StringTestTableFunctions.cpp File Reference

#include "TableFunctionsTesting.h"
#include <iostream>
#include <string>

Include dependency graph for StringTestTableFunctions.cpp:

Go to the source code of this file.

Functions
EXTENSION_NOINLINE_HOST int32_t	ct_binding_str_length__cpu_ (const Column< TextEncodingDict > &input_str, Column< TextEncodingDict > &out_str, Column< int64_t > &out_size)
EXTENSION_NOINLINE_HOST int32_t	ct_binding_str_equals__cpu_ (const ColumnList< TextEncodingDict > &input_strings, Column< TextEncodingDict > &string_if_equal, Column< bool > &strings_are_equal)
EXTENSION_NOINLINE_HOST int32_t	ct_substr__cpu_ (TableFunctionManager &mgr, const Column< TextEncodingDict > &input_str, const Column< int > &pos, const Column< int > &len, Column< TextEncodingDict > &output_substr)
EXTENSION_NOINLINE_HOST int32_t	ct_string_concat__cpu_ (TableFunctionManager &mgr, const ColumnList< TextEncodingDict > &input_strings, const TextEncodingNone &separator, Column< TextEncodingDict > &concatted_string)
EXTENSION_NOINLINE_HOST int32_t	ct_synthesize_new_dict__cpu_ (TableFunctionManager &mgr, const int64_t num_strings, Column< TextEncodingDict > &new_dict_col)
EXTENSION_NOINLINE int32_t	ct_hamming_distance (const TextEncodingNone &str1, const TextEncodingNone &str2, Column< int32_t > &hamming_distance)
template<typename T >
TEMPLATE_NOINLINE int32_t	ct_get_string_chars__template (const Column< T > &indices, const TextEncodingNone &str, const int32_t multiplier, Column< int32_t > &idx, Column< int8_t > &char_bytes)
template TEMPLATE_NOINLINE int32_t	ct_get_string_chars__template (const Column< int16_t > &indices, const TextEncodingNone &str, const int32_t multiplier, Column< int32_t > &idx, Column< int8_t > &char_bytes)
template TEMPLATE_NOINLINE int32_t	ct_get_string_chars__template (const Column< int32_t > &indices, const TextEncodingNone &str, const int32_t multiplier, Column< int32_t > &idx, Column< int8_t > &char_bytes)
EXTENSION_NOINLINE_HOST int32_t	ct_string_to_chars__cpu_ (const TextEncodingNone &input, Column< int32_t > &char_idx, Column< int8_t > &char_bytes)

Function Documentation

EXTENSION_NOINLINE_HOST int32_t ct_binding_str_equals__cpu_	(	const ColumnList< TextEncodingDict > &	input_strings,
		Column< TextEncodingDict > &	string_if_equal,
		Column< bool > &	strings_are_equal
	)

Definition at line 40 of file StringTestTableFunctions.cpp.

References ColumnList< TextEncodingDict >::numCols(), set_output_row_size(), Column< TextEncodingDict >::setNull(), and ColumnList< TextEncodingDict >::size().

                                                                     {
  const int64_t num_rows = input_strings.size();
  const int64_t num_cols = input_strings.numCols();
  set_output_row_size(num_rows);
  for (int64_t r = 0; r < num_rows; r++) {
    bool are_equal = true;
    if (num_cols > 0) {
      std::string first_str = input_strings[0].getString(r);
      for (int64_t c = 1; c != num_cols; ++c) {
        if (input_strings[c].getString(r) != first_str) {
          are_equal = false;
          break;
        }
      }
      strings_are_equal[r] = are_equal;
      if (are_equal && num_cols > 0) {
        string_if_equal[r] = input_strings[0][r];
      } else {
        string_if_equal.setNull(r);
      }
    }
  }
  return num_rows;
}

Here is the call graph for this function:

EXTENSION_NOINLINE_HOST int32_t ct_binding_str_length__cpu_	(	const Column< TextEncodingDict > &	input_str,
		Column< TextEncodingDict > &	out_str,
		Column< int64_t > &	out_size
	)

Definition at line 26 of file StringTestTableFunctions.cpp.

References Column< TextEncodingDict >::getString(), set_output_row_size(), Column< T >::size(), and Column< TextEncodingDict >::size().

                                                               {
  const int64_t num_rows = input_str.size();
  set_output_row_size(num_rows);
  for (int64_t i = 0; i < num_rows; i++) {
    out_str[i] = input_str[i];
    const std::string str = input_str.getString(i);
    out_size[i] = str.size();
  }
  return num_rows;
}

Here is the call graph for this function:

template<typename T >

TEMPLATE_NOINLINE int32_t ct_get_string_chars__template	(	const Column< T > &	indices,
		const TextEncodingNone &	str,
		const int32_t	multiplier,
		Column< int32_t > &	idx,
		Column< int8_t > &	char_bytes
	)

Definition at line 152 of file StringTestTableFunctions.cpp.

References TextEncodingNone::size(), and Column< T >::size().

                                                                                    {
  const int32_t str_len = str.size();
  // Note: we assume RowMultiplier is 1 for this test, was to make running on
  // GPU easy Todo: Provide Constant RowMultiplier interface
  if (multiplier != 1) {
    return 0;
  }
  const int32_t num_input_rows = indices.size();
  const int32_t num_output_rows = num_input_rows * multiplier;

#ifdef __CUDACC__
  const int32_t start = threadIdx.x + blockDim.x * blockIdx.x;
  const int32_t step = blockDim.x * gridDim.x;
#else
  const int32_t start = 0;
  const int32_t step = 1;
#endif

  for (int32_t i = start; i < num_output_rows; i += step) {
    idx[i] = indices[i % num_output_rows];
    char_bytes[i] = str[i % str_len];  // index < str_len ? str[i] : 0;
  }
  return num_output_rows;
}

Here is the call graph for this function:

template TEMPLATE_NOINLINE int32_t ct_get_string_chars__template	(	const Column< int16_t > &	indices,
		const TextEncodingNone &	str,
		const int32_t	multiplier,
		Column< int32_t > &	idx,
		Column< int8_t > &	char_bytes
	)

template TEMPLATE_NOINLINE int32_t ct_get_string_chars__template	(	const Column< int32_t > &	indices,
		const TextEncodingNone &	str,
		const int32_t	multiplier,
		Column< int32_t > &	idx,
		Column< int8_t > &	char_bytes
	)

EXTENSION_NOINLINE int32_t ct_hamming_distance	(	const TextEncodingNone &	str1,
		const TextEncodingNone &	str2,
		Column< int32_t > &	hamming_distance
	)

Definition at line 125 of file StringTestTableFunctions.cpp.

References Column< T >::ptr_, and TextEncodingNone::size().

                                                                                  {
  const int32_t str_len = str1.size() <= str2.size() ? str1.size() : str2.size();

#ifdef __CUDACC__
  const int32_t start = threadIdx.x + blockDim.x * blockIdx.x;
  const int32_t step = blockDim.x * gridDim.x;
  int32_t* output_ptr = hamming_distance.ptr_;
#else
  const int32_t start = 0;
  const int32_t step = 1;
#endif

  int32_t num_chars_unequal = 0;
  for (int32_t i = start; i < str_len; i += step) {
    num_chars_unequal += (str1[i] != str2[i]) ? 1 : 0;
  }
#ifdef __CUDACC__
  atomicAdd(output_ptr, num_chars_unequal);
#else
  hamming_distance[0] = num_chars_unequal;
#endif
  return 1;
}

Here is the call graph for this function:

EXTENSION_NOINLINE_HOST int32_t ct_string_concat__cpu_	(	TableFunctionManager &	mgr,
		const ColumnList< TextEncodingDict > &	input_strings,
		const TextEncodingNone &	separator,
		Column< TextEncodingDict > &	concatted_string
	)

Definition at line 85 of file StringTestTableFunctions.cpp.

References Column< TextEncodingDict >::getOrAddTransient(), TextEncodingNone::getString(), ColumnList< TextEncodingDict >::numCols(), TableFunctionManager::set_output_row_size(), Column< TextEncodingDict >::setNull(), and ColumnList< TextEncodingDict >::size().

                                                                           {
  const int64_t num_rows = input_strings.size();
  const int64_t num_cols = input_strings.numCols();
  const std::string separator_str{separator.getString()};
  mgr.set_output_row_size(num_rows);
  for (int64_t row_idx = 0; row_idx < num_rows; row_idx++) {
    if (num_cols > 0) {
      std::string concatted_output{input_strings[0].getString(row_idx)};
      for (int64_t col_idx = 1; col_idx < num_cols; ++col_idx) {
        concatted_output += separator_str;
        concatted_output += input_strings[col_idx].getString(row_idx);
      }
      const TextEncodingDict concatted_str_id =
          concatted_string.getOrAddTransient(concatted_output);
      concatted_string[row_idx] = concatted_str_id;
    } else {
      concatted_string.setNull(row_idx);
    }
  }
  return num_rows;
}

Here is the call graph for this function:

EXTENSION_NOINLINE_HOST int32_t ct_string_to_chars__cpu_	(	const TextEncodingNone &	input,
		Column< int32_t > &	char_idx,
		Column< int8_t > &	char_bytes
	)

Definition at line 200 of file StringTestTableFunctions.cpp.

References TextEncodingNone::getString(), and set_output_row_size().

                                                                                     {
  const std::string str{input.getString()};
  const int64_t str_size(str.size());
  set_output_row_size(str_size);
  for (int32_t i = 0; i < str_size; ++i) {
    char_idx[i] = i;
    char_bytes[i] = str[i];
  }
  return str_size;
}

Here is the call graph for this function:

EXTENSION_NOINLINE_HOST int32_t ct_substr__cpu_	(	TableFunctionManager &	mgr,
		const Column< TextEncodingDict > &	input_str,
		const Column< int > &	pos,
		const Column< int > &	len,
		Column< TextEncodingDict > &	output_substr
	)

Definition at line 68 of file StringTestTableFunctions.cpp.

References Column< TextEncodingDict >::getOrAddTransient(), Column< TextEncodingDict >::getString(), TableFunctionManager::set_output_row_size(), Column< TextEncodingDict >::size(), and substring().

                                                                 {
  const int64_t num_rows = input_str.size();
  mgr.set_output_row_size(num_rows);
  for (int64_t row_idx = 0; row_idx < num_rows; row_idx++) {
    const std::string input_string{input_str.getString(row_idx)};
    const std::string substring = input_string.substr(pos[row_idx], len[row_idx]);
    const TextEncodingDict substr_id = output_substr.getOrAddTransient(substring);
    output_substr[row_idx] = substr_id;
  }
  return num_rows;
}

Here is the call graph for this function:

EXTENSION_NOINLINE_HOST int32_t ct_synthesize_new_dict__cpu_	(	TableFunctionManager &	mgr,
		const int64_t	num_strings,
		Column< TextEncodingDict > &	new_dict_col
	)

Definition at line 111 of file StringTestTableFunctions.cpp.

References Column< TextEncodingDict >::getOrAddTransient(), TableFunctionManager::set_output_row_size(), and to_string().

                                                                             {
  mgr.set_output_row_size(num_strings);
  for (int32_t s = 0; s < num_strings; ++s) {
    const std::string new_string = "String_" + std::to_string(s);
    const int32_t string_id = new_dict_col.getOrAddTransient(new_string);
    new_dict_col[s] = string_id;
  }
  return num_strings;
}

Here is the call graph for this function: