Double space to JSON array converter

This double space to JSON array converter applies a single delimiter: a run of two or more consecutive space characters. Everything else about the behaviour follows from that, and all of it is worth stating precisely, because the value of a delimiter tool is entirely in knowing what it will and will not cut.

The choice of two-or-more rather than exactly-two is what makes the route useful on column-aligned text. Alignment produced by padding is uneven by nature: one column is separated by two spaces, the next by nine. A rule keyed to an exact count would split inconsistently across the same line. A rule keyed to a run splits every gap and no single space.

The conversion is computed in the browser. No text is transmitted, there is no rate limit, no character ceiling and no account.

Five properties define the result. Each is observable and each is a deliberate limit rather than an oversight.

Split

Every run of two or more space characters becomes a boundary. A single space never does.

Trim

Each resulting segment has leading and trailing whitespace removed before it is emitted.

Discard empties

Segments that trim to an empty string are removed, so a deliberate empty entry is not representable.

String typing only

Every value is emitted as a quoted string. Numeric and boolean inference is not performed.

Standard serialisation

The array is serialised with proper escaping and printed with a two-space indent, one element per line.

There are no options to set, so the procedure is fixed.

Step 1

Paste the column-aligned source into the input.

Step 2

Convert. The split, trim and serialisation run locally and return immediately.

Step 3

Verify the element count against the expected column count before using the output.

Consider a row reading apples, a gap of two spaces, Los Angeles, a gap of nine spaces, then 40. Three columns, two gaps of different widths, and one internal single space.

The rule fires on both gaps because both are runs of two or more, and it does not fire on the space inside Los Angeles. Trimming removes the padding that was swept into each segment. Nothing trims to empty, so nothing is discarded. Serialisation quotes all three values, including 40, and prints them one per line under a two-space indent. Three elements out, in source order, with the phrase intact.

Visual alignment says nothing about the codepoints producing it. Tab characters, no-break spaces and narrow no-break spaces all render as horizontal gaps and none of them satisfies a rule written against the ordinary space. A tabbed table therefore converts to exactly one element, which is the correct behaviour for the rule and a surprise for the user.

Diagnosis is straightforward: if the element count is one, the gap is not made of spaces. Terminal output and spreadsheet exports skew heavily towards tabs. Text extracted from a PDF skews heavily towards padded spaces, which is why PDF extraction is the strongest use case for this particular route.

The delimiter for this route is defined over space characters only. A newline is therefore not a boundary: it survives inside whichever segment contains it, and appears in the serialised output as an escape sequence rather than as a break between elements.

The practical consequence for a multi-row table is that row structure is lost while the newlines themselves are preserved inside values. Convert row by row if the rows are meaningful. If the data is one item per line to begin with, a line-break split is the correct operation and this is not the route for it.

Competing text-to-JSON pages bundle several transformations behind one interface. This route implements one, and the omissions are worth naming.

No object output

Key and value parsing into a JSON object is not implemented. The result is always a flat array.

No nesting

There is no grouping into sub-arrays and no row-to-record mapping.

No header row handling

CSV-style conversion with the first row as keys is a different operation and is not performed here.

No deduplication

Repeated values are all retained. Removing duplicates is a separate transformation.

No file handling

Input arrives by paste and output leaves by copy. There is no upload and no download.

No configurable delimiter

This slug's delimiter is fixed in code. Only the custom delimiter route renders a delimiter field.

The recurring situation is data that was formatted for reading and now has to be parsed.

PDF and document extraction

Extracted text pads columns with spaces far more often than it uses tabs, which matches this rule directly.

Terminal captures

Padded command output, headed for a fixture or a configuration file.

Specification to enumeration

A column of permitted values in a written spec, becoming a union type or a validation list.

Ad hoc shaping

A short list already on screen that needs to be valid JSON without opening an editor.