Skip to main content

Output Formats

The global --format (-F) flag picks how commands print rows. It applies to query and the inspection commands; lynxdb import uses its own --format for the input format.

lynxdb query 'source=main level=ERROR | stats count() as errors by host' --format csv

Formats​

FormatOutput
autoThe default. On a terminal: a table, or a bare value for a one-row, one-column result. Otherwise: json
jsonOne JSON object per row, one row per line
ndjsonSame as json
tableTable with box-drawing borders and a row count
boxSame layout as table
asciiTable drawn with plain ASCII characters, for terminals without box drawing
markdownMarkdown table
verticalOne block per row, one field value line per field
line, GSame as vertical
csvRFC 4180 CSV with a header row
tsvTab-separated values with a header row
rawThe _raw text of each row; rows without _raw print as tab-separated field=value pairs

A \G at the end of a query switches auto to vertical, as in a MySQL client:

lynxdb query 'source=main level=ERROR | head 1 \G'

Examples​

The query for every example:

lynxdb query 'source=main level=ERROR | stats count() as errors by host | sort - errors | head 3' --format <format>

json:

{"errors":67,"host":"api-02"}
{"errors":58,"host":"web-02"}
{"errors":57,"host":"web-03"}

table:

┌────────┬────────┐
│ host │ errors │
├────────┼────────┤
│ api-02 │ 67 │
│ web-02 │ 58 │
│ web-03 │ 57 │
└────────┴────────┘
(3 rows)

ascii:

+--------+--------+
| host | errors |
+--------+--------+
| api-02 | 67 |
| web-02 | 58 |
| web-03 | 57 |
+--------+--------+
(3 rows)

markdown:

| host   | errors |
|--------|--------|
| api-02 | 67 |
| web-02 | 58 |
| web-03 | 57 |
(3 rows)

csv:

host,errors
api-02,67
web-02,58
web-03,57

vertical, for wide event rows:

lynxdb query 'source=main level=ERROR | fields host, path, status, response_time | head 1' --format vertical
  record 1
host api-02
path /static/app.js
response_time 11
status 200

raw prints the original lines, ready for another tool or for a local query:

lynxdb query 'source=main status>=500 | head 2' --format raw
2025-01-15T08:00:45.63Z host=api-02 level=ERROR method=DELETE path=/api/v1/orders status=500 user=bob response_time=42 bytes=10811 msg="DELETE /api/v1/orders completed"
2025-01-15T08:01:10.09Z host=web-01 level=ERROR method=POST path=/api/v1/users status=503 user=alice response_time=344 bytes=45705 msg="POST /api/v1/users completed"

The table, CSV and vertical formats order columns the same way, not in the order of the query's fields or stats clause: the fields _time, _raw, index, source, sourcetype and host first, then the other fields alphabetically. JSON objects list their keys alphabetically.

Human table flags​

These global flags shape table, box, ascii, markdown, vertical and terminal auto output:

FlagDescription
--max-rows <n>Show at most n rows
--max-width <n>Fit tables into n columns (0 = terminal width)
--null-value <text>Print this for null or empty values
--compactLess padding
--theme <name>auto, dark, light or plain
--no-colorNo color; NO_COLOR with any value does the same

Data on stdout, everything else on stderr​

Rows go to stdout. Statistics, hints, warnings and errors go to stderr, so a redirect or a pipe receives only data:

lynxdb query 'source=main | stats count() by host' --format csv > hosts.csv
lynxdb query 'source=main level=ERROR | fields host, path' --since 1h | jq -r .path | sort | uniq -c

When stdout is not a terminal, a query also writes its statistics to stderr as one JSON line, {"stats":{...}}, with the total and execution times in milliseconds (total_ms, exec_ms) and the rows returned (result_rows). --no-stats drops it and --quiet drops everything except data.

Commands other than query​

status, health, indexes, jobs, fields, auth list, config get and config list-profiles print a styled view on a terminal and plain rows otherwise. --format json gives each command's JSON document or one JSON object per item; csv, tsv, markdown and the table formats print the same data as rows.

See also​