bromlife 0.1.0 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 15d41d66995405636a1ba18481d39f546cb3131260e9415601b57f9efba42510
-  data.tar.gz: 100cde67bad24de79e6adb76b3b6090f88351853f49d30175ad7887e5347926e
+  metadata.gz: 2880bb24a47333a5ec7d44ed5d7e82ac8e1e924cb85cd41dfdf6b8964629d44b
+  data.tar.gz: bfdd977bd83af03898b5ecc4b55213e62e7ec81309044ba66d45e053dad348b8
 SHA512:
-  metadata.gz: eff5b0aca2bb023585b3ddcfab2b75f76018dd84c217dd740bf4f8afe80db1236624da93033fb5252a310d03e079f646e2154272fe8e49b34e1256561d65b4e4
-  data.tar.gz: 257232dfce6fdbc2f818bb4bb2b386437505350b3183838558bc557d5b683073984ca09633afef512a37c04f2f530ef0d884ff0e8c1c2016c598b35f4c850732
+  metadata.gz: ee3d465c11aefaf3911b7af9c93d2c82ef9ea6da4f557c45c5ec452ea23bcbcca89e7d856b6d0055b334489566e8f7cca90df70bf72b76746185a7a1ac861e88
+  data.tar.gz: 1ad39aa6d3e17aa24889dc3052f2fa141d16f592acbc951ca9937361a224717c0d3183d552c29036c1378e4955c9e578b34976f0628af77204183903c130ff57

data/ext/bromlife/extconf.rb

@@ -17,8 +17,7 @@
 
 require "mkmf"
 
-have_library("bromlife")
-have_func("oml_to_json", "bromlife.h")
+have_library("bromlife") or raise "no bromlife library"
 
 # Makes all symbols private by default to avoid unintended conflict
 # with other gems. To explicitly export symbols you can use RUBY_FUNC_EXPORTED

data/ext/bromlife/rb_bromlife.c

@@ -15,36 +15,31 @@
  */
 
 #include "rb_bromlife.h"
+
 #include <bromlife.h>
+
 #include <ruby.h>
+
 #include <stddef.h>
 #include <stdint.h>
 
-VALUE rb_mBromlife, rb_eError, rb_cParser, rb_cNodeIter, rb_cElement;
+/* Prefix "rbb_" means RuBy Bromlife. */
 
-/**
- * call-seq:
- *   oml_to_dumped_json (source) -> JSON String
- */
-static VALUE
-rb_oml_m_oml_to_dumped_json (VALUE self, VALUE source)
-{
-  (void)self;
-  char *bytes = rb_string_value_ptr (&source);
-  long len = RSTRING_LEN (source);
-  size_t json_len;
-  uint8_t *json_bytes = oml_to_json ((uint8_t *)bytes, (size_t)len, &json_len);
-  if (!json_bytes)
-    rb_raise (rb_eError, "failed to convert to JSON");
-  return rb_str_new ((char *)json_bytes, json_len);
-}
+VALUE rb_mBromlife, rb_eError, rb_cParser;
 
 /* parser */
 
+struct rbb_parser
+{
+  struct oml_parser parser;
+  uint8_t *bytes;
+};
+
 static void
 rb_oml_parser_free (void *parser)
 {
-  oml_parser_free (parser);
+  struct rbb_parser *p = parser;
+  ruby_xfree (p);
 }
 
 static const rb_data_type_t rb_oml_parser_type
@@ -55,137 +50,66 @@ static const rb_data_type_t rb_oml_parser_type
 static VALUE
 rb_oml_parser_alloc (VALUE klass)
 {
-  return TypedData_Wrap_Struct (klass, &rb_oml_parser_type, NULL);
+  return rb_data_typed_object_wrap (klass, NULL, &rb_oml_parser_type);
 }
 
 static VALUE
-rb_oml_parser_m_initialize (VALUE self)
+rb_oml_parser_m_initialize (VALUE self, VALUE input)
 {
-  OmlParser *parser = oml_parser_create ();
-  if (!parser)
-    rb_raise (rb_eError, "failed to create OML parser");
+  char *bytes = rb_string_value_ptr (&input);
+  size_t len = RSTRING_LEN (input);
+  struct rbb_parser *parser = ruby_xmalloc (sizeof (struct rbb_parser));
+  *parser = (struct rbb_parser){ .parser = oml_parser (bytes, len),
+                                 .bytes = bytes };
   RTYPEDDATA_DATA (self) = parser;
   return RUBY_Qnil;
 }
 
-/**
- * call-seq:
- *   feed (String source)
- */
-static VALUE
-rb_oml_parser_m_feed (VALUE self, VALUE source)
+static inline VALUE
+rbb_raise (const char *message)
 {
-  char *bytes = rb_string_value_ptr (&source);
-  long len = RSTRING_LEN (source);
-  if (!oml_parser_feed (RTYPEDDATA_DATA (self), (uint8_t *)bytes, len))
-    rb_raise (rb_eError, "failed to feed to parser");
+  rb_raise (rb_eError, "%s", message);
   return RUBY_Qnil;
 }
 
-static const rb_data_type_t rb_oml_node_iter_type;
-/**
- * call-seq:
- *   finish -> Bromlife::NodeIter
- */
-static VALUE
-rb_oml_parser_m_finish (VALUE self)
-{
-  OmlNodeIter *iter = oml_parser_finish (RTYPEDDATA_DATA (self));
-  if (!iter)
-    rb_raise (rb_eError, "failed to finish parser");
-  return TypedData_Wrap_Struct (rb_cNodeIter, &rb_oml_node_iter_type, iter);
-}
-
-/* node iterator */
-
-static void
-rb_oml_node_iter_free (void *iter)
-{
-  oml_node_iter_free (iter);
-}
-
-static const rb_data_type_t rb_oml_node_iter_type
-    = { .wrap_struct_name = "OML node iterator",
-        .function = { .dfree = rb_oml_node_iter_free },
-        .flags = RUBY_TYPED_FREE_IMMEDIATELY };
-
 static VALUE
-rb_oml_node_iter_alloc (VALUE klass)
+rb_oml_parser_m_next (VALUE self)
 {
-  return TypedData_Wrap_Struct (klass, &rb_oml_node_iter_type, NULL);
-}
-
-static const rb_data_type_t rb_oml_element_type;
-/**
- * call-seq:
- *   next -> String or Bromlife::Element or nil
- */
-static VALUE
-rb_oml_node_iter_m_next (VALUE self)
-{
-  const OmlNode *node = oml_node_iter_next (RTYPEDDATA_DATA (self));
-  if (!node)
-    return RUBY_Qnil;
-  if (oml_node_is_element (node))
-    {
-      const OmlElement *element = oml_node_to_element (node);
-      if (!element)
-        rb_raise (rb_eError, "unreachable");
-      return TypedData_Wrap_Struct (rb_cElement, &rb_oml_element_type,
-                                    (void *)element);
-    }
-  else if (oml_node_is_text (node))
-    {
-      size_t len;
-      const uint8_t *bytes = oml_node_to_text (node, &len);
-      if (!bytes)
-        rb_raise (rb_eError, "unreachable");
-      return rb_str_new_static ((const char *)bytes, len);
-    }
-  else
-    rb_raise (rb_eError, "unreachable");
-}
-
-/* element */
-
-/* Elements are managed by the root node iterator.
- */
-static const rb_data_type_t rb_oml_element_type
-    = { .wrap_struct_name = "OML element",
-        .function = { .dfree = NULL },
-        .flags = RUBY_TYPED_FREE_IMMEDIATELY };
-
-static VALUE
-rb_oml_element_alloc (VALUE klass)
-{
-  return TypedData_Wrap_Struct (klass, &rb_oml_element_type, NULL);
-}
-
-/**
- * call-seq:
- *   label -> String
- */
-static VALUE
-rb_oml_element_m_label (VALUE self)
-{
-  size_t len;
-  const uint8_t *bytes = oml_element_get_label (RTYPEDDATA_DATA (self), &len);
-  if (!bytes)
-    rb_raise (rb_eError, "failed to get label");
-  return rb_str_new_static ((const char *)bytes, len);
-}
-
-/**
- * call-seq:
- *   children -> Bromlife::NodeIter
- */
-static VALUE
-rb_oml_element_m_children (VALUE self)
-{
-  OmlNodeIter *iter = oml_element_get_children (RTYPEDDATA_DATA (self));
-  if (!iter)
-    rb_raise (rb_eError, "failed to get children");
-  return TypedData_Wrap_Struct (rb_cNodeIter, &rb_oml_node_iter_type, iter);
+  struct rbb_parser *parser = rb_check_typeddata (self, &rb_oml_parser_type);
+  struct oml_event event = oml_next (&parser->parser);
+  if (oml_event_is_empty (event))
+    rb_raise (rb_eStopIteration, "no more event");
+
+  VALUE type = oml_event_is_null (event)    ? rb_id2sym (rb_intern ("null"))
+               : oml_event_is_left (event)  ? rb_id2sym (rb_intern ("left"))
+               : oml_event_is_right (event) ? rb_id2sym (rb_intern ("right"))
+               : oml_event_is_body (event)  ? rb_id2sym (rb_intern ("body"))
+                                            : rbb_raise ("unreachable");
+  VALUE subtype
+      = oml_event_is_normal (event)    ? rb_id2sym (rb_intern ("normal"))
+        : oml_event_is_ignored (event) ? rb_id2sym (rb_intern ("ignored"))
+        : oml_event_is_vocab (event)   ? rb_id2sym (rb_intern ("vocab"))
+        : oml_event_is_word (event)    ? rb_id2sym (rb_intern ("word"))
+        : oml_event_is_label (event)   ? rb_id2sym (rb_intern ("label"))
+                                       : RUBY_Qnil;
+
+  size_t len = oml_event_len (event);
+  VALUE str
+      = len == 0
+            ? RUBY_Qnil
+            : rb_utf8_str_new (parser->bytes + oml_event_offset (event), len);
+
+  VALUE label = RUBY_Qnil;
+  if (oml_event_is_left (event) && oml_event_is_normal (event))
+    label = rb_utf8_str_new (parser->bytes + oml_event_label_offset (event),
+                             oml_event_label_len (event));
+
+  VALUE ary = rb_ary_new ();
+  rb_ary_push (ary, type);
+  rb_ary_push (ary, subtype);
+  rb_ary_push (ary, str);
+  rb_ary_push (ary, label);
+  return ary;
 }
 
 RUBY_FUNC_EXPORTED void
@@ -196,23 +120,10 @@ Init_bromlife (void)
   rb_define_const (rb_mBromlife, "LIBBROMLIFE_VERSION",
                    rb_int2num_inline (OML_VERSION));
 
-  rb_define_module_function (rb_mBromlife, "oml_to_dumped_json",
-                             rb_oml_m_oml_to_dumped_json, 1);
-
   rb_cParser = rb_define_class_under (rb_mBromlife, "Parser", rb_cObject);
   rb_define_alloc_func (rb_cParser, rb_oml_parser_alloc);
-  rb_define_method (rb_cParser, "initialize", rb_oml_parser_m_initialize, 0);
-  rb_define_method (rb_cParser, "feed", rb_oml_parser_m_feed, 1);
-  rb_define_method (rb_cParser, "finish", rb_oml_parser_m_finish, 0);
-
-  rb_cElement = rb_define_class_under (rb_mBromlife, "Element", rb_cObject);
-  rb_define_alloc_func (rb_cElement, rb_oml_element_alloc);
-  rb_define_method (rb_cElement, "label", rb_oml_element_m_label, 0);
-  rb_define_method (rb_cElement, "children", rb_oml_element_m_children, 0);
-
-  rb_cNodeIter = rb_define_class_under (rb_mBromlife, "NodeIter", rb_cObject);
-  rb_define_alloc_func (rb_cNodeIter, rb_oml_node_iter_alloc);
-  rb_define_method (rb_cNodeIter, "next", rb_oml_node_iter_m_next, 0);
+  rb_define_method (rb_cParser, "initialize", rb_oml_parser_m_initialize, 1);
+  rb_define_method (rb_cParser, "next", rb_oml_parser_m_next, 0);
 
   rb_eError = rb_define_class_under (rb_mBromlife, "Error", rb_eStandardError);
 }

data/lib/bromlife/version.rb

@@ -18,5 +18,5 @@
 #++
 
 module Bromlife
-  VERSION = "0.1.0"
+  VERSION = "0.2.0"
 end

data/lib/bromlife.rb

@@ -21,4 +21,56 @@ require_relative "bromlife/version"
 require "bromlife/bromlife"
 
 module Bromlife
+  def self.dom_parse(input)
+    parser = Parser.new(input)
+    result = []
+    current = result
+    loop do
+      event = parser.next
+      event => [type, subtype, str, label]
+      case [type, subtype]
+      in [:left, :vocab | :word | :label | :ignored] \
+        | [:right, :vocab | :word | :label | :ignored] \
+        | [:body, :label | :ignored] # nop
+      in [:body, :normal]
+        current << str
+      in [:left, :normal]
+        elem = Element.new(label:, parent: current)
+        current << elem
+        current = elem
+      in [:right, :normal]
+        current = current.parent
+      end
+    end
+    result
+  end
+
+  class Element
+    attr_reader :parent, :children
+    def initialize(label:, parent:)
+      @label = label
+      @parent = parent
+      @children = []
+    end
+
+    def <<(node)
+      @children << node
+    end
+
+    def deconstruct_keys(*_keys)
+      # TODO: optimize with keys param.
+      { label: @label, children: @children }
+    end
+
+    def inspect
+      "#<Bromlife::Element @label=#{ @label.inspect } @children=#{ @children.inspect }>"
+    end
+
+    def to_h
+      {
+        :label => @label,
+        :children => @children.map { |child| child.is_a?(String) ? child : child.to_h }
+      }
+    end
+  end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bromlife
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.0
 platform: ruby
 authors:
 - gemmaro
@@ -45,7 +45,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 4.0.10
+rubygems_version: 4.0.6
 specification_version: 4
 summary: O Markup Language parser library
 test_files: []