| 1 | // =============================================================== // |
|---|
| 2 | // // |
|---|
| 3 | // File : adChangeKey.cxx // |
|---|
| 4 | // Purpose : Changekey management // |
|---|
| 5 | // // |
|---|
| 6 | // Coded by Elmar Pruesse and Ralf Westram in May 2009 // |
|---|
| 7 | // Institute of Microbiology (Technical University Munich) // |
|---|
| 8 | // http://www.arb-home.de/ // |
|---|
| 9 | // // |
|---|
| 10 | // =============================================================== // |
|---|
| 11 | |
|---|
| 12 | #include <arbdbt.h> |
|---|
| 13 | #include <arb_global_defs.h> |
|---|
| 14 | #include "gb_local.h" |
|---|
| 15 | |
|---|
| 16 | GBDATA *GBT_get_changekey(GBDATA *gb_main, const char *key, const char *change_key_path) { |
|---|
| 17 | // get the container of an item key description |
|---|
| 18 | |
|---|
| 19 | // @@@ check if search for CHANGEKEY_NAME should be case-sensitive! |
|---|
| 20 | GBDATA *gb_key = NULp; |
|---|
| 21 | GBDATA *gb_key_data = GB_search(gb_main, change_key_path, |
|---|
| 22 | GB_CREATE_CONTAINER); |
|---|
| 23 | |
|---|
| 24 | if (gb_key_data) { |
|---|
| 25 | GBDATA *gb_key_name = GB_find_string(gb_key_data, CHANGEKEY_NAME, key, GB_IGNORE_CASE, SEARCH_GRANDCHILD); |
|---|
| 26 | if (gb_key_name) { |
|---|
| 27 | gb_key = GB_get_father(gb_key_name); |
|---|
| 28 | } |
|---|
| 29 | } |
|---|
| 30 | return gb_key; |
|---|
| 31 | } |
|---|
| 32 | |
|---|
| 33 | GB_TYPES GBT_get_type_of_changekey(GBDATA *gb_main, const char *field_name, const char *change_key_path) { |
|---|
| 34 | // get the type of an item key |
|---|
| 35 | GB_TYPES type = GB_NONE; |
|---|
| 36 | GBDATA *gbd = GBT_get_changekey(gb_main, field_name, change_key_path); |
|---|
| 37 | |
|---|
| 38 | if (gbd) { |
|---|
| 39 | long *typePtr = GBT_read_int(gbd, CHANGEKEY_TYPE); |
|---|
| 40 | if (typePtr) { |
|---|
| 41 | type = (GB_TYPES)*typePtr; |
|---|
| 42 | } |
|---|
| 43 | } |
|---|
| 44 | |
|---|
| 45 | return type; |
|---|
| 46 | } |
|---|
| 47 | |
|---|
| 48 | static GB_ERROR gbt_set_type_of_changekey(GBDATA *gb_main, const char *field_name, GB_TYPES type, const char *change_key_path) { |
|---|
| 49 | GB_ERROR error = NULp; |
|---|
| 50 | GBDATA *gbd = GBT_get_changekey(gb_main, field_name, change_key_path); |
|---|
| 51 | |
|---|
| 52 | if (!gbd) { |
|---|
| 53 | error = GBS_global_string("Can't set type of nonexistent changekey \"%s\"", field_name); |
|---|
| 54 | } |
|---|
| 55 | else { |
|---|
| 56 | error = GBT_write_int(gbd, CHANGEKEY_TYPE, type); |
|---|
| 57 | } |
|---|
| 58 | return error; |
|---|
| 59 | } |
|---|
| 60 | |
|---|
| 61 | GBDATA *GBT_searchOrCreate_itemfield_according_to_changekey(GBDATA *gb_item, const char *field_name, const char *change_key_path) { |
|---|
| 62 | /*! search or create an item entry. |
|---|
| 63 | * If the entry exists, the result is identical to GB_search(gb_item, field_name, GB_FIND). |
|---|
| 64 | * If the entry does not exist, an entry with the type stored in the changekey-table will be created. |
|---|
| 65 | * @return created itemfield or NULp in case of error (which is exported in that case) |
|---|
| 66 | */ |
|---|
| 67 | |
|---|
| 68 | gb_assert(!GB_have_error()); |
|---|
| 69 | GBDATA *gb_entry = GB_search(gb_item, field_name, GB_FIND); |
|---|
| 70 | if (!gb_entry) { |
|---|
| 71 | GB_clear_error(); |
|---|
| 72 | |
|---|
| 73 | GB_TYPES type = GBT_get_type_of_changekey(GB_get_root(gb_item), field_name, change_key_path); |
|---|
| 74 | if (type == GB_NONE) { |
|---|
| 75 | GB_export_errorf("Cannot create field '%s' (no type information available)", field_name); |
|---|
| 76 | } |
|---|
| 77 | else { |
|---|
| 78 | gb_entry = GB_search(gb_item, field_name, type); |
|---|
| 79 | } |
|---|
| 80 | } |
|---|
| 81 | gb_assert(gb_entry || GB_have_error()); |
|---|
| 82 | return gb_entry; |
|---|
| 83 | } |
|---|
| 84 | |
|---|
| 85 | |
|---|
| 86 | GB_ERROR GBT_add_new_changekey_to_keypath(GBDATA *gb_main, const char *name, GB_TYPES type, const char *keypath) { |
|---|
| 87 | /*! Add a new or check an existing changekey for items using 'keypath' |
|---|
| 88 | * @param gb_main database |
|---|
| 89 | * @param name name of changekey. may be hierarchical (e.g. 'ali_16s/data') |
|---|
| 90 | * @param type field type of the key (yields error on mismatch!) |
|---|
| 91 | * @param keypath location in database where changekeys are stored. |
|---|
| 92 | * @return error |
|---|
| 93 | */ |
|---|
| 94 | GB_ERROR error = NULp; |
|---|
| 95 | GBDATA *gb_key = GBT_get_changekey(gb_main, name, keypath); |
|---|
| 96 | const char *c = GB_first_non_key_char(name); |
|---|
| 97 | |
|---|
| 98 | if (c) { |
|---|
| 99 | char *new_name = ARB_strdup(name); |
|---|
| 100 | |
|---|
| 101 | *(char*)GB_first_non_key_char(new_name) = 0; |
|---|
| 102 | |
|---|
| 103 | if (*c == '/') error = GBT_add_new_changekey_to_keypath(gb_main, new_name, GB_DB, keypath); |
|---|
| 104 | else error = GBS_global_string("Cannot add '%s' to your key list (illegal character '%c')", name, *c); |
|---|
| 105 | |
|---|
| 106 | free(new_name); |
|---|
| 107 | } |
|---|
| 108 | |
|---|
| 109 | if (!error) { |
|---|
| 110 | if (!gb_key) { // create new key |
|---|
| 111 | GBDATA *gb_key_data = GB_search(gb_main, keypath, GB_CREATE_CONTAINER); |
|---|
| 112 | gb_key = gb_key_data ? GB_create_container(gb_key_data, CHANGEKEY) : NULp; |
|---|
| 113 | |
|---|
| 114 | if (!gb_key) error = GB_await_error(); |
|---|
| 115 | else { |
|---|
| 116 | error = GBT_write_string(gb_key, CHANGEKEY_NAME, name); |
|---|
| 117 | if (!error) error = GBT_write_int(gb_key, CHANGEKEY_TYPE, type); |
|---|
| 118 | } |
|---|
| 119 | } |
|---|
| 120 | else { // check type of existing key |
|---|
| 121 | long *elem_type = GBT_read_int(gb_key, CHANGEKEY_TYPE); |
|---|
| 122 | |
|---|
| 123 | if (!elem_type) error = GB_await_error(); |
|---|
| 124 | else if (*elem_type != type) error = GBS_global_string("Key '%s' exists, but has different type", name); |
|---|
| 125 | } |
|---|
| 126 | } |
|---|
| 127 | |
|---|
| 128 | gb_assert(gb_key || error); |
|---|
| 129 | |
|---|
| 130 | return error; |
|---|
| 131 | } |
|---|
| 132 | |
|---|
| 133 | GB_ERROR GBT_add_new_species_changekey(GBDATA *gb_main, const char *name, GB_TYPES type) { |
|---|
| 134 | return GBT_add_new_changekey_to_keypath(gb_main, name, type, CHANGE_KEY_PATH); |
|---|
| 135 | } |
|---|
| 136 | |
|---|
| 137 | GB_ERROR GBT_add_new_gene_changekey(GBDATA *gb_main, const char *name, GB_TYPES type) { |
|---|
| 138 | return GBT_add_new_changekey_to_keypath(gb_main, name, type, CHANGE_KEY_PATH_GENES); |
|---|
| 139 | } |
|---|
| 140 | |
|---|
| 141 | GB_ERROR GBT_add_new_experiment_changekey(GBDATA *gb_main, const char *name, GB_TYPES type) { |
|---|
| 142 | return GBT_add_new_changekey_to_keypath(gb_main, name, type, CHANGE_KEY_PATH_EXPERIMENTS); |
|---|
| 143 | } |
|---|
| 144 | |
|---|
| 145 | GB_ERROR GBT_add_new_changekey(GBDATA *gb_main, const char *name, int type) { |
|---|
| 146 | // goes to header: __ATTR__DEPRECATED_TODO("obsolete. use GBT_add_new_species_changekey() instead") |
|---|
| 147 | // |
|---|
| 148 | // this method is required for sina 1.3 binary backward compatibility |
|---|
| 149 | // (needed as long as we support ubuntu 10.04) |
|---|
| 150 | // |
|---|
| 151 | // Note: was broken by [19140]. |
|---|
| 152 | return GBT_add_new_species_changekey(gb_main, name, GB_TYPES(type)); |
|---|
| 153 | } |
|---|
| 154 | |
|---|
| 155 | GB_ERROR GBT_add_alignment_changekeys(GBDATA *gb_main, const char *ali) { |
|---|
| 156 | /*! add changekeys for alignment 'ali' |
|---|
| 157 | */ |
|---|
| 158 | char *dataField = GBS_global_string_copy("%s/data", ali); |
|---|
| 159 | GB_ERROR error = GBT_add_new_species_changekey(gb_main, dataField, GB_STRING); |
|---|
| 160 | free(dataField); |
|---|
| 161 | return error; |
|---|
| 162 | } |
|---|
| 163 | |
|---|
| 164 | static GB_ERROR write_int_converted(GBDATA *gbfield, const char *data, bool trimmed, size_t *rounded) { |
|---|
| 165 | // 'rounded' is incremented each time a float value was |
|---|
| 166 | // converted to integer (even if float value can convert lossless, e.g. "4.0"). |
|---|
| 167 | |
|---|
| 168 | char *end = NULp; |
|---|
| 169 | unsigned long i = strtoul(data, &end, 10); |
|---|
| 170 | GB_ERROR error = NULp; |
|---|
| 171 | |
|---|
| 172 | if (end == data || end[0] != 0) { |
|---|
| 173 | if (trimmed) { |
|---|
| 174 | // fallback: convert to double and round |
|---|
| 175 | |
|---|
| 176 | double d = strtod(data, &end); |
|---|
| 177 | if (end == data || end[0] != 0) { |
|---|
| 178 | error = GBS_global_string("cannot convert '%s' to rounded numeric value", data); |
|---|
| 179 | } |
|---|
| 180 | else { |
|---|
| 181 | (*rounded)++; |
|---|
| 182 | i = d>0 ? (int)(d+0.5) : (int)(d-0.5); |
|---|
| 183 | error = GB_write_int(gbfield, i); |
|---|
| 184 | if (error) error = GBS_global_string("write error (%s)", error); |
|---|
| 185 | } |
|---|
| 186 | } |
|---|
| 187 | else { |
|---|
| 188 | char *trimmed_data = GBS_trim(data); |
|---|
| 189 | error = write_int_converted(gbfield, trimmed_data, true, rounded); |
|---|
| 190 | free(trimmed_data); |
|---|
| 191 | } |
|---|
| 192 | } |
|---|
| 193 | else { |
|---|
| 194 | error = GB_write_int(gbfield, i); |
|---|
| 195 | if (error) error = GBS_global_string("write error (%s)", error); |
|---|
| 196 | } |
|---|
| 197 | |
|---|
| 198 | return error; |
|---|
| 199 | } |
|---|
| 200 | |
|---|
| 201 | static GB_ERROR write_float_converted(GBDATA *gbfield, const char *data, bool trimmed) { |
|---|
| 202 | char *end = NULp; |
|---|
| 203 | float f = strtof(data, &end); |
|---|
| 204 | GB_ERROR error = NULp; |
|---|
| 205 | |
|---|
| 206 | if (end == data || end[0] != 0) { |
|---|
| 207 | if (trimmed) { |
|---|
| 208 | error = GBS_global_string("cannot convert '%s' to numeric value", data); |
|---|
| 209 | } |
|---|
| 210 | else { |
|---|
| 211 | char *trimmed_data = GBS_trim(data); |
|---|
| 212 | error = write_float_converted(gbfield, trimmed_data, true); |
|---|
| 213 | free(trimmed_data); |
|---|
| 214 | } |
|---|
| 215 | } |
|---|
| 216 | else { |
|---|
| 217 | error = GB_write_float(gbfield, f); |
|---|
| 218 | if (error) error = GBS_global_string("write error (%s)", error); |
|---|
| 219 | } |
|---|
| 220 | |
|---|
| 221 | return error; |
|---|
| 222 | } |
|---|
| 223 | |
|---|
| 224 | GB_ERROR GBT_write_int_converted(GBDATA *gb_int_writable, const char *data, size_t *rounded) { |
|---|
| 225 | // Note: has functional similarities to GB_write_autoconv_string |
|---|
| 226 | return write_int_converted(gb_int_writable, data, false, rounded); |
|---|
| 227 | } |
|---|
| 228 | GB_ERROR GBT_write_float_converted(GBDATA *gb_float_writable, const char *data) { |
|---|
| 229 | // Note: has functional similarities to GB_write_autoconv_string |
|---|
| 230 | return write_float_converted(gb_float_writable, data, false); |
|---|
| 231 | } |
|---|
| 232 | |
|---|
| 233 | GB_ERROR GBT_convert_changekey(GBDATA *gb_main, const char *name, GB_TYPES target_type) { |
|---|
| 234 | GB_ERROR error = GB_push_transaction(gb_main); |
|---|
| 235 | bool need_convert = true; |
|---|
| 236 | |
|---|
| 237 | if (!error) { |
|---|
| 238 | GBDATA *gbkey = GBT_get_changekey(gb_main, name, CHANGE_KEY_PATH); |
|---|
| 239 | if (gbkey) { |
|---|
| 240 | GB_TYPES source_type = (GB_TYPES)*GBT_read_int(gbkey, CHANGEKEY_TYPE); |
|---|
| 241 | if (source_type == target_type) need_convert = false; |
|---|
| 242 | } |
|---|
| 243 | else { |
|---|
| 244 | if (!name[0] || strcmp(name, NO_FIELD_SELECTED) == 0) { |
|---|
| 245 | error = "Please select field to convert"; |
|---|
| 246 | } |
|---|
| 247 | else { |
|---|
| 248 | error = GBS_global_string("Unknown changekey '%s'", name); |
|---|
| 249 | } |
|---|
| 250 | } |
|---|
| 251 | } |
|---|
| 252 | |
|---|
| 253 | if (!error && need_convert) { |
|---|
| 254 | GBDATA *gbspec = GBT_first_species(gb_main); |
|---|
| 255 | size_t rounded = 0; |
|---|
| 256 | |
|---|
| 257 | for (; gbspec; gbspec = GBT_next_species(gbspec)) { |
|---|
| 258 | GBDATA *gbfield = GB_entry(gbspec, name); |
|---|
| 259 | |
|---|
| 260 | // If entry does not exist, no need to convert (sparse population is valid => 'NULp' value) |
|---|
| 261 | if (gbfield) { |
|---|
| 262 | char *data = GB_read_as_string(gbfield); |
|---|
| 263 | if (!data) { |
|---|
| 264 | error = GBS_global_string("read error (%s)", GB_await_error()); |
|---|
| 265 | } |
|---|
| 266 | else { |
|---|
| 267 | error = GB_delete(gbfield); |
|---|
| 268 | if (!error) { |
|---|
| 269 | gbfield = GB_create(gbspec, name, target_type); |
|---|
| 270 | if (!gbfield) { |
|---|
| 271 | error = GBS_global_string("create error (%s)", GB_await_error()); |
|---|
| 272 | } |
|---|
| 273 | else { |
|---|
| 274 | switch (target_type) { |
|---|
| 275 | case GB_INT: |
|---|
| 276 | error = GBT_write_int_converted(gbfield, data, &rounded); |
|---|
| 277 | break; |
|---|
| 278 | |
|---|
| 279 | case GB_FLOAT: |
|---|
| 280 | error = GBT_write_float_converted(gbfield, data); |
|---|
| 281 | break; |
|---|
| 282 | |
|---|
| 283 | case GB_STRING: |
|---|
| 284 | error = GB_write_string(gbfield, data); |
|---|
| 285 | if (error) error = GBS_global_string("write error (%s)", error); |
|---|
| 286 | break; |
|---|
| 287 | |
|---|
| 288 | default: |
|---|
| 289 | error = "Conversion is not possible"; |
|---|
| 290 | break; |
|---|
| 291 | } |
|---|
| 292 | } |
|---|
| 293 | } |
|---|
| 294 | free(data); |
|---|
| 295 | } |
|---|
| 296 | } |
|---|
| 297 | if (error) break; |
|---|
| 298 | } |
|---|
| 299 | |
|---|
| 300 | if (error && gbspec) { |
|---|
| 301 | const char *spname = GBT_get_name_or_description(gbspec); |
|---|
| 302 | error = GBS_global_string("%s for species '%s'", error, spname); |
|---|
| 303 | } |
|---|
| 304 | |
|---|
| 305 | if (!error) error = gbt_set_type_of_changekey(gb_main, name, target_type, CHANGE_KEY_PATH); |
|---|
| 306 | if (!error && rounded>0) { |
|---|
| 307 | GB_warningf("%zi values were rounded (loss of precision)", rounded); |
|---|
| 308 | } |
|---|
| 309 | } |
|---|
| 310 | |
|---|
| 311 | if (error) error = GBS_global_string("GBT_convert: %s", error); |
|---|
| 312 | |
|---|
| 313 | return GB_end_transaction(gb_main, error); |
|---|
| 314 | } |
|---|
| 315 | |
|---|