feat(workflow-operator): export the source operators as Python - #8341

Open

kz930 wants to merge 36 commits into

apache:mainfrom

kz930:feat/standalone-sources

Commits on Sep 1, 2026

feat(workflow-compiling-service): export a workflow as a standalone Python script

kz930
and
claude
committed
5b8dd57
test(workflow-compiling-service): run an operator both ways and compare the files

kz930
and
claude
committed
a7558eb
test(workflow-compiling-service): verify a generated script against the operator it came from

kz930
and
claude
committed
cba725f

Commits on Sep 2, 2026

Merge remote-tracking branch 'upstream/main' into feat/standalone-sources
kz930
committed
97e05ec
feat(workflow-operator): export the source operators as Python

kz930
and
claude
committed
41724ff
ci: give the verify spec a job with an interpreter, and keep it out of the one without

kz930
and
claude
committed
bc8fc17
Merge remote-tracking branch 'myfork/feat/standalone-verify-harness' into feat/standalone-sources
kz930
committed
9be0dbd
test(verify): pin the no-generator case to an operator that can never have one

kz930
and
claude
committed
883ea72
chore: leave the harness to the change that introduces it

kz930
and
claude
committed
5a44c3c
Merge upstream/main
kz930
committed
3f07020
chore: leave the two verify files to #8327 as well

kz930
and
claude
committed
df10bd8

Commits on Sep 9, 2026

fix(operator): slice the raw lines before converting them

kz930
and
claude
committed
73375b9

Commits on Sep 10, 2026

fix(operator): read a CSV the way the parser does, not the way pandas does

kz930
and
claude
committed
cb23c1e
fix(operator): take the row's first string field as the file to scan

kz930
and
claude
committed
c2e2eb3
fix(operator): decode a fetched body by replacement, as the executor does

kz930
and
claude
committed
1c0405f
fix(operator): slice a JSONL file's lines before parsing them

kz930
and
claude
committed
208816d
fix(operator): take a CSV's column names from the schema

kz930
and
claude
committed
566aa15
test(operator): assert the column names the CSV reader now writes

kz930
and
claude
committed
3d12fae

Commits on Sep 11, 2026

Merge remote-tracking branch 'upstream/main' into HEAD
kz930
committed
14b5571

Commits on Sep 15, 2026

Merge branch 'main' into feat/standalone-sources
kz930
authored
832b914

Commits on Sep 18, 2026

fix(operator): read an Arrow file into the dtypes that keep its nulls

kz930
and
claude
committed
edf46a4
test(operator): pin that an Arrow read keeps a missing value apart from a NaN

kz930
and
claude
committed
fe5e8c0
fix(workflow-operator): follow the native readers in the exported sources

kz930
and
claude
committed
32173ef
Merge remote-tracking branch 'myfork/feat/standalone-sources' into feat/standalone-sources
kz930
committed
292ddbe
feat(workflow-operator): clamp the exported scan window at zero

kz930
and
claude
committed
d938d35
fix(workflow-operator): give a flattened array the executor's column names

kz930
and
claude
committed
a2290cd

Commits on Sep 19, 2026

fix(operator): order the exported JSONL read the way the schema does

kz930
and
claude
committed
2ef4c15
fix(operator): read an Arrow file's narrow numeric widths on both paths

kz930
and
claude
committed
46e62a8
fix(operator): read an Arrow file's unsigned integers as the numbers they count to

kz930
and
claude
committed
b310855
fix(workflow-operator): count the end of an exported scan window in Long

kz930
and
claude
committed
3700e19
fix(operator): decode a file scan with the charset its Encoding field names

kz930
and
claude
committed
d8a9d13
fix(operator): name the file each line came from when a file scan was asked to

kz930
and
claude
committed
95ab40c
fix(operator): type a flattened JSONL timestamp as the schema declares it

kz930
and
claude
committed
f3d0db2
fix(operator): keep a file's columns when its scan window asks for no rows

kz930
and
claude
committed
3d1a4be

Commits on Sep 21, 2026

fix(operator): read every column an Arrow file states, index or not

kz930
and
claude
committed
ce52ee2
fix(operator): read an Arrow file's columns under the names it states

kz930
and
claude
committed
ce2e484