russianstemmer.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

                    break lab0;
                }
                // ], line 110
                bra = cursor;
                switch(among_var) {
                    case 0:
                        cursor = limit - v_1;
                        break lab0;
                    case 1:
                        // (, line 115
                        // or, line 115
                        lab1: do {
                            v_2 = limit - cursor;
                            lab2: do {
                                // literal, line 115
                                if (!(eq_s_b(1, "\u0430")))
                                {
                                    break lab2;
                                }
                                break lab1;
                            } while (false);
                            cursor = limit - v_2;
                            // literal, line 115
                            if (!(eq_s_b(1, "\u044F")))
                            {
                                cursor = limit - v_1;
                                break lab0;
                            }
                        } while (false);
                        // delete, line 115
                        slice_del();
                        break;
                    case 2:
                        // (, line 122
                        // delete, line 122
                        slice_del();
                        break;
                }
            } while (false);
            return true;
        }

        private boolean r_reflexive() {
            int among_var;
            // (, line 128
            // [, line 129
            ket = cursor;
            // substring, line 129
            among_var = find_among_b(a_3, 2);
            if (among_var == 0)
            {
                return false;
            }
            // ], line 129
            bra = cursor;
            switch(among_var) {
                case 0:
                    return false;
                case 1:
                    // (, line 132
                    // delete, line 132
                    slice_del();
                    break;
            }
            return true;
        }

        private boolean r_verb() {
            int among_var;
            int v_1;
            // (, line 136
            // [, line 137
            ket = cursor;
            // substring, line 137
            among_var = find_among_b(a_4, 46);
            if (among_var == 0)
            {
                return false;
            }
            // ], line 137
            bra = cursor;
            switch(among_var) {
                case 0:
                    return false;
                case 1:
                    // (, line 143
                    // or, line 143
                    lab0: do {
                        v_1 = limit - cursor;
                        lab1: do {
                            // literal, line 143
                            if (!(eq_s_b(1, "\u0430")))
                            {
                                break lab1;
                            }
                            break lab0;
                        } while (false);
                        cursor = limit - v_1;
                        // literal, line 143
                        if (!(eq_s_b(1, "\u044F")))
                        {
                            return false;
                        }
                    } while (false);
                    // delete, line 143
                    slice_del();
                    break;
                case 2:
                    // (, line 151
                    // delete, line 151
                    slice_del();
                    break;
            }
            return true;
        }

        private boolean r_noun() {
            int among_var;
            // (, line 159
            // [, line 160
            ket = cursor;
            // substring, line 160
            among_var = find_among_b(a_5, 36);
            if (among_var == 0)
            {
                return false;
            }
            // ], line 160
            bra = cursor;
            switch(among_var) {
                case 0:
                    return false;
                case 1:
                    // (, line 167
                    // delete, line 167
                    slice_del();
                    break;
            }
            return true;
        }

        private boolean r_derivational() {
            int among_var;
            // (, line 175
            // [, line 176
            ket = cursor;
            // substring, line 176
            among_var = find_among_b(a_6, 2);
            if (among_var == 0)
            {
                return false;
            }
            // ], line 176
            bra = cursor;
            // call R2, line 176
            if (!r_R2())
            {
                return false;
            }
            switch(among_var) {
                case 0:
                    return false;
                case 1:
                    // (, line 179
                    // delete, line 179
                    slice_del();
                    break;
            }
            return true;
        }

        private boolean r_tidy_up() {
            int among_var;
            // (, line 183
            // [, line 184
            ket = cursor;
            // substring, line 184
            among_var = find_among_b(a_7, 4);
            if (among_var == 0)
            {
                return false;
            }
            // ], line 184
            bra = cursor;
            switch(among_var) {
                case 0:
                    return false;
                case 1:
                    // (, line 188
                    // delete, line 188
                    slice_del();
                    // [, line 189
                    ket = cursor;
                    // literal, line 189
                    if (!(eq_s_b(1, "\u043D")))
                    {
                        return false;
                    }
                    // ], line 189
                    bra = cursor;
                    // literal, line 189
                    if (!(eq_s_b(1, "\u043D")))
                    {
                        return false;
                    }
                    // delete, line 189
                    slice_del();
                    break;
                case 2:
                    // (, line 192
                    // literal, line 192
                    if (!(eq_s_b(1, "\u043D")))
                    {
                        return false;
                    }
                    // delete, line 192
                    slice_del();
                    break;
                case 3:
                    // (, line 194
                    // delete, line 194
                    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;
            int v_9;
            int v_10;
            // (, line 199
            // do, line 201
            v_1 = cursor;
            lab0: do {
                // call mark_regions, line 201
                if (!r_mark_regions())
                {
                    break lab0;
                }
            } while (false);
            cursor = v_1;
            // backwards, line 202
            limit_backward = cursor; cursor = limit;
            // setlimit, line 202
            v_2 = limit - cursor;
            // tomark, line 202
            if (cursor < I_pV)
            {
                return false;
            }
            cursor = I_pV;
            v_3 = limit_backward;
            limit_backward = cursor;
            cursor = limit - v_2;
            // (, line 202
            // do, line 203
            v_4 = limit - cursor;
            lab1: do {
                // (, line 203
                // or, line 204
                lab2: do {
                    v_5 = limit - cursor;
                    lab3: do {
                        // call perfective_gerund, line 204
                        if (!r_perfective_gerund())
                        {
                            break lab3;
                        }
                        break lab2;
                    } while (false);
                    cursor = limit - v_5;
                    // (, line 205
                    // try, line 205
                    v_6 = limit - cursor;
                    lab4: do {
                        // call reflexive, line 205
                        if (!r_reflexive())
                        {
                            cursor = limit - v_6;
                            break lab4;
                        }
                    } while (false);
                    // or, line 206
                    lab5: do {
                        v_7 = limit - cursor;
                        lab6: do {
                            // call adjectival, line 206
                            if (!r_adjectival())
                            {
                                break lab6;
                            }
                            break lab5;
                        } while (false);
                        cursor = limit - v_7;
                        lab7: do {
                            // call verb, line 206
                            if (!r_verb())
                            {
                                break lab7;
                            }
                            break lab5;
                        } while (false);
                        cursor = limit - v_7;
                        // call noun, line 206
                        if (!r_noun())
                        {
                            break lab1;
                        }
                    } while (false);
                } while (false);
            } while (false);
            cursor = limit - v_4;
            // try, line 209
            v_8 = limit - cursor;
            lab8: do {
                // (, line 209
                // [, line 209
                ket = cursor;
                // literal, line 209
                if (!(eq_s_b(1, "\u0438")))
                {
                    cursor = limit - v_8;
                    break lab8;
                }
                // ], line 209
                bra = cursor;
                // delete, line 209
                slice_del();
            } while (false);
            // do, line 212
            v_9 = limit - cursor;
            lab9: do {
                // call derivational, line 212
                if (!r_derivational())
                {
                    break lab9;
                }
            } while (false);
            cursor = limit - v_9;
            // do, line 213
            v_10 = limit - cursor;
            lab10: do {
                // call tidy_up, line 213
                if (!r_tidy_up())
                {
                    break lab10;
                }
            } while (false);
            cursor = limit - v_10;
            limit_backward = v_3;
            cursor = limit_backward;            return true;
        }

}