Skip to content

Commit

Permalink
Merge pull request #452 from jpmorganchase/port-view
Browse files Browse the repository at this point in the history 
Port view to C++
  • Loading branch information
@texodus
texodus authored Feb 27, 2019
2 parents 9b3b1a3 + 0268586 commit 711071e
Show file tree
Hide file tree
Showing 14 changed files with 1,040 additions and 827 deletions.
163 changes: 163 additions & 0 deletions cpp/perspective/src/cpp/base.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -221,6 +221,38 @@ get_dtype_descr(t_dtype dtype) {
return std::string("dummy");
}

std::string
dtype_to_str(t_dtype dtype) {
std::stringstream str_dtype;
switch (dtype) {
case DTYPE_FLOAT32:
case DTYPE_FLOAT64: {
str_dtype << "float";
} break;
case DTYPE_INT8:
case DTYPE_INT16:
case DTYPE_INT32:
case DTYPE_INT64: {
str_dtype << "integer";
} break;
case DTYPE_BOOL: {
str_dtype << "boolean";
} break;
case DTYPE_DATE: {
str_dtype << "date";
} break;
case DTYPE_TIME: {
str_dtype << "datetime";
} break;
case DTYPE_STR: {
str_dtype << "string";
} break;
default: { PSP_COMPLAIN_AND_ABORT("Cannot convert unknown dtype to string!"); }
}

return str_dtype.str();
}

std::string
filter_op_to_str(t_filter_op op) {
switch (op) {
Expand Down Expand Up @@ -280,6 +312,137 @@ filter_op_to_str(t_filter_op op) {
return "";
}

t_filter_op
str_to_filter_op(std::string str) {
if (str == "<") {
return t_filter_op::FILTER_OP_LT;
} else if (str == "<=") {
return t_filter_op::FILTER_OP_LTEQ;
} else if (str == ">") {
return t_filter_op::FILTER_OP_GT;
} else if (str == ">=") {
return t_filter_op::FILTER_OP_GTEQ;
} else if (str == "==") {
return t_filter_op::FILTER_OP_EQ;
} else if (str == "!=") {
return t_filter_op::FILTER_OP_NE;
} else if (str == "begins with" || str == "startswith") {
return t_filter_op::FILTER_OP_BEGINS_WITH;
} else if (str == "ends with" || str == "endswith") {
return t_filter_op::FILTER_OP_ENDS_WITH;
} else if (str == "in") {
return t_filter_op::FILTER_OP_IN;
} else if (str == "contains") {
return t_filter_op::FILTER_OP_CONTAINS;
} else if (str == "not in") {
return t_filter_op::FILTER_OP_NOT_IN;
} else if (str == "&" || str == "and") {
return t_filter_op::FILTER_OP_AND;
} else if (str == "|") {
return t_filter_op::FILTER_OP_OR;
} else if (str == "is nan" || str == "is_nan") {
return t_filter_op::FILTER_OP_IS_NAN;
} else if (str == "is not nan" || str == "!is_nan") {
return t_filter_op::FILTER_OP_IS_NOT_NAN;
} else if (str == "is not None") {
return t_filter_op::FILTER_OP_IS_VALID;
} else if (str == "is None") {
return t_filter_op::FILTER_OP_IS_NOT_VALID;
} else {
PSP_COMPLAIN_AND_ABORT("Encountered unknown filter operation.");
// use and as default
return t_filter_op::FILTER_OP_AND;
}
}

t_sorttype
str_to_sorttype(std::string str) {
if (str == "none") {
return SORTTYPE_NONE;
} else if (str == "asc" || str == "col asc") {
return SORTTYPE_ASCENDING;
} else if (str == "desc" || str == "col desc") {
return SORTTYPE_DESCENDING;
} else if (str == "asc abs" || str == "col asc abs") {
return SORTTYPE_ASCENDING_ABS;
} else if (str == "desc abs" || str == "col desc abs") {
return SORTTYPE_DESCENDING_ABS;
} else {
PSP_COMPLAIN_AND_ABORT("Encountered unknown sort type string");
return SORTTYPE_DESCENDING;
}
}

t_aggtype
str_to_aggtype(std::string str) {
if (str == "distinct count" || str == "distinctcount" || str == "distinct"
|| str == "distinct_count") {
return t_aggtype::AGGTYPE_DISTINCT_COUNT;
} else if (str == "sum") {
return t_aggtype::AGGTYPE_SUM;
} else if (str == "mul") {
return t_aggtype::AGGTYPE_MUL;
} else if (str == "avg" || str == "mean") {
return t_aggtype::AGGTYPE_MEAN;
} else if (str == "count") {
return t_aggtype::AGGTYPE_COUNT;
} else if (str == "weighted mean" || str == "weighted_mean") {
return t_aggtype::AGGTYPE_WEIGHTED_MEAN;
} else if (str == "unique") {
return t_aggtype::AGGTYPE_UNIQUE;
} else if (str == "any") {
return t_aggtype::AGGTYPE_ANY;
} else if (str == "median") {
return t_aggtype::AGGTYPE_MEDIAN;
} else if (str == "join") {
return t_aggtype::AGGTYPE_JOIN;
} else if (str == "div") {
return t_aggtype::AGGTYPE_SCALED_DIV;
} else if (str == "add") {
return t_aggtype::AGGTYPE_SCALED_ADD;
} else if (str == "dominant") {
return t_aggtype::AGGTYPE_DOMINANT;
} else if (str == "first by index" || str == "first") {
return t_aggtype::AGGTYPE_FIRST;
} else if (str == "last by index") {
return t_aggtype::AGGTYPE_LAST;
} else if (str == "py_agg") {
return t_aggtype::AGGTYPE_PY_AGG;
} else if (str == "and") {
return t_aggtype::AGGTYPE_AND;
} else if (str == "or") {
return t_aggtype::AGGTYPE_OR;
} else if (str == "last" || str == "last_value") {
return t_aggtype::AGGTYPE_LAST_VALUE;
} else if (str == "high" || str == "high_water_mark") {
return t_aggtype::AGGTYPE_HIGH_WATER_MARK;
} else if (str == "low" || str == "low_water_mark") {
return t_aggtype::AGGTYPE_LOW_WATER_MARK;
} else if (str == "sub abs") {
return t_aggtype::AGGTYPE_SUM_ABS;
} else if (str == "sum not null" || str == "sum_not_null") {
return t_aggtype::AGGTYPE_SUM_NOT_NULL;
} else if (str == "mean by count" || str == "mean_by_count") {
return t_aggtype::AGGTYPE_MEAN_BY_COUNT;
} else if (str == "identity") {
return t_aggtype::AGGTYPE_IDENTITY;
} else if (str == "distinct leaf" || str == "distinct_leaf") {
return t_aggtype::AGGTYPE_DISTINCT_LEAF;
} else if (str == "pct sum parent" || str == "pct_sum_parent") {
return t_aggtype::AGGTYPE_PCT_SUM_PARENT;
} else if (str == "pct sum grand total" || str == "pct_sum_grand_total") {
return t_aggtype::AGGTYPE_PCT_SUM_GRAND_TOTAL;
} else if (str.find("udf_combiner_") != std::string::npos) {
return t_aggtype::AGGTYPE_UDF_COMBINER;
} else if (str.find("udf_reducer_") != std::string::npos) {
return t_aggtype::AGGTYPE_UDF_REDUCER;
} else {
PSP_COMPLAIN_AND_ABORT("Encountered unknown aggregate operation.");
// use any as default
return t_aggtype::AGGTYPE_ANY;
}
}

std::string
get_status_descr(t_status status) {
switch (status) {
Expand Down
57 changes: 57 additions & 0 deletions cpp/perspective/src/cpp/config.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -83,6 +83,26 @@ t_config::t_config(const std::vector<t_pivot>& row_pivots,
setup(detail_columns, sort_pivot, sort_pivot_by);
}

// view config
t_config::t_config(const std::vector<std::string>& row_pivots,
const std::vector<std::string>& col_pivots, const std::vector<t_aggspec>& aggregates,
const std::vector<t_sortspec>& sortspecs, t_filter_op combiner,
const std::vector<t_fterm>& fterms, const std::vector<std::string>& col_names,
bool column_only)
: m_column_only(column_only)
, m_sortspecs(sortspecs)
, m_aggregates(aggregates)
, m_detail_columns(col_names)
, m_combiner(combiner)
, m_fterms(fterms) {
for (const auto& p : row_pivots) {
m_row_pivots.push_back(t_pivot(p));
}
for (const auto& p : col_pivots) {
m_col_pivots.push_back(t_pivot(p));
}
};

t_config::t_config(
const std::vector<t_pivot>& row_pivots, const std::vector<t_aggspec>& aggregates)
: m_row_pivots(row_pivots)
Expand Down Expand Up @@ -120,6 +140,20 @@ t_config::t_config(const std::vector<std::string>& row_pivots,

{}

t_config::t_config(const std::vector<t_pivot>& row_pivots,
const std::vector<t_pivot>& col_pivots, const std::vector<t_aggspec>& aggregates,
const t_totals totals, t_filter_op combiner, const std::vector<t_fterm>& fterms)
: m_row_pivots(row_pivots)
, m_col_pivots(col_pivots)
, m_aggregates(aggregates)
, m_totals(totals)
, m_combiner(combiner)
, m_fterms(fterms)
, m_handle_nan_sort(true)
, m_fmode(FMODE_SIMPLE_CLAUSES) {
setup(m_detail_columns, std::vector<std::string>{}, std::vector<std::string>{});
}

t_config::t_config(const std::vector<std::string>& row_pivots,
const std::vector<std::string>& col_pivots, const std::vector<t_aggspec>& aggregates,
const t_totals totals, t_filter_op combiner, const std::vector<t_fterm>& fterms)
Expand Down Expand Up @@ -168,6 +202,19 @@ t_config::t_config(const std::vector<std::string>& row_pivots, const t_aggspec&
setup(m_detail_columns, std::vector<std::string>{}, std::vector<std::string>{});
}

t_config::t_config(const std::vector<t_pivot>& row_pivots,
const std::vector<t_aggspec>& aggregates, t_filter_op combiner,
const std::vector<t_fterm>& fterms)
: m_row_pivots(row_pivots)
, m_aggregates(aggregates)
, m_totals(TOTALS_BEFORE)
, m_combiner(combiner)
, m_fterms(fterms)
, m_handle_nan_sort(true)
, m_fmode(FMODE_SIMPLE_CLAUSES) {
setup(m_detail_columns, std::vector<std::string>{}, std::vector<std::string>{});
}

t_config::t_config(const std::vector<std::string>& row_pivots,
const std::vector<t_aggspec>& aggregates, t_filter_op combiner,
const std::vector<t_fterm>& fterms)
Expand Down Expand Up @@ -348,6 +395,11 @@ t_config::get_num_cpivots() const {
return m_col_pivots.size();
}

bool
t_config::get_column_only() const {
return m_column_only;
}

const std::vector<t_pivot>&
t_config::get_row_pivots() const {
return m_row_pivots;
Expand All @@ -371,6 +423,11 @@ t_config::get_sortby_pairs() const {
return rval;
}

const std::vector<t_sortspec>&
t_config::get_sortspecs() const {
return m_sortspecs;
}

const std::vector<t_aggspec>&
t_config::get_aggregates() const {
return m_aggregates;
Expand Down
Loading

0 comments on commit 711071e

Please sign in to comment.