| 1 | #include <stdio.h> |
|---|
| 2 | #include <string.h> |
|---|
| 3 | #include <ctype.h> |
|---|
| 4 | #include <arbdb.h> |
|---|
| 5 | #include <arbdbt.h> |
|---|
| 6 | #include <aw_root.hxx> |
|---|
| 7 | #include <aw_device.hxx> |
|---|
| 8 | #include <aw_window.hxx> |
|---|
| 9 | #include <awt.hxx> |
|---|
| 10 | #include "merge.hxx" |
|---|
| 11 | |
|---|
| 12 | |
|---|
| 13 | |
|---|
| 14 | #define AWAR_SOURCE_FIELD "/tmp/merge1/chk/source" |
|---|
| 15 | #define AWAR_DEST_FIELD "/tmp/merge1/chk/dest" |
|---|
| 16 | #define AWAR_TOUPPER "/tmp/merge1/chk/ToUpper" |
|---|
| 17 | #define AWAR_EXCLUDE "/tmp/merge1/chk/exclude" |
|---|
| 18 | #define AWAR_CORRECT "/tmp/merge1/chk/correct" |
|---|
| 19 | #define AWAR_ETAG "/tmp/merge1/chk/tag" |
|---|
| 20 | |
|---|
| 21 | |
|---|
| 22 | int gbs_cmp_strings(char *str1,char *str2, int *tab){ /* returns 0 if strings are equal */ |
|---|
| 23 | register char *s1,*s2; |
|---|
| 24 | register int c1,c2; |
|---|
| 25 | s1 = str1; |
|---|
| 26 | s2 = str2; |
|---|
| 27 | int count = 10; |
|---|
| 28 | do { |
|---|
| 29 | do { c1= *(s1++); } while (tab[c1] < 0); |
|---|
| 30 | do { c2= *(s2++); } while (tab[c2] < 0); |
|---|
| 31 | if (tab[c1] != tab[c2]) { /* difference found */ |
|---|
| 32 | return 1; |
|---|
| 33 | } |
|---|
| 34 | count --; |
|---|
| 35 | } while (count && c1 && c2); |
|---|
| 36 | return 0; |
|---|
| 37 | } |
|---|
| 38 | |
|---|
| 39 | |
|---|
| 40 | char *GBS_diff_strings(char *str1,char * &str2, char *exclude , long ToUpper, long correct, |
|---|
| 41 | char **res1, char **res2, long *corrrected){ |
|---|
| 42 | |
|---|
| 43 | char buffer1[256]; |
|---|
| 44 | char buffer2[256]; |
|---|
| 45 | char *dest1 = buffer1; |
|---|
| 46 | char *dest2 = buffer2; |
|---|
| 47 | register char *s1,*s2; |
|---|
| 48 | register int c1,c2; |
|---|
| 49 | int count = 3; |
|---|
| 50 | int tab[256]; |
|---|
| 51 | int i; |
|---|
| 52 | |
|---|
| 53 | s1 = str1; |
|---|
| 54 | s2 = str2; |
|---|
| 55 | *dest1 = 0; |
|---|
| 56 | *dest2 = 0; |
|---|
| 57 | tab[0] = 0; |
|---|
| 58 | char gapchar = '#'; |
|---|
| 59 | if (strlen(exclude)) gapchar = exclude[0]; |
|---|
| 60 | else exclude = 0; |
|---|
| 61 | |
|---|
| 62 | for (i=1;i<256;i++) { |
|---|
| 63 | tab[i] = i; |
|---|
| 64 | if (exclude && strchr(exclude,i)) { |
|---|
| 65 | tab[i] = -1; |
|---|
| 66 | continue; |
|---|
| 67 | } |
|---|
| 68 | if (ToUpper && i>= 'a' && i<= 'z' ){ |
|---|
| 69 | tab[i] = i-'a'+'A'; |
|---|
| 70 | } |
|---|
| 71 | } |
|---|
| 72 | |
|---|
| 73 | do { |
|---|
| 74 | do { c1= *(s1++); } while (tab[c1] < 0); |
|---|
| 75 | do { c2= *(s2++); } while (tab[c2] < 0); |
|---|
| 76 | if (tab[c1] != tab[c2]) { /* difference found */ |
|---|
| 77 | if (correct) { |
|---|
| 78 | /* check subsitution */ |
|---|
| 79 | { |
|---|
| 80 | int c = s2[-1]; |
|---|
| 81 | s2[-1] = s1[-1]; |
|---|
| 82 | if (toupper(c1) == toupper(c2) || |
|---|
| 83 | !gbs_cmp_strings(s1,s2,&tab[0]) ){ |
|---|
| 84 | *corrrected = 1; |
|---|
| 85 | continue; |
|---|
| 86 | } |
|---|
| 87 | s2[-1] = c; |
|---|
| 88 | } |
|---|
| 89 | |
|---|
| 90 | /* check insertion in s2 */ |
|---|
| 91 | if (!gbs_cmp_strings(s1-1,s2,&tab[0]) ){ |
|---|
| 92 | s2[-1] = gapchar; |
|---|
| 93 | do { c2= *(s2++); } while (tab[c2] < 0); /* eat s2 */ |
|---|
| 94 | *corrrected = 1; |
|---|
| 95 | continue; |
|---|
| 96 | } |
|---|
| 97 | /* check deletion in s2 */ |
|---|
| 98 | if ( !gbs_cmp_strings(s1,s2-1,&tab[0]) ){ |
|---|
| 99 | int toins = c1; |
|---|
| 100 | char *toinspos = s2-1; |
|---|
| 101 | if (toinspos > str2) toinspos--; |
|---|
| 102 | if (tab[toinspos[0]]> 0) { /* real insertion */ |
|---|
| 103 | void *str = GBS_stropen(strlen(str2+10)); |
|---|
| 104 | int pos = s2-str2-1; |
|---|
| 105 | GBS_strncat(str,str2,pos); |
|---|
| 106 | GBS_chrcat(str,toins); |
|---|
| 107 | GBS_strcat(str,str2+pos); |
|---|
| 108 | delete str2; |
|---|
| 109 | str2 = GBS_strclose(str,0); |
|---|
| 110 | s2 = str2+pos+1; |
|---|
| 111 | *corrrected = 1; |
|---|
| 112 | continue; |
|---|
| 113 | } |
|---|
| 114 | int side=1; /* 0 = left 1= right */ |
|---|
| 115 | if ( tab[s1[0]]<0 ) side = 0; |
|---|
| 116 | if ( ! side ) { |
|---|
| 117 | while ( toinspos > str2 && |
|---|
| 118 | tab[toinspos[-1]] < 0 ) toinspos--; |
|---|
| 119 | } |
|---|
| 120 | toinspos[0] = toins; |
|---|
| 121 | *corrrected = 1; |
|---|
| 122 | do { c1= *(s1++); } while (tab[c1] < 0); /* eat s1 */ |
|---|
| 123 | continue; |
|---|
| 124 | } |
|---|
| 125 | } |
|---|
| 126 | if (count >=0){ |
|---|
| 127 | sprintf(dest1,"%i ",s1-str1-1); |
|---|
| 128 | sprintf(dest2,"%i ",s2-str2-1); |
|---|
| 129 | dest1 += strlen(dest1); |
|---|
| 130 | dest2 += strlen(dest2); |
|---|
| 131 | } |
|---|
| 132 | count --; |
|---|
| 133 | } |
|---|
| 134 | } while (c1 && c2); |
|---|
| 135 | |
|---|
| 136 | if (c1 || c2) { |
|---|
| 137 | sprintf(dest1,"... %i ",s1-str1-1); |
|---|
| 138 | sprintf(dest2,"... %i ",s2-str2-1); |
|---|
| 139 | dest1 += strlen(dest1); |
|---|
| 140 | dest2 += strlen(dest2); |
|---|
| 141 | } |
|---|
| 142 | if (count<0){ |
|---|
| 143 | sprintf(dest1,"and %i more",1-count); |
|---|
| 144 | sprintf(dest2,"and %i more",1-count); |
|---|
| 145 | dest1 += strlen(dest1); |
|---|
| 146 | dest2 += strlen(dest2); |
|---|
| 147 | } |
|---|
| 148 | if (strlen(buffer1) ) { |
|---|
| 149 | *res1 = strdup(buffer1); |
|---|
| 150 | *res2 = strdup(buffer2); |
|---|
| 151 | }else{ |
|---|
| 152 | *res1 = 0; |
|---|
| 153 | *res2 = 0; |
|---|
| 154 | } |
|---|
| 155 | return 0; |
|---|
| 156 | } |
|---|
| 157 | |
|---|
| 158 | #undef IS_QUERIED |
|---|
| 159 | #define IS_QUERIED(gb_species) (1 & GB_read_usr_private(gb_species)) |
|---|
| 160 | |
|---|
| 161 | void mg_check_field_cb(AW_window *aww){ |
|---|
| 162 | AW_root *root = aww->get_root(); |
|---|
| 163 | GB_ERROR error = 0; |
|---|
| 164 | char *source = root->awar(AWAR_SOURCE_FIELD)->read_string(); |
|---|
| 165 | char *dest = root->awar(AWAR_DEST_FIELD)->read_string(); |
|---|
| 166 | char *exclude = root->awar(AWAR_EXCLUDE)->read_string(); |
|---|
| 167 | long ToUpper = root->awar(AWAR_TOUPPER)->read_int(); |
|---|
| 168 | long correct = root->awar(AWAR_CORRECT)->read_int(); |
|---|
| 169 | char *tag = root->awar(AWAR_ETAG)->read_string(); |
|---|
| 170 | |
|---|
| 171 | if (!strlen(source)) { |
|---|
| 172 | delete source; |
|---|
| 173 | delete dest; |
|---|
| 174 | delete exclude; |
|---|
| 175 | delete tag; |
|---|
| 176 | aw_message("Please select a source field"); |
|---|
| 177 | return; |
|---|
| 178 | } |
|---|
| 179 | if (!strlen(dest)) { |
|---|
| 180 | delete source; |
|---|
| 181 | delete tag; |
|---|
| 182 | delete dest; |
|---|
| 183 | delete exclude; |
|---|
| 184 | aw_message("Please select a dest field"); |
|---|
| 185 | return; |
|---|
| 186 | } |
|---|
| 187 | aw_openstatus("Checking fields"); |
|---|
| 188 | GB_begin_transaction(gb_merge); |
|---|
| 189 | GB_begin_transaction(gb_dest); |
|---|
| 190 | |
|---|
| 191 | GBDATA *gb_species_data1 = GB_search(gb_merge,"species_data",GB_CREATE_CONTAINER); |
|---|
| 192 | GBDATA *gb_species_data2 = GB_search(gb_dest,"species_data",GB_CREATE_CONTAINER); |
|---|
| 193 | |
|---|
| 194 | GBDATA *gb_species1; |
|---|
| 195 | GBDATA *gb_species2; |
|---|
| 196 | GBDATA *gb_name1; |
|---|
| 197 | GBDATA *gb_field1; |
|---|
| 198 | GBDATA *gb_field2; |
|---|
| 199 | GBDATA *gbd; |
|---|
| 200 | int sum_species = 0; |
|---|
| 201 | int species_count = 0; |
|---|
| 202 | // First step: count selected species |
|---|
| 203 | for (gb_species1 = GBT_first_species_rel_species_data(gb_species_data1); |
|---|
| 204 | gb_species1; |
|---|
| 205 | gb_species1 = GBT_next_species(gb_species1)){ |
|---|
| 206 | gbd = GB_search(gb_species1,dest,GB_FIND); |
|---|
| 207 | if (gbd) GB_delete(gbd); |
|---|
| 208 | if (!IS_QUERIED(gb_species1)) continue; |
|---|
| 209 | sum_species ++; |
|---|
| 210 | } |
|---|
| 211 | // Delete all 'dest' fields in gb_database 2 |
|---|
| 212 | for ( gb_species2 = GBT_first_species_rel_species_data(gb_species_data2); |
|---|
| 213 | gb_species2; |
|---|
| 214 | gb_species2 = GBT_next_species(gb_species2)){ |
|---|
| 215 | gbd = GB_search(gb_species2,dest,GB_FIND); |
|---|
| 216 | if (gbd) GB_delete(gbd); |
|---|
| 217 | } |
|---|
| 218 | |
|---|
| 219 | for ( gb_species1 = GBT_first_species_rel_species_data(gb_species_data1); |
|---|
| 220 | gb_species1; |
|---|
| 221 | gb_species1 = GBT_next_species(gb_species1)){ |
|---|
| 222 | gbd = GB_search(gb_species1,dest,GB_FIND); |
|---|
| 223 | if (gbd) GB_delete(gbd); |
|---|
| 224 | |
|---|
| 225 | gb_name1 = GB_find(gb_species1,"name",0,down_level); |
|---|
| 226 | if (!gb_name1) continue; // no name what happend ??? |
|---|
| 227 | |
|---|
| 228 | if (!IS_QUERIED(gb_species1)) continue; |
|---|
| 229 | species_count++; |
|---|
| 230 | if ( (species_count & 0xf) == 0 ){ |
|---|
| 231 | aw_status(species_count/ (double)sum_species); |
|---|
| 232 | } |
|---|
| 233 | gb_species2 = GB_find(gb_species_data2,"name", |
|---|
| 234 | GB_read_char_pntr(gb_name1),down_2_level); |
|---|
| 235 | |
|---|
| 236 | if (!gb_species2) { |
|---|
| 237 | sprintf(AW_ERROR_BUFFER,"WARNING: Species %s not found in DB II", |
|---|
| 238 | GB_read_char_pntr(gb_name1)); |
|---|
| 239 | aw_message(); |
|---|
| 240 | continue; |
|---|
| 241 | }else{ |
|---|
| 242 | gb_species2 = GB_get_father(gb_species2); |
|---|
| 243 | } |
|---|
| 244 | char *s1 = 0; |
|---|
| 245 | char *s2 = 0; |
|---|
| 246 | gb_field1 = GB_search(gb_species1,source,GB_FIND); |
|---|
| 247 | gb_field2 = GB_search(gb_species2,source,GB_FIND); |
|---|
| 248 | |
|---|
| 249 | if ( (!gb_field1) && (!gb_field2) ) continue; |
|---|
| 250 | if (gb_field1 ) s1 = GB_read_as_tagged_string(gb_field1,tag); |
|---|
| 251 | if (gb_field2 ) s2 = GB_read_as_tagged_string(gb_field2,tag); |
|---|
| 252 | if (!s1 && !s2) continue; |
|---|
| 253 | |
|---|
| 254 | { |
|---|
| 255 | char *positions1; |
|---|
| 256 | char *positions2; |
|---|
| 257 | if (s1 && s2){ |
|---|
| 258 | long corrected= 0; |
|---|
| 259 | GBS_diff_strings(s1,s2,exclude,ToUpper,correct,&positions1,&positions2,&corrected); |
|---|
| 260 | if (corrected) { |
|---|
| 261 | error = GB_write_as_string(gb_field2,s2); |
|---|
| 262 | GB_write_flag(gb_species2,1); |
|---|
| 263 | } |
|---|
| 264 | }else{ |
|---|
| 265 | if (s1) positions1 = strdup("missing field in other db"); |
|---|
| 266 | else positions1 = strdup("missing field in this db"); |
|---|
| 267 | if (s2) positions2 = strdup("missing field in other db"); |
|---|
| 268 | else positions2 = strdup("missing field in this db"); |
|---|
| 269 | } |
|---|
| 270 | if (positions1){ |
|---|
| 271 | gbd = GB_search(gb_species2,dest,GB_STRING); |
|---|
| 272 | GB_write_string(gbd,positions2); |
|---|
| 273 | delete positions2; |
|---|
| 274 | gbd = GB_search(gb_species1,dest,GB_STRING); |
|---|
| 275 | GB_write_string(gbd,positions1); |
|---|
| 276 | delete positions1; |
|---|
| 277 | } |
|---|
| 278 | } |
|---|
| 279 | |
|---|
| 280 | delete s1; |
|---|
| 281 | delete s2; |
|---|
| 282 | } |
|---|
| 283 | aw_closestatus(); |
|---|
| 284 | if (error){ |
|---|
| 285 | GB_abort_transaction(gb_merge); |
|---|
| 286 | GB_abort_transaction(gb_dest); |
|---|
| 287 | aw_message(error); |
|---|
| 288 | error = 0; |
|---|
| 289 | }else{ |
|---|
| 290 | GB_commit_transaction(gb_merge); |
|---|
| 291 | GB_commit_transaction(gb_dest); |
|---|
| 292 | } |
|---|
| 293 | delete tag; |
|---|
| 294 | delete source; |
|---|
| 295 | delete dest; |
|---|
| 296 | delete exclude; |
|---|
| 297 | } |
|---|
| 298 | |
|---|
| 299 | |
|---|
| 300 | AW_window *create_mg_check_fields(AW_root *aw_root){ |
|---|
| 301 | AW_window_simple *aws = 0; |
|---|
| 302 | |
|---|
| 303 | aw_root->awar_string(AWAR_SOURCE_FIELD); |
|---|
| 304 | aw_root->awar_string(AWAR_DEST_FIELD,"tmp",AW_ROOT_DEFAULT); |
|---|
| 305 | aw_root->awar_string(AWAR_EXCLUDE,".-",AW_ROOT_DEFAULT); |
|---|
| 306 | aw_root->awar_string(AWAR_ETAG,""); |
|---|
| 307 | aw_root->awar_int(AWAR_TOUPPER); |
|---|
| 308 | aw_root->awar_int(AWAR_CORRECT); |
|---|
| 309 | |
|---|
| 310 | aws = new AW_window_simple; |
|---|
| 311 | aws->init( aw_root, "MERGE_COMPARE_FIELDS","COMPARE DATABASE FIELDS", 600, 0 ); |
|---|
| 312 | aws->load_xfig("merge/seqcheck.fig"); |
|---|
| 313 | |
|---|
| 314 | aws->callback( (AW_CB0)AW_POPDOWN); |
|---|
| 315 | aws->create_button("CLOSE","CLOSE","C"); |
|---|
| 316 | |
|---|
| 317 | aws->at("help"); |
|---|
| 318 | aws->callback(AW_POPUP_HELP,(AW_CL)"checkfield.hlp"); |
|---|
| 319 | aws->create_button("HELP","HELP","H"); |
|---|
| 320 | |
|---|
| 321 | |
|---|
| 322 | aws->at("exclude"); |
|---|
| 323 | aws->create_input_field(AWAR_EXCLUDE); |
|---|
| 324 | |
|---|
| 325 | aws->at("toupper"); |
|---|
| 326 | aws->create_toggle(AWAR_TOUPPER); |
|---|
| 327 | |
|---|
| 328 | aws->at("correct"); |
|---|
| 329 | aws->create_toggle(AWAR_CORRECT); |
|---|
| 330 | |
|---|
| 331 | aws->at("tag"); |
|---|
| 332 | aws->create_input_field(AWAR_ETAG,6); |
|---|
| 333 | |
|---|
| 334 | awt_create_selection_list_on_scandb(gb_dest,aws,AWAR_SOURCE_FIELD, |
|---|
| 335 | AWT_STRING_FILTER, "source",0); |
|---|
| 336 | |
|---|
| 337 | awt_create_selection_list_on_scandb(gb_dest,aws,AWAR_DEST_FIELD, |
|---|
| 338 | (1<<GB_STRING)|(1<<GB_INT), "dest",0); |
|---|
| 339 | |
|---|
| 340 | |
|---|
| 341 | aws->at("go"); |
|---|
| 342 | aws->highlight(); |
|---|
| 343 | aws->callback(mg_check_field_cb); |
|---|
| 344 | aws->create_button("GO","GO"); |
|---|
| 345 | |
|---|
| 346 | return (AW_window *)aws; |
|---|
| 347 | } |
|---|