Commit bf1789d6 authored by Wohlgemuth, Jason's avatar Wohlgemuth, Jason
Browse files

Merge branch 'feat/update-groups' into 'main' 

Add all groups to organization JSON

See merge request !8
parents 8d14ace8 5a79ca99

acorn-lib/assets/constants/organization.json

+2336 −71

File changed.

Preview size limit exceeded, changes collapsed.

acorn-lib/src/schema/mod.rs

+66 −67
Original line number Diff line number Diff line
@@ -22,6 +22,20 @@ use validator::{Validate, ValidationErrorsKind}; 
pub mod validate;

use validate::*;



/// ### Core concepts related to the associated research activity

///

/// Could be used to filter research activity data and/or power data analytics through concept composition

///

/// ### Guidelines for creating keywords

/// - **Shall**

///     - Be officially sanctioned by responsible parties

///     - Be in lower-kebab-case

///     - Be unique relative to other keywords

///     - Contain three or more characters

/// - **Should**

///     - Not be too specific

///     - Be one or two words (ex. `foo` or `foo-bar`)

type Keyword = String;

/// U.S. Classified National Security Information Level

///

/// See [President Executive Order 13526](https://www.archives.gov/isoo/policy-documents/cnsi-eo.html)
@@ -53,6 +67,7 @@ pub enum ClassificationLevel { 
pub enum FuzzyValue {

    #[display("partners")]

    Partner,

    /// See [Keyword]

    #[display("keywords")]

    Keyword,

    #[display("sponsors")]
@@ -297,21 +312,8 @@ pub struct Metadata { 
    /// <div class="warning"><a href="https://code.ornl.gov/research-enablement/acorn/-/blob/main/acorn-lib/assets/constants/sponsors.csv">Full list of sponsors</a></div>

    #[serde(skip_serializing_if = "Option::is_none")]

    pub sponsors: Option<Vec<String>>,

    /// ### Core concepts related to the associated research activity

    ///

    /// Could be used to filter research activity data and/or power data analytics through concept composition

    ///

    /// ### Guidelines for creating keywords

    /// - **Shall**

    ///     - Be officially sanctioned by responsible parties

    ///     - Be in lower-kebab-case

    ///     - Be unique relative to other keywords

    ///     - Contain three or more characters

    /// - **Should**

    ///     - Not be too specific

    ///     - Be one or two words (ex. `foo` or `foo-bar`)

    #[builder(default = Vec::<String>::new())]

    pub keywords: Vec<String>,

    pub keywords: Vec<Keyword>,

    /// ### Software, code languages, and digital tools related to the associated research activity data

    ///

    /// ***Examples***
@@ -328,18 +330,30 @@ pub struct Metadata { 
    pub related: Option<Vec<String>>,

}

#[derive(Clone, Debug, Serialize, Deserialize, Display, Hash, PartialEq, PartialOrd)]

#[display("Organization ({additional_type}) - {name} ({alternative_name})")]

#[display("Organization ({additional_type}) - {name})")]

#[serde(rename_all = "camelCase")]

pub struct Organization {

    /// ### Full name of the organization

    ///

    /// See <https://schema.org/name>

    pub name: String,

    /// ### Research Organization Registry

    ///

    /// See <https://www.ror.org/> for more information

    pub ror: Option<String>,

    pub alternative_name: String,

    /// ### Organization alias (e.g. acronym or nickname)

    ///

    /// See <https://schema.org/alternateName>

    pub alternative_name: Option<String>,

    /// ### Organization sub-type

    ///

    /// See <https://schema.org/additionalType>

    pub additional_type: OrganizationType,

    pub keywords: Option<Vec<String>>,

    pub children: Vec<Organization>,

    pub keywords: Option<Vec<Keyword>>,

    /// ### Distinct part(s) of the associated containing organization

    ///

    /// See <https://schema.org/member>

    pub member: Vec<Organization>,

}

// TODO: Trim whitespace with desaerialize_with derive - https://serde.rs/impl-serialize.html

#[derive(Builder, Clone, Debug, Serialize, Deserialize, Validate)]
@@ -516,10 +530,10 @@ impl Organization { 
    }

    pub fn flatten(&self) -> Vec<Organization> {

        let mut items = vec![self.clone()];

        let directorates = self.children.clone();

        let directorates = self.member.clone();

        for directorate in &directorates {

            items.push(directorate.clone());

            let divisions = directorate.children.clone();

            let divisions = directorate.member.clone();

            for division in &divisions {

                items.push(division.clone());

            }
@@ -535,23 +549,23 @@ impl Organization { 
            .collect::<Vec<_>>()

    }

    pub fn get(&self, organization_type: OrganizationType) -> Vec<Organization> {

        let items = self.children.clone();

        let items = self.member.clone();

        Organization::_get(items, organization_type)

    }

    pub fn get_all(&self, organization_type: OrganizationType) -> Vec<Organization> {

        let items = self.flatten();

        Organization::_get(items, organization_type)

    }

    pub fn get_all_name_pairs(&self) -> Vec<(String, String)> {

    pub fn get_all_name_pairs(&self) -> Vec<(String, Option<String>)> {

        self.flatten()

            .into_iter()

            .map(|x| (x.alternative_name.clone(), x.name.clone()))

            .filter(|(abbr, name)| !(abbr.is_empty() || name.is_empty()))

            .collect::<Vec<(String, String)>>()

            .map(|x| (x.name.clone(), x.alternative_name.clone()))

            .filter(|(name, alias)| !(name.is_empty() || alias.is_none()))

            .collect::<Vec<(String, Option<String>)>>()

    }

    pub fn visit<F: Copy + for<'a> Fn(&'a mut Organization)>(&mut self, f: F) {

        f(self);

        for child in self.children.iter_mut() {

        for child in self.member.iter_mut() {

            child.visit(f);

        }

    }
@@ -684,7 +698,10 @@ impl ResearchActivity { 
            },

            | FuzzyValue::Technology => self.meta.technology,

        };

        let mut data: Vec<_> = values.into_iter().flat_map(|x| resolve_from_csv(format!("{}", value_type), x)).collect();

        let mut data: Vec<_> = values

            .into_iter()

            .flat_map(|x| resolve_from_csv_asset(format!("{}", value_type), x))

            .collect();

        data.sort();

        data.dedup();

        data
@@ -729,7 +746,7 @@ impl ResearchActivity { 
        clone.meta.partners = Some(self.clone().resolve(FuzzyValue::Partner));

        clone.meta.sponsors = Some(self.clone().resolve(FuzzyValue::Sponsor));

        clone.meta.technology = self.clone().resolve(FuzzyValue::Technology);

        clone.contact.organization = match resolve_organization(self.clone().contact.organization) {

        clone.contact.organization = match resolve_from_organization_json(self.clone().contact.organization) {

            | Some(value) => value,

            | None => "".to_string(),

        };
@@ -802,43 +819,12 @@ fn print_resolution(output: Option<String>, value: String, name: String) { 
        }

    };

}

fn resolve_organization(value: String) -> Option<String> {

    let data = Organization::load()[0]

        .get_all_name_pairs()

        .into_iter()

        .map(|(key, value)| (key.to_string(), value.to_string()))

        .collect::<HashMap<String, String>>();

    let value = value.trim().to_string().replace("&", "and");

    let sanitized = sanitize(value.clone());

    let output = match sanitized.len() {

        | 3..=5 if data.contains_key(&value.to_uppercase()) => data.get(&value.to_uppercase()).cloned(),

        | _ => {

            let pattern = Pattern::parse(&sanitized, CaseMatching::Ignore, Normalization::Smart);

            let mut matcher = Matcher::new(Config::DEFAULT.match_paths());

            let output = pattern

                .match_list(data.clone().into_values().collect::<Vec<String>>(), &mut matcher)

                .first()

                .map(|(x, _)| x.to_string());

            output

        }

    };

    let label = "Organization";

    match output.clone() {

        | Some(resolved) => {

            if resolved.eq(&value.to_string()) {

                trace!("=> {} {} = \"{}\"", Label::using(), label, value.clone());

            } else {

                debug!(input = value.clone(), resolved, "=> {} {}", Label::found(), label);

fn resolve_from_csv_asset(name: String, value: String) -> Option<String> {

    let data = Constant::csv(format!("{}.csv", name).as_str());

    resolve_from_list_of_lists(value, data)

}

        }

        | None => {

            debug!(value = value.clone(), "=> {} {}", Label::not_found(), label);

        }

    };

    output

}

fn resolve_from_csv(name: String, value: String) -> Option<String> {

    let output = Constant::csv(format!("{}.csv", name).as_str())

fn resolve_from_list_of_lists(value: String, data: Vec<Vec<String>>) -> Option<String> {

    let output = data

        .into_iter()

        .flat_map(|values| {

            let sanitized = sanitize(value.clone());
@@ -860,13 +846,26 @@ fn resolve_from_csv(name: String, value: String) -> Option<String> { 
        })

        .max_by_key(|(_, score)| *score)

        .map(|(x, _)| x.to_string());



    print_resolution(output.clone(), value, name);

    print_resolution(output.clone(), value, "organization".to_string());

    output

}

fn resolve_from_organization_json(value: String) -> Option<String> {

    let data = Organization::load()[0]

        .get_all_name_pairs()

        .into_iter()

        .map(|(name, alias)| {

            let alternative_name = match alias {

                | Some(x) => x.to_string(),

                | None => name.clone(),

            };

            vec![name, alternative_name]

        })

        .collect::<Vec<Vec<String>>>();

    resolve_from_list_of_lists(value, data)

}

fn sanitize(value: String) -> String {

    match Regex::new(r"[-_.]") {

        | Ok(re) => re.replace_all(&value, "").trim().to_string(),

        | Ok(re) => re.replace_all(&value, "").replace("&", "and").trim().to_string(),

        | Err(err) => err.to_string(),

    }

}

acorn-lib/src/schema/tests/mod.rs

+66 −34
Original line number Diff line number Diff line
@@ -67,64 +67,96 @@ fn test_read() { 
    }

}

#[test]

fn test_resolve_from_csv() {

fn test_fuzzy_matching() {

    //

    // Keywords

    //

    let name = "keywords".to_string();

    let exact: String = "critical-infrastructure".into();

    let misspelled: String = "cyb".into();

    assert_eq!(resolve_from_csv(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv(name.clone(), misspelled), Some("cyber".into()));

    assert_eq!(resolve_from_csv(name.clone(), "machine-learn".into()), Some("machine-learning".into()));

    assert_eq!(resolve_from_csv(name.clone(), "Automation".into()), Some("automation".into()));

    assert_eq!(resolve_from_csv(name.clone(), "math".into()), Some("mathematics".into()));

    assert_eq!(resolve_from_csv(name.clone(), "mathematics".into()), Some("mathematics".into()));

    assert_eq!(resolve_from_csv(name.clone(), "ml".into()), Some("machine-learning".into()));

    assert_eq!(resolve_from_csv(name.clone(), "ai".into()), Some("artificial-intelligence".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv_asset(name.clone(), misspelled), Some("cyber".into()));

    assert_eq!(

        resolve_from_csv_asset(name.clone(), "machine-learn".into()),

        Some("machine-learning".into())

    );

    assert_eq!(resolve_from_csv_asset(name.clone(), "Automation".into()), Some("automation".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), "math".into()), Some("mathematics".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), "mathematics".into()), Some("mathematics".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), "ml".into()), Some("machine-learning".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), "ai".into()), Some("artificial-intelligence".into()));

    let exact: String = "high-performance-computing".into();

    assert_eq!(resolve_from_csv(name.clone(), "hpc".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv(name.clone(), "high-performance-computi".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv_asset(name.clone(), "hpc".into()), Some(exact.clone()));

    assert_eq!(

        resolve_from_csv_asset(name.clone(), "high-performance-computi".into()),

        Some(exact.clone())

    );

    assert_eq!(resolve_from_csv_asset(name.clone(), exact.clone()), Some(exact));

    //

    // Organizations

    //

    let exact: String = "Oak Ridge National Laboratory".into();

    assert_eq!(resolve_from_organization_json(exact.clone()), Some(exact.clone()));

    assert_eq!(resolve_from_organization_json("ORNL".into()), Some(exact.clone()));

    assert_eq!(resolve_from_organization_json("Oak Ridge National Laborato".into()), Some(exact.clone()));

    let exact: String = "Geospatial Science and Human Security Division".into();

    assert_eq!(resolve_from_organization_json("GSHS".into()), Some(exact.clone()));

    assert_eq!(resolve_from_organization_json(exact.clone()), Some(exact.clone()));

    assert_eq!(

        resolve_from_organization_json("Geospatial Science & Human Security".into()),

        Some(exact.clone())

    );

    assert_eq!(

        resolve_from_organization_json("Geospatial Science and Human Security".into()),

        Some(exact.clone())

    );

    assert_eq!(

        resolve_from_organization_json("Geospatial Science and Human Security".to_lowercase()),

        Some(exact.clone())

    );

    assert_eq!(resolve_from_organization_json("PSD".into()), Some("Physical Sciences Directorate".into()));

    //

    // Sponsors

    //

    let name = "sponsors".to_string();

    let exact: String = "Oak Ridge National Laboratory".into();

    assert_eq!(resolve_from_csv(name.clone(), "ORNL".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv_asset(name.clone(), "ORNL".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv_asset(name.clone(), exact.clone()), Some(exact));

    let exact: String = "Department of Energy".into();

    assert_eq!(resolve_from_csv(name.clone(), "Dept of Energy".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv(name.clone(), "Dept. of Energy".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv_asset(name.clone(), "Dept of Energy".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv_asset(name.clone(), "Dept. of Energy".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv_asset(name.clone(), exact.clone()), Some(exact));

    let exact: String = "Department of Homeland Security".into();

    assert_eq!(resolve_from_csv(name.clone(), "DHS".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv_asset(name.clone(), "DHS".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv_asset(name.clone(), exact.clone()), Some(exact));

    let exact: String = "Environmental Protection Agency".into();

    assert_eq!(resolve_from_csv(name.clone(), " epa".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv_asset(name.clone(), " epa".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv_asset(name.clone(), exact.clone()), Some(exact));

    let exact: String = "Defense Advanced Research Projects Agency".into();

    assert_eq!(resolve_from_csv(name.clone(), "darpa".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv_asset(name.clone(), "darpa".into()), Some(exact.clone()));

    assert_eq!(resolve_from_csv_asset(name.clone(), exact.clone()), Some(exact));

    //

    // Technology

    //

    let name = "technology".to_string();

    let exact: String = "react".into();

    assert_eq!(resolve_from_csv(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv(name.clone(), "astro".into()), Some("astro".into()));

    assert_eq!(resolve_from_csv(name.clone(), "React.js".into()), Some("react".into()));

    assert_eq!(resolve_from_csv(name.clone(), "ReactJS".into()), Some("react".into()));

    assert_eq!(resolve_from_csv(name.clone(), "rs".into()), Some("rust".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), exact.clone()), Some(exact));

    assert_eq!(resolve_from_csv_asset(name.clone(), "astro".into()), Some("astro".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), "React.js".into()), Some("react".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), "ReactJS".into()), Some("react".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), "rs".into()), Some("rust".into()));

    assert_eq!(

        resolve_from_csv(name.clone(), "Geospatial Data Abstraction Lib".into()),

        resolve_from_csv_asset(name.clone(), "Geospatial Data Abstraction Lib".into()),

        Some("gdal".into())

    );

    assert_eq!(resolve_from_csv(name.clone(), "node.js".into()), Some("javascript".into()));

    assert_eq!(resolve_from_csv(name.clone(), "js".into()), Some("javascript".into()));

    assert_eq!(resolve_from_csv(name.clone(), "language::JavaScript".into()), Some("javascript".into()));

    assert_eq!(resolve_from_csv(name.clone(), "foobarbaz".into()), None);

    assert_eq!(resolve_from_csv(name.clone(), "".into()), None);

    assert_eq!(resolve_from_csv_asset(name.clone(), "node.js".into()), Some("javascript".into()));

    assert_eq!(resolve_from_csv_asset(name.clone(), "js".into()), Some("javascript".into()));

    assert_eq!(

        resolve_from_csv_asset(name.clone(), "language::JavaScript".into()),

        Some("javascript".into())

    );

    assert_eq!(resolve_from_csv_asset(name.clone(), "foobarbaz".into()), None);

    assert_eq!(resolve_from_csv_asset(name.clone(), "".into()), None);

}

#[test]

fn test_validate_attribute_areas() {