Kouhei Sutou
null+****@clear*****
Wed Feb 1 10:42:20 JST 2017
Kouhei Sutou 2017-02-01 10:42:20 +0900 (Wed, 01 Feb 2017) New Revision: b1b6709cecfe57c227adf7ce0ef9e30157c049c8 https://github.com/groonga/groonga/commit/b1b6709cecfe57c227adf7ce0ef9e30157c049c8 Message: select: add columns[LABEL].window.group_keys See test about how to use. Added files: test/command/suite/select/columns/window_function/window_sum/group.expected test/command/suite/select/columns/window_function/window_sum/group.test Modified files: include/groonga/window_function.h lib/grn_window_function.h lib/proc/proc_select.c lib/window_function.c Modified: include/groonga/window_function.h (+2 -0) =================================================================== --- include/groonga/window_function.h 2017-01-31 19:04:36 +0900 (1473b01) +++ include/groonga/window_function.h 2017-02-01 10:42:20 +0900 (500638f) @@ -42,6 +42,8 @@ GRN_API grn_obj *grn_window_get_table(grn_ctx *ctx, typedef struct _grn_window_definition { grn_table_sort_key *sort_keys; size_t n_sort_keys; + grn_table_sort_key *group_keys; + size_t n_group_keys; } grn_window_definition; typedef grn_rc grn_window_function_func(grn_ctx *ctx, Modified: lib/grn_window_function.h (+7 -2) =================================================================== --- lib/grn_window_function.h 2017-01-31 19:04:36 +0900 (71c71e9) +++ lib/grn_window_function.h 2017-02-01 10:42:20 +0900 (7d53977) @@ -1,5 +1,6 @@ +/* -*- c-basic-offset: 2 -*- */ /* - Copyright(C) 2016 Brazil + Copyright(C) 2016-2017 Brazil This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public @@ -20,6 +21,7 @@ struct _grn_window { grn_obj *table; + grn_obj *grouped_table; grn_obj ids; size_t n_ids; ssize_t current_index; @@ -28,7 +30,10 @@ struct _grn_window { size_t n_sort_keys; }; -grn_rc grn_window_init(grn_ctx *ctx, grn_window *window, grn_obj *table); +grn_rc grn_window_init(grn_ctx *ctx, + grn_window *window, + grn_obj *table, + grn_obj *grouped_table); grn_rc grn_window_fin(grn_ctx *ctx, grn_window *window); grn_rc grn_window_set_sort_keys(grn_ctx *ctx, grn_window *window, Modified: lib/proc/proc_select.c (+79 -26) =================================================================== --- lib/proc/proc_select.c 2017-01-31 19:04:36 +0900 (8adb3fe) +++ lib/proc/proc_select.c 2017-02-01 10:42:20 +0900 (4e056f5) @@ -1,6 +1,6 @@ /* -*- c-basic-offset: 2 -*- */ /* - Copyright(C) 2009-2016 Brazil + Copyright(C) 2009-2017 Brazil This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public @@ -67,6 +67,7 @@ typedef struct { grn_select_string value; struct { grn_select_string sort_keys; + grn_select_string group_keys; } window; } grn_column_data; @@ -331,6 +332,7 @@ grn_column_data_init(grn_ctx *ctx, column->flags = GRN_OBJ_COLUMN_SCALAR; GRN_SELECT_INIT_STRING(column->value); GRN_SELECT_INIT_STRING(column->window.sort_keys); + GRN_SELECT_INIT_STRING(column->window.group_keys); return GRN_TRUE; } @@ -341,7 +343,8 @@ grn_column_data_fill(grn_ctx *ctx, grn_obj *type_raw, grn_obj *flags, grn_obj *value, - grn_obj *window_sort_keys) + grn_obj *window_sort_keys, + grn_obj *window_group_keys) { if (type_raw && GRN_TEXT_LEN(type_raw) > 0) { grn_obj *type; @@ -399,6 +402,7 @@ grn_column_data_fill(grn_ctx *ctx, GRN_SELECT_FILL_STRING(column->value, value); GRN_SELECT_FILL_STRING(column->window.sort_keys, window_sort_keys); + GRN_SELECT_FILL_STRING(column->window.group_keys, window_group_keys); return GRN_TRUE; } @@ -426,9 +430,11 @@ grn_column_data_collect(grn_ctx *ctx, grn_obj *value = NULL; struct { grn_obj *sort_keys; + grn_obj *group_keys; } window; window.sort_keys = NULL; + window.group_keys = NULL; grn_hash_cursor_get_value(ctx, cursor, (void **)&column); @@ -456,13 +462,16 @@ grn_column_data_collect(grn_ctx *ctx, GET_VAR(flags); GET_VAR(value); GET_VAR(window.sort_keys); + GET_VAR(window.group_keys); #undef GET_VAR #undef GET_VAR_RAW grn_column_data_fill(ctx, column, - type, flags, value, window.sort_keys); + type, flags, value, + window.sort_keys, + window.group_keys); } grn_hash_cursor_close(ctx, cursor); return GRN_TRUE; @@ -1301,29 +1310,66 @@ grn_select_apply_columns(grn_ctx *ctx, expression, data->filter.condition.expression); - if (column_data->window.sort_keys.length > 0) { + if (column_data->window.sort_keys.length > 0 || + column_data->window.group_keys.length > 0) { grn_window_definition definition; - int n_sort_keys; grn_rc rc; - definition.sort_keys = - grn_table_sort_key_from_str(ctx, - column_data->window.sort_keys.value, - column_data->window.sort_keys.length, - table, &n_sort_keys); - definition.n_sort_keys = n_sort_keys; - if (!definition.sort_keys) { - grn_obj_close(ctx, expression); - grn_obj_close(ctx, column); - GRN_PLUGIN_ERROR(ctx, - GRN_INVALID_ARGUMENT, - "[select][column][%s][%.*s] " - "failed to parse sort keys: %s", - grn_column_stage_name(column_data->stage), - (int)(column_data->label.length), - column_data->label.value, - ctx->errbuf); - break; + if (column_data->window.sort_keys.length > 0) { + int n_sort_keys; + definition.sort_keys = + grn_table_sort_key_from_str(ctx, + column_data->window.sort_keys.value, + column_data->window.sort_keys.length, + table, &n_sort_keys); + definition.n_sort_keys = n_sort_keys; + if (!definition.sort_keys) { + grn_obj_close(ctx, expression); + grn_obj_close(ctx, column); + GRN_PLUGIN_ERROR(ctx, + GRN_INVALID_ARGUMENT, + "[select][column][%s][%.*s] " + "failed to parse sort keys: %s", + grn_column_stage_name(column_data->stage), + (int)(column_data->label.length), + column_data->label.value, + ctx->errbuf); + break; + } + } else { + definition.sort_keys = NULL; + definition.n_sort_keys = 0; + } + + if (column_data->window.group_keys.length > 0) { + int n_group_keys; + definition.group_keys = + grn_table_sort_key_from_str(ctx, + column_data->window.group_keys.value, + column_data->window.group_keys.length, + table, &n_group_keys); + definition.n_group_keys = n_group_keys; + if (!definition.group_keys) { + grn_obj_close(ctx, expression); + grn_obj_close(ctx, column); + if (definition.sort_keys) { + grn_table_sort_key_close(ctx, + definition.sort_keys, + definition.n_sort_keys); + } + GRN_PLUGIN_ERROR(ctx, + GRN_INVALID_ARGUMENT, + "[select][column][%s][%.*s] " + "failed to parse group keys: %s", + grn_column_stage_name(column_data->stage), + (int)(column_data->label.length), + column_data->label.value, + ctx->errbuf); + break; + } + } else { + definition.group_keys = NULL; + definition.n_group_keys = 0; } rc = grn_table_apply_window_function(ctx, @@ -1331,9 +1377,16 @@ grn_select_apply_columns(grn_ctx *ctx, column, &definition, expression); - grn_table_sort_key_close(ctx, - definition.sort_keys, - definition.n_sort_keys); + if (definition.sort_keys) { + grn_table_sort_key_close(ctx, + definition.sort_keys, + definition.n_sort_keys); + } + if (definition.group_keys) { + grn_table_sort_key_close(ctx, + definition.group_keys, + definition.n_group_keys); + } if (rc != GRN_SUCCESS) { grn_obj_close(ctx, expression); grn_obj_close(ctx, column); Modified: lib/window_function.c (+160 -47) =================================================================== --- lib/window_function.c 2017-01-31 19:04:36 +0900 (96482fb) +++ lib/window_function.c 2017-02-01 10:42:20 +0900 (4989ce8) @@ -1,6 +1,6 @@ /* -*- c-basic-offset: 2 -*- */ /* - Copyright(C) 2016 Brazil + Copyright(C) 2016-2017 Brazil This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public @@ -24,11 +24,15 @@ #include <string.h> grn_rc -grn_window_init(grn_ctx *ctx, grn_window *window, grn_obj *table) +grn_window_init(grn_ctx *ctx, + grn_window *window, + grn_obj *table, + grn_obj *grouped_table) { GRN_API_ENTER; window->table = table; + window->grouped_table = grouped_table; GRN_RECORD_INIT(&(window->ids), GRN_OBJ_VECTOR, grn_obj_id(ctx, table)); window->n_ids = 0; window->current_index = 0; @@ -154,8 +158,6 @@ grn_window_set_sort_keys(grn_ctx *ctx, grn_table_sort_key *sort_keys, size_t n_sort_keys) { - grn_obj *sorted; - GRN_API_ENTER; if (!window) { @@ -163,39 +165,71 @@ grn_window_set_sort_keys(grn_ctx *ctx, GRN_API_RETURN(ctx->rc); } - sorted = grn_table_create(ctx, - NULL, 0, NULL, - GRN_OBJ_TABLE_NO_KEY, - NULL, window->table); - if (!sorted) { - ERR(ctx->rc, - "[window][set][sort-keys] " - "failed to create a table to store sorted records: %s", - ctx->errbuf); - GRN_API_RETURN(ctx->rc); - } - - grn_table_sort(ctx, window->table, 0, -1, sorted, sort_keys, n_sort_keys); - if (ctx->rc != GRN_SUCCESS) { - ERR(ctx->rc, - "[window][set][sort-keys] " - "failed to sort: %s", - ctx->errbuf); - grn_obj_unlink(ctx, sorted); - GRN_API_RETURN(ctx->rc); - } - - GRN_BULK_REWIND(&(window->ids)); - GRN_TABLE_EACH_BEGIN(ctx, sorted, cursor, id) { - void *value; - grn_id record_id; + if (sort_keys) { + grn_obj *sorted; + + sorted = grn_table_create(ctx, + NULL, 0, NULL, + GRN_OBJ_TABLE_NO_KEY, + NULL, + window->grouped_table); + if (!sorted) { + ERR(ctx->rc, + "[window][set][sort-keys] " + "failed to create a table to store sorted records: %s", + ctx->errbuf); + GRN_API_RETURN(ctx->rc); + } - grn_table_cursor_get_value(ctx, cursor, &value); - record_id = *((grn_id *)value); - GRN_RECORD_PUT(ctx, &(window->ids), record_id); - } GRN_TABLE_EACH_END(ctx, cursor); + grn_table_sort(ctx, + window->grouped_table, + 0, -1, + sorted, + sort_keys, n_sort_keys); + if (ctx->rc != GRN_SUCCESS) { + ERR(ctx->rc, + "[window][set][sort-keys] " + "failed to sort: %s", + ctx->errbuf); + grn_obj_unlink(ctx, sorted); + GRN_API_RETURN(ctx->rc); + } - grn_obj_unlink(ctx, sorted); + GRN_BULK_REWIND(&(window->ids)); + GRN_TABLE_EACH_BEGIN(ctx, sorted, cursor, id) { + void *value; + grn_id record_id; + + grn_table_cursor_get_value(ctx, cursor, &value); + if (window->table == window->grouped_table) { + record_id = *((grn_id *)value); + } else { + grn_id grouped_record_id; + grouped_record_id = *((grn_id *)value); + grn_table_get_key(ctx, + window->grouped_table, + grouped_record_id, + &record_id, sizeof(grn_id)); + } + GRN_RECORD_PUT(ctx, &(window->ids), record_id); + } GRN_TABLE_EACH_END(ctx, cursor); + + grn_obj_close(ctx, sorted); + } else { + GRN_BULK_REWIND(&(window->ids)); + GRN_TABLE_EACH_BEGIN(ctx, window->grouped_table, cursor, id) { + grn_id record_id; + if (window->table == window->grouped_table) { + record_id = id; + } else { + grn_table_get_key(ctx, + window->grouped_table, + id, + &record_id, sizeof(grn_id)); + } + GRN_RECORD_PUT(ctx, &(window->ids), id); + } GRN_TABLE_EACH_END(ctx, cursor); + } window->n_ids = GRN_BULK_VSIZE(&(window->ids)) / sizeof(grn_id); @@ -300,6 +334,29 @@ grn_expr_call_window_function(grn_ctx *ctx, return rc; } +static grn_rc +grn_table_apply_window_function_per_group(grn_ctx *ctx, + grn_obj *table, + grn_obj *grouped_table, + grn_window_definition *definition, + grn_obj *output_column, + grn_obj *window_function_call) +{ + grn_window window; + + grn_window_init(ctx, &window, table, grouped_table); + grn_window_set_sort_keys(ctx, &window, + definition->sort_keys, + definition->n_sort_keys); + grn_expr_call_window_function(ctx, + output_column, + &window, + window_function_call); + grn_window_fin(ctx, &window); + + return GRN_SUCCESS; +} + grn_rc grn_table_apply_window_function(grn_ctx *ctx, grn_obj *table, @@ -307,7 +364,7 @@ grn_table_apply_window_function(grn_ctx *ctx, grn_window_definition *definition, grn_obj *window_function_call) { - grn_window window; + grn_rc rc; GRN_API_ENTER; @@ -329,16 +386,72 @@ grn_table_apply_window_function(grn_ctx *ctx, GRN_API_RETURN(ctx->rc); } - grn_window_init(ctx, &window, table); - grn_window_set_direction(ctx, &window, window.direction); - grn_window_set_sort_keys(ctx, &window, - definition->sort_keys, - definition->n_sort_keys); - grn_expr_call_window_function(ctx, - output_column, - &window, - window_function_call); - grn_window_fin(ctx, &window); + if (definition->group_keys) { + grn_table_group_result grouped; + grn_obj subrecs; + + grouped.table = NULL; + grouped.key_begin = 0; + grouped.key_end = definition->n_group_keys; + grouped.limit = -1; + grouped.flags = GRN_TABLE_GROUP_CALC_COUNT; + grouped.op = 0; + grouped.max_n_subrecs = grn_table_size(ctx, table); + grouped.calc_target = NULL; + + /* TODO: grn_table_group() should support table output for sub records? */ + GRN_RECORD_INIT(&subrecs, GRN_OBJ_VECTOR, grn_obj_id(ctx, table)); + grn_bulk_reserve(ctx, &subrecs, sizeof(grn_obj *) * grouped.max_n_subrecs); + grn_table_group(ctx, + table, + definition->group_keys, + definition->n_group_keys, + &grouped, + 1); + GRN_TABLE_EACH_BEGIN(ctx, grouped.table, cursor, id) { + grn_obj *grouped_table; + + grouped_table = grn_table_create(ctx, + NULL, 0, + NULL, + GRN_TABLE_HASH_KEY, + table, + NULL); + { + unsigned int i, n; + GRN_BULK_REWIND(&subrecs); + n = grn_table_get_subrecs(ctx, + grouped.table, + id, + (grn_id *)GRN_BULK_HEAD(&subrecs), + NULL, + grouped.max_n_subrecs); + for (i = 0; i < n; i++) { + grn_id subrec_id; + subrec_id = GRN_RECORD_VALUE_AT(&subrecs, i); + grn_table_add(ctx, grouped_table, &subrec_id, sizeof(grn_id), NULL); + } + } + /* TODO: rc handling */ + rc = grn_table_apply_window_function_per_group(ctx, + table, + grouped_table, + definition, + output_column, + window_function_call); + grn_obj_close(ctx, grouped_table); + } GRN_TABLE_EACH_END(ctx, cursor); + + grn_obj_close(ctx, grouped.table); + GRN_OBJ_FIN(ctx, &subrecs); + } else { + rc = grn_table_apply_window_function_per_group(ctx, + table, + table, + definition, + output_column, + window_function_call); + } - GRN_API_RETURN(GRN_SUCCESS); + GRN_API_RETURN(rc); } Added: test/command/suite/select/columns/window_function/window_sum/group.expected (+75 -0) 100644 =================================================================== --- /dev/null +++ test/command/suite/select/columns/window_function/window_sum/group.expected 2017-02-01 10:42:20 +0900 (2847c88) @@ -0,0 +1,75 @@ +table_create Logs TABLE_NO_KEY +[[0,0.0,0.0],true] +column_create Logs item COLUMN_SCALAR ShortText +[[0,0.0,0.0],true] +column_create Logs price COLUMN_SCALAR UInt32 +[[0,0.0,0.0],true] +load --table Logs +[ +{"item": "item1", "price": 666}, +{"item": "item1", "price": 999}, +{"item": "item1", "price": 777}, +{"item": "item2", "price": 111}, +{"item": "item2", "price": 333}, +{"item": "item3", "price": 222} +] +[[0,0.0,0.0],6] +select Logs --columns[sum].stage initial --columns[sum].value 'window_sum(price)' --columns[sum].type UInt32 --columns[sum].window.group_keys item --columns[sum].window.sort_keys price --output_columns 'item, price, sum' --sort_keys item,price +[ + [ + 0, + 0.0, + 0.0 + ], + [ + [ + [ + 6 + ], + [ + [ + "item", + "ShortText" + ], + [ + "price", + "UInt32" + ], + [ + "sum", + "UInt32" + ] + ], + [ + "item1", + 666, + 666 + ], + [ + "item1", + 777, + 1443 + ], + [ + "item1", + 999, + 2442 + ], + [ + "item2", + 111, + 111 + ], + [ + "item2", + 333, + 444 + ], + [ + "item3", + 222, + 222 + ] + ] + ] +] Added: test/command/suite/select/columns/window_function/window_sum/group.test (+22 -0) 100644 =================================================================== --- /dev/null +++ test/command/suite/select/columns/window_function/window_sum/group.test 2017-02-01 10:42:20 +0900 (3a6f148) @@ -0,0 +1,22 @@ +table_create Logs TABLE_NO_KEY +column_create Logs item COLUMN_SCALAR ShortText +column_create Logs price COLUMN_SCALAR UInt32 + +load --table Logs +[ +{"item": "item1", "price": 666}, +{"item": "item1", "price": 999}, +{"item": "item1", "price": 777}, +{"item": "item2", "price": 111}, +{"item": "item2", "price": 333}, +{"item": "item3", "price": 222} +] + +select Logs \ + --columns[sum].stage initial \ + --columns[sum].value 'window_sum(price)' \ + --columns[sum].type UInt32 \ + --columns[sum].window.group_keys item \ + --columns[sum].window.sort_keys price \ + --output_columns 'item, price, sum' \ + --sort_keys item,price -------------- next part -------------- HTML����������������������������...Descargar
