use crate::load_naming::main::loaded_name::loaded_name;
use crate::row_batches::_helpers::flatten::{keys_are_data, leaf_keys};
use crate::row_batches::constants::{FLATTEN_SEPARATOR, SHOWN_OBJECT_KEYS};
use crate::row_batches::models::{Flatten, FlattenObjects};
pub fn flatten_lines(
flatten: &Flatten,
sample: Option<(&[String], &[String])>,
kept: &[String],
) -> Vec<String> {
let setting = match &flatten.objects {
FlattenObjects::None => return Vec::new(),
FlattenObjects::All => format!("every object, {} levels down at most", flatten.max_depth),
FlattenObjects::Paths(paths) => paths.join(", "),
};
let Some((objects, columns)) = sample else {
return vec![format!("{setting}; objects flatten as rows arrive at load")];
};
let mut lines: Vec<String> = vec![setting];
let shown: Vec<(String, Vec<String>)> = objects
.iter()
.map(|object| {
(
object.replace(FLATTEN_SEPARATOR, "."),
leaf_keys(object, (objects, columns)),
)
})
.collect();
let width = shown
.iter()
.map(|(path, _)| path.chars().count())
.max()
.unwrap_or_default();
for (path, keys) in &shown {
let first: Vec<String> = keys
.iter()
.take(SHOWN_OBJECT_KEYS)
.map(|key| loaded_name(flatten.naming, key, None))
.collect();
let more = match keys.len() > SHOWN_OBJECT_KEYS {
true => ", …",
false => "",
};
let count = match keys.len() {
1 => "1 column".to_owned(),
count => format!("{count} columns"),
};
lines.push(format!(
"{path:<width$} → {count} ({}{more})",
first.join(", ")
));
}
if !kept.is_empty() {
lines.push(format!("kept as JSON: {}", kept.join(", ")));
}
let widest = shown
.iter()
.filter(|(_, keys)| keys_are_data(keys))
.max_by_key(|(_, keys)| keys.len());
if let (FlattenObjects::All, Some((path, _))) = (&flatten.objects, widest) {
let name = path.replace('.', FLATTEN_SEPARATOR);
let name = loaded_name(flatten.naming, &name, None);
lines.push(
"⚠ an object whose keys are data adds a column per key on every load;".to_owned(),
);
lines.push(format!(
" keep it as one column with columns = {{ {name} = \"json\" }}"
));
}
lines
}