2d5c21f0d3a8b67dd86f1f9f999f2aa384f67275 hiram Sat Feb 3 13:17:19 2024 -0800 beginning to run FULLTEXT search on asmSummary table refs #23589 diff --git src/hg/inc/asmSummary.h src/hg/inc/asmSummary.h index c1f7d32..05e9019 100644 --- src/hg/inc/asmSummary.h +++ src/hg/inc/asmSummary.h @@ -1,117 +1,124 @@ /* asmSummary.h was originally generated by the autoSql program, which also * generated asmSummary.c and asmSummary.sql. This header links the database and * the RAM representation of objects. */ #ifndef ASMSUMMARY_H #define ASMSUMMARY_H #include "jksql.h" #define ASMSUMMARY_NUM_COLS 37 extern char *asmSummaryCommaSepFieldNames; struct asmSummary /* NCBI assembly_summary data, see: https://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/README_assembly_summary.txt */ { struct asmSummary *next; /* Next in singly linked list. */ char *assemblyAccession; /* www.ncbi.nlm.nih.gov/assembly/?term=xxx */ char *bioproject; /* www.ncbi.nlm.nih.gov/bioproject/?term=xxx */ char *biosample; /* www.ncbi.nlm.nih.gov/biosample/?term=xxx */ char *wgsMaster; /* www.ncbi.nlm.nih.gov/nuccore/xxx */ char *refseqCategory; /* representative or reference */ unsigned *taxId; /* www.ncbi.nlm.nih.gov/taxonomy/?term=xxx */ unsigned *speciesTaxid; /* www.ncbi.nlm.nih.gov/taxonomy/?term=xxx */ char *organismName; /* binomial scientific name */ char *infraspecificName; /* strain/cultivar/ecotype/breed */ char *isolate; /* source of sample */ char *versionStatus; /* latest/suppressed/replaced */ char *assemblyLevel; /* Contig/Scaffold/Complete Genome/Chromosome */ char *releaseType; /* Major/Minor/Patch */ char *genomeRep; /* Full/Partial */ char *seqRelDate; /* date YYYY/MM/DD sequence released to INSDC */ char *asmName; /* submitter supplied name */ char *asmSubmitter; /* institution submitting assembly */ char *gbrsPairedAsm; /* GenBank<->RefSeq GCA/GCF relationship */ char *pairedAsmComp; /* identical/different for GCA<->GCF relationship */ char *ftpPath; /* ftp.ncbi.nlm.nih.gov/genomes/all/GCx/012/345/678/asmId */ char *excludedFromRefseq; /* noted reason for exclusion from RefSeq */ char *relationToTypeMaterial; /* note of assembly relation to sample */ char *assemblyType; /* haploid/diploid/haploid-with-alt-loci/alternate-pseudohaplotype */ char *phyloGroup; /* bacteria/viral/archaea/fungi/metagenomes/invertebrate/other/vertebrate_other/plant/vertebrate_mammalian/protozoa */ long long *genomeSize; /* total length of all top-level sequences in the primary assembly */ long long *genomeSizeUngapped; /* genome length not counting gaps (gap == 10 or more Ns) */ float *gcPercent; /* GC percent */ unsigned short *repliconCount; /* total number of chromosomes, organelle genomes and plasmids */ unsigned *scaffoldCount; /* number of scaffolds: placed, unlocalzes, unplace, alternate loci and patch */ unsigned *contigCount; /* number of conts in the primary assembly */ char *annotationProvider; /* the group that provied the annotation on the assembly */ char *annotationName; /* the name of the annotation */ char *annotationDate; /* annotation date YYYY/MM/DD */ char *totalGeneCount; /* total gene count in annotation */ unsigned *proteinCodingGeneCount; /* protein coding gene count in annotation */ char *nonCodingGeneCount; /* non coding gene count in annotation */ char *pubmedId; /* comma separated list of PubMed ID(s) */ }; void asmSummaryStaticLoadWithNull(char **row, struct asmSummary *ret); /* Load a row from asmSummary table into ret. The contents of ret will * be replaced at the next call to this function. */ struct asmSummary *asmSummaryLoadByQuery(struct sqlConnection *conn, char *query); /* Load all asmSummary 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 asmSummaryFreeList(). */ void asmSummarySaveToDb(struct sqlConnection *conn, struct asmSummary *el, char *tableName, int updateSize); /* Save asmSummary 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 asmSummary *asmSummaryLoadWithNull(char **row); /* Load a asmSummary from row fetched with select * from asmSummary * from database. Dispose of this with asmSummaryFree(). */ struct asmSummary *asmSummaryLoadAll(char *fileName); /* Load all asmSummary from whitespace-separated file. * Dispose of this with asmSummaryFreeList(). */ struct asmSummary *asmSummaryLoadAllByChar(char *fileName, char chopper); /* Load all asmSummary from chopper separated file. * Dispose of this with asmSummaryFreeList(). */ #define asmSummaryLoadAllByTab(a) asmSummaryLoadAllByChar(a, '\t'); /* Load all asmSummary from tab separated file. * Dispose of this with asmSummaryFreeList(). */ struct asmSummary *asmSummaryCommaIn(char **pS, struct asmSummary *ret); /* Create a asmSummary out of a comma separated string. * This will fill in ret if non-null, otherwise will * return a new asmSummary */ void asmSummaryFree(struct asmSummary **pEl); /* Free a single dynamically allocated asmSummary such as created * with asmSummaryLoad(). */ void asmSummaryFreeList(struct asmSummary **pList); /* Free a list of dynamically allocated asmSummary's */ void asmSummaryOutput(struct asmSummary *el, FILE *f, char sep, char lastSep); /* Print out asmSummary. Separate fields with sep. Follow last field with lastSep. */ #define asmSummaryTabOut(el,f) asmSummaryOutput(el,f,'\t','\n'); /* Print out asmSummary as a line in a tab-separated file. */ #define asmSummaryCommaOut(el,f) asmSummaryOutput(el,f,',',','); /* Print out asmSummary as a comma separated list including final comma. */ void asmSummaryJsonOutput(struct asmSummary *el, FILE *f); /* Print out asmSummary in JSON format. */ /* -------------------------------- End autoSql Generated Code -------------------------------- */ +struct asmSummary *asmSummaryFullText(struct sqlConnection *conn, char *words, long long rowLimit, long long *totalMatch); +/* perform a FULLTEXT search on the asmSummary table with the list + * of words string (may be only a single word) + * return is a list of items found up to rowLimit, or NULL if none found + * also returning totalMatch to understand if it is more than the rowLimit + */ + #endif /* ASMSUMMARY_H */