923 add support for categories for software

TODOs.md

@@ -0,0 +1,7 @@
+
+TODOs for category feature
+==========================
+
+- add backend test? how?
+- how should be abbreviate some of these long category names?
+- the API calling functions have a lot of copy&paste code. Let's create some generic helper functions for API calls

database/007-create-keyword-and-category.sql

@@ -0,0 +1,153 @@
+-- SPDX-FileCopyrightText: 2022 Ewan Cahen (Netherlands eScience Center) <[email protected]>
+-- SPDX-FileCopyrightText: 2022 Netherlands eScience Center
+-- SPDX-FileCopyrightText: 2023 Felix Mühlbauer (GFZ) <[email protected]>
+-- SPDX-FileCopyrightText: 2023 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
+--
+-- SPDX-License-Identifier: Apache-2.0
+-- SPDX-License-Identifier: EUPL-1.2
+
+--------------
+-- Keywords --
+--------------
+
+CREATE TABLE keyword (
+	id UUID DEFAULT gen_random_uuid() PRIMARY KEY,
+	value CITEXT UNIQUE CHECK (value ~ '^\S+( \S+)*$')
+);
+
+CREATE FUNCTION sanitise_insert_keyword() RETURNS TRIGGER LANGUAGE plpgsql AS
+$$
+BEGIN
+	NEW.id = gen_random_uuid();
+	return NEW;
+END
+$$;
+
+CREATE TRIGGER sanitise_insert_keyword BEFORE INSERT ON keyword FOR EACH ROW EXECUTE PROCEDURE sanitise_insert_keyword();
+
+
+CREATE FUNCTION sanitise_update_keyword() RETURNS TRIGGER LANGUAGE plpgsql AS
+$$
+BEGIN
+	NEW.id = OLD.id;
+	return NEW;
+END
+$$;
+
+CREATE TRIGGER sanitise_update_keyword BEFORE UPDATE ON keyword FOR EACH ROW EXECUTE PROCEDURE sanitise_update_keyword();
+
+
+CREATE TABLE keyword_for_software (
+	software UUID references software (id),
+	keyword UUID references keyword (id),
+	PRIMARY KEY (software, keyword)
+);
+
+CREATE TABLE keyword_for_project (
+	project UUID references project (id),
+	keyword UUID references keyword (id),
+	PRIMARY KEY (project, keyword)
+);
+
+-- ADD basic keywords from topics and tags
+INSERT into keyword (value)
+VALUES
+	('Big data'),
+	('GPU'),
+	('High performance computing'),
+	('Image processing'),
+	('Inter-operability & linked data'),
+	('Machine learning'),
+	('Multi-scale & multi model simulations'),
+	('Optimized data handling'),
+	('Real time data analysis'),
+	('Text analysis & natural language processing'),
+	('Visualization'),
+	('Workflow technologies');
+
+----------------
+-- Categories --
+----------------
+
+CREATE TABLE category (
+	id UUID DEFAULT gen_random_uuid() PRIMARY KEY,
+	parent UUID REFERENCES category DEFAULT NULL,
+	short_name VARCHAR NOT NULL,
+	name VARCHAR NOT NULL
+	-- We need postgresql 15 for "NULLS NOT DISTINCT" :(
+	-- UNIQUE NULLS NOT DISTINCT (parent, short_name),
+	-- UNIQUE NULLS NOT DISTINCT (parent, name)
+);
+-- workaround: for postgrsql <15 for "NULLS NOT DISTINCT"
+CREATE UNIQUE INDEX category_parent_short_name_key1 ON category (parent, short_name) WHERE parent IS NOT NULL;
+CREATE UNIQUE INDEX category_parent_short_name_key2 ON category (short_name) WHERE parent IS NULL;
+CREATE UNIQUE INDEX category_parent_name_key1 ON category (parent, name) WHERE parent IS NOT NULL;
+CREATE UNIQUE INDEX category_parent_name_key2 ON category (name) WHERE parent IS NULL;
+-- workaround END
+
+CREATE TABLE category_for_software (
+	software_id UUID references software (id),
+	category_id UUID references category (id),
+	PRIMARY KEY (software_id, category_id)
+);
+
+CREATE FUNCTION category_path(category_id UUID)
+RETURNS TABLE (like category)
+LANGUAGE SQL STABLE AS
+$$
+	WITH RECURSIVE cat_path AS (
+		SELECT *, 1 AS r_index
+			FROM category WHERE id = category_id
+	UNION ALL
+		SELECT category.*, cat_path.r_index+1
+			FROM category
+			JOIN cat_path
+		ON category.id = cat_path.parent
+	)
+	-- TODO: How can we reverse the output rows without injecting a new column
+	-- Now we have to list all columns of `category` explicitely
+	-- I want to have something like `* without 'r_index'` to be independant from modifications of `category`
+	SELECT id, parent, short_name, name
+	FROM cat_path
+	ORDER BY r_index DESC;
+$$;
+
+-- returns a list of `category` entries traversing from the tree root to entry with `category_id`
+CREATE FUNCTION category_path_expanded(category_id UUID)
+RETURNS JSON
+LANGUAGE SQL STABLE AS
+$$
+	SELECT json_agg(row_to_json) AS path FROM (SELECT row_to_json(category_path(category_id))) AS cats;
+$$;
+
+
+CREATE FUNCTION category_paths_by_software_expanded(software_id UUID)
+RETURNS JSON
+LANGUAGE SQL STABLE AS
+$$
+	WITH
+		cat_ids AS
+		(SELECT category_id FROM category_for_software AS c4s WHERE c4s.software_id = category_paths_by_software_expanded.software_id),
+	paths as
+		(SELECT category_path_expanded(category_id) AS path FROM cat_ids)
+	SELECT
+		CASE WHEN EXISTS(SELECT 1 FROM cat_ids) THEN (SELECT json_agg(path) FROM paths)
+		ELSE '[]'::json
+		END as result
+$$;
+
+
+CREATE FUNCTION available_categories_expanded()
+RETURNS JSON
+LANGUAGE SQL STABLE AS
+$$
+	WITH
+	cat_ids AS
+		(SELECT id AS category_id FROM category AS node WHERE NOT EXISTS (SELECT 1 FROM category AS sub WHERE node.id = sub.parent)),
+	paths AS
+		(SELECT category_path_expanded(category_id) AS path FROM cat_ids)
+	SELECT
+		CASE WHEN EXISTS(SELECT 1 FROM cat_ids) THEN (SELECT json_agg(path) AS result FROM paths)
+		ELSE '[]'::json
+		END
+$$

database/020-row-level-security.sql

@@ -2,8 +2,11 @@
 -- SPDX-FileCopyrightText: 2021 - 2023 Netherlands eScience Center
 -- SPDX-FileCopyrightText: 2022 - 2023 Dusan Mijatovic (dv4all)
 -- SPDX-FileCopyrightText: 2022 - 2023 dv4all
+-- SPDX-FileCopyrightText: 2023 Felix Mühlbauer (GFZ) <[email protected]>
+-- SPDX-FileCopyrightText: 2023 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 --
 -- SPDX-License-Identifier: Apache-2.0
+-- SPDX-License-Identifier: EUPL-1.2
 
 -- maintainer tables
 ALTER TABLE maintainer_for_software ENABLE ROW LEVEL SECURITY;
@@ -273,6 +276,27 @@ CREATE POLICY admin_all_rights ON testimonial TO rsd_admin
 	USING (TRUE)
 	WITH CHECK (TRUE);
 
+-- categories for software
+
+ALTER TABLE category_for_software ENABLE ROW LEVEL SECURITY;
+
+-- allow everybody to read metadata of published software
+CREATE POLICY anyone_can_read ON category_for_software
+	FOR SELECT
+	TO rsd_web_anon, rsd_user
+	USING (EXISTS(SELECT 1 FROM software WHERE id = software_id AND is_published = TRUE));
+
+-- allow software maintainers to have read/write access to their software
+CREATE POLICY maintainer_all_rights ON category_for_software
+	TO rsd_user
+	USING (software_id IN (SELECT * FROM software_of_current_maintainer()));
+
+-- allow admins to have full read/write access
+CREATE POLICY admin_all_rights ON category_for_software
+	TO rsd_admin
+	USING (TRUE);
+
+
 -- keywords
 ALTER TABLE keyword ENABLE ROW LEVEL SECURITY;
 

database/999-add-hgf-categories.sql

@@ -0,0 +1,56 @@
+-- SPDX-FileCopyrightText: 2023 Felix Mühlbauer (GFZ) <[email protected]>
+-- SPDX-FileCopyrightText: 2023 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
+--
+-- SPDX-License-Identifier: EUPL-1.2
+
+-- THIS FILE SHOULD BE MOVED TO THE DEPLOMENT REPO AND USED AS A MIGRATION FILE
+
+CREATE FUNCTION add_category(parent_name varchar, parent_short_name varchar, sub_categories varchar[][])
+RETURNS TABLE (LIKE category)
+LANGUAGE plpgsql
+AS $$
+DECLARE parent_id uuid;
+DECLARE category varchar[];
+BEGIN
+  INSERT INTO category(short_name, name) VALUES (parent_short_name, parent_name)
+  RETURNING id INTO parent_id;
+  FOREACH category SLICE 1 IN ARRAY sub_categories LOOP
+    INSERT INTO category(parent, short_name, name) VALUES (parent_id, category[1], category[2]);
+  END LOOP;
+END
+$$;
+
+SELECT add_category('Research fields', 'Research fields',
+  ARRAY[
+    ['Energy', 'Energy'],
+    ['E&E', 'Earth and Environment'],
+    ['Health', 'Health'],
+    ['Information', 'Information'],
+    ['AST', 'Aeronautics, Space and Transport'],
+    ['Matter', 'Matter']
+  ]
+);
+
+SELECT add_category('POF IV', 'Program-oriented Funding IV',
+  ARRAY[
+    ['Energy System Design', 'Energy System Design (Energy)'],
+    ['Materials and Technologies for the Energy Transition', 'Materials and Technologies for the Energy Transition (Energy)'],
+    ['Fusion', 'Fusion (Energy)'],
+    ['Nuclear Waste Management, Safety, and Radiation Research', 'Nuclear Waste Management, Safety, and Radiation Research (Energy)'],
+    ['Changing Earth – Sustaining our Future', 'Changing Earth – Sustaining our Future (Earth and Environment)'],
+    ['Cancer Research', 'Cancer Research (Health)'],
+    ['Environment-related and Metabolic Diseases', 'Environment-related and Metabolic Diseases (Health)'],
+    ['Systemic Medicine and Cardiovascular Diseases', 'Systemic Medicine and Cardiovascular Diseases (Health)'],
+    ['Infection Research', 'Infection Research (Health)'],
+    ['Neurodegenerative Diseases', 'Neurodegenerative Diseases (Health)'],
+    ['Engineering Digital Futures', 'Engineering Digital Futures (Information)'],
+    ['Natural, Artificial, and Cognitive Information Processing', 'Natural, Artificial, and Cognitive Information Processing (Information)'],
+    ['Materials Systems Engineering', 'Materials Systems Engineering (Information)'],
+    ['Aeronautics', 'Aeronautics (Aeronautics, Space, and Transport)'],
+    ['Space', 'Space (Aeronautics, Space, and Transport)'],
+    ['Transport', 'Transport (Aeronautics, Space, and Transport)'],
+    ['Matter and the Universe', 'Matter and the Universe (Matter)'],
+    ['Matter and Technologies', 'Matter and Technologies (Matter)'],
+    ['From Matter to Materials and Life', 'From Matter to Materials and Life (Matter)']
+  ]
+);

frontend/components/software/AboutSection.tsx

@@ -1,35 +1,38 @@
 // SPDX-FileCopyrightText: 2021 - 2023 Dusan Mijatovic (dv4all)
-// SPDX-FileCopyrightText: 2021 - 2023 dv4all
+// SPDX-FileCopyrightText: 2022 - 2023 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 // SPDX-FileCopyrightText: 2022 Christian Meeßen (GFZ) <[email protected]>
-// SPDX-FileCopyrightText: 2022 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
+// SPDX-FileCopyrightText: 2023 Felix Mühlbauer (GFZ) <[email protected]>
 //
 // SPDX-License-Identifier: Apache-2.0
+// SPDX-License-Identifier: EUPL-1.2
 
-import {License, ProgramingLanguages, CodePlatform, KeywordForSoftware} from '../../types/SoftwareTypes'
+import {License, ProgramingLanguages, CodePlatform, KeywordForSoftware, CategoriesForSoftware} from '../../types/SoftwareTypes'
 import PageContainer from '../layout/PageContainer'
 import AboutStatement from './AboutStatement'
 import SoftwareKeywords from './SoftwareKeywords'
 import AboutLanguages from './AboutLanguages'
 import AboutLicense from './AboutLicense'
 import AboutSourceCode from './AboutSourceCode'
 import SoftwareLogo from './SoftwareLogo'
+import { SoftwareCategoriesWithHeadline } from './SoftwareCategories'
 
 type AboutSectionType = {
   brand_name: string
   description: string
   description_type: 'link' | 'markdown'
   keywords: KeywordForSoftware[]
+  categories: CategoriesForSoftware
   licenses: License[]
   repository: string | null
   platform: CodePlatform
-  languages: ProgramingLanguages,
+  languages: ProgramingLanguages
   image_id: string | null
 }
 
 
 export default function AboutSection(props:AboutSectionType) {
   const {
-    brand_name = '', description = '', keywords, licenses,
+    brand_name = '', description = '', keywords, categories, licenses,
     repository, languages, platform, description_type = 'markdown',
     image_id
   } = props
@@ -58,6 +61,7 @@ export default function AboutSection(props:AboutSectionType) {
       </div>
       <div className="flex-1">
         {getSoftwareLogo()}
+        <SoftwareCategoriesWithHeadline categories={categories} />
         <SoftwareKeywords keywords={keywords || []} />
         <AboutLanguages languages={languages} platform={platform} />
         <AboutLicense license={license || []} />