|
| 1 | +/* |
| 2 | + * Copyright (C) 2021 Liquidaty and zsv contributors. All rights reserved. |
| 3 | + * This file is part of zsv/lib, distributed under the license defined at |
| 4 | + * https://opensource.org/licenses/MIT |
| 5 | + */ |
| 6 | + |
| 7 | +#include <assert.h> |
| 8 | +#include <errno.h> |
| 9 | +#include "../external/sqlite3/sqlite3.h" |
| 10 | +#include <zsv/ext/implementation.h> |
| 11 | +#include <zsv/ext/sheet.h> |
| 12 | +#include <zsv/utils/writer.h> |
| 13 | +#include <zsv/utils/file.h> |
| 14 | +#include <zsv/utils/prop.h> |
| 15 | +#include "file.h" |
| 16 | +#include "handlers_internal.h" |
| 17 | +#include "../curses.h" |
| 18 | +#include "../sql_internal.h" |
| 19 | + |
| 20 | +struct pivot_row { |
| 21 | + char *value; // to do: this will be the drill-down criteria |
| 22 | +}; |
| 23 | + |
| 24 | +struct pivot_data { |
| 25 | + char *value_sql; // the sql expression entered by the user e.g. City |
| 26 | + char *data_filename; |
| 27 | + struct { |
| 28 | + struct pivot_row *data; // for each row, the value of the sql expression e.g. New York |
| 29 | + size_t capacity; |
| 30 | + size_t used; |
| 31 | + } rows; |
| 32 | +}; |
| 33 | + |
| 34 | +static void pivot_data_delete(void *h) { |
| 35 | + struct pivot_data *pd = h; |
| 36 | + if (pd) { |
| 37 | + for (size_t i = 0; i < pd->rows.used; i++) |
| 38 | + free(pd->rows.data[i].value); |
| 39 | + free(pd->rows.data); |
| 40 | + free(pd->value_sql); |
| 41 | + free(pd->data_filename); |
| 42 | + free(pd); |
| 43 | + } |
| 44 | +} |
| 45 | + |
| 46 | +static struct pivot_data *pivot_data_new(const char *data_filename, const char *value_sql) { |
| 47 | + struct pivot_data *pd = calloc(1, sizeof(*pd)); |
| 48 | + if (pd && (pd->value_sql = strdup(value_sql)) && (pd->data_filename = strdup(data_filename))) |
| 49 | + return pd; |
| 50 | + pivot_data_delete(pd); |
| 51 | + return NULL; |
| 52 | +} |
| 53 | + |
| 54 | +#define ZSV_MYSHEET_PIVOT_DATA_ROWS_INITIAL 32 |
| 55 | +static int pivot_data_grow(struct pivot_data *pd) { |
| 56 | + if (pd->rows.used == pd->rows.capacity) { |
| 57 | + size_t new_capacity = pd->rows.capacity == 0 ? ZSV_MYSHEET_PIVOT_DATA_ROWS_INITIAL : pd->rows.capacity * 2; |
| 58 | + struct pivot_row *new_data = realloc(pd->rows.data, new_capacity * sizeof(*pd->rows.data)); |
| 59 | + if (!new_data) |
| 60 | + return ENOMEM; |
| 61 | + pd->rows.data = new_data; |
| 62 | + pd->rows.capacity = new_capacity; |
| 63 | + } |
| 64 | + return 0; |
| 65 | +} |
| 66 | + |
| 67 | +static int add_pivot_row(struct pivot_data *pd, const char *value, size_t len) { |
| 68 | + int err = pivot_data_grow(pd); |
| 69 | + char *value_dup = NULL; |
| 70 | + if (!err && value && len) { |
| 71 | + value_dup = malloc(len + 1); |
| 72 | + if (value_dup) { |
| 73 | + memcpy(value_dup, value, len); |
| 74 | + value_dup[len] = '\0'; |
| 75 | + } |
| 76 | + } |
| 77 | + pd->rows.data[pd->rows.used++].value = value_dup; |
| 78 | + return err; |
| 79 | +} |
| 80 | + |
| 81 | +static struct pivot_row *get_pivot_row_data(struct pivot_data *pd, size_t row_ix) { |
| 82 | + if (pd && row_ix < pd->rows.used) |
| 83 | + return &pd->rows.data[row_ix]; |
| 84 | + return NULL; |
| 85 | +} |
| 86 | + |
| 87 | +// TO DO: return zsvsheet_status |
| 88 | +static enum zsv_ext_status get_cell_attrs(void *pdh, int *attrs, size_t start_row, size_t row_count, size_t cols) { |
| 89 | + struct pivot_data *pd = pdh; |
| 90 | + size_t end_row = start_row + row_count; |
| 91 | + int attr = 0; |
| 92 | + |
| 93 | +#ifdef A_BOLD |
| 94 | + attr |= A_BOLD; |
| 95 | +#endif |
| 96 | +// Absent on Mac OSX 13 |
| 97 | +#ifdef A_ITALIC |
| 98 | + attr |= A_ITALIC; |
| 99 | +#endif |
| 100 | + |
| 101 | + if (end_row > pd->rows.used) |
| 102 | + end_row = pd->rows.used; |
| 103 | + for (size_t i = start_row; i < end_row; i++) |
| 104 | + attrs[i * cols] = attr; |
| 105 | + return zsv_ext_status_ok; |
| 106 | +} |
| 107 | + |
| 108 | +static void pivot_on_header_cell(void *ctx, size_t col_ix, const char *colname) { |
| 109 | + (void)colname; |
| 110 | + if (col_ix == 0) |
| 111 | + add_pivot_row(ctx, NULL, 0); |
| 112 | +} |
| 113 | + |
| 114 | +static void pivot_on_data_cell(void *ctx, size_t col_ix, const char *text, size_t len) { |
| 115 | + if (col_ix == 0) |
| 116 | + add_pivot_row(ctx, text, len); |
| 117 | +} |
| 118 | + |
| 119 | +static zsvsheet_status zsv_sqlite3_to_csv(zsvsheet_proc_context_t pctx, struct zsv_sqlite3_db *zdb, const char *sql, |
| 120 | + void *ctx, void (*on_header_cell)(void *, size_t, const char *), |
| 121 | + void (*on_data_cell)(void *, size_t, const char *, size_t len)) { |
| 122 | + const char *err_msg = NULL; |
| 123 | + zsvsheet_status zst = zsvsheet_status_error; |
| 124 | + sqlite3_stmt *stmt = NULL; |
| 125 | + |
| 126 | + if ((zdb->rc = sqlite3_prepare_v2(zdb->db, sql, -1, &stmt, NULL)) == SQLITE_OK) { |
| 127 | + char *tmp_fn = zsv_get_temp_filename("zsv_mysheet_ext_XXXXXXXX"); |
| 128 | + struct zsv_csv_writer_options writer_opts = zsv_writer_get_default_opts(); |
| 129 | + zsv_csv_writer cw = NULL; |
| 130 | + if (!tmp_fn) |
| 131 | + zst = zsvsheet_status_memory; |
| 132 | + else if (!(writer_opts.stream = fopen(tmp_fn, "wb"))) { |
| 133 | + zst = zsvsheet_status_error; |
| 134 | + err_msg = strerror(errno); |
| 135 | + } else if (!(cw = zsv_writer_new(&writer_opts))) |
| 136 | + zst = zsvsheet_status_memory; |
| 137 | + else { |
| 138 | + zst = zsvsheet_status_ok; |
| 139 | + unsigned char cw_buff[1024]; |
| 140 | + zsv_writer_set_temp_buff(cw, cw_buff, sizeof(cw_buff)); |
| 141 | + |
| 142 | + int col_count = sqlite3_column_count(stmt); |
| 143 | + // write header row |
| 144 | + for (int i = 0; i < col_count; i++) { |
| 145 | + const char *colname = sqlite3_column_name(stmt, i); |
| 146 | + zsv_writer_cell(cw, !i, (const unsigned char *)colname, colname ? strlen(colname) : 0, 1); |
| 147 | + if (on_header_cell) |
| 148 | + on_header_cell(ctx, i, colname); |
| 149 | + } |
| 150 | + |
| 151 | + // write sql results |
| 152 | + while (sqlite3_step(stmt) == SQLITE_ROW) { |
| 153 | + for (int i = 0; i < col_count; i++) { |
| 154 | + const unsigned char *text = sqlite3_column_text(stmt, i); |
| 155 | + int len = text ? sqlite3_column_bytes(stmt, i) : 0; |
| 156 | + zsv_writer_cell(cw, !i, text, len, 1); |
| 157 | + if (on_data_cell) |
| 158 | + on_data_cell(ctx, i, (const char *)text, len); |
| 159 | + } |
| 160 | + } |
| 161 | + } |
| 162 | + if (cw) |
| 163 | + zsv_writer_delete(cw); |
| 164 | + if (writer_opts.stream) |
| 165 | + fclose(writer_opts.stream); |
| 166 | + |
| 167 | + if (tmp_fn && zsv_file_exists(tmp_fn)) { |
| 168 | + struct zsvsheet_ui_buffer_opts uibopts = {0}; |
| 169 | + uibopts.data_filename = tmp_fn; |
| 170 | + zst = zsvsheet_open_file_opts(pctx, &uibopts); |
| 171 | + } else { |
| 172 | + if (zst == zsvsheet_status_ok) { |
| 173 | + zst = zsvsheet_status_error; // to do: make this more specific |
| 174 | + if (!err_msg && zdb && zdb->rc != SQLITE_OK) |
| 175 | + err_msg = sqlite3_errmsg(zdb->db); |
| 176 | + } |
| 177 | + } |
| 178 | + if (zst != zsvsheet_status_ok) |
| 179 | + free(tmp_fn); |
| 180 | + } |
| 181 | + if (stmt) |
| 182 | + sqlite3_finalize(stmt); |
| 183 | + if (err_msg) |
| 184 | + zsvsheet_set_status(ctx, "Error: %s", err_msg); |
| 185 | + return zst; |
| 186 | +} |
| 187 | + |
| 188 | +zsvsheet_status pivot_drill_down(zsvsheet_proc_context_t ctx) { |
| 189 | + enum zsvsheet_status zst = zsvsheet_status_ok; |
| 190 | + zsvsheet_buffer_t buff = zsvsheet_buffer_current(ctx); |
| 191 | + struct pivot_data *pd; |
| 192 | + struct zsvsheet_rowcol rc; |
| 193 | + if (zsvsheet_buffer_get_ctx(buff, (void **)&pd) != zsv_ext_status_ok || |
| 194 | + zsvsheet_buffer_get_selected_cell(buff, &rc) != zsvsheet_status_ok) { |
| 195 | + return zsvsheet_status_error; |
| 196 | + } |
| 197 | + struct pivot_row *pr = get_pivot_row_data(pd, rc.row); |
| 198 | + if (pd && pd->data_filename && pd->value_sql && pr) { |
| 199 | + struct zsv_sqlite3_dbopts dbopts = {0}; |
| 200 | + sqlite3_str *sql_str = NULL; |
| 201 | + struct zsv_sqlite3_db *zdb = zsv_sqlite3_db_new(&dbopts); |
| 202 | + |
| 203 | + if (!zdb || !(sql_str = sqlite3_str_new(zdb->db))) |
| 204 | + zst = zsvsheet_status_memory; |
| 205 | + else if (zdb->rc == SQLITE_OK && zsv_sqlite3_add_csv(zdb, pd->data_filename, NULL, NULL) == SQLITE_OK) { |
| 206 | + if (zsvsheet_buffer_info(buff).has_row_num) |
| 207 | + sqlite3_str_appendf(sql_str, "select *"); |
| 208 | + else |
| 209 | + sqlite3_str_appendf(sql_str, "select rowid as [Row #], *"); |
| 210 | + sqlite3_str_appendf(sql_str, " from data where %s = %Q", pd->value_sql, pr->value); |
| 211 | + fprintf(stderr, "SQL: %s\n", sqlite3_str_value(sql_str)); |
| 212 | + zst = zsv_sqlite3_to_csv(ctx, zdb, sqlite3_str_value(sql_str), NULL, NULL, NULL); |
| 213 | + } |
| 214 | + |
| 215 | + if (sql_str) |
| 216 | + sqlite3_free(sqlite3_str_finish(sql_str)); |
| 217 | + if (zdb) { |
| 218 | + if (zst != zsvsheet_status_ok) { |
| 219 | + // to do: consolidate this with same code in sql.c |
| 220 | + if (zdb->err_msg) |
| 221 | + fprintf(stderr, "Error: %s\n", zdb->err_msg); |
| 222 | + else if (!zdb->db) |
| 223 | + fprintf(stderr, "Error (unable to open db, code %i): %s\n", zdb->rc, sqlite3_errstr(zdb->rc)); |
| 224 | + else if (zdb->rc != SQLITE_OK) |
| 225 | + fprintf(stderr, "Error (code %i): %s\n", zdb->rc, sqlite3_errstr(zdb->rc)); |
| 226 | + } |
| 227 | + zsv_sqlite3_db_delete(zdb); |
| 228 | + } |
| 229 | + } |
| 230 | + return zst; |
| 231 | +} |
| 232 | + |
| 233 | +/** |
| 234 | + * Here we define a custom command for the zsv `sheet` feature |
| 235 | + */ |
| 236 | +static enum zsvsheet_status zsvsheet_pivot_handler(struct zsvsheet_proc_context *ctx, char *prompt_expr) { |
| 237 | + char result_buffer[256] = {0}; |
| 238 | + const char *expr; |
| 239 | + struct zsvsheet_rowcol rc; |
| 240 | + int ch = zsvsheet_ext_keypress(ctx); |
| 241 | + if (ch < 0) |
| 242 | + return zsvsheet_status_error; |
| 243 | + |
| 244 | + zsvsheet_buffer_t buff = zsvsheet_buffer_current(ctx); |
| 245 | + const char *data_filename = NULL; |
| 246 | + if (buff) |
| 247 | + data_filename = zsvsheet_buffer_data_filename(buff); |
| 248 | + |
| 249 | + if (!data_filename) { // TO DO: check that the underlying data is a tabular file and we know how to parse |
| 250 | + zsvsheet_set_status(ctx, "Pivot table only available for tabular data buffers"); |
| 251 | + return zsvsheet_status_ok; |
| 252 | + } |
| 253 | + |
| 254 | + switch (ctx->proc_id) { |
| 255 | + case zsvsheet_builtin_proc_pivot_expr: |
| 256 | + zsvsheet_ext_prompt(ctx, result_buffer, sizeof(result_buffer), "Pivot table: Enter group-by SQL expr"); |
| 257 | + if (*result_buffer == '\0') |
| 258 | + return zsvsheet_status_ok; |
| 259 | + expr = result_buffer; |
| 260 | + break; |
| 261 | + case zsvsheet_builtin_proc_pivot_cur_col: |
| 262 | + if (zsvsheet_buffer_get_selected_cell(buff, &rc) != zsvsheet_status_ok) |
| 263 | + return zsvsheet_status_error; |
| 264 | + expr = zsvsheet_ui_buffer_get_header(buff, rc.col); |
| 265 | + assert(expr); |
| 266 | + break; |
| 267 | + default: |
| 268 | + assert(0); |
| 269 | + return zsvsheet_status_error; |
| 270 | + } |
| 271 | + |
| 272 | + enum zsvsheet_status zst = zsvsheet_status_ok; |
| 273 | + struct zsv_sqlite3_dbopts dbopts = {0}; |
| 274 | + struct zsv_sqlite3_db *zdb = zsv_sqlite3_db_new(&dbopts); |
| 275 | + sqlite3_str *sql_str = NULL; |
| 276 | + struct pivot_data *pd = NULL; |
| 277 | + if (!zdb || !(sql_str = sqlite3_str_new(zdb->db))) |
| 278 | + zst = zsvsheet_status_memory; |
| 279 | + else if (zdb->rc == SQLITE_OK && zsv_sqlite3_add_csv(zdb, data_filename, NULL, NULL) == SQLITE_OK) { |
| 280 | + sqlite3_str_appendf(sql_str, "select %s as value, count(1) as Count from data group by %s", expr, expr); |
| 281 | + if (!(pd = pivot_data_new(data_filename, expr))) |
| 282 | + zst = zsvsheet_status_memory; |
| 283 | + else { |
| 284 | + zst = zsv_sqlite3_to_csv(ctx, zdb, sqlite3_str_value(sql_str), pd, pivot_on_header_cell, pivot_on_data_cell); |
| 285 | + if (zst == zsvsheet_status_ok) { |
| 286 | + buff = zsvsheet_buffer_current(ctx); |
| 287 | + zsvsheet_buffer_set_ctx(buff, pd, pivot_data_delete); |
| 288 | + zsvsheet_buffer_set_cell_attrs(buff, get_cell_attrs); |
| 289 | + zsvsheet_buffer_on_newline(buff, pivot_drill_down); |
| 290 | + pd = NULL; // so that it isn't cleaned up below |
| 291 | + } |
| 292 | + } |
| 293 | + // TO DO: add param to ext_sheet_open_file to set filename vs data_filename, and set buffer type or proc owner |
| 294 | + // TO DO: add way to attach custom context, and custom context destructor, to the new buffer |
| 295 | + // TO DO: add cell highlighting |
| 296 | + } |
| 297 | + |
| 298 | + zsv_sqlite3_db_delete(zdb); |
| 299 | + if (sql_str) |
| 300 | + sqlite3_free(sqlite3_str_finish(sql_str)); |
| 301 | + pivot_data_delete(pd); |
| 302 | + return zst; |
| 303 | +} |
0 commit comments