cc3c8382f4e4564b4f6381acf87c7c7528d31aa0 angie Wed Jan 7 12:10:56 2015 -0800 Added annoFormatTabSetColumnVis so that columns can be selectively included in output. diff --git src/lib/annoFormatTab.c src/lib/annoFormatTab.c index a8f1d45..e7dbf7a 100644 --- src/lib/annoFormatTab.c +++ src/lib/annoFormatTab.c @@ -1,71 +1,127 @@ /* annoFormatTab -- collect fields from all inputs and print them out, tab-separated. */ /* Copyright (C) 2013 The Regents of the University of California * See README in this or parent directory for licensing information. */ #include "annoFormatTab.h" #include "annoGratorQuery.h" #include "dystring.h" struct annoFormatTab { - struct annoFormatter formatter; - char *fileName; - FILE *f; + struct annoFormatter formatter; // External interface + char *fileName; // Output file name, can be "stdout" + FILE *f; // Output file handle + struct hash *columnVis; // Hash of columns that have been explicitly selected + // or deselected by user. boolean needHeader; // TRUE if we should print out the header }; -static void printHeaderColumns(FILE *f, struct annoStreamer *source, boolean isFirst) +static void makeFullColumnName(char *fullName, size_t size, char *sourceName, char *colName) +/* If sourceName is non-full, make fullName sourceName.colName, otherwise just colName. */ +{ +if (sourceName) + safef(fullName, size, "%s.%s", sourceName, colName); +else + safecpy(fullName, size, colName); +} + +void annoFormatTabSetColumnVis(struct annoFormatter *vSelf, char *sourceName, char *colName, + boolean enabled) +/* Explicitly include or exclude column in output. sourceName must be the same + * as the corresponding annoStreamer source's name. */ +{ +struct annoFormatTab *self = (struct annoFormatTab *)vSelf; +if (! self->columnVis) + self->columnVis = hashNew(0); +char fullName[PATH_LEN]; +makeFullColumnName(fullName, sizeof(fullName), sourceName, colName); +hashAddInt(self->columnVis, fullName, enabled); +} + +static boolean columnIsIncluded(struct annoFormatTab *self, char *sourceName, char *colName) +// Return TRUE if column has not been explicitly deselected. +{ +if (self->columnVis) + { + char fullName[PATH_LEN]; + makeFullColumnName(fullName, sizeof(fullName), sourceName, colName); + int vis = hashIntValDefault(self->columnVis, fullName, 1); + if (vis == 0) + return FALSE; + } +return TRUE; +} + +static void printHeaderColumns(struct annoFormatTab *self, struct annoStreamer *source, + boolean isFirst) /* Print names of included columns from this source. */ { +FILE *f = self->f; +char *sourceName = source->name; if (source->rowType == arWig) { // Fudge in the row's chrom, start, end as output columns even though they're not in autoSql if (isFirst) { + if (sourceName) + fprintf(f, "#%s.chrom", sourceName); + else fputs("#chrom", f); isFirst = FALSE; } + if (sourceName) + fprintf(f, "\t%s.start\t%s.end", sourceName, sourceName); + else fputs("\tstart\tend", f); } struct asColumn *col; int i; for (col = source->asObj->columnList, i = 0; col != NULL; col = col->next, i++) { - if (isFirst && i == 0) + if (columnIsIncluded(self, sourceName, col->name)) + { + if (isFirst) + { fputc('#', f); + isFirst = FALSE; + } else fputc('\t', f); + if (sourceName) + fprintf(f, "%s.%s", sourceName, col->name); + else fputs(col->name, f); } } +} static void aftInitialize(struct annoFormatter *vSelf, struct annoStreamer *primary, struct annoStreamer *integrators) /* Print header, regardless of whether we get any data after this. */ { struct annoFormatTab *self = (struct annoFormatTab *)vSelf; if (self->needHeader) { char *primaryHeader = primary->getHeader(primary); if (isNotEmpty(primaryHeader)) fprintf(self->f, "# Header from primary input:\n%s", primaryHeader); - printHeaderColumns(self->f, primary, TRUE); + printHeaderColumns(self, primary, TRUE); struct annoStreamer *grator; for (grator = integrators; grator != NULL; grator = grator->next) - printHeaderColumns(self->f, grator, FALSE); + printHeaderColumns(self, grator, FALSE); fputc('\n', self->f); self->needHeader = FALSE; } } static double wigRowAvg(struct annoRow *row) /* Return the average value of floats in row->data. */ { float *vector = row->data; int len = row->end - row->start; double sum = 0.0; int i; for (i = 0; i < len; i++) sum += vector[i]; return sum / (double)len; @@ -114,59 +170,66 @@ //#*** config options: avg? more stats? list of values? boolean doAvg = FALSE; if (doAvg) words = wordsFromWigRowAvg(row); else words = wordsFromWigRowVals(row); } else errAbort("annoFormatTab: unrecognized row type %d from source %s", source->rowType, source->name); if (retFreeWhenDone != NULL) *retFreeWhenDone = freeWhenDone; return words; } -static void printColumns(FILE *f, struct annoStreamer *streamer, struct annoRow *row, - boolean isFirst) +static void printColumns(struct annoFormatTab *self, struct annoStreamer *streamer, + struct annoRow *row, boolean isFirst) /* Print columns in streamer's row (if NULL, print the right number of empty fields). */ { +FILE *f = self->f; +char *sourceName = streamer->name; boolean freeWhenDone = FALSE; char **words = wordsFromRow(row, streamer, &freeWhenDone); if (streamer->rowType == arWig) { // Fudge in the row's chrom, start, end as output columns even though they're not in autoSql if (isFirst) { if (row != NULL) fputs(row->chrom, f); isFirst = FALSE; } if (row != NULL) fprintf(f, "\t%u\t%u", row->start, row->end); else fputs("\t\t", f); } -int colCount = slCount(streamer->asObj->columnList); +struct asColumn *col; int i; -for (i = 0; i < colCount; i++) +for (col = streamer->asObj->columnList, i = 0; col != NULL; col = col->next, i++) { - if (!isFirst || i > 0) + if (columnIsIncluded(self, sourceName, col->name)) + { + if (isFirst) + isFirst = FALSE; + else fputc('\t', f); if (words != NULL) fputs((words[i] ? words[i] : ""), f); } + } if (freeWhenDone) { freeMem(words[0]); freeMem(words); } } static void aftComment(struct annoFormatter *fSelf, char *content) /* Print out a comment line. */ { if (strchr(content, '\n')) errAbort("aftComment: no multi-line input"); struct annoFormatTab *self = (struct annoFormatTab *)fSelf; fprintf(self->f, "# %s\n", content); } @@ -178,35 +241,35 @@ { struct annoFormatTab *self = (struct annoFormatTab *)vSelf; // Got one row from primary; what's the largest # of rows from any grator? int maxRows = 1; int iG; for (iG = 0; iG < gratorCount; iG++) { int gratorRowCount = slCount(gratorData[iG].rowList); if (gratorRowCount > maxRows) maxRows = gratorRowCount; } // Print out enough rows to make sure that all grator rows are included. int iR; for (iR = 0; iR < maxRows; iR++) { - printColumns(self->f, primaryData->streamer, primaryData->rowList, TRUE); + printColumns(self, primaryData->streamer, primaryData->rowList, TRUE); for (iG = 0; iG < gratorCount; iG++) { struct annoRow *gratorRow = slElementFromIx(gratorData[iG].rowList, iR); - printColumns(self->f, gratorData[iG].streamer, gratorRow, FALSE); + printColumns(self, gratorData[iG].streamer, gratorRow, FALSE); } fputc('\n', self->f); } } static void aftClose(struct annoFormatter **pVSelf) /* Close file handle, free self. */ { if (pVSelf == NULL) return; struct annoFormatTab *self = *(struct annoFormatTab **)pVSelf; freeMem(self->fileName); carefulClose(&(self->f)); annoFormatterFree(pVSelf); }