romanianstemmer.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

                    // (, line 85
                    // <-, line 85
                    slice_from("at");
                    break;
                case 7:
                    // (, line 87
                    // <-, line 87
                    slice_from("a\u0163i");
                    break;
            }
            return true;
        }

        private boolean r_combo_suffix() {
            int among_var;
            int v_1;
            // test, line 91
            v_1 = limit - cursor;
            // (, line 91
            // [, line 92
            ket = cursor;
            // substring, line 92
            among_var = find_among_b(a_2, 46);
            if (among_var == 0)
            {
                return false;
            }
            // ], line 92
            bra = cursor;
            // call R1, line 92
            if (!r_R1())
            {
                return false;
            }
            // (, line 92
            switch(among_var) {
                case 0:
                    return false;
                case 1:
                    // (, line 100
                    // <-, line 101
                    slice_from("abil");
                    break;
                case 2:
                    // (, line 103
                    // <-, line 104
                    slice_from("ibil");
                    break;
                case 3:
                    // (, line 106
                    // <-, line 107
                    slice_from("iv");
                    break;
                case 4:
                    // (, line 112
                    // <-, line 113
                    slice_from("ic");
                    break;
                case 5:
                    // (, line 117
                    // <-, line 118
                    slice_from("at");
                    break;
                case 6:
                    // (, line 121
                    // <-, line 122
                    slice_from("it");
                    break;
            }
            // set standard_suffix_removed, line 125
            B_standard_suffix_removed = true;
            cursor = limit - v_1;
            return true;
        }

        private boolean r_standard_suffix() {
            int among_var;
            int v_1;
            // (, line 129
            // unset standard_suffix_removed, line 130
            B_standard_suffix_removed = false;
            // repeat, line 131
            replab0: while(true)
            {
                v_1 = limit - cursor;
                lab1: do {
                    // call combo_suffix, line 131
                    if (!r_combo_suffix())
                    {
                        break lab1;
                    }
                    continue replab0;
                } while (false);
                cursor = limit - v_1;
                break replab0;
            }
            // [, line 132
            ket = cursor;
            // substring, line 132
            among_var = find_among_b(a_3, 62);
            if (among_var == 0)
            {
                return false;
            }
            // ], line 132
            bra = cursor;
            // call R2, line 132
            if (!r_R2())
            {
                return false;
            }
            // (, line 132
            switch(among_var) {
                case 0:
                    return false;
                case 1:
                    // (, line 148
                    // delete, line 149
                    slice_del();
                    break;
                case 2:
                    // (, line 151
                    // literal, line 152
                    if (!(eq_s_b(1, "\u0163")))
                    {
                        return false;
                    }
                    // ], line 152
                    bra = cursor;
                    // <-, line 152
                    slice_from("t");
                    break;
                case 3:
                    // (, line 155
                    // <-, line 156
                    slice_from("ist");
                    break;
            }
            // set standard_suffix_removed, line 160
            B_standard_suffix_removed = true;
            return true;
        }

        private boolean r_verb_suffix() {
            int among_var;
            int v_1;
            int v_2;
            int v_3;
            // setlimit, line 164
            v_1 = limit - cursor;
            // tomark, line 164
            if (cursor < I_pV)
            {
                return false;
            }
            cursor = I_pV;
            v_2 = limit_backward;
            limit_backward = cursor;
            cursor = limit - v_1;
            // (, line 164
            // [, line 165
            ket = cursor;
            // substring, line 165
            among_var = find_among_b(a_4, 94);
            if (among_var == 0)
            {
                limit_backward = v_2;
                return false;
            }
            // ], line 165
            bra = cursor;
            switch(among_var) {
                case 0:
                    limit_backward = v_2;
                    return false;
                case 1:
                    // (, line 200
                    // or, line 200
                    lab0: do {
                        v_3 = limit - cursor;
                        lab1: do {
                            if (!(out_grouping_b(g_v, 97, 259)))
                            {
                                break lab1;
                            }
                            break lab0;
                        } while (false);
                        cursor = limit - v_3;
                        // literal, line 200
                        if (!(eq_s_b(1, "u")))
                        {
                            limit_backward = v_2;
                            return false;
                        }
                    } while (false);
                    // delete, line 200
                    slice_del();
                    break;
                case 2:
                    // (, line 214
                    // delete, line 214
                    slice_del();
                    break;
            }
            limit_backward = v_2;
            return true;
        }

        private boolean r_vowel_suffix() {
            int among_var;
            // (, line 218
            // [, line 219
            ket = cursor;
            // substring, line 219
            among_var = find_among_b(a_5, 5);
            if (among_var == 0)
            {
                return false;
            }
            // ], line 219
            bra = cursor;
            // call RV, line 219
            if (!r_RV())
            {
                return false;
            }
            switch(among_var) {
                case 0:
                    return false;
                case 1:
                    // (, line 220
                    // delete, line 220
                    slice_del();
                    break;
            }
            return true;
        }

        public boolean stem() {
            int v_1;
            int v_2;
            int v_3;
            int v_4;
            int v_5;
            int v_6;
            int v_7;
            int v_8;
            // (, line 225
            // do, line 226
            v_1 = cursor;
            lab0: do {
                // call prelude, line 226
                if (!r_prelude())
                {
                    break lab0;
                }
            } while (false);
            cursor = v_1;
            // do, line 227
            v_2 = cursor;
            lab1: do {
                // call mark_regions, line 227
                if (!r_mark_regions())
                {
                    break lab1;
                }
            } while (false);
            cursor = v_2;
            // backwards, line 228
            limit_backward = cursor; cursor = limit;
            // (, line 228
            // do, line 229
            v_3 = limit - cursor;
            lab2: do {
                // call step_0, line 229
                if (!r_step_0())
                {
                    break lab2;
                }
            } while (false);
            cursor = limit - v_3;
            // do, line 230
            v_4 = limit - cursor;
            lab3: do {
                // call standard_suffix, line 230
                if (!r_standard_suffix())
                {
                    break lab3;
                }
            } while (false);
            cursor = limit - v_4;
            // do, line 231
            v_5 = limit - cursor;
            lab4: do {
                // (, line 231
                // or, line 231
                lab5: do {
                    v_6 = limit - cursor;
                    lab6: do {
                        // Boolean test standard_suffix_removed, line 231
                        if (!(B_standard_suffix_removed))
                        {
                            break lab6;
                        }
                        break lab5;
                    } while (false);
                    cursor = limit - v_6;
                    // call verb_suffix, line 231
                    if (!r_verb_suffix())
                    {
                        break lab4;
                    }
                } while (false);
            } while (false);
            cursor = limit - v_5;
            // do, line 232
            v_7 = limit - cursor;
            lab7: do {
                // call vowel_suffix, line 232
                if (!r_vowel_suffix())
                {
                    break lab7;
                }
            } while (false);
            cursor = limit - v_7;
            cursor = limit_backward;            // do, line 234
            v_8 = cursor;
            lab8: do {
                // call postlude, line 234
                if (!r_postlude())
                {
                    break lab8;
                }
            } while (false);
            cursor = v_8;
            return true;
        }

}