zeekayandHanzo Dev 0b00f44cf2 split: place every chunk in its document ...

Every splitter cuts on byte spans of the document and then threw them
away, so a chunk knew its text and its position in the list but not where
in the document it came from, and a caller citing it had to search for it.
Chunk now has Start and End, the byte range its text was cut from, so
Text is doc.Text[Start:End]; the pipeline's one-chunk default spans the
whole document. A test holds that for every splitter, overlapping or not.

The JSON and CSV exports carry every field of an assertion and read it
back, so they carry the two offsets too; CSV appends them as columns, so a
table written before still reads.

Co-authored-by: Hanzo Dev <[email protected]>

2026-09-23 00:41:28 -07:00

..

embed.go

semantic: the pipeline, end to end

2026-09-08 12:57:59 -07:00

markup.go

semantic: the pipeline, end to end

2026-09-08 12:57:59 -07:00

new.go

semantic: the pipeline, end to end

2026-09-08 12:57:59 -07:00

scan.go

semantic: the pipeline, end to end

2026-09-08 12:57:59 -07:00

size.go

Build on Go 1.27

2026-09-17 06:19:09 -07:00

split_test.go

split: place every chunk in its document

2026-09-23 00:41:28 -07:00

split.go

split: place every chunk in its document

2026-09-23 00:41:28 -07:00

text.go

semantic: the pipeline, end to end

2026-09-08 12:57:59 -07:00