datagen/
schema.rs

1use std::fs::File;
2use serde_yaml::Error;
3use crate::DType;
4
5#[derive(Debug, PartialEq, Serialize, Deserialize, Clone)]
6pub struct Schema {
7    pub name: String,
8    pub dataset: DataSet,
9}
10
11#[derive(Debug, PartialEq, Serialize, Deserialize, Clone)]
12pub struct DataSet {
13    pub name: String,
14    pub columns: Vec<Column>,
15}
16
17#[derive(Debug, PartialEq, Serialize, Deserialize, Clone)]
18pub struct Column {
19    pub name: String,
20    pub not_null: Option<bool>,
21    pub dtype: DType,
22    pub one_of: Option<Vec<String>>,
23    pub min: Option<String>,
24    pub max: Option<String>,
25    pub mean: Option<f64>,
26    pub std: Option<f64>,
27    pub format: Option<String>
28
29}
30
31impl Schema {
32    pub fn from(s: &str) -> Result<Schema, Error> {
33        serde_yaml::from_str(s)
34    }
35
36    pub fn from_path(path: String) -> Result<Schema, Error> {
37        let file = File::open(path).expect("Unable to open the file");
38        serde_yaml::from_reader(file)
39    }
40}
41
42#[cfg(test)]
43mod tests {
44    use crate::schema::Schema;
45
46    #[test]
47    fn derive_struct_from_yaml() {
48        let yaml = r#"name: person_schema
49dataset:
50    name: person_table
51    columns:
52        -
53            name: id
54            not_null: false
55            dtype: int
56        -
57            name: name
58            dtype: string
59        -
60            name: age
61            dtype: int
62        -
63            name: adult
64            default: 'false'
65            dtype: boolean
66        -
67            name: gender
68            dtype: string
69"#;
70
71        let schema = Schema::from(&yaml);
72        pretty_assertions::assert_eq!(format ! ("{:?}", schema.unwrap()), r#"Schema { name: "person_schema", dataset: DataSet { name: "person_table", columns: [Column { name: "id", not_null: Some(false), dtype: Int, one_of: None, min: None, max: None, mean: None, std: None, format: None }, Column { name: "name", not_null: None, dtype: String, one_of: None, min: None, max: None, mean: None, std: None, format: None }, Column { name: "age", not_null: None, dtype: Int, one_of: None, min: None, max: None, mean: None, std: None, format: None }, Column { name: "adult", not_null: None, dtype: Boolean, one_of: None, min: None, max: None, mean: None, std: None, format: None }, Column { name: "gender", not_null: None, dtype: String, one_of: None, min: None, max: None, mean: None, std: None, format: None }] } }"#);
73    }
74
75    #[test]
76    fn derive_struct_from_file() {
77        let file_path = "./test_data/schema_simple.yaml".to_string();
78        let schema = Schema::from_path(file_path);
79        pretty_assertions::assert_eq!(format!("{:?}", schema.unwrap()), r#"Schema { name: "person_schema", dataset: DataSet { name: "person_table", columns: [Column { name: "id", not_null: Some(false), dtype: Int, one_of: None, min: None, max: None, mean: None, std: None, format: None }, Column { name: "name", not_null: None, dtype: Name, one_of: None, min: None, max: None, mean: None, std: None, format: None }, Column { name: "age", not_null: None, dtype: Age, one_of: None, min: None, max: None, mean: None, std: None, format: None }, Column { name: "adult", not_null: None, dtype: Boolean, one_of: None, min: None, max: None, mean: None, std: None, format: None }, Column { name: "gender", not_null: None, dtype: String, one_of: Some(["M", "F"]), min: None, max: None, mean: None, std: None, format: None }, Column { name: "date", not_null: None, dtype: Date, one_of: None, min: Some("01/01/2014"), max: Some("03/01/2014"), mean: None, std: None, format: Some("%d/%m/%Y") }] } }"#);
80    }
81}
datagen/schema.rs

datagen/
schema.rs
