98874f1656202cd5781466a5d9b25f47da1f104d hiram Mon Feb 13 16:03:53 2023 -0800 a bit closer to correct NCBI links, not perfect, but better than it was no redmine diff --git src/hg/lib/asmEquivalent.c src/hg/lib/asmEquivalent.c index 3971aadd..7942c2c 100644 --- src/hg/lib/asmEquivalent.c +++ src/hg/lib/asmEquivalent.c @@ -1,307 +1,303 @@ /* asmEquivalent.c was originally generated by the autoSql program, which also * generated asmEquivalent.h and asmEquivalent.sql. This module links the database and * the RAM representation of objects. */ #include "common.h" #include "linefile.h" #include "dystring.h" #include "jksql.h" #include "asmEquivalent.h" #include "hdb.h" char *asmEquivalentCommaSepFieldNames = "source,destination,sourceAuthority,destinationAuthority,matchCount,sourceCount,destinationCount"; /* definitions for sourceAuthority column */ static char *values_sourceAuthority[] = {"ensembl", "ucsc", "genbank", "refseq", NULL}; static struct hash *valhash_sourceAuthority = NULL; /* definitions for destinationAuthority column */ static char *values_destinationAuthority[] = {"ensembl", "ucsc", "genbank", "refseq", NULL}; static struct hash *valhash_destinationAuthority = NULL; void asmEquivalentStaticLoad(char **row, struct asmEquivalent *ret) /* Load a row from asmEquivalent table into ret. The contents of ret will * be replaced at the next call to this function. */ { ret->source = row[0]; ret->destination = row[1]; ret->sourceAuthority = sqlEnumParse(row[2], values_sourceAuthority, &valhash_sourceAuthority); ret->destinationAuthority = sqlEnumParse(row[3], values_destinationAuthority, &valhash_destinationAuthority); ret->matchCount = sqlLongLong(row[4]); ret->sourceCount = sqlLongLong(row[5]); ret->destinationCount = sqlLongLong(row[6]); } struct asmEquivalent *asmEquivalentLoadByQuery(struct sqlConnection *conn, char *query) /* Load all asmEquivalent from table that satisfy the query given. * Where query is of the form 'select * from example where something=something' * or 'select example.* from example, anotherTable where example.something = * anotherTable.something'. * Dispose of this with asmEquivalentFreeList(). */ { struct asmEquivalent *list = NULL, *el; struct sqlResult *sr; char **row; sr = sqlGetResult(conn, query); while ((row = sqlNextRow(sr)) != NULL) { el = asmEquivalentLoad(row); slAddHead(&list, el); } slReverse(&list); sqlFreeResult(&sr); return list; } void asmEquivalentSaveToDb(struct sqlConnection *conn, struct asmEquivalent *el, char *tableName, int updateSize) /* Save asmEquivalent as a row to the table specified by tableName. * As blob fields may be arbitrary size updateSize specifies the approx size * of a string that would contain the entire query. Arrays of native types are * converted to comma separated strings and loaded as such, User defined types are * inserted as NULL. This function automatically escapes quoted strings for mysql. */ { struct dyString *update = dyStringNew(updateSize); sqlDyStringPrintf(update, "insert into %s values ( '%s','%s',%d,%d,%lld,%lld,%lld)", tableName, el->source, el->destination, el->sourceAuthority, el->destinationAuthority, el->matchCount, el->sourceCount, el->destinationCount); sqlUpdate(conn, update->string); dyStringFree(&update); } struct asmEquivalent *asmEquivalentLoad(char **row) /* Load a asmEquivalent from row fetched with select * from asmEquivalent * from database. Dispose of this with asmEquivalentFree(). */ { struct asmEquivalent *ret; AllocVar(ret); ret->source = cloneString(row[0]); ret->destination = cloneString(row[1]); ret->sourceAuthority = sqlEnumParse(row[2], values_sourceAuthority, &valhash_sourceAuthority); ret->destinationAuthority = sqlEnumParse(row[3], values_destinationAuthority, &valhash_destinationAuthority); ret->matchCount = sqlLongLong(row[4]); ret->sourceCount = sqlLongLong(row[5]); ret->destinationCount = sqlLongLong(row[6]); return ret; } struct asmEquivalent *asmEquivalentLoadAll(char *fileName) /* Load all asmEquivalent from a whitespace-separated file. * Dispose of this with asmEquivalentFreeList(). */ { struct asmEquivalent *list = NULL, *el; struct lineFile *lf = lineFileOpen(fileName, TRUE); char *row[7]; while (lineFileRow(lf, row)) { el = asmEquivalentLoad(row); slAddHead(&list, el); } lineFileClose(&lf); slReverse(&list); return list; } struct asmEquivalent *asmEquivalentLoadAllByChar(char *fileName, char chopper) /* Load all asmEquivalent from a chopper separated file. * Dispose of this with asmEquivalentFreeList(). */ { struct asmEquivalent *list = NULL, *el; struct lineFile *lf = lineFileOpen(fileName, TRUE); char *row[7]; while (lineFileNextCharRow(lf, chopper, row, ArraySize(row))) { el = asmEquivalentLoad(row); slAddHead(&list, el); } lineFileClose(&lf); slReverse(&list); return list; } struct asmEquivalent *asmEquivalentCommaIn(char **pS, struct asmEquivalent *ret) /* Create a asmEquivalent out of a comma separated string. * This will fill in ret if non-null, otherwise will * return a new asmEquivalent */ { char *s = *pS; if (ret == NULL) AllocVar(ret); ret->source = sqlStringComma(&s); ret->destination = sqlStringComma(&s); ret->sourceAuthority = sqlEnumComma(&s, values_sourceAuthority, &valhash_sourceAuthority); ret->destinationAuthority = sqlEnumComma(&s, values_destinationAuthority, &valhash_destinationAuthority); ret->matchCount = sqlLongLongComma(&s); ret->sourceCount = sqlLongLongComma(&s); ret->destinationCount = sqlLongLongComma(&s); *pS = s; return ret; } void asmEquivalentFree(struct asmEquivalent **pEl) /* Free a single dynamically allocated asmEquivalent such as created * with asmEquivalentLoad(). */ { struct asmEquivalent *el; if ((el = *pEl) == NULL) return; freeMem(el->source); freeMem(el->destination); freez(pEl); } void asmEquivalentFreeList(struct asmEquivalent **pList) /* Free a list of dynamically allocated asmEquivalent's */ { struct asmEquivalent *el, *next; for (el = *pList; el != NULL; el = next) { next = el->next; asmEquivalentFree(&el); } *pList = NULL; } void asmEquivalentOutput(struct asmEquivalent *el, FILE *f, char sep, char lastSep) /* Print out asmEquivalent. Separate fields with sep. Follow last field with lastSep. */ { if (sep == ',') fputc('"',f); fprintf(f, "%s", el->source); if (sep == ',') fputc('"',f); fputc(sep,f); if (sep == ',') fputc('"',f); fprintf(f, "%s", el->destination); if (sep == ',') fputc('"',f); fputc(sep,f); if (sep == ',') fputc('"',f); sqlEnumPrint(f, el->sourceAuthority, values_sourceAuthority); if (sep == ',') fputc('"',f); fputc(sep,f); if (sep == ',') fputc('"',f); sqlEnumPrint(f, el->destinationAuthority, values_destinationAuthority); if (sep == ',') fputc('"',f); fputc(sep,f); fprintf(f, "%lld", el->matchCount); fputc(sep,f); fprintf(f, "%lld", el->sourceCount); fputc(sep,f); fprintf(f, "%lld", el->destinationCount); fputc(lastSep,f); } void asmEquivalentJsonOutput(struct asmEquivalent *el, FILE *f) /* Print out asmEquivalent in JSON format. */ { fputc('{',f); fputc('"',f); fprintf(f,"source"); fputc('"',f); fputc(':',f); fputc('"',f); fprintf(f, "%s", el->source); fputc('"',f); fputc(',',f); fputc('"',f); fprintf(f,"destination"); fputc('"',f); fputc(':',f); fputc('"',f); fprintf(f, "%s", el->destination); fputc('"',f); fputc(',',f); fputc('"',f); fprintf(f,"sourceAuthority"); fputc('"',f); fputc(':',f); fputc('"',f); sqlEnumPrint(f, el->sourceAuthority, values_sourceAuthority); fputc('"',f); fputc(',',f); fputc('"',f); fprintf(f,"destinationAuthority"); fputc('"',f); fputc(':',f); fputc('"',f); sqlEnumPrint(f, el->destinationAuthority, values_destinationAuthority); fputc('"',f); fputc(',',f); fputc('"',f); fprintf(f,"matchCount"); fputc('"',f); fputc(':',f); fprintf(f, "%lld", el->matchCount); fputc(',',f); fputc('"',f); fprintf(f,"sourceCount"); fputc('"',f); fputc(':',f); fprintf(f, "%lld", el->sourceCount); fputc(',',f); fputc('"',f); fprintf(f,"destinationCount"); fputc('"',f); fputc(':',f); fprintf(f, "%lld", el->destinationCount); fputc('}',f); } /* -------------------------------- End autoSql Generated Code -------------------------------- */ -char *asmEquivalentUcscToNCBI(char *ucscName) -/* check if there is a RefSeq/GenBank equivalent to this UCSC assembly name. - * If RefSeq exists, return that first, else if GenBank than return that. +char *asmEquivalentUcscToNCBI(char *ucscName, char *authority) +/* check if there is an "authority" equivalent to this UCSC assembly name. + * where 'authority' in this case is either 'refseq' or 'genbank' * No checking of sequence match counts in this first implementation, * therefore, could be a fuzzy match, and since it is returning only the * first one, it might not be the best match. Could add more specifics * later to get better match. */ { char *ret = NULL; if (ucscName == NULL) return ret; + struct sqlConnection *conn = hAllocConn("hgFixed"); if (!conn) return ret; if (!sqlTableExists(conn, "asmEquivalent")) { hFreeConn(&conn); return ret; } char buffer[4096]; -sqlSafef(buffer, sizeof buffer, "SELECT destination FROM asmEquivalent WHERE sourceAuthority='ucsc' AND destinationAuthority='refseq' AND source='%s' LIMIT 1", ucscName); +sqlSafef(buffer, sizeof buffer, "SELECT destination FROM asmEquivalent WHERE sourceAuthority='ucsc' AND destinationAuthority='%s' AND source='%s' LIMIT 1", authority, ucscName); char *sqlAnswer = sqlQuickString(conn, buffer); -if (isEmpty(sqlAnswer)) - { - sqlSafef(buffer, sizeof buffer, "SELECT destination FROM asmEquivalent WHERE sourceAuthority='ucsc' AND destinationAuthority='genbank' AND source='%s' LIMIT 1", ucscName); - sqlAnswer = sqlQuickString(conn, buffer); - } hFreeConn(&conn); /* if there is a result, for example: GCA_000001405.28_GRCh38.p13 * want to return only: GCA_000001405.28 */ if (isNotEmpty(sqlAnswer)) { char *words[3]; int wordCount = 0; wordCount = chopString(sqlAnswer, "_", words, ArraySize(words)); if (3 != wordCount) /* something wrong with this answer */ return ret; safef(buffer, sizeof(buffer), "%s_%s", words[0], words[1]); ret = cloneString(buffer); } return ret; }