ctddump 0.24.2

Convert oceanographic CTD (Conductivity, Temperature, Depth) data from NetCDF to Parquet or YAML
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
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250

//! Parquet report: aggregate a ctddump Parquet file at the global, platform, or
//! profile level and emit a summary table.

use std::collections::HashSet;
use std::error::Error;
use std::path::Path;

use polars::prelude::*;

use crate::cli::{ReportFormat, ReportLevel};
use crate::report::format;

/// Summarise `src` at `level` and write the report in `format` to `dest` (or stdout).
pub fn run(
    level: ReportLevel,
    format: ReportFormat,
    src: &Path,
    dest: Option<&Path>,
) -> Result<(), Box<dyn Error>> {
    let lf = LazyFrame::scan_parquet(src, ScanArgsParquet::default())
        .map_err(|e| format!("Cannot scan {}: {}", src.display(), e))?;

    // When a `markdup` `is_dup` column is present, the report also summarises
    // duplicate counts; otherwise the duplicate columns are omitted entirely.
    let has_dup = lf.clone().slice(0, 0).collect()?.schema().index_of("is_dup").is_some();

    let df = match level {
        ReportLevel::Global => global_report(lf, has_dup)?,
        ReportLevel::Platform => platform_report(lf, has_dup)?,
        ReportLevel::Profile => profile_report(lf, has_dup)?,
    };

    format::write_report(&df, format, dest)?;
    Ok(())
}

/// Observation-level aggregates for `temp`/`psal`/`pres`: NA count (null or NaN)
/// plus min/max/mean over the valid (non-NaN) values. Reused by every level.
fn measure_aggs() -> Vec<Expr> {
    let mut v = Vec::new();
    for name in ["temp", "psal", "pres"] {
        v.push(
            col(name)
                .is_null()
                .or(col(name).is_nan())
                .sum()
                .alias(format!("na_{name}").as_str()),
        );
        let valid = col(name).filter(col(name).is_not_nan());
        v.push(valid.clone().min().alias(format!("{name}_min").as_str()));
        v.push(valid.clone().max().alias(format!("{name}_max").as_str()));
        v.push(valid.mean().alias(format!("{name}_mean").as_str()));
    }
    v
}

/// Columns in report order, after the level-specific identifier columns.
fn measure_cols() -> Vec<Expr> {
    let mut v = Vec::new();
    for name in ["temp", "psal", "pres"] {
        v.push(col(format!("na_{name}").as_str()));
        v.push(col(format!("{name}_min").as_str()));
        v.push(col(format!("{name}_max").as_str()));
        v.push(col(format!("{name}_mean").as_str()));
    }
    v
}

/// Geographic bounding-box aggregates: min/max of `longitude` and `latitude`
/// over the valid (non-NaN) positions. Used by the global and platform levels.
fn geo_aggs() -> Vec<Expr> {
    let mut v = Vec::new();
    for name in ["longitude", "latitude"] {
        let valid = col(name).filter(col(name).is_not_nan());
        v.push(valid.clone().min().alias(format!("{name}_min").as_str()));
        v.push(valid.max().alias(format!("{name}_max").as_str()));
    }
    v
}

fn geo_cols() -> Vec<Expr> {
    let mut v = Vec::new();
    for name in ["longitude", "latitude"] {
        v.push(col(format!("{name}_min").as_str()));
        v.push(col(format!("{name}_max").as_str()));
    }
    v
}

/// Per-profile roll-up of the profile-level QC flags into per-platform "good"
/// counts (flag `== "1"`), plus the profile count. With `has_dup`, also counts
/// duplicated profiles per platform (`is_dup` is profile-level).
fn platform_qc(lf: LazyFrame, has_dup: bool) -> LazyFrame {
    let mut prof_agg = vec![col("time_qc").first(), col("position_qc").first()];
    let mut plat_agg = vec![
        len().alias("n_profiles"),
        col("time_qc").eq(lit("1")).sum().alias("time_qc_good"),
        col("position_qc").eq(lit("1")).sum().alias("position_qc_good"),
    ];
    if has_dup {
        prof_agg.push(col("is_dup").first());
        plat_agg.push(col("is_dup").sum().alias("dup_profiles"));
    }
    lf.group_by([col("platform_code"), col("profile_no")])
        .agg(prof_agg)
        .group_by([col("platform_code")])
        .agg(plat_agg)
}

fn platform_report(lf: LazyFrame, has_dup: bool) -> Result<DataFrame, Box<dyn Error>> {
    let obs = lf.clone().group_by([col("platform_code")]).agg({
        let mut v = vec![len().alias("n_obs")];
        v.extend(measure_aggs());
        v.extend(geo_aggs());
        v
    });
    let qc = platform_qc(lf, has_dup);

    let mut select_cols = vec![
        col("platform_code"),
        col("n_profiles"),
        col("n_obs"),
        col("time_qc_good"),
        col("position_qc_good"),
    ];
    if has_dup {
        select_cols.push(col("dup_profiles"));
    }
    select_cols.extend(geo_cols());
    select_cols.extend(measure_cols());

    let df = obs
        .join(
            qc,
            [col("platform_code")],
            [col("platform_code")],
            JoinArgs::new(JoinType::Left),
        )
        .sort(vec!["platform_code"], SortMultipleOptions::default())
        .select(select_cols)
        .collect()?;
    Ok(df)
}

fn profile_report(lf: LazyFrame, has_dup: bool) -> Result<DataFrame, Box<dyn Error>> {
    let agg = {
        let mut v = vec![
            col("profile_timestamp").first().alias("profile_timestamp"),
            col("longitude").first().alias("longitude"),
            col("latitude").first().alias("latitude"),
            len().alias("n_obs"),
            col("time_qc").first().alias("time_qc"),
            col("position_qc").first().alias("position_qc"),
        ];
        if has_dup {
            v.push(col("is_dup").first().alias("is_dup"));
        }
        v.extend(measure_aggs());
        v
    };

    let mut select_cols = vec![
        col("platform_code"),
        col("profile_no"),
        col("profile_timestamp"),
        col("longitude"),
        col("latitude"),
        col("n_obs"),
        col("time_qc"),
        col("position_qc"),
    ];
    if has_dup {
        select_cols.push(col("is_dup"));
    }
    select_cols.extend(measure_cols());

    let df = lf
        .group_by([col("platform_code"), col("profile_no")])
        .agg(agg)
        .sort(
            vec!["platform_code", "profile_no"],
            SortMultipleOptions::default(),
        )
        .select(select_cols)
        .collect()?;
    Ok(df)
}

fn global_report(lf: LazyFrame, has_dup: bool) -> Result<DataFrame, Box<dyn Error>> {
    // Whole-file observation aggregates (one row).
    let base = lf
        .clone()
        .select({
            let mut v = vec![len().alias("n_obs")];
            v.extend(measure_aggs());
            v.extend(geo_aggs());
            v
        })
        .collect()?;

    // Profile-level reduction for platform/profile counts and QC "good" counts.
    let profiles = lf
        .group_by([col("platform_code"), col("profile_no")])
        .agg({
            let mut v = vec![col("time_qc").first(), col("position_qc").first()];
            if has_dup {
                v.push(col("is_dup").first());
            }
            v
        })
        .collect()?;

    let n_profiles = profiles.height() as u32;
    let time_qc_good = count_eq(profiles.column("time_qc")?.str()?, "1");
    let position_qc_good = count_eq(profiles.column("position_qc")?.str()?, "1");

    let mut platforms = HashSet::new();
    for s in profiles.column("platform_code")?.str()?.into_iter().flatten() {
        platforms.insert(s.to_string());
    }
    let n_platforms = platforms.len() as u32;

    let mut cols: Vec<Series> = vec![
        Series::new("n_platforms".into(), vec![n_platforms]),
        Series::new("n_profiles".into(), vec![n_profiles]),
        base.column("n_obs")?.clone(),
        Series::new("time_qc_good".into(), vec![time_qc_good]),
        Series::new("position_qc_good".into(), vec![position_qc_good]),
    ];
    if has_dup {
        let dup_profiles = profiles.column("is_dup")?.bool()?.sum().unwrap_or(0);
        cols.push(Series::new("dup_profiles".into(), vec![dup_profiles]));
    }
    for name in ["longitude", "latitude"] {
        cols.push(base.column(format!("{name}_min").as_str())?.clone());
        cols.push(base.column(format!("{name}_max").as_str())?.clone());
    }
    for name in ["temp", "psal", "pres"] {
        cols.push(base.column(format!("na_{name}").as_str())?.clone());
        cols.push(base.column(format!("{name}_min").as_str())?.clone());
        cols.push(base.column(format!("{name}_max").as_str())?.clone());
        cols.push(base.column(format!("{name}_mean").as_str())?.clone());
    }
    Ok(DataFrame::new(cols)?)
}

/// Count entries equal to `val` in a string column.
fn count_eq(ca: &StringChunked, val: &str) -> u32 {
    ca.into_iter().filter(|o| *o == Some(val)).count() as u32
}