opendp 0.4.0

A library of differential privacy algorithms for the statistical analysis of sensitive private data.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213

use std::cmp::Ordering;
use std::collections::HashMap;
use std::fmt::Debug;
use std::hash::Hash;
use std::iter::repeat;

use crate::core::{Function, StabilityRelation, Transformation};
use crate::data::Column;
use crate::dom::{AllDomain, MapDomain, VectorDomain};
use crate::error::*;
use crate::dist::SymmetricDistance;
use crate::traits::CheckNull;

pub type DataFrame<K> = HashMap<K, Column>;
pub type DataFrameDomain<K> = MapDomain<AllDomain<K>, AllDomain<Column>>;

/// ensure all rows have `len` number of cells
fn conform_records<'a>(len: usize, records: &[Vec<&'a str>]) -> Vec<Vec<&'a str>> {
    records.iter().map(|record| match record.len().cmp(&len) {
        Ordering::Less => // record is too short; pad with empty strings
            record.clone().into_iter().chain(repeat("").take(len - record.len())).collect(),
        Ordering::Equal => // record is just right
            record.clone(),
        Ordering::Greater => // record is too long; slice down
            record[0..len].to_vec()
    }).collect()
}

fn create_dataframe<K: Eq + Hash>(col_names: Vec<K>, records: &[Vec<&str>]) -> DataFrame<K> {
    // make data rectangular
    let records = conform_records(col_names.len(), &records);

    // transpose and collect into dataframe
    col_names.into_iter().enumerate()
        .map(|(i, col_name)| (
            col_name,
            Column::new(records.iter().map(|record| record[i].to_string()).collect())))
        .collect()
}

fn create_dataframe_domain<K: Eq + Hash + CheckNull>() -> DataFrameDomain<K> {
    MapDomain::new(AllDomain::new(), AllDomain::new())
}

pub fn make_create_dataframe<K>(
    col_names: Vec<K>
) -> Fallible<Transformation<VectorDomain<VectorDomain<AllDomain<String>>>, DataFrameDomain<K>, SymmetricDistance, SymmetricDistance>>
    where K: 'static + Eq + Hash + Clone + CheckNull {
    Ok(Transformation::new(
        VectorDomain::new(VectorDomain::new_all()),
        create_dataframe_domain(),
        Function::new(move |arg: &Vec<Vec<String>>| -> DataFrame<K> {
            let arg: Vec<_>  = arg.iter().map(|e| vec_string_to_str(e)).collect();
            create_dataframe(col_names.clone(), &arg)
        }),
        SymmetricDistance::default(),
        SymmetricDistance::default(),
        StabilityRelation::new_from_constant(1)
    ))
}

fn split_dataframe<K: Hash + Eq>(separator: &str, col_names: Vec<K>, s: &str) -> DataFrame<K> {
    let lines = split_lines(s);
    let records = split_records(separator, &lines);
    let records = conform_records(col_names.len(), &records);
    create_dataframe(col_names, &records)
}

pub fn make_split_dataframe<K>(
    separator: Option<&str>, col_names: Vec<K>
) -> Fallible<Transformation<AllDomain<String>, DataFrameDomain<K>, SymmetricDistance, SymmetricDistance>>
    where K: 'static + Hash + Eq + Clone + CheckNull {
    let separator = separator.unwrap_or(",").to_owned();
    Ok(Transformation::new(
        AllDomain::new(),
        create_dataframe_domain(),
        Function::new(move |arg: &String| split_dataframe(&separator, col_names.clone(), &arg)),
        SymmetricDistance::default(),
        SymmetricDistance::default(),
        StabilityRelation::new_from_constant(1)))
}

pub fn make_select_column<K, TOA>(key: K) -> Fallible<Transformation<DataFrameDomain<K>, VectorDomain<AllDomain<TOA>>, SymmetricDistance, SymmetricDistance>>
    where K: 'static + Eq + Hash + Debug + CheckNull,
          TOA: 'static + Debug + Clone + PartialEq + CheckNull {
    Ok(Transformation::new(
        create_dataframe_domain(),
        VectorDomain::new_all(),
        Function::new_fallible(move |arg: &DataFrame<K>| -> Fallible<Vec<TOA>> {
            // retrieve column from dataframe and handle error
            arg.get(&key).ok_or_else(|| err!(FailedFunction, "column does not exist: {:?}", key))?
                // cast down to &Vec<T>
                .as_form::<Vec<TOA>>().map(|c| c.clone())
        }),
        SymmetricDistance::default(),
        SymmetricDistance::default(),
        StabilityRelation::new_from_constant(1)))
}

fn vec_string_to_str(src: &[String]) -> Vec<&str> {
    src.iter().map(|e| e.as_str()).collect()
}

fn vec_str_to_string(src: Vec<&str>) -> Vec<String> {
    src.into_iter().map(|e| e.to_owned()).collect()
}

fn split_lines(s: &str) -> Vec<&str> {
    s.lines().collect()
}

/// A [`Transformation`] that takes a `String` and splits it into a `Vec<String>` of its lines.
pub fn make_split_lines() -> Fallible<Transformation<AllDomain<String>, VectorDomain<AllDomain<String>>, SymmetricDistance, SymmetricDistance>> {
    Ok(Transformation::new(
        AllDomain::<String>::new(),
        VectorDomain::new_all(),
        Function::new(|arg: &String| -> Vec<String> {
            arg.lines().map(|v| v.to_owned()).collect()
        }),
        SymmetricDistance::default(),
        SymmetricDistance::default(),
        StabilityRelation::new_from_constant(1)))
}

fn split_records<'a>(separator: &str, lines: &[&'a str]) -> Vec<Vec<&'a str>> {
    fn split<'a>(line: &'a str, separator: &str) -> Vec<&'a str> {
        line.split(separator).into_iter().map(|e| e.trim()).collect()
    }
    lines.iter().map(|e| split(e, separator)).collect()
}

pub fn make_split_records(separator: Option<&str>) -> Fallible<Transformation<VectorDomain<AllDomain<String>>, VectorDomain<VectorDomain<AllDomain<String>>>, SymmetricDistance, SymmetricDistance>> {
    let separator = separator.unwrap_or(",").to_owned();
    Ok(Transformation::new(
        VectorDomain::new_all(),
        VectorDomain::new(VectorDomain::new_all()),
        // move is necessary because it captures `separator`
        Function::new(move |arg: &Vec<String>| -> Vec<Vec<String>> {
            let arg = vec_string_to_str(arg);
            let ret = split_records(&separator, &arg);
            ret.into_iter().map(vec_str_to_string).collect()
        }),
        SymmetricDistance::default(),
        SymmetricDistance::default(),
        StabilityRelation::new_from_constant(1)))
}


#[cfg(test)]
mod tests {
    use crate::error::ExplainUnwrap;
    use super::*;

    #[test]
    fn test_make_split_lines() {
        let transformation = make_split_lines().unwrap_test();
        let arg = "ant\nbat\ncat\n".to_owned();
        let ret = transformation.invoke(&arg).unwrap_test();
        assert_eq!(ret, vec!["ant".to_owned(), "bat".to_owned(), "cat".to_owned()]);
    }

    #[test]
    fn test_make_split_records() {
        let transformation = make_split_records(None).unwrap_test();
        let arg = vec!["ant, foo".to_owned(), "bat, bar".to_owned(), "cat, baz".to_owned()];
        let ret = transformation.invoke(&arg).unwrap_test();
        assert_eq!(ret, vec![
            vec!["ant".to_owned(), "foo".to_owned()],
            vec!["bat".to_owned(), "bar".to_owned()],
            vec!["cat".to_owned(), "baz".to_owned()],
        ]);
    }

    #[test]
    fn test_make_create_dataframe() {
        let transformation = make_create_dataframe::<u32>(vec![0, 1]).unwrap_test();
        let arg = vec![
            vec!["ant".to_owned(), "foo".to_owned()],
            vec!["bat".to_owned(), "bar".to_owned()],
            vec!["cat".to_owned(), "baz".to_owned()],
        ];
        let ret = transformation.invoke(&arg).unwrap_test();
        let expected: DataFrame<u32> = vec![
            (0, Column::new(vec!["ant".to_owned(), "bat".to_owned(), "cat".to_owned()])),
            (1, Column::new(vec!["foo".to_owned(), "bar".to_owned(), "baz".to_owned()])),
        ].into_iter().collect();
        assert_eq!(ret, expected);
    }

    #[test]
    fn test_make_split_dataframe() {
        let transformation = make_split_dataframe::<String>(None, vec!["0".to_string(), "1".to_string()]).unwrap_test();
        let arg = "ant, foo\nbat, bar\ncat, baz".to_owned();
        let ret = transformation.invoke(&arg).unwrap_test();
        let expected: DataFrame<String> = vec![
            ("0".to_owned(), Column::new(vec!["ant".to_owned(), "bat".to_owned(), "cat".to_owned()])),
            ("1".to_owned(), Column::new(vec!["foo".to_owned(), "bar".to_owned(), "baz".to_owned()])),
        ].into_iter().collect();
        assert_eq!(ret, expected);
    }

    #[test]
    fn test_make_select_column() {
        let transformation = make_select_column::<String, String>("1".to_owned()).unwrap_test();
        let arg: DataFrame<String> = vec![
            ("0".to_owned(), Column::new(vec!["ant".to_owned(), "bat".to_owned(), "cat".to_owned()])),
            ("1".to_owned(), Column::new(vec!["foo".to_owned(), "bar".to_owned(), "baz".to_owned()])),
        ].into_iter().collect();
        let ret = transformation.invoke(&arg).unwrap_test();
        let expected = vec!["foo".to_owned(), "bar".to_owned(), "baz".to_owned()];
        assert_eq!(ret, expected);
    }
}