CSV options

phibble, phit, and phpolars each read a CSV file, and the three name the reader's options three ways. phibble's DataFrame::fromCsv() follows readr's read_delim() and takes readr's option names. Phit\read_csv() is the native reader that phibble calls when the extension is loaded, and it takes the same names in readr's own spelling. Phpolars\read_csv() wraps Polars' reader and takes py-polars' names.

A reader who learns fromCsv(nMax: 5) in phibble needs read_csv(nRows: 5) in phpolars. The table below covers every option of both readers.

Why the names differ

Convention 3 of the family's API conventions keeps free functions snake_case and members camelCase. It then exempts a package that follows another library's API closely, so that package keeps that library's names. phpolars holds one of those exemptions because it wraps Polars. Renaming its options toward readr would put phpolars' parameter names at odds with Polars' own documentation, which is what the exemption prevents. The conventions are in the family repository, at docs/api-conventions.md.

phibble and phit spell the readr options differently

phibble's reader takes PHP parameters, so its names are camelCase: colNames, nMax, and trimWs. Phit\read_csv() takes an array of options, and its keys are readr's snake_case names: col_names, n_max, and trim_ws. Phibble\Io\NativeCsv translates between the two before it calls the extension. Each spelling names the same readr option, so the table gives each its own column.

Phit\read_csv() accepts fifteen of these options. phibble handles column selection, name repair, factor columns, and the readr type letters that the extension has no type code for, applying them in PHP to the result the extension returns.

The options

What it controls	phibble	phit	phpolars
The field separator	delimiter	delimiter	separator
Whether the first row holds the column names	colNames as a boolean	col_names	hasHeader
Column names supplied rather than read	colNames as a list	absent	newColumns
The strings that read as missing	na	na	nullValues
The type of every column	types	types	schema
The type of some columns, the rest guessed	types as a map	types	schemaOverrides
The columns to keep, in order	select	absent	columns
Rows skipped before the header	skip	skip	skipRows
Rows skipped after the header	absent	absent	skipRowsAfterHeader
The most rows to read	nMax	n_max	nRows
Rows read to guess the types	guessMax	guess_max	inferSchemaLength
The quote character	quote	quote	quoteChar
A doubled quote stands for one quote	escapeDouble	escape_double	absent
A backslash escapes the next character	escapeBackslash	escape_backslash	absent
The prefix that marks a comment line	comment	comment	commentPrefix
Whitespace trimmed from each field	trimWs	trim_ws	absent
Empty rows dropped	skipEmptyRows	skip_empty_rows	absent
The decimal mark	decimalMark	decimal_mark	decimalComma
The digit grouping mark	groupingMark	grouping_mark	absent
How duplicate and empty column names are repaired	nameRepair	absent	absent
The line ending	absent	absent	eolChar
Dates and datetimes parsed while reading	absent	absent	tryParseDates
The input encoding	absent	absent	encoding
Parse errors tolerated rather than raised	absent	absent	ignoreErrors
A missing text field read as an empty string	absent	absent	missingUtf8IsEmptyString
Rows with too many fields truncated	absent	absent	truncateRaggedLines
A row number column, by name and first value	absent	absent	rowIndexName, rowIndexOffset
An empty file raises rather than returning nothing	absent	absent	raiseIfEmpty
The source path expanded as a glob	absent	absent	glob

An option marked absent has no counterpart in that reader.

Where a match is partial

• types covers both of the Polars options next to it. It takes readr's compact letter string, a list of specifications by position, or a map by column name with .default for the rest. phpolars splits the two cases: schema names a Polars data type for every column, and schemaOverrides names one for some columns and leaves the rest to type inference.
• colNames takes true, false, or a list of names. Phit\read_csv() reads col_names as a boolean only, so phibble supplies a list itself. phpolars splits the two cases into hasHeader and newColumns.
• decimalMark and decimal_mark take the character. decimalComma is a boolean that selects a comma, and Polars has no option for the grouping mark.
• columns is a parameter of Phpolars\read_csv() rather than one of Polars' reader options, and it selects from the data frame the reader returns. phibble's select keeps the named columns in the order given.
• phibble guesses dates and datetimes as part of guessing a column's type, so it has no option to turn that on. It collects parse failures and reports them as problems rather than offering a flag to ignore them.
• encoding has no counterpart because neither readr-named reader inspects the encoding. Both pass bytes through unchanged and validate nothing, so the bytes of a file in any encoding reach the column values as written. Both strip a UTF-8 byte-order mark from the start of the file and leave every other byte alone.