tublub

Convert or view tabular data files — CSV, JSON, XLSX, YAML, and more — from the command line.

tublub is a thin wrapper around Tablib. Tablib reads and writes plenty of formats, but it won't open the files for you — you have to know which ones need binary mode, special newline handling, and so on. tublub bakes that per-format knowledge in, so you just point it at a file and it works.

$ tublub --list-formats
Available formats: json xlsx xls yaml csv tsv ods dbf html jira latex rst sql cli

$ tublub input.json
Username |Identifier|First name|Last name
---------|----------|----------|---------
booker12 |9012      |Rachel    |Booker
grey07   |2070      |Laura     |Grey
jenkins46|9346      |Mary      |Jenkins
johnson81|4081      |Craig     |Johnson
smith79  |5079      |Jamie     |Smith

$ tublub input.json output.xlsx

$ file output.xlsx
output.xlsx: Microsoft Excel 2007+

$ tublub -t csv input.json
Username,Identifier,First name,Last name
booker12,9012,Rachel,Booker
grey07,2070,Laura,Grey
jenkins46,9346,Mary,Jenkins
johnson81,4081,Craig,Johnson
smith79,5079,Jamie,Smith

$ tublub -o book.xlsx sales.csv users.json regions.tsv
Saved 'book.xlsx', 3 sheets (xlsx)

With -o, all positional arguments are inputs. Two or more inputs become sheets in a single multi-sheet file (XLSX, ODS, JSON, YAML, ...) — every sheet of every input, so multi-sheet inputs are expanded rather than truncated. Sheets keep their own names, and a whole-file input is named after its file stem; names are only changed when two would collide.

Multi-sheet input

Three rules cover the whole surface:

1. -s/--sheet picks sheets; selection never changes where output may go. One selected sheet behaves exactly like a single-sheet input does; several behave like a multi-sheet workbook — printed with headings, saved with -o, exported with -t — wherever the output format can hold multiple sheets. --all-sheets is sugar for "select everything".
2. You can select exactly what -l/--list-sheets shows. Indexed lines ([0] people ...) are selectable by index or title; a bare line means the input has no sheet structure and nothing to select.
3. Commas are for index lists; repeat -s for names; name: forces a title. So -s 0,2 takes two indices, -s people -s cities takes two titles, and -s name:2024 picks the sheet titled 2024 rather than index 2024.

-l/--list-sheets inspects an input without converting it:

$ tublub -l book.xlsx
[0] people  2 rows x 2 cols
[1] cities  2 rows x 2 cols

$ tublub -l people.csv
2 rows x 2 cols

An input without sheet structure (CSV, TSV, records-shaped JSON, ...) prints one bare N rows x M cols line — there is no index or title to select.

Pick one sheet and it prints like any single-sheet file; pick several and each gets a heading:

$ tublub -s cities book.xlsx
city          pop
Stockholm  975904
Goteborg   583056

$ tublub -s 0,1 book.xlsx
=== people (2 rows) ===
name      age
Alice      30
Bob        41

=== cities (2 rows) ===
city          pop
Stockholm  975904
Goteborg   583056

Converting without any selection converts every sheet, as long as the target format can hold them:

$ tublub book.xlsx out.ods
Saved 'out.ods', 2 sheets (ods)

Single-sheet formats (CSV, TSV, DBF, ...) get the first sheet and always say so on stderr, even when stderr is redirected — tublub never splits one input across several output files:

$ tublub -t csv book.xlsx
book.xlsx: format 'csv' cannot hold all 2 sheets; converting only the first (use -s to choose)
name,age
Alice,30
Bob,41

Asking for every sheet explicitly stays strict: --all-sheets -o out.csv errors instead of quietly dropping data.

All of this works on piped input too, either as an explicit - or implicitly whenever stdin is not a terminal:

$ cat book.xlsx | tublub -l -
[0] people  2 rows x 2 cols
[1] cities  2 rows x 2 cols

News and Changes

Please see the changelog for more details.