| 1 | // =============================================================== // |
|---|
| 2 | // // |
|---|
| 3 | // File : MG_checkfield.cxx // |
|---|
| 4 | // Purpose : // |
|---|
| 5 | // // |
|---|
| 6 | // Institute of Microbiology (Technical University Munich) // |
|---|
| 7 | // http://www.arb-home.de/ // |
|---|
| 8 | // // |
|---|
| 9 | // =============================================================== // |
|---|
| 10 | |
|---|
| 11 | #include "merge.hxx" |
|---|
| 12 | |
|---|
| 13 | #include <item_sel_list.h> |
|---|
| 14 | |
|---|
| 15 | #include <aw_awar.hxx> |
|---|
| 16 | #include <aw_root.hxx> |
|---|
| 17 | #include <aw_msg.hxx> |
|---|
| 18 | #include <arb_progress.h> |
|---|
| 19 | |
|---|
| 20 | #include <arbdbt.h> |
|---|
| 21 | #include <arb_strbuf.h> |
|---|
| 22 | |
|---|
| 23 | #include <cctype> |
|---|
| 24 | |
|---|
| 25 | #define AWAR_CHECK AWAR_MERGE_TMP "chk/" |
|---|
| 26 | |
|---|
| 27 | #define AWAR_SOURCE_FIELD AWAR_CHECK "source" |
|---|
| 28 | #define AWAR_DEST_FIELD AWAR_CHECK "dest" |
|---|
| 29 | #define AWAR_TOUPPER AWAR_CHECK "ToUpper" |
|---|
| 30 | #define AWAR_EXCLUDE AWAR_CHECK "exclude" |
|---|
| 31 | #define AWAR_CORRECT AWAR_CHECK "correct" |
|---|
| 32 | #define AWAR_ETAG AWAR_CHECK "tag" |
|---|
| 33 | |
|---|
| 34 | |
|---|
| 35 | static int gbs_cmp_strings(char *str1, char *str2, int *tab) { // returns 0 if strings are equal |
|---|
| 36 | char *s1, *s2; |
|---|
| 37 | int c1, c2; |
|---|
| 38 | s1 = str1; |
|---|
| 39 | s2 = str2; |
|---|
| 40 | int count = 10; |
|---|
| 41 | do { |
|---|
| 42 | do { c1 = *(s1++); } while (tab[c1] < 0); |
|---|
| 43 | do { c2 = *(s2++); } while (tab[c2] < 0); |
|---|
| 44 | if (tab[c1] != tab[c2]) { // difference found |
|---|
| 45 | return 1; |
|---|
| 46 | } |
|---|
| 47 | count --; |
|---|
| 48 | } while (count && c1 && c2); |
|---|
| 49 | return 0; |
|---|
| 50 | } |
|---|
| 51 | |
|---|
| 52 | |
|---|
| 53 | static char *GBS_diff_strings(char *str1, char * &str2, char *exclude, long ToUpper, long correct, |
|---|
| 54 | char **res1, char **res2, long *corrrected) { |
|---|
| 55 | |
|---|
| 56 | char buffer1[256]; |
|---|
| 57 | char buffer2[256]; |
|---|
| 58 | char *dest1 = buffer1; |
|---|
| 59 | char *dest2 = buffer2; |
|---|
| 60 | char *s1, *s2; |
|---|
| 61 | int c1, c2; |
|---|
| 62 | int count = 3; |
|---|
| 63 | int tab[256]; |
|---|
| 64 | int i; |
|---|
| 65 | |
|---|
| 66 | s1 = str1; |
|---|
| 67 | s2 = str2; |
|---|
| 68 | *dest1 = 0; |
|---|
| 69 | *dest2 = 0; |
|---|
| 70 | tab[0] = 0; |
|---|
| 71 | char gapchar = '#'; |
|---|
| 72 | if (strlen(exclude)) gapchar = exclude[0]; |
|---|
| 73 | else exclude = 0; |
|---|
| 74 | |
|---|
| 75 | for (i=1; i<256; i++) { |
|---|
| 76 | tab[i] = i; |
|---|
| 77 | if (exclude && strchr(exclude, i)) { |
|---|
| 78 | tab[i] = -1; |
|---|
| 79 | continue; |
|---|
| 80 | } |
|---|
| 81 | if (ToUpper && i >= 'a' && i <= 'z') { |
|---|
| 82 | tab[i] = i-'a'+'A'; |
|---|
| 83 | } |
|---|
| 84 | } |
|---|
| 85 | |
|---|
| 86 | do { |
|---|
| 87 | do { c1 = *(s1++); } while (tab[c1] < 0); |
|---|
| 88 | do { c2 = *(s2++); } while (tab[c2] < 0); |
|---|
| 89 | if (tab[c1] != tab[c2]) { // difference found |
|---|
| 90 | if (correct) { |
|---|
| 91 | // check substitution |
|---|
| 92 | { |
|---|
| 93 | int c = s2[-1]; |
|---|
| 94 | s2[-1] = s1[-1]; |
|---|
| 95 | if (toupper(c1) == toupper(c2) || |
|---|
| 96 | !gbs_cmp_strings(s1, s2, &tab[0])) { |
|---|
| 97 | *corrrected = 1; |
|---|
| 98 | continue; |
|---|
| 99 | } |
|---|
| 100 | s2[-1] = c; |
|---|
| 101 | } |
|---|
| 102 | |
|---|
| 103 | // check insertion in s2 |
|---|
| 104 | if (!gbs_cmp_strings(s1-1, s2, &tab[0])) { |
|---|
| 105 | s2[-1] = gapchar; |
|---|
| 106 | do { c2 = *(s2++); } while (tab[c2] < 0); // eat s2 |
|---|
| 107 | *corrrected = 1; |
|---|
| 108 | continue; |
|---|
| 109 | } |
|---|
| 110 | // check deletion in s2 |
|---|
| 111 | if (!gbs_cmp_strings(s1, s2-1, &tab[0])) { |
|---|
| 112 | int toins = c1; |
|---|
| 113 | char *toinspos = s2-1; |
|---|
| 114 | if (toinspos > str2) toinspos--; |
|---|
| 115 | if (tab[(unsigned char)toinspos[0]]> 0) { // real insertion |
|---|
| 116 | GBS_strstruct *str = GBS_stropen(strlen(str2+10)); |
|---|
| 117 | int pos = s2-str2-1; |
|---|
| 118 | GBS_strncat(str, str2, pos); |
|---|
| 119 | GBS_chrcat(str, toins); |
|---|
| 120 | GBS_strcat(str, str2+pos); |
|---|
| 121 | delete str2; |
|---|
| 122 | str2 = GBS_strclose(str); |
|---|
| 123 | s2 = str2+pos+1; |
|---|
| 124 | *corrrected = 1; |
|---|
| 125 | continue; |
|---|
| 126 | } |
|---|
| 127 | int side=1; // 0 = left 1= right |
|---|
| 128 | if (tab[(unsigned char)s1[0]]<0) side = 0; |
|---|
| 129 | if (! side) { |
|---|
| 130 | while (toinspos > str2 && |
|---|
| 131 | tab[(unsigned char)toinspos[-1]] < 0) toinspos--; |
|---|
| 132 | } |
|---|
| 133 | toinspos[0] = toins; |
|---|
| 134 | *corrrected = 1; |
|---|
| 135 | do { c1 = *(s1++); } while (tab[c1] < 0); // eat s1 |
|---|
| 136 | continue; |
|---|
| 137 | } |
|---|
| 138 | } |
|---|
| 139 | if (count >= 0) { |
|---|
| 140 | sprintf(dest1, "%ti ", s1-str1-1); |
|---|
| 141 | sprintf(dest2, "%ti ", s2-str2-1); |
|---|
| 142 | dest1 += strlen(dest1); |
|---|
| 143 | dest2 += strlen(dest2); |
|---|
| 144 | } |
|---|
| 145 | count --; |
|---|
| 146 | } |
|---|
| 147 | } while (c1 && c2); |
|---|
| 148 | |
|---|
| 149 | if (c1 || c2) { |
|---|
| 150 | sprintf(dest1, "... %ti ", s1-str1-1); |
|---|
| 151 | sprintf(dest2, "... %ti ", s2-str2-1); |
|---|
| 152 | dest1 += strlen(dest1); |
|---|
| 153 | dest2 += strlen(dest2); |
|---|
| 154 | } |
|---|
| 155 | if (count<0) { |
|---|
| 156 | sprintf(dest1, "and %i more", 1-count); |
|---|
| 157 | sprintf(dest2, "and %i more", 1-count); |
|---|
| 158 | dest1 += strlen(dest1); |
|---|
| 159 | dest2 += strlen(dest2); |
|---|
| 160 | } |
|---|
| 161 | if (strlen(buffer1)) { |
|---|
| 162 | *res1 = strdup(buffer1); |
|---|
| 163 | *res2 = strdup(buffer2); |
|---|
| 164 | } |
|---|
| 165 | else { |
|---|
| 166 | *res1 = 0; |
|---|
| 167 | *res2 = 0; |
|---|
| 168 | } |
|---|
| 169 | return 0; |
|---|
| 170 | } |
|---|
| 171 | |
|---|
| 172 | int mg_count_queried(GBDATA *gb_main) { |
|---|
| 173 | int queried = 0; |
|---|
| 174 | for (GBDATA *gb_spec = GBT_first_species(gb_main); |
|---|
| 175 | gb_spec; |
|---|
| 176 | gb_spec = GBT_next_species(gb_spec)) |
|---|
| 177 | { |
|---|
| 178 | if (IS_QUERIED_SPECIES(gb_spec)) queried++; |
|---|
| 179 | } |
|---|
| 180 | return queried; |
|---|
| 181 | } |
|---|
| 182 | |
|---|
| 183 | static void mg_check_field_cb(AW_window *aww) { |
|---|
| 184 | AW_root *root = aww->get_root(); |
|---|
| 185 | GB_ERROR error = 0; |
|---|
| 186 | char *source = root->awar(AWAR_SOURCE_FIELD)->read_string(); |
|---|
| 187 | char *dest = root->awar(AWAR_DEST_FIELD)->read_string(); |
|---|
| 188 | char *exclude = root->awar(AWAR_EXCLUDE)->read_string(); |
|---|
| 189 | long ToUpper = root->awar(AWAR_TOUPPER)->read_int(); |
|---|
| 190 | long correct = root->awar(AWAR_CORRECT)->read_int(); |
|---|
| 191 | char *tag = root->awar(AWAR_ETAG)->read_string(); |
|---|
| 192 | |
|---|
| 193 | if (source[0] == 0) { |
|---|
| 194 | error = "Please select a source field"; |
|---|
| 195 | } |
|---|
| 196 | else if (dest[0] == 0) { |
|---|
| 197 | error = "Please select a dest field"; |
|---|
| 198 | } |
|---|
| 199 | else { |
|---|
| 200 | error = GB_begin_transaction(GLOBAL_gb_src); |
|---|
| 201 | |
|---|
| 202 | if (!error) { |
|---|
| 203 | error = GB_begin_transaction(GLOBAL_gb_dst); |
|---|
| 204 | |
|---|
| 205 | GBDATA *gb_src_species_data = GBT_get_species_data(GLOBAL_gb_src); |
|---|
| 206 | GBDATA *gb_dst_species_data = GBT_get_species_data(GLOBAL_gb_dst); |
|---|
| 207 | |
|---|
| 208 | GBDATA *gb_src_species; |
|---|
| 209 | GBDATA *gb_dst_species; |
|---|
| 210 | |
|---|
| 211 | // First step: count selected species |
|---|
| 212 | arb_progress progress("Checking fields", mg_count_queried(GLOBAL_gb_src)); |
|---|
| 213 | |
|---|
| 214 | // Delete all 'dest' fields in target database |
|---|
| 215 | for (gb_dst_species = GBT_first_species_rel_species_data(gb_dst_species_data); |
|---|
| 216 | gb_dst_species && !error; |
|---|
| 217 | gb_dst_species = GBT_next_species(gb_dst_species)) |
|---|
| 218 | { |
|---|
| 219 | GBDATA *gbd = GB_search(gb_dst_species, dest, GB_FIND); |
|---|
| 220 | if (gbd) error = GB_delete(gbd); |
|---|
| 221 | } |
|---|
| 222 | |
|---|
| 223 | for (gb_src_species = GBT_first_species_rel_species_data(gb_src_species_data); |
|---|
| 224 | gb_src_species && !error; |
|---|
| 225 | gb_src_species = GBT_next_species(gb_src_species)) |
|---|
| 226 | { |
|---|
| 227 | { |
|---|
| 228 | GBDATA *gbd = GB_search(gb_src_species, dest, GB_FIND); |
|---|
| 229 | if (gbd) error = GB_delete(gbd); |
|---|
| 230 | } |
|---|
| 231 | |
|---|
| 232 | if (!error) { |
|---|
| 233 | if (IS_QUERIED_SPECIES(gb_src_species)) { |
|---|
| 234 | const char *src_name = GBT_read_name(gb_src_species); |
|---|
| 235 | gb_dst_species = GB_find_string(gb_dst_species_data, "name", src_name, GB_IGNORE_CASE, SEARCH_GRANDCHILD); |
|---|
| 236 | if (!gb_dst_species) { |
|---|
| 237 | aw_message(GBS_global_string("WARNING: Species %s not found in target DB", src_name)); |
|---|
| 238 | } |
|---|
| 239 | else { |
|---|
| 240 | gb_dst_species = GB_get_father(gb_dst_species); |
|---|
| 241 | |
|---|
| 242 | GBDATA *gb_src_field = GB_search(gb_src_species, source, GB_FIND); |
|---|
| 243 | GBDATA *gb_dst_field = GB_search(gb_dst_species, source, GB_FIND); |
|---|
| 244 | |
|---|
| 245 | char *src_val = gb_src_field ? GB_read_as_tagged_string(gb_src_field, tag) : 0; |
|---|
| 246 | char *dst_val = gb_dst_field ? GB_read_as_tagged_string(gb_dst_field, tag) : 0; |
|---|
| 247 | |
|---|
| 248 | if (src_val || dst_val) { |
|---|
| 249 | char *src_positions = 0; |
|---|
| 250 | char *dst_positions = 0; |
|---|
| 251 | |
|---|
| 252 | if (src_val && dst_val) { |
|---|
| 253 | long corrected = 0; |
|---|
| 254 | GBS_diff_strings(src_val, dst_val, exclude, ToUpper, correct, &src_positions, &dst_positions, &corrected); |
|---|
| 255 | if (corrected) { |
|---|
| 256 | error = GB_write_as_string(gb_dst_field, dst_val); |
|---|
| 257 | if (!error) GB_write_flag(gb_dst_species, 1); |
|---|
| 258 | } |
|---|
| 259 | } |
|---|
| 260 | else { |
|---|
| 261 | src_positions = GBS_global_string_copy("field missing in %s DB", src_val ? "other" : "this"); |
|---|
| 262 | dst_positions = GBS_global_string_copy("field missing in %s DB", dst_val ? "other" : "this"); |
|---|
| 263 | } |
|---|
| 264 | |
|---|
| 265 | if (src_positions && !error) { |
|---|
| 266 | error = GBT_write_string(gb_dst_species, dest, dst_positions); |
|---|
| 267 | if (!error) error = GBT_write_string(gb_src_species, dest, src_positions); |
|---|
| 268 | } |
|---|
| 269 | |
|---|
| 270 | free(dst_positions); |
|---|
| 271 | free(src_positions); |
|---|
| 272 | } |
|---|
| 273 | |
|---|
| 274 | free(dst_val); |
|---|
| 275 | free(src_val); |
|---|
| 276 | } |
|---|
| 277 | progress.inc_and_check_user_abort(error); |
|---|
| 278 | } |
|---|
| 279 | } |
|---|
| 280 | } |
|---|
| 281 | |
|---|
| 282 | error = GB_end_transaction(GLOBAL_gb_src, error); |
|---|
| 283 | error = GB_end_transaction(GLOBAL_gb_dst, error); |
|---|
| 284 | } |
|---|
| 285 | } |
|---|
| 286 | if (error) aw_message(error); |
|---|
| 287 | |
|---|
| 288 | free(tag); |
|---|
| 289 | free(exclude); |
|---|
| 290 | free(dest); |
|---|
| 291 | free(source); |
|---|
| 292 | } |
|---|
| 293 | |
|---|
| 294 | |
|---|
| 295 | AW_window *create_mg_check_fields_window(AW_root *aw_root) { |
|---|
| 296 | aw_root->awar_string(AWAR_SOURCE_FIELD); |
|---|
| 297 | aw_root->awar_string(AWAR_DEST_FIELD, "tmp", AW_ROOT_DEFAULT); |
|---|
| 298 | aw_root->awar_string(AWAR_EXCLUDE, ".-", AW_ROOT_DEFAULT); |
|---|
| 299 | aw_root->awar_string(AWAR_ETAG, ""); |
|---|
| 300 | aw_root->awar_int(AWAR_TOUPPER); |
|---|
| 301 | aw_root->awar_int(AWAR_CORRECT); |
|---|
| 302 | |
|---|
| 303 | AW_window_simple *aws = new AW_window_simple; |
|---|
| 304 | aws->init(aw_root, "MERGE_COMPARE_FIELDS", "COMPARE DATABASE FIELDS"); |
|---|
| 305 | aws->load_xfig("merge/seqcheck.fig"); |
|---|
| 306 | |
|---|
| 307 | aws->callback((AW_CB0)AW_POPDOWN); |
|---|
| 308 | aws->create_button("CLOSE", "CLOSE", "C"); |
|---|
| 309 | |
|---|
| 310 | aws->at("help"); |
|---|
| 311 | aws->callback(makeHelpCallback("checkfield.hlp")); |
|---|
| 312 | aws->create_button("HELP", "HELP", "H"); |
|---|
| 313 | |
|---|
| 314 | |
|---|
| 315 | aws->at("exclude"); |
|---|
| 316 | aws->create_input_field(AWAR_EXCLUDE); |
|---|
| 317 | |
|---|
| 318 | aws->at("toupper"); |
|---|
| 319 | aws->create_toggle(AWAR_TOUPPER); |
|---|
| 320 | |
|---|
| 321 | aws->at("correct"); |
|---|
| 322 | aws->create_toggle(AWAR_CORRECT); |
|---|
| 323 | |
|---|
| 324 | aws->at("tag"); |
|---|
| 325 | aws->create_input_field(AWAR_ETAG, 6); |
|---|
| 326 | |
|---|
| 327 | create_selection_list_on_itemfields(GLOBAL_gb_dst, aws, AWAR_SOURCE_FIELD, true, FIELD_FILTER_STRING, "source", 0, SPECIES_get_selector(), 20, 10, SF_STANDARD, NULL); |
|---|
| 328 | create_selection_list_on_itemfields(GLOBAL_gb_dst, aws, AWAR_DEST_FIELD, true, (1<<GB_STRING)|(1<<GB_INT), "dest", 0, SPECIES_get_selector(), 20, 10, SF_STANDARD, NULL); |
|---|
| 329 | |
|---|
| 330 | #if defined(WARN_TODO) |
|---|
| 331 | #warning check code above. Maybe one call has to get GLOBAL_gb_src ? |
|---|
| 332 | #endif |
|---|
| 333 | |
|---|
| 334 | |
|---|
| 335 | aws->at("go"); |
|---|
| 336 | aws->highlight(); |
|---|
| 337 | aws->callback(mg_check_field_cb); |
|---|
| 338 | aws->create_button("GO", "GO"); |
|---|
| 339 | |
|---|
| 340 | return aws; |
|---|
| 341 | } |
|---|
