englishstemmer.java

一个很不错的词频统计程序,目前只支持英文,中文的本人正在修改中.改好后上传给大家分享

// This file was generated automatically by the Snowball to Java compiler

package org.tartarus.snowball.ext;
import org.tartarus.snowball.SnowballProgram;
import org.tartarus.snowball.Among;

/**
 * Generated class implementing code defined by a snowball script.
 */
public class englishStemmer extends SnowballProgram {

        private Among a_0[] = {
            new Among ( "arsen", -1, -1, "", this),
            new Among ( "commun", -1, -1, "", this),
            new Among ( "gener", -1, -1, "", this)
        };

        private Among a_1[] = {
            new Among ( "'", -1, 1, "", this),
            new Among ( "'s'", 0, 1, "", this),
            new Among ( "'s", -1, 1, "", this)
        };

        private Among a_2[] = {
            new Among ( "ied", -1, 2, "", this),
            new Among ( "s", -1, 3, "", this),
            new Among ( "ies", 1, 2, "", this),
            new Among ( "sses", 1, 1, "", this),
            new Among ( "ss", 1, -1, "", this),
            new Among ( "us", 1, -1, "", this)
        };

        private Among a_3[] = {
            new Among ( "", -1, 3, "", this),
            new Among ( "bb", 0, 2, "", this),
            new Among ( "dd", 0, 2, "", this),
            new Among ( "ff", 0, 2, "", this),
            new Among ( "gg", 0, 2, "", this),
            new Among ( "bl", 0, 1, "", this),
            new Among ( "mm", 0, 2, "", this),
            new Among ( "nn", 0, 2, "", this),
            new Among ( "pp", 0, 2, "", this),
            new Among ( "rr", 0, 2, "", this),
            new Among ( "at", 0, 1, "", this),
            new Among ( "tt", 0, 2, "", this),
            new Among ( "iz", 0, 1, "", this)
        };

        private Among a_4[] = {
            new Among ( "ed", -1, 2, "", this),
            new Among ( "eed", 0, 1, "", this),
            new Among ( "ing", -1, 2, "", this),
            new Among ( "edly", -1, 2, "", this),
            new Among ( "eedly", 3, 1, "", this),
            new Among ( "ingly", -1, 2, "", this)
        };

        private Among a_5[] = {
            new Among ( "anci", -1, 3, "", this),
            new Among ( "enci", -1, 2, "", this),
            new Among ( "ogi", -1, 13, "", this),
            new Among ( "li", -1, 16, "", this),
            new Among ( "bli", 3, 12, "", this),
            new Among ( "abli", 4, 4, "", this),
            new Among ( "alli", 3, 8, "", this),
            new Among ( "fulli", 3, 14, "", this),
            new Among ( "lessli", 3, 15, "", this),
            new Among ( "ousli", 3, 10, "", this),
            new Among ( "entli", 3, 5, "", this),
            new Among ( "aliti", -1, 8, "", this),
            new Among ( "biliti", -1, 12, "", this),
            new Among ( "iviti", -1, 11, "", this),
            new Among ( "tional", -1, 1, "", this),
            new Among ( "ational", 14, 7, "", this),
            new Among ( "alism", -1, 8, "", this),
            new Among ( "ation", -1, 7, "", this),
            new Among ( "ization", 17, 6, "", this),
            new Among ( "izer", -1, 6, "", this),
            new Among ( "ator", -1, 7, "", this),
            new Among ( "iveness", -1, 11, "", this),
            new Among ( "fulness", -1, 9, "", this),
            new Among ( "ousness", -1, 10, "", this)
        };

        private Among a_6[] = {
            new Among ( "icate", -1, 4, "", this),
            new Among ( "ative", -1, 6, "", this),
            new Among ( "alize", -1, 3, "", this),
            new Among ( "iciti", -1, 4, "", this),
            new Among ( "ical", -1, 4, "", this),
            new Among ( "tional", -1, 1, "", this),
            new Among ( "ational", 5, 2, "", this),
            new Among ( "ful", -1, 5, "", this),
            new Among ( "ness", -1, 5, "", this)
        };

        private Among a_7[] = {
            new Among ( "ic", -1, 1, "", this),
            new Among ( "ance", -1, 1, "", this),
            new Among ( "ence", -1, 1, "", this),
            new Among ( "able", -1, 1, "", this),
            new Among ( "ible", -1, 1, "", this),
            new Among ( "ate", -1, 1, "", this),
            new Among ( "ive", -1, 1, "", this),
            new Among ( "ize", -1, 1, "", this),
            new Among ( "iti", -1, 1, "", this),
            new Among ( "al", -1, 1, "", this),
            new Among ( "ism", -1, 1, "", this),
            new Among ( "ion", -1, 2, "", this),
            new Among ( "er", -1, 1, "", this),
            new Among ( "ous", -1, 1, "", this),
            new Among ( "ant", -1, 1, "", this),
            new Among ( "ent", -1, 1, "", this),
            new Among ( "ment", 15, 1, "", this),
            new Among ( "ement", 16, 1, "", this)
        };

        private Among a_8[] = {
            new Among ( "e", -1, 1, "", this),
            new Among ( "l", -1, 2, "", this)
        };

        private Among a_9[] = {
            new Among ( "succeed", -1, -1, "", this),
            new Among ( "proceed", -1, -1, "", this),
            new Among ( "exceed", -1, -1, "", this),
            new Among ( "canning", -1, -1, "", this),
            new Among ( "inning", -1, -1, "", this),
            new Among ( "earring", -1, -1, "", this),
            new Among ( "herring", -1, -1, "", this),
            new Among ( "outing", -1, -1, "", this)
        };

        private Among a_10[] = {
            new Among ( "andes", -1, -1, "", this),
            new Among ( "atlas", -1, -1, "", this),
            new Among ( "bias", -1, -1, "", this),
            new Among ( "cosmos", -1, -1, "", this),
            new Among ( "dying", -1, 3, "", this),
            new Among ( "early", -1, 9, "", this),
            new Among ( "gently", -1, 7, "", this),
            new Among ( "howe", -1, -1, "", this),
            new Among ( "idly", -1, 6, "", this),
            new Among ( "lying", -1, 4, "", this),
            new Among ( "news", -1, -1, "", this),
            new Among ( "only", -1, 10, "", this),
            new Among ( "singly", -1, 11, "", this),
            new Among ( "skies", -1, 2, "", this),
            new Among ( "skis", -1, 1, "", this),
            new Among ( "sky", -1, -1, "", this),
            new Among ( "tying", -1, 5, "", this),
            new Among ( "ugly", -1, 8, "", this)
        };

        private static final char g_v[] = {17, 65, 16, 1 };

        private static final char g_v_WXY[] = {1, 17, 65, 208, 1 };

        private static final char g_valid_LI[] = {55, 141, 2 };

        private boolean B_Y_found;
        private int I_p2;
        private int I_p1;

        private void copy_from(englishStemmer other) {
            B_Y_found = other.B_Y_found;
            I_p2 = other.I_p2;
            I_p1 = other.I_p1;
            super.copy_from(other);
        }

        private boolean r_prelude() {
            int v_1;
            int v_2;
            int v_3;
            int v_4;
            int v_5;
            // (, line 25
            // unset Y_found, line 26
            B_Y_found = false;
            // do, line 27
            v_1 = cursor;
            lab0: do {
                // (, line 27
                // [, line 27
                bra = cursor;
                // literal, line 27
                if (!(eq_s(1, "'")))
                {
                    break lab0;
                }
                // ], line 27
                ket = cursor;
                // delete, line 27
                slice_del();
            } while (false);
            cursor = v_1;
            // do, line 28
            v_2 = cursor;
            lab1: do {
                // (, line 28
                // [, line 28
                bra = cursor;
                // literal, line 28
                if (!(eq_s(1, "y")))
                {
                    break lab1;
                }
                // ], line 28
                ket = cursor;
                // <-, line 28
                slice_from("Y");
                // set Y_found, line 28
                B_Y_found = true;
            } while (false);
            cursor = v_2;
            // do, line 29
            v_3 = cursor;
            lab2: do {
                // repeat, line 29
                replab3: while(true)
                {
                    v_4 = cursor;
                    lab4: do {
                        // (, line 29
                        // goto, line 29
                        golab5: while(true)
                        {
                            v_5 = cursor;
                            lab6: do {
                                // (, line 29
                                if (!(in_grouping(g_v, 97, 121)))
                                {
                                    break lab6;
                                }
                                // [, line 29
                                bra = cursor;
                                // literal, line 29
                                if (!(eq_s(1, "y")))
                                {
                                    break lab6;
                                }
                                // ], line 29
                                ket = cursor;
                                cursor = v_5;
                                break golab5;
                            } while (false);
                            cursor = v_5;
                            if (cursor >= limit)
                            {
                                break lab4;
                            }
                            cursor++;
                        }
                        // <-, line 29
                        slice_from("Y");
                        // set Y_found, line 29
                        B_Y_found = true;
                        continue replab3;
                    } while (false);
                    cursor = v_4;
                    break replab3;
                }
            } while (false);
            cursor = v_3;
            return true;
        }

        private boolean r_mark_regions() {
            int v_1;
            int v_2;
            // (, line 32
            I_p1 = limit;
            I_p2 = limit;
            // do, line 35
            v_1 = cursor;
            lab0: do {
                // (, line 35
                // or, line 41
                lab1: do {
                    v_2 = cursor;
                    lab2: do {
                        // among, line 36
                        if (find_among(a_0, 3) == 0)
                        {
                            break lab2;
                        }
                        break lab1;
                    } while (false);
                    cursor = v_2;
                    // (, line 41
                    // gopast, line 41
                    golab3: while(true)
                    {
                        lab4: do {
                            if (!(in_grouping(g_v, 97, 121)))
                            {
                                break lab4;
                            }
                            break golab3;
                        } while (false);
                        if (cursor >= limit)
                        {
                            break lab0;
                        }
                        cursor++;
                    }
                    // gopast, line 41
                    golab5: while(true)
                    {
                        lab6: do {
                            if (!(out_grouping(g_v, 97, 121)))
                            {
                                break lab6;
                            }
                            break golab5;
                        } while (false);
                        if (cursor >= limit)
                        {
                            break lab0;
                        }
                        cursor++;
                    }
                } while (false);
                // setmark p1, line 42
                I_p1 = cursor;
                // gopast, line 43
                golab7: while(true)
                {
                    lab8: do {
                        if (!(in_grouping(g_v, 97, 121)))
                        {
                            break lab8;
                        }
                        break golab7;
                    } while (false);
                    if (cursor >= limit)
                    {
                        break lab0;
                    }
                    cursor++;
                }
                // gopast, line 43
                golab9: while(true)
                {
                    lab10: do {
                        if (!(out_grouping(g_v, 97, 121)))
                        {
                            break lab10;
                        }
                        break golab9;
                    } while (false);
                    if (cursor >= limit)
                    {
                        break lab0;
                    }
                    cursor++;
                }
                // setmark p2, line 43
                I_p2 = cursor;
            } while (false);
            cursor = v_1;
            return true;
        }

        private boolean r_shortv() {
            int v_1;
            // (, line 49
            // or, line 51
            lab0: do {
                v_1 = limit - cursor;
                lab1: do {
                    // (, line 50
                    if (!(out_grouping_b(g_v_WXY, 89, 121)))
                    {
                        break lab1;
                    }
                    if (!(in_grouping_b(g_v, 97, 121)))
                    {
                        break lab1;
                    }
                    if (!(out_grouping_b(g_v, 97, 121)))
                    {
                        break lab1;
                    }
                    break lab0;
                } while (false);
                cursor = limit - v_1;
                // (, line 52
                if (!(out_grouping_b(g_v, 97, 121)))
                {
                    return false;
                }
                if (!(in_grouping_b(g_v, 97, 121)))
                {
                    return false;
                }
                // atlimit, line 52
                if (cursor > limit_backward)
                {
                    return false;
                }
            } while (false);
            return true;
        }

        private boolean r_R1() {
            if (!(I_p1 <= cursor))
            {
                return false;
            }
            return true;
        }

        private boolean r_R2() {
            if (!(I_p2 <= cursor))
            {
                return false;
            }
            return true;
        }

        private boolean r_Step_1a() {
            int among_var;
            int v_1;
            int v_2;
            // (, line 58
            // try, line 59
            v_1 = limit - cursor;
            lab0: do {
                // (, line 59
                // [, line 60
                ket = cursor;
                // substring, line 60
                among_var = find_among_b(a_1, 3);
                if (among_var == 0)
                {
                    cursor = limit - v_1;
                    break lab0;