Parse-once cache for multi-column external series files (CSV / TSV / PCSWMM TSF). More...

#include <string>
#include <string_view>
#include <unordered_map>
#include <vector>

Include dependency graph for MultiColumnSeriesFile.hpp:

This graph shows which files directly or indirectly include this file:

Classes
struct	openswmm::input::ParsedSeriesFile
	One fully parsed multi-column series file (raw file-unit values). More...

class	openswmm::input::MultiColumnFileCache
	Per-resolve-pass cache keyed by resolved path. More...

Namespaces
namespace	openswmm

namespace	openswmm::input

Enumerations
enum class	openswmm::input::SeriesFileStatus { openswmm::input::OK , openswmm::input::OPEN_FAILED , openswmm::input::FORMAT_FAILED }
	Why a parse failed (or that it did not). More...

Functions
bool	openswmm::input::parse_series_datetime (const std::string &cell, double &out)
	Parse a full datetime cell.

bool	openswmm::input::parse_multicolumn_series_file (const std::string &abs_path, ParsedSeriesFile &out, std::vector< std::string > &errors, SeriesFileStatus *status=nullptr)
	Parse a multi-column series file (format auto-detected).

bool	openswmm::input::looks_like_multicolumn_series_file (const std::string &abs_path)
	Cheap content sniff: does this file look multi-column?

bool	openswmm::input::split_series_file_token (std::string_view token, std::string &path_out, std::string &column_out)
	Split a `path[:column]` FILE token into its path and column parts.

long	openswmm::input::multicolumn_parse_count_total () noexcept
	Process-wide count of multi-column file parses (test hook).

Detailed Description

Parse-once cache for multi-column external series files (CSV / TSV / PCSWMM TSF).

Implements the eager parse-once design of plans/MULTICOLUMN_SERIES_SINGLE_READ_2026-08-17.md (§3.1): each external file referenced by any number of rain gages and/or FILE-backed [TIMESERIES] tables is read from disk exactly once per resolve pass; consumers copy their selected column out of the shared ParsedSeriesFile and the cache is freed at end of pass.

Values are stored RAW (file units, no rain-type or unit conversion) — each consumer applies its own transform on copy-out, exactly as it did when it read the file itself.

Format auto-detection:

First content line starting with IDs: → PCSWMM TSF (tab-delimited; line 1 tokens 1..N are the column names, the next two lines — parameter and units rows — are skipped; datetimes are 12-hour MM/DD/YYYY hh:mm:ss AM/PM).
Otherwise CSV/TSV: delimiter sniffed from the header row (tab when tabs dominate, else comma); row 1 = headers, column 0 = time.

Lines are read with std::getline, so there is no fixed line-width limit (replaces the 4096-byte fgets buffers that silently split wide rows — plan gap P2).

See also: src/engine/input/PostParseResolver.cpp — the two consumers

Author: Caleb Buahin caleb.nosp@m..bua.nosp@m.hin@g.nosp@m.mail.nosp@m..com

License\n Apache-2.0

Classes

Namespaces

Enumerations

Functions

Detailed Description