Examples with BufferedString water.parser.BufferedString used on opensource projects

Example 6 with BufferedString

use of water.parser.BufferedString in project h2o-3 by h2oai.

the class AstBinOp method frame_op_frame.

/**
   * Auto-widen: If one frame has only 1 column, auto-widen that 1 column to
   * the rest.  Otherwise the frames must have the same column count, and
   * auto-widen element-by-element.  Short-cut if one frame has zero
   * columns.
   */
private ValFrame frame_op_frame(Frame lf, Frame rt) {
    if (lf.numRows() != rt.numRows()) {
        // special case for broadcasting a single row of data across a frame
        if (lf.numRows() == 1 || rt.numRows() == 1) {
            if (lf.numCols() != rt.numCols())
                throw new IllegalArgumentException("Frames must have same columns, found " + lf.numCols() + " columns and " + rt.numCols() + " columns.");
            return frame_op_row(lf, rt);
        } else
            throw new IllegalArgumentException("Frames must have same rows, found " + lf.numRows() + " rows and " + rt.numRows() + " rows.");
    }
    if (lf.numCols() == 0)
        return new ValFrame(lf);
    if (rt.numCols() == 0)
        return new ValFrame(rt);
    if (lf.numCols() == 1 && rt.numCols() > 1)
        return vec_op_frame(lf.vecs()[0], rt);
    if (rt.numCols() == 1 && lf.numCols() > 1)
        return frame_op_vec(lf, rt.vecs()[0]);
    if (lf.numCols() != rt.numCols())
        throw new IllegalArgumentException("Frames must have same columns, found " + lf.numCols() + " columns and " + rt.numCols() + " columns.");
    Frame res = new MRTask() {

        @Override
        public void map(Chunk[] chks, NewChunk[] cress) {
            BufferedString lfstr = new BufferedString();
            BufferedString rtstr = new BufferedString();
            assert (cress.length << 1) == chks.length;
            for (int c = 0; c < cress.length; c++) {
                Chunk clf = chks[c];
                Chunk crt = chks[c + cress.length];
                NewChunk cres = cress[c];
                if (clf.vec().isString())
                    for (int i = 0; i < clf._len; i++) cres.addNum(str_op(clf.atStr(lfstr, i), crt.atStr(rtstr, i)));
                else
                    for (int i = 0; i < clf._len; i++) cres.addNum(op(clf.atd(i), crt.atd(i)));
            }
        }
    }.doAll(lf.numCols(), Vec.T_NUM, new Frame(lf).add(rt)).outputFrame(lf._names, null);
    // Cleanup categorical misuse
    return cleanCategorical(lf, res);
}

Example 7 with BufferedString

use of water.parser.BufferedString in project h2o-3 by h2oai.

the class AstBinOp method scalar_op_frame.

/**
   * Auto-widen the scalar to every element of the frame
   */
private ValFrame scalar_op_frame(final String str, Frame fr) {
    Frame res = new MRTask() {

        @Override
        public void map(Chunk[] chks, NewChunk[] cress) {
            BufferedString vstr = new BufferedString();
            for (int c = 0; c < chks.length; c++) {
                Chunk chk = chks[c];
                NewChunk cres = cress[c];
                Vec vec = chk.vec();
                // String Vectors: apply str_op as BufferedStrings to all elements
                if (vec.isString()) {
                    final BufferedString conStr = new BufferedString(str);
                    for (int i = 0; i < chk._len; i++) cres.addNum(str_op(conStr, chk.atStr(vstr, i)));
                } else if (vec.isCategorical()) {
                    // categorical Vectors: convert string to domain value; apply op (not
                    // str_op).  Not sure what the "right" behavior here is, can
                    // easily argue that should instead apply str_op to the categorical
                    // string domain value - except that this whole operation only
                    // makes sense for EQ/NE, and is much faster when just comparing
                    // doubles vs comparing strings.
                    final double d = (double) ArrayUtils.find(vec.domain(), str);
                    for (int i = 0; i < chk._len; i++) cres.addNum(op(d, chk.atd(i)));
                } else {
                    // mixing string and numeric
                    // false or true only
                    final double d = op(1, 2);
                    for (int i = 0; i < chk._len; i++) cres.addNum(d);
                }
            }
        }
    }.doAll(fr.numCols(), Vec.T_NUM, fr).outputFrame(fr._names, null);
    return new ValFrame(res);
}

Example 8 with BufferedString

use of water.parser.BufferedString in project h2o-3 by h2oai.

the class AstAsDate method apply.

@Override
public ValFrame apply(Env env, Env.StackHelp stk, AstRoot[] asts) {
    Frame fr = stk.track(asts[1].exec(env)).getFrame();
    Vec vec = fr.vecs()[0];
    if (fr.vecs().length != 1 || !(vec.isCategorical() || vec.isString()))
        throw new IllegalArgumentException("as.Date requires a single column of factors or strings");
    final String format = asts[2].exec(env).getStr();
    if (format.isEmpty())
        throw new IllegalArgumentException("as.Date requires a non-empty format string");
    // check the format string more?
    final String[] dom = vec.domain();
    final boolean isStr = dom == null && vec.isString();
    assert isStr || dom != null : "as.Date error: domain is null, but vec is not String";
    Frame fr2 = new MRTask() {

        private transient DateTimeFormatter _fmt;

        @Override
        public void setupLocal() {
            _fmt = ParseTime.forStrptimePattern(format).withZone(ParseTime.getTimezone());
        }

        @Override
        public void map(Chunk c, NewChunk nc) {
            //done on each node in lieu of rewriting DateTimeFormatter as Iced
            String date;
            BufferedString tmpStr = new BufferedString();
            for (int i = 0; i < c._len; ++i) {
                if (!c.isNA(i)) {
                    if (isStr)
                        date = c.atStr(tmpStr, i).toString();
                    else
                        date = dom[(int) c.at8(i)];
                    nc.addNum(DateTime.parse(date, _fmt).getMillis(), 0);
                } else
                    nc.addNA();
            }
        }
    }.doAll(1, Vec.T_NUM, fr).outputFrame(fr._names, null);
    return new ValFrame(fr2);
}

Example 9 with BufferedString

use of water.parser.BufferedString in project h2o-3 by h2oai.

the class AstReplaceAll method replaceAllStringCol.

private Vec replaceAllStringCol(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).asciiReplaceAll(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).replaceAll(pattern, replacement));
                        else
                            newChk.addStr(chk.atStr(tmpStr, i).toString().replaceAll(pattern, replacement));
                    }
                }
            }
        }
    }.doAll(new byte[] { Vec.T_STR }, vec).outputFrame().anyVec();
}

Example 10 with BufferedString

use of water.parser.BufferedString 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();
}