finnishstemmer.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

            // test, line 162
            v_3 = limit - cursor;
            if (!(in_grouping_b(g_V1, 97, 246)))
            {
                limit_backward = v_2;
                return false;
            }
            cursor = limit - v_3;
            // delete, line 163
            slice_del();
            limit_backward = v_2;
            // setlimit, line 165
            v_4 = limit - cursor;
            // tomark, line 165
            if (cursor < I_p2)
            {
                return false;
            }
            cursor = I_p2;
            v_5 = limit_backward;
            limit_backward = cursor;
            cursor = limit - v_4;
            // (, line 165
            // [, line 165
            ket = cursor;
            // substring, line 165
            among_var = find_among_b(a_9, 2);
            if (among_var == 0)
            {
                limit_backward = v_5;
                return false;
            }
            // ], line 165
            bra = cursor;
            limit_backward = v_5;
            switch(among_var) {
                case 0:
                    return false;
                case 1:
                    // (, line 167
                    // not, line 167
                    {
                        v_6 = limit - cursor;
                        lab0: do {
                            // literal, line 167
                            if (!(eq_s_b(2, "po")))
                            {
                                break lab0;
                            }
                            return false;
                        } while (false);
                        cursor = limit - v_6;
                    }
                    break;
            }
            // delete, line 170
            slice_del();
            return true;
        }

        private boolean r_tidy() {
            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;
            // (, line 172
            // setlimit, line 173
            v_1 = limit - cursor;
            // tomark, line 173
            if (cursor < I_p1)
            {
                return false;
            }
            cursor = I_p1;
            v_2 = limit_backward;
            limit_backward = cursor;
            cursor = limit - v_1;
            // (, line 173
            // do, line 174
            v_3 = limit - cursor;
            lab0: do {
                // (, line 174
                // and, line 174
                v_4 = limit - cursor;
                // call LONG, line 174
                if (!r_LONG())
                {
                    break lab0;
                }
                cursor = limit - v_4;
                // (, line 174
                // [, line 174
                ket = cursor;
                // next, line 174
                if (cursor <= limit_backward)
                {
                    break lab0;
                }
                cursor--;
                // ], line 174
                bra = cursor;
                // delete, line 174
                slice_del();
            } while (false);
            cursor = limit - v_3;
            // do, line 175
            v_5 = limit - cursor;
            lab1: do {
                // (, line 175
                // [, line 175
                ket = cursor;
                if (!(in_grouping_b(g_AEI, 97, 228)))
                {
                    break lab1;
                }
                // ], line 175
                bra = cursor;
                if (!(out_grouping_b(g_V1, 97, 246)))
                {
                    break lab1;
                }
                // delete, line 175
                slice_del();
            } while (false);
            cursor = limit - v_5;
            // do, line 176
            v_6 = limit - cursor;
            lab2: do {
                // (, line 176
                // [, line 176
                ket = cursor;
                // literal, line 176
                if (!(eq_s_b(1, "j")))
                {
                    break lab2;
                }
                // ], line 176
                bra = cursor;
                // or, line 176
                lab3: do {
                    v_7 = limit - cursor;
                    lab4: do {
                        // literal, line 176
                        if (!(eq_s_b(1, "o")))
                        {
                            break lab4;
                        }
                        break lab3;
                    } while (false);
                    cursor = limit - v_7;
                    // literal, line 176
                    if (!(eq_s_b(1, "u")))
                    {
                        break lab2;
                    }
                } while (false);
                // delete, line 176
                slice_del();
            } while (false);
            cursor = limit - v_6;
            // do, line 177
            v_8 = limit - cursor;
            lab5: do {
                // (, line 177
                // [, line 177
                ket = cursor;
                // literal, line 177
                if (!(eq_s_b(1, "o")))
                {
                    break lab5;
                }
                // ], line 177
                bra = cursor;
                // literal, line 177
                if (!(eq_s_b(1, "j")))
                {
                    break lab5;
                }
                // delete, line 177
                slice_del();
            } while (false);
            cursor = limit - v_8;
            limit_backward = v_2;
            // goto, line 179
            golab6: while(true)
            {
                v_9 = limit - cursor;
                lab7: do {
                    if (!(out_grouping_b(g_V1, 97, 246)))
                    {
                        break lab7;
                    }
                    cursor = limit - v_9;
                    break golab6;
                } while (false);
                cursor = limit - v_9;
                if (cursor <= limit_backward)
                {
                    return false;
                }
                cursor--;
            }
            // [, line 179
            ket = cursor;
            // next, line 179
            if (cursor <= limit_backward)
            {
                return false;
            }
            cursor--;
            // ], line 179
            bra = cursor;
            // -> x, line 179
            S_x = slice_to(S_x);
            // name x, line 179
            if (!(eq_v_b(S_x)))
            {
                return false;
            }
            // delete, line 179
            slice_del();
            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;
            // (, line 183
            // do, line 185
            v_1 = cursor;
            lab0: do {
                // call mark_regions, line 185
                if (!r_mark_regions())
                {
                    break lab0;
                }
            } while (false);
            cursor = v_1;
            // unset ending_removed, line 186
            B_ending_removed = false;
            // backwards, line 187
            limit_backward = cursor; cursor = limit;
            // (, line 187
            // do, line 188
            v_2 = limit - cursor;
            lab1: do {
                // call particle_etc, line 188
                if (!r_particle_etc())
                {
                    break lab1;
                }
            } while (false);
            cursor = limit - v_2;
            // do, line 189
            v_3 = limit - cursor;
            lab2: do {
                // call possessive, line 189
                if (!r_possessive())
                {
                    break lab2;
                }
            } while (false);
            cursor = limit - v_3;
            // do, line 190
            v_4 = limit - cursor;
            lab3: do {
                // call case_ending, line 190
                if (!r_case_ending())
                {
                    break lab3;
                }
            } while (false);
            cursor = limit - v_4;
            // do, line 191
            v_5 = limit - cursor;
            lab4: do {
                // call other_endings, line 191
                if (!r_other_endings())
                {
                    break lab4;
                }
            } while (false);
            cursor = limit - v_5;
            // or, line 192
            lab5: do {
                v_6 = limit - cursor;
                lab6: do {
                    // (, line 192
                    // Boolean test ending_removed, line 192
                    if (!(B_ending_removed))
                    {
                        break lab6;
                    }
                    // do, line 192
                    v_7 = limit - cursor;
                    lab7: do {
                        // call i_plural, line 192
                        if (!r_i_plural())
                        {
                            break lab7;
                        }
                    } while (false);
                    cursor = limit - v_7;
                    break lab5;
                } while (false);
                cursor = limit - v_6;
                // do, line 192
                v_8 = limit - cursor;
                lab8: do {
                    // call t_plural, line 192
                    if (!r_t_plural())
                    {
                        break lab8;
                    }
                } while (false);
                cursor = limit - v_8;
            } while (false);
            // do, line 193
            v_9 = limit - cursor;
            lab9: do {
                // call tidy, line 193
                if (!r_tidy())
                {
                    break lab9;
                }
            } while (false);
            cursor = limit - v_9;
            cursor = limit_backward;            return true;
        }

}