// read all records in all files,
// divide into groups based on GDS hash
// each group has an arraylist of all records that belong to it.
// for each group, run rectlizer to derive the coordinates and variables
public List<Group> makeAggregatedGroups(
ArrayList<String> filenames, CollectionManager.Force force, Formatter f) throws IOException {
Map<Integer, Group> gdsMap = new HashMap<Integer, Group>();
f.format("GribCollection %s: makeAggregatedGroups%n", gc.getName());
int total = 0;
int fileno = 0;
for (CollectionManager dcm : collections) {
// dcm.scanIfNeeded(); // LOOK ??
f.format(" dcm= %s%n", dcm);
Map<Integer, Integer> gdsConvert = (Map<Integer, Integer>) dcm.getAuxInfo("gdsHash");
for (MFile mfile : dcm.getFiles()) {
// f.format("%3d: %s%n", fileno, mfile.getPath());
filenames.add(mfile.getPath());
Grib2Index index = new Grib2Index();
try {
if (!index.readIndex(
mfile.getPath(),
mfile.getLastModified(),
force)) { // heres where the index date is checked against the data file
index.makeIndex(mfile.getPath(), f);
f.format(
" Index written: %s == %d records %n",
mfile.getName() + Grib2Index.IDX_EXT, index.getRecords().size());
} else if (debug) {
f.format(
" Index read: %s == %d records %n",
mfile.getName() + Grib2Index.IDX_EXT, index.getRecords().size());
}
} catch (IOException ioe) {
f.format(
"GribCollectionBuilder: reading/Creating gbx9 index failed err=%s%n skipping %s%n",
ioe.getMessage(), mfile.getPath() + Grib2Index.IDX_EXT);
continue;
}
for (Grib2Record gr : index.getRecords()) {
gr.setFile(fileno); // each record tracks which file it belongs to
int gdsHash =
gr.getGDSsection().getGDS().hashCode(); // use GDS hash code to group records
if (gdsConvert != null
&& gdsConvert.get(gdsHash)
!= null) { // allow external config to muck with gdsHash. Why? because of error in
// encoding
gdsHash = (Integer) gdsConvert.get(gdsHash); // and we need exact hash matching
}
Group g = gdsMap.get(gdsHash);
if (g == null) {
g = new Group(gr.getGDSsection(), gdsHash);
gdsMap.put(gdsHash, g);
}
g.records.add(gr);
total++;
}
fileno++;
}
}
f.format(" total grib records= %d%n", total);
Grib2Rectilyser.Counter c = new Grib2Rectilyser.Counter();
List<Group> result = new ArrayList<Group>(gdsMap.values());
for (Group g : result) {
g.rect = new Grib2Rectilyser(g.records, g.gdsHash);
f.format(" GDS hash %d == ", g.gdsHash);
g.rect.make(f, c);
}
f.format(
" Rectilyser: nvars=%d records unique=%d total=%d dups=%d (%f) %n",
c.vars, c.recordsUnique, c.records, c.dups, ((float) c.dups) / c.records);
return result;
}
// consistency check on variables : compare each variable to corresponding one in proto
// also set the groupno and partno for each partition
private boolean checkPartitions(TimePartition.Partition canon, Formatter f) throws IOException {
List<TimePartition.Partition> partitions = tp.getPartitions();
int npart = partitions.size();
boolean ok = true;
// for each group in canonical Partition
GribCollection canonGc = canon.makeGribCollection(f);
for (GribCollection.GroupHcs firstGroup : canonGc.getGroups()) {
String gname = firstGroup.getId();
if (trace) f.format(" Check Group %s%n", gname);
// hash proto variables for quick lookup
Map<Integer, GribCollection.VariableIndex> check =
new HashMap<Integer, GribCollection.VariableIndex>(firstGroup.varIndex.size());
List<GribCollection.VariableIndex> varIndexP =
new ArrayList<GribCollection.VariableIndex>(firstGroup.varIndex.size());
for (GribCollection.VariableIndex vi : firstGroup.varIndex) {
TimePartition.VariableIndexPartitioned vip = tp.makeVariableIndexPartitioned(vi, npart);
varIndexP.add(vip);
check.put(vi.cdmHash, vip); // replace with its evil twin
}
firstGroup.varIndex = varIndexP; // replace with its evil twin
// for each partition
for (int partno = 0; partno < npart; partno++) {
TimePartition.Partition tpp = partitions.get(partno);
if (trace) f.format(" Check Partition %s%n", tpp.getName());
// get corresponding group
GribCollection gc = tpp.makeGribCollection(f);
int groupIdx = gc.findGroupIdxById(firstGroup.getId());
if (groupIdx < 0) {
f.format(" Cant find group %s in partition %s%n", gname, tpp.getName());
ok = false;
continue;
}
GribCollection.GroupHcs group = gc.getGroup(groupIdx);
// for each variable in partition group
for (int varIdx = 0; varIdx < group.varIndex.size(); varIdx++) {
GribCollection.VariableIndex vi2 = group.varIndex.get(varIdx);
if (trace) f.format(" Check variable %s%n", vi2);
int flag = 0;
GribCollection.VariableIndex vi1 = check.get(vi2.cdmHash); // compare with proto variable
if (vi1 == null) {
f.format(
" WARN Cant find variable %s from %s in proto - ignoring that variable%n",
vi2, tpp.getName());
continue; // we can tolerate this
}
// compare vert coordinates
VertCoord vc1 = vi1.getVertCoord();
VertCoord vc2 = vi2.getVertCoord();
if ((vc1 == null) != (vc2 == null)) {
f.format(
" ERR Vert coordinates existence on variable %s in %s doesnt match%n",
vi2, tpp.getName());
ok = false;
} else if ((vc1 != null) && !vc1.equalsData(vc2)) {
f.format(
" WARN Vert coordinates values on variable %s in %s dont match%n",
vi2, tpp.getName());
f.format(" canon vc = %s%n", vc1);
f.format(" this vc = %s%n", vc2);
flag |= TimePartition.VERT_COORDS_DIFFER;
}
// compare ens coordinates
EnsCoord ec1 = vi1.getEnsCoord();
EnsCoord ec2 = vi2.getEnsCoord();
if ((ec1 == null) != (ec2 == null)) {
f.format(
" ERR Ensemble coordinates existence on variable %s in %s doesnt match%n",
vi2, tpp.getName());
ok = false;
} else if ((ec1 != null) && !ec1.equalsData(ec2)) {
f.format(
" WARN Ensemble coordinates values on variable %s in %s dont match%n",
vi2, tpp.getName());
f.format(" canon ec = %s%n", ec1);
f.format(" this ec = %s%n", ec2);
flag |= TimePartition.ENS_COORDS_DIFFER;
}
((TimePartition.VariableIndexPartitioned) vi1)
.setPartitionIndex(partno, groupIdx, varIdx, flag);
} // loop over variable
} // loop over partition
} // loop over group
if (ok) f.format(" Partition check: vert, ens coords OK%n");
return ok;
}
// read all records in all files,
// divide into groups based on GDS hash
// each group has an arraylist of all records that belong to it.
// for each group, run rectlizer to derive the coordinates and variables
public List<Group> makeAggregatedGroups(
List<String> filenames, CollectionManager.Force force, Formatter f) throws IOException {
Map<Integer, Group> gdsMap = new HashMap<Integer, Group>();
boolean intvMerge = mergeIntvDefault;
f.format("GribCollection %s: makeAggregatedGroups%n", gc.getName());
int total = 0;
int fileno = 0;
for (CollectionManager dcm : collections) {
f.format(" dcm= %s%n", dcm);
FeatureCollectionConfig.GribConfig config =
(FeatureCollectionConfig.GribConfig)
dcm.getAuxInfo(FeatureCollectionConfig.AUX_GRIB_CONFIG);
Map<Integer, Integer> gdsConvert = (config != null) ? config.gdsHash : null;
FeatureCollectionConfig.GribIntvFilter intvMap = (config != null) ? config.intvFilter : null;
intvMerge =
(config == null) || (config.intvMerge == null) ? mergeIntvDefault : config.intvMerge;
for (MFile mfile : dcm.getFiles()) {
// f.format("%3d: %s%n", fileno, mfile.getPath());
filenames.add(mfile.getPath());
Grib2Index index = null;
try {
index =
(Grib2Index)
GribIndex.readOrCreateIndexFromSingleFile(
false, !isSingleFile, mfile, config, force, f);
} catch (IOException ioe) {
logger.warn(
"GribCollectionBuilder {}: reading/Creating gbx9 index failed err={}",
gc.getName(),
ioe.getMessage());
f.format(
"GribCollectionBuilder: reading/Creating gbx9 index failed err=%s%n skipping %s%n",
ioe.getMessage(), mfile.getPath() + GribIndex.IDX_EXT);
continue;
}
for (Grib2Record gr : index.getRecords()) {
if (this.tables == null) {
Grib2SectionIdentification ids =
gr.getId(); // so all records must use the same table (!)
this.tables =
Grib2Customizer.factory(
ids.getCenter_id(),
ids.getSubcenter_id(),
ids.getMaster_table_version(),
ids.getLocal_table_version());
if (config != null)
tables.setTimeUnitConverter(
config
.getTimeUnitConverter()); // LOOK doesnt really work with multiple collections
}
if (intvMap != null && filterTinv(gr, intvMap, f)) continue; // skip
gr.setFile(fileno); // each record tracks which file it belongs to
int gdsHash =
gr.getGDSsection().getGDS().hashCode(); // use GDS hash code to group records
if (gdsConvert != null
&& gdsConvert.get(gdsHash)
!= null) // allow external config to muck with gdsHash. Why? because of error in
// encoding
gdsHash = (Integer) gdsConvert.get(gdsHash); // and we need exact hash matching
Group g = gdsMap.get(gdsHash);
if (g == null) {
g = new Group(gr.getGDSsection(), gdsHash);
gdsMap.put(gdsHash, g);
}
g.records.add(gr);
total++;
}
fileno++;
}
}
f.format(" total grib records= %d%n", total);
Grib2Rectilyser.Counter c = new Grib2Rectilyser.Counter(); // debugging
List<Group> result = new ArrayList<Group>(gdsMap.values());
for (Group g : result) {
g.rect = new Grib2Rectilyser(tables, g.records, g.gdsHash, intvMerge);
f.format(" GDS hash %d == ", g.gdsHash);
g.rect.make(f, c, filenames);
}
f.format(
" Rectilyser: nvars=%d records unique=%d total=%d dups=%d (%f) %n",
c.vars, c.recordsUnique, c.records, c.dups, ((float) c.dups) / c.records);
return result;
}