = NULL;
+  }
+  if (options & XML_PARSE_HUGE) {
+    if (ctxt->dict != NULL) {
+      xmlDictSetLimit(ctxt->dict, 0);
+    }
+  }
+
+  ctxt->linenumbers = 1;
+
+  return (options & ~allMask);
+}
+#endif
+
+#ifndef HAVE_XMLCTXTGETOPTIONS
+int
+xmlCtxtGetOptions(xmlParserCtxtPtr ctxt)
+{
+  return (ctxt->options);
+}
+#endif
+
+#ifndef HAVE_XMLSWITCHENCODINGNAME
+int
+xmlSwitchEncodingName(xmlParserCtxtPtr ctxt, const char *encoding)
+{
+  if (ctxt == NULL) {
+    return (-1);
+  }
+
+  xmlCharEncodingHandlerPtr handler = xmlFindCharEncodingHandler(encoding);
+  if (handler == NULL) {
+    return (-1);
+  }
+
+  return (xmlSwitchToEncoding(ctxt, handler));
+}
+#endif

data/nokogiri-1.19.4/ext/nokogiri/nokogiri.c

@@ -0,0 +1,297 @@
+#include <nokogiri.h>
+
+VALUE mNokogiri ;
+VALUE mNokogiriGumbo ;
+VALUE mNokogiriHtml4 ;
+VALUE mNokogiriHtml4Sax ;
+VALUE mNokogiriHtml5 ;
+VALUE mNokogiriXml ;
+VALUE mNokogiriXmlSax ;
+VALUE mNokogiriXmlXpath ;
+VALUE mNokogiriXslt ;
+
+VALUE cNokogiriSyntaxError;
+VALUE cNokogiriXmlCharacterData;
+VALUE cNokogiriXmlElement;
+VALUE cNokogiriXmlXpathSyntaxError;
+
+void noko_init_xml_attr(void);
+void noko_init_xml_attribute_decl(void);
+void noko_init_xml_cdata(void);
+void noko_init_xml_comment(void);
+void noko_init_xml_document(void);
+void noko_init_xml_document_fragment(void);
+void noko_init_xml_dtd(void);
+void noko_init_xml_element_content(void);
+void noko_init_xml_element_decl(void);
+void noko_init_xml_encoding_handler(void);
+void noko_init_xml_entity_decl(void);
+void noko_init_xml_entity_reference(void);
+void noko_init_xml_namespace(void);
+void noko_init_xml_node(void);
+void noko_init_xml_node_set(void);
+void noko_init_xml_processing_instruction(void);
+void noko_init_xml_reader(void);
+void noko_init_xml_relax_ng(void);
+void noko_init_xml_sax_parser(void);
+void noko_init_xml_sax_parser_context(void);
+void noko_init_xml_sax_push_parser(void);
+void noko_init_xml_schema(void);
+void noko_init_xml_syntax_error(void);
+void noko_init_xml_text(void);
+void noko_init_xml_xpath_context(void);
+void noko_init_xslt_stylesheet(void);
+void noko_init_html_document(void);
+void noko_init_html_element_description(void);
+void noko_init_html_entity_lookup(void);
+void noko_init_html_sax_parser_context(void);
+void noko_init_html_sax_push_parser(void);
+void noko_init_html4_sax_parser(void);
+void noko_init_gumbo(void);
+void noko_init_test_global_handlers(void);
+
+static ID id_read, id_write, id_external_encoding;
+
+
+static VALUE
+noko_io_read_check(VALUE val)
+{
+  VALUE *args = (VALUE *)val;
+  return rb_funcall(args[0], id_read, 1, args[1]);
+}
+
+
+static VALUE
+noko_io_read_failed(VALUE arg, VALUE exc)
+{
+  return Qundef;
+}
+
+
+int
+noko_io_read(void *io, char *c_buffer, int c_buffer_len)
+{
+  VALUE rb_io = (VALUE)io;
+  VALUE rb_read_string, rb_args[2];
+  size_t n_bytes_read, safe_len;
+
+  rb_args[0] = rb_io;
+  rb_args[1] = INT2NUM(c_buffer_len);
+
+  rb_read_string = rb_rescue(noko_io_read_check, (VALUE)rb_args, noko_io_read_failed, 0);
+
+  if (NIL_P(rb_read_string)) { return 0; }
+  if (rb_read_string == Qundef) { return -1; }
+  if (TYPE(rb_read_string) != T_STRING) { return -1; }
+
+  n_bytes_read = (size_t)RSTRING_LEN(rb_read_string);
+  safe_len = (n_bytes_read > (size_t)c_buffer_len) ? (size_t)c_buffer_len : n_bytes_read;
+  memcpy(c_buffer, StringValuePtr(rb_read_string), safe_len);
+
+  return (int)safe_len;
+}
+
+
+static VALUE
+noko_io_write_check(VALUE rb_args)
+{
+  VALUE rb_io = ((VALUE *)rb_args)[0];
+  VALUE rb_output = ((VALUE *)rb_args)[1];
+  return rb_funcall(rb_io, id_write, 1, rb_output);
+}
+
+
+static VALUE
+noko_io_write_failed(VALUE arg, VALUE exc)
+{
+  return Qundef;
+}
+
+
+int
+noko_io_write(void *io, char *c_buffer, int c_buffer_len)
+{
+  VALUE rb_args[2], rb_n_bytes_written;
+  VALUE rb_io = (VALUE)io;
+  VALUE rb_enc = Qnil;
+  rb_encoding *io_encoding;
+
+  if (rb_respond_to(rb_io, id_external_encoding)) {
+    rb_enc = rb_funcall(rb_io, id_external_encoding, 0);
+  }
+  io_encoding = RB_NIL_P(rb_enc) ? rb_ascii8bit_encoding() : rb_to_encoding(rb_enc);
+
+  rb_args[0] = rb_io;
+  rb_args[1] = rb_enc_str_new(c_buffer, (long)c_buffer_len, io_encoding);
+
+  rb_n_bytes_written = rb_rescue(noko_io_write_check, (VALUE)rb_args, noko_io_write_failed, 0);
+  if (rb_n_bytes_written == Qundef) { return -1; }
+
+  return NUM2INT(rb_n_bytes_written);
+}
+
+
+int
+noko_io_close(void *io)
+{
+  return 0;
+}
+
+
+#if defined(_WIN32) && !defined(NOKOGIRI_PACKAGED_LIBRARIES)
+#  define NOKOGIRI_WINDOWS_DLLS 1
+#else
+#  define NOKOGIRI_WINDOWS_DLLS 0
+#endif
+
+//
+//   |      dlls || true    | false   |
+//   | nlmm      ||         |         |
+//   |-----------++---------+---------|
+//   | NULL      || default | ruby    |
+//   | "random"  || default | ruby    |
+//   | "ruby"    || ruby    | ruby    |
+//   | "default" || default | default |
+//
+//   We choose *not* to use Ruby's memory management functions with windows DLLs because of this
+//   issue: https://github.com/sparklemotion/nokogiri/issues/2241
+//
+static void
+set_libxml_memory_management(void)
+{
+  const char *nlmm = getenv("NOKOGIRI_LIBXML_MEMORY_MANAGEMENT");
+  if (nlmm) {
+    if (strcmp(nlmm, "default") == 0) {
+      goto libxml_uses_default_memory_management;
+    } else if (strcmp(nlmm, "ruby") == 0) {
+      goto libxml_uses_ruby_memory_management;
+    }
+  }
+  if (NOKOGIRI_WINDOWS_DLLS) {
+libxml_uses_default_memory_management:
+    rb_const_set(mNokogiri, rb_intern("LIBXML_MEMORY_MANAGEMENT"), NOKOGIRI_STR_NEW2("default"));
+    return;
+  } else {
+libxml_uses_ruby_memory_management:
+    rb_const_set(mNokogiri, rb_intern("LIBXML_MEMORY_MANAGEMENT"), NOKOGIRI_STR_NEW2("ruby"));
+    xmlMemSetup((xmlFreeFunc)ruby_xfree, (xmlMallocFunc)ruby_xmalloc, (xmlReallocFunc)ruby_xrealloc, ruby_strdup);
+    return;
+  }
+}
+
+
+void
+Init_nokogiri(void)
+{
+  mNokogiri         = rb_define_module("Nokogiri");
+  mNokogiriGumbo    = rb_define_module_under(mNokogiri, "Gumbo");
+  mNokogiriHtml4    = rb_define_module_under(mNokogiri, "HTML4");
+  mNokogiriHtml4Sax = rb_define_module_under(mNokogiriHtml4, "SAX");
+  mNokogiriHtml5    = rb_define_module_under(mNokogiri, "HTML5");
+  mNokogiriXml      = rb_define_module_under(mNokogiri, "XML");
+  mNokogiriXmlSax   = rb_define_module_under(mNokogiriXml, "SAX");
+  mNokogiriXmlXpath = rb_define_module_under(mNokogiriXml, "XPath");
+  mNokogiriXslt     = rb_define_module_under(mNokogiri, "XSLT");
+
+  set_libxml_memory_management(); /* must be before any function calls that might invoke xmlInitParser() */
+  xmlInitParser();
+  exsltRegisterAll();
+
+  rb_const_set(mNokogiri, rb_intern("LIBXML_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION));
+  rb_const_set(mNokogiri, rb_intern("LIBXML_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xmlParserVersion));
+
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXSLT_DOTTED_VERSION));
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xsltEngineVersion));
+
+#ifdef NOKOGIRI_PACKAGED_LIBRARIES
+  rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qtrue);
+#  ifdef NOKOGIRI_PRECOMPILED_LIBRARIES
+  rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qtrue);
+#  else
+  rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
+#  endif
+  rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATCHES), " "));
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATCHES), " "));
+#else
+  rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qfalse);
+  rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
+  rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), Qnil);
+#endif
+
+#ifdef LIBXML_ICONV_ENABLED
+  rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qtrue);
+#else
+  rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qfalse);
+#endif
+
+  rb_const_set(mNokogiri, rb_intern("LIBXML_ZLIB_ENABLED"),
+               xmlHasFeature(XML_WITH_ZLIB) == 1 ? Qtrue : Qfalse);
+
+  rb_const_set(mNokogiri, rb_intern("LIBXML_HTTP_ENABLED"),
+               xmlHasFeature(XML_WITH_HTTP) == 1 ? Qtrue : Qfalse);
+
+#ifdef NOKOGIRI_OTHER_LIBRARY_VERSIONS
+  rb_const_set(mNokogiri, rb_intern("OTHER_LIBRARY_VERSIONS"), NOKOGIRI_STR_NEW2(NOKOGIRI_OTHER_LIBRARY_VERSIONS));
+#endif
+
+  if (xsltExtModuleFunctionLookup((const xmlChar *)"date-time", EXSLT_DATE_NAMESPACE)) {
+    rb_const_set(mNokogiri, rb_intern("LIBXSLT_DATETIME_ENABLED"), Qtrue);
+  } else {
+    rb_const_set(mNokogiri, rb_intern("LIBXSLT_DATETIME_ENABLED"), Qfalse);
+  }
+
+  cNokogiriSyntaxError = rb_define_class_under(mNokogiri, "SyntaxError", rb_eStandardError);
+  noko_init_xml_syntax_error();
+  assert(cNokogiriXmlSyntaxError);
+  cNokogiriXmlXpathSyntaxError = rb_define_class_under(mNokogiriXmlXpath, "SyntaxError", cNokogiriXmlSyntaxError);
+
+  noko_init_xml_element_content();
+  noko_init_xml_encoding_handler();
+  noko_init_xml_namespace();
+  noko_init_xml_node_set();
+  noko_init_xml_reader();
+
+  noko_init_xml_sax_parser();
+  noko_init_html4_sax_parser();
+
+  noko_init_xml_xpath_context();
+  noko_init_xslt_stylesheet();
+  noko_init_html_element_description();
+  noko_init_html_entity_lookup();
+
+  noko_init_xml_schema();
+  noko_init_xml_relax_ng();
+
+  noko_init_xml_sax_parser_context();
+  noko_init_html_sax_parser_context();
+
+  noko_init_xml_sax_push_parser();
+  noko_init_html_sax_push_parser();
+
+  noko_init_xml_node();
+  noko_init_xml_attr();
+  noko_init_xml_attribute_decl();
+  noko_init_xml_dtd();
+  noko_init_xml_element_decl();
+  noko_init_xml_entity_decl();
+  noko_init_xml_entity_reference();
+  noko_init_xml_processing_instruction();
+  assert(cNokogiriXmlNode);
+  cNokogiriXmlElement = rb_define_class_under(mNokogiriXml, "Element", cNokogiriXmlNode);
+  cNokogiriXmlCharacterData = rb_define_class_under(mNokogiriXml, "CharacterData", cNokogiriXmlNode);
+  noko_init_xml_comment();
+  noko_init_xml_text();
+  noko_init_xml_cdata();
+
+  noko_init_xml_document_fragment();
+  noko_init_xml_document();
+  noko_init_html_document();
+  noko_init_gumbo();
+
+  noko_init_test_global_handlers();
+
+  id_read = rb_intern("read");
+  id_write = rb_intern("write");
+  id_external_encoding = rb_intern("external_encoding");
+}