fast_trie 0.3.5

data/ext/trie/trie.h

@@ -0,0 +1,40 @@
+#include "darray.h"
+#include "tail.h"
+
+typedef struct _Trie {
+    DArray     *da;
+    Tail       *tail;
+} Trie;
+
+typedef struct _TrieState {
+    const Trie *trie;       /**< the corresponding trie */
+    TrieIndex   index;      /**< index in double-array/tail structures */
+    short       suffix_idx; /**< suffix character offset, if in suffix */
+    short       is_suffix;  /**< whether it is currently in suffix part */
+} TrieState;
+
+
+#define trie_da_is_separate(da,s)      (da_get_base ((da), (s)) < 0)
+#define trie_da_get_tail_index(da,s)   (-da_get_base ((da), (s)))
+#define trie_da_set_tail_index(da,s,v) (da_set_base ((da), (s), -(v)))
+#define trie_state_is_terminal(s) trie_state_is_walkable((s),TRIE_CHAR_TERM)
+
+
+Trie* trie_new();
+void trie_free(Trie *trie);
+static Bool trie_branch_in_branch (Trie *trie, TrieIndex sep_node, const TrieChar *suffix, TrieData data);
+static Bool trie_branch_in_tail(Trie *trie, TrieIndex sep_node, const TrieChar *suffix, TrieData data);
+Bool trie_store (Trie *trie, const TrieChar *key, TrieData data);
+Bool trie_retrieve (const Trie *trie, const TrieChar *key, TrieData *o_data);
+Bool trie_delete (Trie *trie, const TrieChar *key);
+TrieState * trie_root (const Trie *trie);
+static TrieState * trie_state_new (const Trie *trie, TrieIndex index, short suffix_idx, short is_suffix);
+TrieState * trie_state_clone (const TrieState *s);
+void trie_state_free (TrieState *s);
+void trie_state_rewind (TrieState *s);
+Bool trie_state_walk (TrieState *s, TrieChar c);
+Bool trie_state_is_walkable (const TrieState *s, TrieChar c);
+Bool trie_state_is_leaf (const TrieState *s);
+TrieData trie_state_get_data (const TrieState *s);
+
+

data/ext/trie/triedefs.h

@@ -0,0 +1,73 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+/*
+ * triedefs.h - General typedefs for trie
+ * Created: 2006-08-11
+ * Author:  Theppitak Karoonboonyanan <thep@linux.thai.net>
+ */
+
+#ifndef __TRIEDEFS_H
+#define __TRIEDEFS_H
+
+#include "typedefs.h"
+
+/**
+ * @file triedefs.h
+ * @brief General typedefs for trie
+ */
+
+/**
+ * @brief Trie IO modes
+ */
+typedef enum {
+    TRIE_IO_READ   = 0x01,
+    TRIE_IO_WRITE  = 0x02,
+    TRIE_IO_CREATE = 0x04
+} TrieIOMode;
+
+/**
+ * @brief Trie character type for alphabet
+ */
+typedef uint32         AlphaChar;
+
+/**
+ * @brief Error value for alphabet character
+ */
+#define ALPHA_CHAR_ERROR   (~(AlphaChar)0)
+
+/**
+ * @brief Trie character type for key
+ */
+typedef unsigned char  TrieChar;
+/**
+ * @brief Trie terminator character
+ */
+#define TRIE_CHAR_TERM    '\0'
+#define TRIE_CHAR_MAX     255
+
+/**
+ * @brief Type of Trie index
+ */
+typedef int32          TrieIndex;
+/**
+ * @brief Trie error index
+ */
+#define TRIE_INDEX_ERROR  0
+/**
+ * @brief Maximum trie index value
+ */
+#define TRIE_INDEX_MAX    0x7fffffff
+
+/**
+ * @brief Type of value associated to trie entries
+ */
+typedef unsigned long TrieData;
+/**
+ * @brief Trie error data
+ */
+#define TRIE_DATA_ERROR  -1
+
+#endif  /* __TRIEDEFS_H */
+
+/*
+vi:ts=4:ai:expandtab
+*/

data/ext/trie/typedefs.h

@@ -0,0 +1,113 @@
+/* -*- Mode: C; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
+/*
+ * typedefs.h - general types
+ * Created : 11 Aug 2006
+ * Author  : Theppitak Karoonboonyanan <thep@linux.thai.net>
+ */
+
+#ifndef __TYPEDEFS_H
+#define __TYPEDEFS_H
+
+#include <limits.h>
+
+typedef enum { FALSE = 0, TRUE = 1 } Bool;
+
+# if UCHAR_MAX == 0xff
+#   ifndef UINT8_TYPEDEF
+#     define UINT8_TYPEDEF
+      typedef unsigned char  uint8;
+#   endif /* UINT8_TYPEDEF */
+# endif /* UCHAR_MAX */
+
+# if SCHAR_MAX == 0x7f
+#   ifndef INT8_TYPEDEF
+#     define INT8_TYPEDEF
+      typedef signed char    int8;
+#   endif /* INT8_TYPEDEF */
+# endif /* SCHAR_MAX */
+
+# if UINT_MAX == 0xffff
+#   ifndef UINT16_TYPEDEF
+#     define UINT16_TYPEDEF
+      typedef unsigned int   uint16;
+#   endif /* UINT16_TYPEDEF */
+# endif /* UINT_MAX */
+
+# if INT_MAX == 0x7fff
+#   ifndef INT16_TYPEDEF
+#     define INT16_TYPEDEF
+      typedef int            int16;
+#   endif /* INT16_TYPEDEF */
+# endif /* INT_MAX */
+
+# if USHRT_MAX == 0xffff
+#   ifndef UINT16_TYPEDEF
+#     define UINT16_TYPEDEF
+      typedef unsigned short uint16;
+#   endif /* UINT16_TYPEDEF */
+# endif /* USHRT_MAX */
+
+# if SHRT_MAX == 0x7fff
+#   ifndef INT16_TYPEDEF
+#     define INT16_TYPEDEF
+      typedef short          int16;
+#   endif /* INT16_TYPEDEF */
+# endif /* SHRT_MAX */
+
+# if UINT_MAX == 0xffffffff
+#   ifndef UINT32_TYPEDEF
+#     define UINT32_TYPEDEF
+      typedef unsigned int   uint32;
+#   endif /* UINT32_TYPEDEF */
+# endif /* UINT_MAX */
+
+# if INT_MAX == 0x7fffffff
+#   ifndef INT32_TYPEDEF
+#     define INT32_TYPEDEF
+      typedef int            int32;
+#   endif /* INT32_TYPEDEF */
+# endif /* INT_MAX */
+
+# if ULONG_MAX == 0xffffffff
+#   ifndef UINT32_TYPEDEF
+#     define UINT32_TYPEDEF
+      typedef unsigned long  uint32;
+#   endif /* UINT32_TYPEDEF */
+# endif /* ULONG_MAX */
+
+# if LONG_MAX == 0x7fffffff
+#   ifndef INT32_TYPEDEF
+#     define INT32_TYPEDEF
+      typedef long           int32;
+#   endif /* INT32_TYPEDEF */
+# endif /* LONG_MAX */
+
+# ifndef UINT8_TYPEDEF
+#   error "uint8 type is undefined!"
+# endif
+# ifndef INT8_TYPEDEF
+#   error "int8 type is undefined!"
+# endif
+# ifndef UINT16_TYPEDEF
+#   error "uint16 type is undefined!"
+# endif
+# ifndef INT16_TYPEDEF
+#   error "int16 type is undefined!"
+# endif
+# ifndef UINT32_TYPEDEF
+#   error "uint32 type is undefined!"
+# endif
+# ifndef INT32_TYPEDEF
+#   error "int32 type is undefined!"
+# endif
+
+typedef uint8  byte;
+typedef uint16 word;
+typedef uint32 dword;
+
+
+#endif /* __TYPEDEFS_H */
+
+/*
+vi:ts=4:ai:expandtab
+*/

data/lib/trie.rb

@@ -0,0 +1 @@
+require File.dirname(__FILE__) + '/../ext/trie'

data/spec/trie_spec.rb

@@ -0,0 +1,266 @@
+require File.dirname(__FILE__) + '/../ext/trie/trie'
+
+describe Trie do
+  before :each do
+    @trie = Trie.new;
+    @trie.add('rocket')
+    @trie.add('rock')
+    @trie.add('frederico')
+  end
+  
+  #describe :path do
+  #  it 'returns the correct path' do
+  #    @trie.path.should == TRIE_PATH
+  #  end
+  #end
+
+  describe :has_key? do
+    it 'returns true for words in the trie' do
+      @trie.has_key?('rocket').should be_true
+    end
+
+    it 'returns nil for words that are not in the trie' do
+      @trie.has_key?('not_in_the_trie').should be_nil
+    end
+  end
+
+  describe :get do
+    it 'returns -1 for words in the trie without a weight' do
+      @trie.get('rocket').should == -1
+    end
+
+    it 'returns nil if the word is not in the trie' do
+      @trie.get('not_in_the_trie').should be_nil
+    end
+  end
+
+  describe :add do
+    it 'adds a word to the trie' do
+      @trie.add('forsooth').should == true
+      @trie.get('forsooth').should == -1
+    end
+
+    it 'adds a word with a weight to the trie' do
+      @trie.add('chicka',123).should == true
+      @trie.get('chicka').should == 123
+    end
+
+    it 'adds values greater than 16-bit allows' do
+      @trie.add('chicka', 72_000).should == true
+      @trie.get('chicka').should == 72_000
+    end
+
+    it 'adds a word with a non-numeric value to the trie' do
+      @trie.add('doot', 'Heeey').should == true
+      @trie.get('doot').should == 'Heeey'
+    end
+  end
+
+  describe :delete do
+    it 'deletes a word from the trie' do
+      @trie.delete('rocket').should == true
+      @trie.has_key?('rocket').should be_nil
+    end
+  end
+
+  describe :children do
+    it 'returns all words beginning with a given prefix' do
+      children = @trie.children('roc')
+      children.size.should == 2
+      children.should include('rock')
+      children.should include('rocket')
+    end
+
+    it 'returns blank array if prefix does not exist' do
+      @trie.children('ajsodij').should == []
+    end
+
+    it 'includes the prefix if the prefix is a word' do
+      children = @trie.children('rock')
+      children.size.should == 2
+      children.should include('rock')
+      children.should include('rocket')
+    end
+
+    it 'returns blank array if prefix is nil' do
+      @trie.children(nil).should == []
+    end
+  end
+
+  describe :children_with_values do
+    before :each do
+      @trie.add('abc',2)
+      @trie.add('abcd',4)
+    end
+
+    it 'returns all words with values beginning with a given prefix' do
+      children = @trie.children_with_values('ab')
+      children.size.should == 2
+      children.should include(['abc',2])
+      children.should include(['abcd',4])
+    end
+
+    it 'returns nil if prefix does not exist' do
+      @trie.children_with_values('ajsodij').should == []
+    end
+
+    it 'includes the prefix if the prefix is a word' do
+      children = @trie.children_with_values('abc')
+      children.size.should == 2
+      children.should include(['abc',2])
+      children.should include(['abcd',4])
+    end
+
+    it 'returns blank array if prefix is nil' do
+      @trie.children_with_values(nil).should == []
+    end
+  end
+
+  #describe :walk_to_terminal do
+  #  it 'returns the first word found along a path' do
+  #    @trie.add 'anderson'
+  #    @trie.add 'andreas'
+  #    @trie.add 'and'
+
+  #    @trie.walk_to_terminal('anderson').should == 'and'
+  #  end
+
+  #  it 'returns the first word and value along a path' do
+  #    @trie.add 'anderson'
+  #    @trie.add 'andreas'
+  #    @trie.add 'and', 15
+
+  #    @trie.walk_to_terminal('anderson',true).should == ['and', 15]
+  #  end
+  #end
+
+  describe :root do
+    it 'returns a TrieNode' do
+      @trie.root.should be_an_instance_of(TrieNode)
+    end
+
+    it 'returns a different TrieNode each time' do
+      @trie.root.should_not == @trie.root
+    end
+  end
+
+  #describe :save do
+  #  it 'saves the trie to disk such that another trie can be spawned which will read succesfully' do
+  #    @trie.add('omgwtf',123)
+  #    @trie.save
+  #
+  #    trie2 = Trie.new(TRIE_PATH)
+  #    trie2.get('omgwtf').should == 123
+  #  end
+  #end
+end
+
+describe TrieNode do
+  before :each do
+    @trie = Trie.new;
+    @trie.add('rocket',1)
+    @trie.add('rock',2)
+    @trie.add('frederico',3)
+    @node = @trie.root
+  end
+  
+  describe :state do
+    it 'returns the most recent state character' do
+      @node.walk!('r')
+      @node.state.should == 'r'
+      @node.walk!('o')
+      @node.state.should == 'o'
+    end
+
+    it 'is nil when no walk has occurred' do
+      @node.state.should == nil
+    end
+  end
+
+  describe :full_state do
+    it 'returns the current string' do
+      @node.walk!('r').walk!('o').walk!('c')
+      @node.full_state.should == 'roc'
+    end
+
+    it 'is a blank string when no walk has occurred' do
+      @node.full_state.should == ''
+    end
+  end
+  
+  describe :walk! do
+    it 'returns the updated object when the walk succeeds' do
+      other = @node.walk!('r')
+      other.should == @node
+    end
+
+    it 'returns nil when the walk fails' do
+      @node.walk!('q').should be_nil
+    end
+  end
+
+  describe :walk do
+    it 'returns a new node object when the walk succeeds' do
+      other = @node.walk('r')
+      other.should != @node
+    end
+
+    it 'returns nil when the walk fails' do
+      @node.walk('q').should be_nil
+    end
+  end
+
+
+  describe :value do
+    it 'returns nil when the node is not terminal' do
+      @node.walk!('r')
+      @node.value.should be_nil
+    end
+
+    it 'returns a value when the node is terminal' do
+      @node.walk!('r').walk!('o').walk!('c').walk!('k')
+      @node.value.should == 2
+    end
+  end
+
+  describe :terminal? do
+    it 'returns true when the node is a word end' do
+      @node.walk!('r').walk!('o').walk!('c').walk!('k')
+      @node.should be_terminal
+    end
+
+    it 'returns nil when the node is not a word end' do
+      @node.walk!('r').walk!('o').walk!('c')
+      @node.should_not be_terminal
+    end
+  end
+
+  describe :leaf? do
+    it 'returns true when this is the end of a branch of the trie' do
+      @node.walk!('r').walk!('o').walk!('c').walk!('k').walk!('e').walk!('t')
+      @node.should be_leaf
+    end
+
+    it 'returns nil when there are more splits on this branch' do
+      @node.walk!('r').walk!('o').walk!('c').walk!('k')
+      @node.should_not be_leaf
+    end
+  end
+
+  describe :clone do
+    it 'creates a new instance of this node which is not this node' do
+      new_node = @node.clone
+      new_node.should_not == @node
+    end
+
+    it 'matches the state of the current node' do
+      new_node = @node.clone
+      new_node.state.should == @node.state
+    end
+
+    it 'matches the full_state of the current node' do
+      new_node = @node.clone
+      new_node.full_state.should == @node.full_state
+    end
+  end
+end