diff --git a/drs/src/cordex.rs b/drs/src/cordex.rs
index 23e1326dd03bbbca08de9e038928a905de32476c..46be62f6febc0dd0f0e4ed687823e5716b009438 100644
--- a/drs/src/cordex.rs
+++ b/drs/src/cordex.rs
@@ -16,7 +16,9 @@
//! <rcm_model_name>_<rcm_version_id>_<frequency>[_<start_time>-<end_time>].nc
//! ```
//!
-//! Exceptions to the Cordex standard:
+//! # Exceptions to the Cordex standard:
+//! * According to the spec, `activity` and `product` must be `cordex` and `output` respectively. This allows
+//! arbitrary values for these elements.
//! * This allows frequency values of `1hr` in addition to the normal values
use std::str::FromStr;
@@ -27,13 +29,6 @@ use thiserror::Error;
use crate::{cmip5::EnsembleMember, parser::parse_cordex};
-/// As per the documentation, <activity> must be `"CORDEX"`
-///
-/// In netcdf file attributes, called `project_id`.
-pub const ACTIVITY_VALUE: &str = "cordex";
-/// As per the documentation, <product> must be `"output"`
-pub const PRODUCT_VALUE: &str = "output";
-
/// Error parsing a cordex style path
#[derive(Debug, Error)]
#[error("error parsing CORDEX path: {reason}")]
@@ -82,6 +77,10 @@ impl<'a> Cordex<'a> {
/// Where relevant, name in attributes is referring to the key within the netcdf file's metadata for that field.
#[derive(Debug, PartialEq, Eq)]
pub struct PathMetadata<'a> {
+ /// Name of the activity for this dataset. In the spec this must be `cordex` but this is not enforced here.
+ pub activity: &'a str,
+ /// Name of the product of this dataset. In the spec this must be `output` but this is not enforced here.
+ pub product: &'a str,
/// The name assigned to each of the CORDEX regions and includes a flag for resolution.
///
/// Possible values come from the name column(?) of Tables 1 and 2 in the source document.
diff --git a/drs/src/parser/cordex.rs b/drs/src/parser/cordex.rs
index 61e09d905b424ec1eb6229991bbf6e7431740866..991eb6907ebd6bee8d7754023f5188d319b71454 100644
--- a/drs/src/parser/cordex.rs
+++ b/drs/src/parser/cordex.rs
@@ -3,13 +3,13 @@ use std::str::FromStr;
use chrono::NaiveDateTime;
use nom::{
bytes::complete::tag,
- combinator::{eof, map_res, opt, verify},
+ combinator::{eof, map_res, opt},
sequence::{preceded, terminated},
IResult,
};
use crate::{
- cordex::{Frequency, PathMetadata, ACTIVITY_VALUE, PRODUCT_VALUE},
+ cordex::{Frequency, PathMetadata},
parser::{
cmip5::parse_ensemble,
common::{name_segment, name_sep, parse_time, parse_version, path_sep},
@@ -24,8 +24,8 @@ pub(crate) fn parse_cordex(i: &str) -> IResult<&str, PathMetadata> {
<gcm_model_name>/<cmip5_experiement_name>/<cmip5_ensemble_member>/
<rcm_model_name>/<rcm_version_id>/<frequency>/<variable_name>
*/
- let (i, _activity) = terminated(parse_activity, path_sep)(i)?;
- let (i, _product) = terminated(parse_product, path_sep)(i)?;
+ let (i, activity) = path_segment(i)?;
+ let (i, product) = path_segment(i)?;
let (i, domain) = path_segment(i)?;
let (i, institution) = path_segment(i)?;
let (i, gcm_model_name) = path_segment(i)?;
@@ -59,6 +59,8 @@ pub(crate) fn parse_cordex(i: &str) -> IResult<&str, PathMetadata> {
Ok((
i,
PathMetadata {
+ activity,
+ product,
domain,
institution,
gcm_model_name,
@@ -75,14 +77,6 @@ pub(crate) fn parse_cordex(i: &str) -> IResult<&str, PathMetadata> {
))
}
-fn parse_activity(i: &str) -> IResult<&str, &str> {
- verify(word, |activity: &str| activity == ACTIVITY_VALUE)(i)
-}
-
-fn parse_product(i: &str) -> IResult<&str, &str> {
- verify(word, |product: &str| product == PRODUCT_VALUE)(i)
-}
-
fn parse_frequency(i: &str) -> IResult<&str, Frequency> {
map_res(word, Frequency::from_str)(i)
}
@@ -129,6 +123,8 @@ mod tests {
"rsdt_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_19910101-19951231.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -153,6 +149,8 @@ mod tests {
"prw_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_19960101-20001231.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -177,6 +175,8 @@ mod tests {
"ta500_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_20060101-20101231.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -201,6 +201,8 @@ mod tests {
"tauu_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_19910101-19951231.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -225,6 +227,8 @@ mod tests {
"rlut_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_20010101-20051231.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -249,6 +253,8 @@ mod tests {
"ua500_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_19960101-20001231.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -273,6 +279,8 @@ mod tests {
"tauv_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_20110101-20111130.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -297,6 +305,8 @@ mod tests {
"hus850_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_20110101-20111130.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -321,6 +331,8 @@ mod tests {
"evspsbl_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_20010101-20051231.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -345,6 +357,8 @@ mod tests {
"va200_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_19900101-19901231.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -369,6 +383,8 @@ mod tests {
"sfcWindmax_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_20110101-20111130.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
@@ -393,6 +409,8 @@ mod tests {
"zmla_AFR-44_ECMWF-ERAINT_evaluation_r1i1p1_MOHC-HadRM3P_v1_day_20010101-20051231.nc"
),
PathMetadata {
+ activity: "cordex",
+ product: "output",
domain: "AFR-44",
institution: "MOHC",
gcm_model_name: "ECMWF-ERAINT",
diff --git a/freva/src/drs/metadata.rs b/freva/src/drs/metadata.rs
index 009f4b7d0338c1a5464cd8c372d40e524977d659..e5a6fb82c36c773471015fb0a38812a1e28b49d8 100644
--- a/freva/src/drs/metadata.rs
+++ b/freva/src/drs/metadata.rs
@@ -12,8 +12,7 @@ use std::{
};
use camino::{Utf8Path, Utf8PathBuf};
-use drs::{cmip5::Cmip5, cordex::Cordex};
-use drs::{cmip6::Cmip6, cordex};
+use drs::{cmip5::Cmip5, cmip6::Cmip6, cordex::Cordex};
use netcdf::AttrValue;
use thiserror::Error;
use tracing::{debug, error};
@@ -312,8 +311,8 @@ impl<'a> From<Cordex<'a>> for Metadata {
let mip_table = "atmos".to_owned(); // all cordex data is atmos
Self {
path: cordex.path.to_owned(),
- activity: cordex::ACTIVITY_VALUE.to_owned(),
- product: cordex::PRODUCT_VALUE.to_owned(),
+ activity: m.activity.to_owned(),
+ product: m.product.to_owned(),
institute: m.institution.to_owned(),
model: m.gcm_model_name.to_owned(),
experiment: m.cmip5_experiment_name.to_owned(),