signet::forge::ColumnIndex Struct Reference

Per-page min/max statistics for predicate pushdown. More...

#include <column_index.hpp>

Public Types
enum class	BoundaryOrder : int32_t { UNORDERED = 0 , ASCENDING = 1 , DESCENDING = 2 }
	Ordering of min values across pages, used to short-circuit filtering. More...

Public Member Functions
bool	valid () const
	Check if deserialization was successful.
void	serialize (thrift::CompactEncoder &enc) const
	Serialize this ColumnIndex to a Thrift compact encoder.
void	deserialize (thrift::CompactDecoder &dec)
	Deserialize this ColumnIndex from a Thrift compact decoder.
std::vector< size_t >	filter_pages (const std::string &min_val, const std::string &max_val, PhysicalType physical_type=PhysicalType::BYTE_ARRAY) const
	Filter pages by a value range for predicate pushdown.

Public Attributes
bool	valid_ = true
	False if deserialization failed (M-V7).
std::vector< bool >	null_pages
	True if the corresponding page is all nulls.
std::vector< std::string >	min_values
	Binary-encoded minimum value per page.
std::vector< std::string >	max_values
	Binary-encoded maximum value per page.
BoundaryOrder	boundary_order = BoundaryOrder::UNORDERED
	Boundary order of min values.
std::vector< int64_t >	null_counts
	Null count per page (optional).

Detailed Description

Per-page min/max statistics for predicate pushdown.

Stores binary-encoded min/max values for each data page in a column chunk, along with null-page flags, boundary ordering, and optional null counts. Readers use filter_pages() to eliminate pages whose value ranges do not overlap the query predicate.

See also

OffsetIndex (companion for page offsets)

ColumnIndexBuilder (builder pattern for constructing during writes)

Definition at line 147 of file column_index.hpp.

Member Enumeration Documentation

BoundaryOrder

enum class signet::forge::ColumnIndex::BoundaryOrder : int32_t

strong

Ordering of min values across pages, used to short-circuit filtering.

Enumerator
UNORDERED	Min values have no particular order.
ASCENDING	Min values are non-decreasing across pages.
DESCENDING	Min values are non-increasing across pages.

Definition at line 154 of file column_index.hpp.

Member Function Documentation

deserialize()

void signet::forge::ColumnIndex::deserialize ( thrift::CompactDecoder &  dec )

inline

Deserialize this ColumnIndex from a Thrift compact decoder.

Parameters

dec	The decoder to read from.

Definition at line 215 of file column_index.hpp.

filter_pages()

std::vector< size_t > signet::forge::ColumnIndex::filter_pages ( const std::string &  min_val, const std::string &  max_val, PhysicalType  physical_type = PhysicalType::BYTE_ARRAY  ) const

inline

Filter pages by a value range for predicate pushdown.

Given a range [min_val, max_val] (binary-encoded, same encoding as min_values/max_values), returns page indices that might contain matching data. A page is excluded only if its max is strictly less than min_val or its min is strictly greater than max_val. All-null pages are always excluded.

Parameters

min_val	Lower bound of the query range (binary-encoded).
max_val	Upper bound of the query range (binary-encoded).
physical_type	Physical type for typed comparison (default: BYTE_ARRAY = lexicographic).

Returns

A vector of page indices (0-based) that may contain matches.

Definition at line 294 of file column_index.hpp.

serialize()

void signet::forge::ColumnIndex::serialize ( thrift::CompactEncoder &  enc ) const

inline

Serialize this ColumnIndex to a Thrift compact encoder.

Parameters

enc	The encoder to write to.

Definition at line 168 of file column_index.hpp.

valid()

bool signet::forge::ColumnIndex::valid ( ) const

inline

Check if deserialization was successful.

Definition at line 164 of file column_index.hpp.

Member Data Documentation

boundary_order

BoundaryOrder signet::forge::ColumnIndex::boundary_order = BoundaryOrder::UNORDERED

Boundary order of min values.

Definition at line 159 of file column_index.hpp.

max_values

std::vector<std::string> signet::forge::ColumnIndex::max_values

Binary-encoded maximum value per page.

Definition at line 151 of file column_index.hpp.

min_values

std::vector<std::string> signet::forge::ColumnIndex::min_values

Binary-encoded minimum value per page.

Definition at line 150 of file column_index.hpp.

null_counts

std::vector<int64_t> signet::forge::ColumnIndex::null_counts

Null count per page (optional).

Definition at line 161 of file column_index.hpp.

null_pages

std::vector<bool> signet::forge::ColumnIndex::null_pages

True if the corresponding page is all nulls.

Definition at line 149 of file column_index.hpp.

valid_

bool signet::forge::ColumnIndex::valid_ = true

False if deserialization failed (M-V7).

Definition at line 148 of file column_index.hpp.

---

The documentation for this struct was generated from the following file:

• include/signet/column_index.hpp