openswmm::TableData Struct Reference

SoA collection of all time series and curves in the model. More...

#include <TableData.hpp>

Collaboration diagram for openswmm::TableData:

Public Member Functions
std::size_t	count () const noexcept
	Count stored time-series and curve tables.
Table &	operator[] (int idx)
	Borrow a mutable stored table.
const Table &	operator[] (int idx) const
	Borrow an immutable stored table.
int	add (const std::string &id, TableType type)
	Append an empty table and index its copied name.
void	rebuild_index ()
	Rebuilds by_name from tables.
int	find_by_kind (std::string_view name, bool want_timeseries) const noexcept
	Find the first indexed table matching a case-insensitive name and kind.
void	reset_cursors () noexcept
	Reset all cursors (call before re-running a simulation).

Public Attributes
std::vector< Table >	tables
	All tables in index order.
std::unordered_map< std::string, std::vector< int >, CiHash, CiEqual >	by_name
	Case-insensitive name → table indices, ascending.

Detailed Description

SoA collection of all time series and curves in the model.

Indexed by table index (integer). Lookup by name is done via the SimulationContext name-to-index hash map.

Member Function Documentation

add()

int openswmm::TableData::add ( const std::string & id, TableType type )

inline

Append an empty table and index its copied name.

Name index uses ASCII-insensitive equality and appends index in increasing order. tables push precedes map insertion; allocation failure may leave the lookup index incomplete. Existing table references may be invalidated.

Parameters

[in]	id	Borrowed table ID copied verbatim; duplicate and case-equivalent names are allowed.
[in]	type	Table type stored without enum validation.

Returns

New table index.

Here is the caller graph for this function:

count()

std::size_t openswmm::TableData::count ( ) const

inlinenoexcept

Count stored time-series and curve tables.

Returns

Owned table vector size, independent of name-index synchronization.

Here is the caller graph for this function:

find_by_kind()

int openswmm::TableData::find_by_kind ( std::string_view name, bool want_timeseries ) const

inlinenoexcept

Find the first indexed table matching a case-insensitive name and kind.

Normally lowest matching index because add/rebuild_index keep ordered lists. Caller must maintain a coherent map and valid stored indices after direct edits.

Parameters

[in]	name	Borrowed name, matched through the ASCII-insensitive name map.
[in]	want_timeseries	True accepts TIMESERIES only; false accepts every non-TIMESERIES type.

Returns

First matching indexed row, or -1.

Here is the caller graph for this function:

operator[]() [1/2]

Table & openswmm::TableData::operator[] ( int idx )

inline

Borrow a mutable stored table.

Caller rebuilds the name index after changing table IDs or structure; mutable access is not lifecycle/thread guarded.

Parameters

[in]

idx

Existing index in [0,count()); unchecked, including negatives.

Returns

Owned table reference; vector growth/erasure or container destruction can invalidate it.

operator[]() [2/2]

const Table & openswmm::TableData::operator[] ( int idx ) const

inline

Borrow an immutable stored table.

Parameters

[in]

idx

Existing table index; unchecked.

Returns

Owned reference, invalidated by vector relocation/erasure or destruction.

rebuild_index()

void openswmm::TableData::rebuild_index ( )

inline

Rebuilds by_name from tables.

Required after any structural change that is not an append — in practice only ObjectDeleter's erase-and-renumber, which is already O(n). Anything that mutates tables directly without calling this leaves lookups stale.

Here is the caller graph for this function:

reset_cursors()

void openswmm::TableData::reset_cursors ( )

inlinenoexcept

Reset all cursors (call before re-running a simulation).

Member Data Documentation

by_name

std::unordered_map<std::string, std::vector<int>, CiHash, CiEqual> openswmm::TableData::by_name

Case-insensitive name → table indices, ascending.

Replaces the linear ieq scan that find_timeseries() / find_curve() used to run over every table. That scan was the load path's worst complexity defect: [TIMESERIES] and [CURVES] rows repeat their object's name on every row in standard SWMM files, so parsing a model with many long series cost O(total_rows x n_tables) string compares before anything else happened.

A name maps to a list because legacy keeps TSERIES and CURVE in separate hash tables, so one name may legitimately denote both a timeseries and a curve. Lists are in ascending index order (add() only appends), which is what makes a hash hit return the same winner the old first-match scan did — the duplicate-name parity the name-case tests pin.

tables

std::vector<Table> openswmm::TableData::tables

All tables in index order.

---

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

• /home/runner/work/openswmm.engine/openswmm.engine/src/engine/data/TableData.hpp