[FEA] Support `num_rows_per_source` in table metadata after a `cudf::io::read_json`
Open
cuIO
feature request
libcudf
Python
- Dominant language
- C++
- Stars
- 9.8k
- Forks
- 1.1k
- Avg merge
- 3d 6m
- Merged PRs (30d)
- 278
Description
**Is your feature request related to a problem? Please describe.**
After doing a `read_json` call, I'd like to be able to get the number of rows in each source in the output tables metadata. Eg.
```python
import pylibcudf as plc
df = plc.io.json.read_json(
plc.io.json.JsonReaderOptions.builder(
plc.io.types.SourceInfo(["df.json"])
).lines(True).build()
)
print(df.num_rows_per_source)
```
```
[]
```
Right now the list is empty.
**Describe the solution you'd like**
Populate this field in the libcudf implementation.
**Additional context**
xref #18057 (polars feature that needs this)
Contributor guide
Assessment
This issue has not been assessed yet.