frenchstemmer.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

                    v_3 = limit - cursor;
                    lab0: do {
                        // (, line 191
                        // [, line 191
                        ket = cursor;
                        // literal, line 191
                        if (!(eq_s_b(1, "e")))
                        {
                            cursor = limit - v_3;
                            break lab0;
                        }
                        // ], line 191
                        bra = cursor;
                        // delete, line 191
                        slice_del();
                    } while (false);
                    break;
            }
            limit_backward = v_2;
            return true;
        }

        private boolean r_residual_suffix() {
            int among_var;
            int v_1;
            int v_2;
            int v_3;
            int v_4;
            int v_5;
            // (, line 198
            // try, line 199
            v_1 = limit - cursor;
            lab0: do {
                // (, line 199
                // [, line 199
                ket = cursor;
                // literal, line 199
                if (!(eq_s_b(1, "s")))
                {
                    cursor = limit - v_1;
                    break lab0;
                }
                // ], line 199
                bra = cursor;
                // test, line 199
                v_2 = limit - cursor;
                if (!(out_grouping_b(g_keep_with_s, 97, 232)))
                {
                    cursor = limit - v_1;
                    break lab0;
                }
                cursor = limit - v_2;
                // delete, line 199
                slice_del();
            } while (false);
            // setlimit, line 200
            v_3 = limit - cursor;
            // tomark, line 200
            if (cursor < I_pV)
            {
                return false;
            }
            cursor = I_pV;
            v_4 = limit_backward;
            limit_backward = cursor;
            cursor = limit - v_3;
            // (, line 200
            // [, line 201
            ket = cursor;
            // substring, line 201
            among_var = find_among_b(a_7, 7);
            if (among_var == 0)
            {
                limit_backward = v_4;
                return false;
            }
            // ], line 201
            bra = cursor;
            switch(among_var) {
                case 0:
                    limit_backward = v_4;
                    return false;
                case 1:
                    // (, line 202
                    // call R2, line 202
                    if (!r_R2())
                    {
                        limit_backward = v_4;
                        return false;
                    }
                    // or, line 202
                    lab1: do {
                        v_5 = limit - cursor;
                        lab2: do {
                            // literal, line 202
                            if (!(eq_s_b(1, "s")))
                            {
                                break lab2;
                            }
                            break lab1;
                        } while (false);
                        cursor = limit - v_5;
                        // literal, line 202
                        if (!(eq_s_b(1, "t")))
                        {
                            limit_backward = v_4;
                            return false;
                        }
                    } while (false);
                    // delete, line 202
                    slice_del();
                    break;
                case 2:
                    // (, line 204
                    // <-, line 204
                    slice_from("i");
                    break;
                case 3:
                    // (, line 205
                    // delete, line 205
                    slice_del();
                    break;
                case 4:
                    // (, line 206
                    // literal, line 206
                    if (!(eq_s_b(2, "gu")))
                    {
                        limit_backward = v_4;
                        return false;
                    }
                    // delete, line 206
                    slice_del();
                    break;
            }
            limit_backward = v_4;
            return true;
        }

        private boolean r_un_double() {
            int v_1;
            // (, line 211
            // test, line 212
            v_1 = limit - cursor;
            // among, line 212
            if (find_among_b(a_8, 5) == 0)
            {
                return false;
            }
            cursor = limit - v_1;
            // [, line 212
            ket = cursor;
            // next, line 212
            if (cursor <= limit_backward)
            {
                return false;
            }
            cursor--;
            // ], line 212
            bra = cursor;
            // delete, line 212
            slice_del();
            return true;
        }

        private boolean r_un_accent() {
            int v_3;
            // (, line 215
            // atleast, line 216
            {
                int v_1 = 1;
                // atleast, line 216
                replab0: while(true)
                {
                    lab1: do {
                        if (!(out_grouping_b(g_v, 97, 251)))
                        {
                            break lab1;
                        }
                        v_1--;
                        continue replab0;
                    } while (false);
                    break replab0;
                }
                if (v_1 > 0)
                {
                    return false;
                }
            }
            // [, line 217
            ket = cursor;
            // or, line 217
            lab2: do {
                v_3 = limit - cursor;
                lab3: do {
                    // literal, line 217
                    if (!(eq_s_b(1, "\u00E9")))
                    {
                        break lab3;
                    }
                    break lab2;
                } while (false);
                cursor = limit - v_3;
                // literal, line 217
                if (!(eq_s_b(1, "\u00E8")))
                {
                    return false;
                }
            } while (false);
            // ], line 217
            bra = cursor;
            // <-, line 217
            slice_from("e");
            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;
            int v_11;
            // (, line 221
            // do, line 223
            v_1 = cursor;
            lab0: do {
                // call prelude, line 223
                if (!r_prelude())
                {
                    break lab0;
                }
            } while (false);
            cursor = v_1;
            // do, line 224
            v_2 = cursor;
            lab1: do {
                // call mark_regions, line 224
                if (!r_mark_regions())
                {
                    break lab1;
                }
            } while (false);
            cursor = v_2;
            // backwards, line 225
            limit_backward = cursor; cursor = limit;
            // (, line 225
            // do, line 227
            v_3 = limit - cursor;
            lab2: do {
                // (, line 227
                // or, line 237
                lab3: do {
                    v_4 = limit - cursor;
                    lab4: do {
                        // (, line 228
                        // and, line 233
                        v_5 = limit - cursor;
                        // (, line 229
                        // or, line 229
                        lab5: do {
                            v_6 = limit - cursor;
                            lab6: do {
                                // call standard_suffix, line 229
                                if (!r_standard_suffix())
                                {
                                    break lab6;
                                }
                                break lab5;
                            } while (false);
                            cursor = limit - v_6;
                            lab7: do {
                                // call i_verb_suffix, line 230
                                if (!r_i_verb_suffix())
                                {
                                    break lab7;
                                }
                                break lab5;
                            } while (false);
                            cursor = limit - v_6;
                            // call verb_suffix, line 231
                            if (!r_verb_suffix())
                            {
                                break lab4;
                            }
                        } while (false);
                        cursor = limit - v_5;
                        // try, line 234
                        v_7 = limit - cursor;
                        lab8: do {
                            // (, line 234
                            // [, line 234
                            ket = cursor;
                            // or, line 234
                            lab9: do {
                                v_8 = limit - cursor;
                                lab10: do {
                                    // (, line 234
                                    // literal, line 234
                                    if (!(eq_s_b(1, "Y")))
                                    {
                                        break lab10;
                                    }
                                    // ], line 234
                                    bra = cursor;
                                    // <-, line 234
                                    slice_from("i");
                                    break lab9;
                                } while (false);
                                cursor = limit - v_8;
                                // (, line 235
                                // literal, line 235
                                if (!(eq_s_b(1, "\u00E7")))
                                {
                                    cursor = limit - v_7;
                                    break lab8;
                                }
                                // ], line 235
                                bra = cursor;
                                // <-, line 235
                                slice_from("c");
                            } while (false);
                        } while (false);
                        break lab3;
                    } while (false);
                    cursor = limit - v_4;
                    // call residual_suffix, line 238
                    if (!r_residual_suffix())
                    {
                        break lab2;
                    }
                } while (false);
            } while (false);
            cursor = limit - v_3;
            // do, line 243
            v_9 = limit - cursor;
            lab11: do {
                // call un_double, line 243
                if (!r_un_double())
                {
                    break lab11;
                }
            } while (false);
            cursor = limit - v_9;
            // do, line 244
            v_10 = limit - cursor;
            lab12: do {
                // call un_accent, line 244
                if (!r_un_accent())
                {
                    break lab12;
                }
            } while (false);
            cursor = limit - v_10;
            cursor = limit_backward;            // do, line 246
            v_11 = cursor;
            lab13: do {
                // call postlude, line 246
                if (!r_postlude())
                {
                    break lab13;
                }
            } while (false);
            cursor = v_11;
            return true;
        }

}