Examples with MRTask water.MRTask used on opensource projects

Example 16 with MRTask

use of water.MRTask in project h2o-3 by h2oai.

the class AstReplaceFirst method replaceFirstStringCol.

private Vec replaceFirstStringCol(Vec vec, String pat, String rep, boolean ic) {
    final String pattern = pat;
    final String replacement = rep;
    final boolean ignoreCase = ic;
    return new MRTask() {

        @Override
        public void map(Chunk chk, NewChunk newChk) {
            if (// all NAs
            chk instanceof C0DChunk)
                for (int i = 0; i < chk.len(); i++) newChk.addNA();
            else {
                //        if (((CStrChunk)chk)._isAllASCII) { // fast-path operations
                //          ((CStrChunk) chk).asciiReplaceFirst(newChk);
                //        } else { //UTF requires Java string methods for accuracy
                BufferedString tmpStr = new BufferedString();
                for (int i = 0; i < chk._len; i++) {
                    if (chk.isNA(i))
                        newChk.addNA();
                    else {
                        if (ignoreCase)
                            newChk.addStr(chk.atStr(tmpStr, i).toString().toLowerCase(Locale.ENGLISH).replaceFirst(pattern, replacement));
                        else
                            newChk.addStr(chk.atStr(tmpStr, i).toString().replaceFirst(pattern, replacement));
                    }
                }
            }
        }
    }.doAll(new byte[] { Vec.T_STR }, vec).outputFrame().anyVec();
}

Example 17 with MRTask

use of water.MRTask in project h2o-3 by h2oai.

the class AstStrSplit method strSplitCategoricalCol.

private Vec[] strSplitCategoricalCol(Vec vec, String splitRegEx) {
    final String[] old_domains = vec.domain();
    final String[][] new_domains = newDomains(old_domains, splitRegEx);
    final String regex = splitRegEx;
    return new MRTask() {

        @Override
        public void map(Chunk[] cs, NewChunk[] ncs) {
            Chunk c = cs[0];
            for (int i = 0; i < c._len; ++i) {
                int cnt = 0;
                if (!c.isNA(i)) {
                    int idx = (int) c.at8(i);
                    String s = old_domains[idx];
                    String[] ss = s.split(regex);
                    for (String s1 : ss) {
                        int n_idx = Arrays.asList(new_domains[cnt]).indexOf(s1);
                        if (n_idx == -1)
                            ncs[cnt++].addNA();
                        else
                            ncs[cnt++].addNum(n_idx);
                    }
                }
                if (cnt < ncs.length)
                    for (; cnt < ncs.length; ++cnt) ncs[cnt].addNA();
            }
        }
    }.doAll(new_domains.length, Vec.T_CAT, new Frame(vec)).outputFrame(null, null, new_domains).vecs();
}

Example 18 with MRTask

use of water.MRTask in project h2o-3 by h2oai.

the class AstSetTimeZone method apply.

@Override
public ValNum apply(Env env, Env.StackHelp stk, AstRoot[] asts) {
    final String tz = asts[1].exec(env).getStr();
    Set<String> idSet = DateTimeZone.getAvailableIDs();
    if (!idSet.contains(tz))
        throw new IllegalArgumentException("Unacceptable timezone " + tz + " given.  For a list of acceptable names, use listTimezone().");
    new MRTask() {

        @Override
        public void setupLocal() {
            ParseTime.setTimezone(tz);
        }
    }.doAllNodes();
    return new ValNum(Double.NaN);
}

Example 19 with MRTask

use of water.MRTask in project h2o-3 by h2oai.

the class AstBinOp method vec_op_frame.

private ValFrame vec_op_frame(Vec vec, Frame fr) {
    // Already checked for same rows, non-zero frame
    Frame rt = new Frame(fr);
    rt.add("", vec);
    Frame res = new MRTask() {

        @Override
        public void map(Chunk[] chks, NewChunk[] cress) {
            assert cress.length == chks.length - 1;
            Chunk clf = chks[cress.length];
            for (int c = 0; c < cress.length; c++) {
                Chunk crt = chks[c];
                NewChunk cres = cress[c];
                for (int i = 0; i < clf._len; i++) cres.addNum(op(clf.atd(i), crt.atd(i)));
            }
        }
    }.doAll(fr.numCols(), Vec.T_NUM, rt).outputFrame(fr._names, null);
    // Cleanup categorical misuse
    return cleanCategorical(fr, res);
}

Example 20 with MRTask

use of water.MRTask in project h2o-3 by h2oai.

the class AstBinOp method frame_op_row.

private ValFrame frame_op_row(Frame lf, Frame row) {
    final double[] rawRow = new double[row.numCols()];
    for (int i = 0; i < rawRow.length; ++i) // is numberlike, if not then NaN
    rawRow[i] = row.vec(i).isNumeric() || row.vec(i).isTime() ? row.vec(i).at(0) : Double.NaN;
    Frame res = new MRTask() {

        @Override
        public void map(Chunk[] chks, NewChunk[] cress) {
            for (int c = 0; c < cress.length; c++) {
                Chunk clf = chks[c];
                NewChunk cres = cress[c];
                for (int r = 0; r < clf._len; ++r) {
                    if (clf.vec().isString())
                        // TODO: improve
                        cres.addNum(Double.NaN);
                    else
                        cres.addNum(op(clf.atd(r), rawRow[c]));
                }
            }
        }
    }.doAll(lf.numCols(), Vec.T_NUM, lf).outputFrame(lf._names, null);
    return cleanCategorical(lf, res);
}