diff --git a/js/codemeta_generation.js b/js/codemeta_generation.js index 42ffecd..a052e1a 100644 --- a/js/codemeta_generation.js +++ b/js/codemeta_generation.js @@ -1,295 +1,295 @@ /** * Copyright (C) 2019-2020 The Software Heritage developers * See the AUTHORS file at the top-level directory of this distribution * License: GNU Affero General Public License version 3, or any later version * See top-level LICENSE file for more information */ "use strict"; const CODEMETA_CONTEXT_URL = 'https://doi.org/10.5063/schema/codemeta-2.0'; const SPDX_PREFIX = 'https://spdx.org/licenses/'; const loadContextData = async () => { const contextResponse = await fetch("../data/contexts/codemeta-2.0.jsonld"); const context = await contextResponse.json(); return { [CODEMETA_CONTEXT_URL]: context } } const getJsonldCustomLoader = contexts => { return url => { const xhrDocumentLoader = jsonld.documentLoaders.xhr(); if (url in contexts) { return { contextUrl: null, document: contexts[url], documentUrl: url }; } return xhrDocumentLoader(url); } }; const initJsonldLoader = contexts => { jsonld.documentLoader = getJsonldCustomLoader(contexts); }; function emptyToUndefined(v) { if (v == null || v == "") return undefined; else return v; } function getIfSet(query) { return emptyToUndefined(document.querySelector(query).value); } function setIfDefined(query, value) { if (value !== undefined) { document.querySelector(query).value = value; } } function getLicenses() { let selectedLicenses = Array.from(document.getElementById("selected-licenses").children); return selectedLicenses.map(licenseDiv => SPDX_PREFIX + licenseDiv.children[0].innerText); } // Names of codemeta properties with a matching HTML field name const directCodemetaFields = [ 'codeRepository', 'contIntegration', 'dateCreated', 'datePublished', 'dateModified', 'downloadUrl', 'issueTracker', 'name', 'version', 'identifier', 'description', 'applicationCategory', 'releaseNotes', 'funding', 'developmentStatus', 'isPartOf', 'referencePublication' ]; const splittedCodemetaFields = [ ['keywords', ','], ['programmingLanguage', ','], ['runtimePlatform', ','], ['operatingSystem', ','], ['softwareRequirements', '\n'], ['relatedLink', '\n'], ] // Names of codemeta properties with a matching HTML field name, // in a Person object const directPersonCodemetaFields = [ 'givenName', 'familyName', 'email', 'affiliation', ]; function generateShortOrg(fieldName) { var affiliation = getIfSet(fieldName); if (affiliation !== undefined) { if (isUrl(affiliation)) { return { "@type": "Organization", "@id": affiliation, }; } else { return { "@type": "Organization", "name": affiliation, }; } } else { return undefined; } } function generatePerson(idPrefix) { var doc = { "@type": "Person", } var id = getIfSet(`#${idPrefix}_id`); if (id !== undefined) { doc["@id"] = id; } directPersonCodemetaFields.forEach(function (item, index) { doc[item] = getIfSet(`#${idPrefix}_${item}`); }); doc["affiliation"] = generateShortOrg(`#${idPrefix}_affiliation`) return doc; } function generatePersons(prefix) { var persons = []; var nbPersons = getNbPersons(prefix); for (let personId = 1; personId <= nbPersons; personId++) { persons.push(generatePerson(`${prefix}_${personId}`)); } return persons; } function buildDoc() { var doc = { "@context": CODEMETA_CONTEXT_URL, "@type": "SoftwareSourceCode", }; let licenses = getLicenses(); if (licenses.length > 0) { doc["license"] = licenses; } // Generate most fields directCodemetaFields.forEach(function (item, index) { doc[item] = getIfSet('#' + item) }); doc["funder"] = generateShortOrg('#funder', doc["affiliation"]); // Generate simple fields parsed simply by splitting splittedCodemetaFields.forEach(function (item, index) { const id = item[0]; const separator = item[1]; const value = getIfSet('#' + id); if (value !== undefined) { doc[id] = value.split(separator).map(trimSpaces); } }); // Generate dynamic fields var authors = generatePersons('author'); if (authors.length > 0) { doc["author"] = authors; } var contributors = generatePersons('contributor'); if (contributors.length > 0) { doc["contributor"] = contributors; } return doc; } async function generateCodemeta() { var inputForm = document.querySelector('#inputForm'); var codemetaText, errorHTML; if (inputForm.checkValidity()) { var doc = buildDoc(); const expanded = await jsonld.expand(doc); const compacted = await jsonld.compact(expanded, CODEMETA_CONTEXT_URL); codemetaText = JSON.stringify(compacted, null, 4); errorHTML = ""; } else { codemetaText = ""; errorHTML = "invalid input (see error above)"; inputForm.reportValidity(); } document.querySelector('#codemetaText').innerText = codemetaText; setError(errorHTML); // Run validator on the exported value, for extra validation. // If this finds a validation, it means there is a bug in our code (either // generation or validation), and the generation MUST NOT generate an // invalid codemeta file, regardless of user input. if (codemetaText && !validateDocument(JSON.parse(codemetaText))) { alert('Bug detected! The data you wrote is correct; but for some reason, it seems we generated an invalid codemeta.json. Please report this bug at https://github.com/codemeta/codemeta-generator/issues/new and copy-paste the generated codemeta.json file. Thanks!'); } if (codemetaText) { // For restoring the form state on page reload sessionStorage.setItem('codemetaText', codemetaText); } } // Imports a single field (name or @id) from an Organization. function importShortOrg(fieldName, doc) { if (doc !== undefined) { // Use @id if set, else use name setIfDefined(fieldName, doc["name"]); - setIfDefined(fieldName, doc["@id"]); + setIfDefined(fieldName, getDocumentId(doc)); } } function importPersons(prefix, legend, docs) { if (docs === undefined) { return; } docs.forEach(function (doc, index) { var personId = addPerson(prefix, legend); - setIfDefined(`#${prefix}_${personId}_id`, doc['@id']); + setIfDefined(`#${prefix}_${personId}_id`, getDocumentId(doc)); directPersonCodemetaFields.forEach(function (item, index) { setIfDefined(`#${prefix}_${personId}_${item}`, doc[item]); }); importShortOrg(`#${prefix}_${personId}_affiliation`, doc['affiliation']) }) } -function importCodemeta() { +async function importCodemeta() { var inputForm = document.querySelector('#inputForm'); - var doc = parseAndValidateCodemeta(false); + var doc = await parseAndValidateCodemeta(false); resetForm(); if (doc['license'] !== undefined) { if (typeof doc['license'] === 'string') { doc['license'] = [doc['license']]; } doc['license'].forEach(l => { if (l.indexOf(SPDX_PREFIX) !== 0) { return; } let licenseId = l.substring(SPDX_PREFIX.length); insertLicenseElement(licenseId); }); } directCodemetaFields.forEach(function (item, index) { setIfDefined('#' + item, doc[item]); }); importShortOrg('#funder', doc["funder"]); // Import simple fields by joining on their separator splittedCodemetaFields.forEach(function (item, index) { const id = item[0]; const separator = item[1]; let value = doc[id]; if (value !== undefined) { if (Array.isArray(value)) { value = value.join(separator); } setIfDefined('#' + id, value); } }); importPersons('author', 'Author', doc['author']) importPersons('contributor', 'Contributor', doc['contributor']) } function loadStateFromStorage() { var codemetaText = sessionStorage.getItem('codemetaText') if (codemetaText) { document.querySelector('#codemetaText').innerText = codemetaText; importCodemeta(); } } diff --git a/js/validation/index.js b/js/validation/index.js index 546fab0..563a2e4 100644 --- a/js/validation/index.js +++ b/js/validation/index.js @@ -1,104 +1,106 @@ /** * Copyright (C) 2020 The Software Heritage developers * See the AUTHORS file at the top-level directory of this distribution * License: GNU Affero General Public License version 3, or any later version * See top-level LICENSE file for more information */ /* * Reads a Codemeta file and shows human-friendly errors on it. * * This validator intentionaly does not use a schema, in order to show errors * that are easy to understand for users with no understanding of JSON-LD. */ function validateDocument(doc) { if (!Array.isArray(doc) && typeof doc != 'object') { setError("Document must be an object (starting and ending with { and }), not ${typeof doc}.") return false; } // TODO: validate id/@id context = doc["@context"]; if (context == "https://doi.org/10.5063/schema/codemeta-2.0") { // Correct } else if (Array.isArray(context) && context.includes("https://doi.org/10.5063/schema/codemeta-2.0")) { if (context.length !== 1) { setError(`Multiple values in @context are not supported (@context should be "https://doi.org/10.5063/schema/codemeta-2.0", not ${JSON.stringify(context)})`); return false; } } else { setError(`@context must be "https://doi.org/10.5063/schema/codemeta-2.0", not ${JSON.stringify(context)}`); return false; } // TODO: check there is either type or @type but not both var type = getDocumentType(doc); if (type === undefined) { setError("Missing type (must be SoftwareSourceCode or SoftwareApplication).") return false; } else if (!isCompactTypeEqual(type, "SoftwareSourceCode") && !isCompactTypeEqual(type, "SoftwareApplication")) { // Check this before other fields, as a wrong type error is more // understandable than "invalid field". setError(`Wrong document type: must be "SoftwareSourceCode"/"SoftwareApplication", not ${JSON.stringify(type)}`) return false; } else { return Object.entries(doc).every((entry) => { var fieldName = entry[0]; var subdoc = entry[1]; if (fieldName == "@context") { // Was checked before return true; } else if (fieldName == "type" || fieldName == "@type") { // Was checked before return true; } else { var validator = softwareFieldValidators[fieldName]; if (validator === undefined) { // TODO: find if it's a field that belongs to another type, // and suggest that to the user setError(`Unknown field "${fieldName}".`) return false; } else { return validator(fieldName, subdoc); } } }); } } -function parseAndValidateCodemeta(showPopup) { +async function parseAndValidateCodemeta(showPopup) { var codemetaText = document.querySelector('#codemetaText').innerText; - var doc; + let parsed, doc; try { - doc = JSON.parse(codemetaText); + parsed = JSON.parse(codemetaText); } catch (e) { setError(`Could not read codemeta document because it is not valid JSON (${e}). Check for missing or extra quote, colon, or bracket characters.`); return; } setError(""); - var isValid = validateDocument(doc); + var isValid = validateDocument(parsed); if (showPopup) { if (isValid) { alert('Document is valid!') } else { alert('Document is invalid.'); } } + const expanded = await jsonld.expand(parsed); + doc = await jsonld.compact(expanded, CODEMETA_CONTEXT_URL); return doc; } diff --git a/js/validation/things.js b/js/validation/things.js index 496cd01..30fe016 100644 --- a/js/validation/things.js +++ b/js/validation/things.js @@ -1,309 +1,313 @@ /** * Copyright (C) 2020-2021 The Software Heritage developers * See the AUTHORS file at the top-level directory of this distribution * License: GNU Affero General Public License version 3, or any later version * See top-level LICENSE file for more information */ /* * Validators for codemeta objects derived from http://schema.org/Thing. */ function getDocumentType(doc) { // TODO: check there is at most one. // FIXME: is the last variant allowed? return doc["type"] || doc["@type"] || doc["codemeta:type"] } +function getDocumentId(doc) { + return doc["id"] || doc["@id"]; +} + function isCompactTypeEqual(type, compactedType) { // FIXME: are all variants allowed? return (type == `${compactedType}` || type == `schema:${compactedType}` || type == `codemeta:${compactedType}` || type == `http://schema.org/${compactedType}` ); } function noValidation(fieldName, doc) { return true; } // Validates subtypes of Thing, or URIs // // typeFieldValidators is a map: {type => {fieldName => fieldValidator}} function validateThingOrId(parentFieldName, typeFieldValidators, doc) { var acceptedTypesString = Object.keys(typeFieldValidators).join('/'); if (typeof doc == 'string') { if (!isUrl(doc)) { setError(`"${parentFieldName}" must be an URL or a ${acceptedTypesString} object, not: ${JSON.stringify(doc)}`); return false; } else { return true; } } else if (!Array.isArray(doc) && typeof doc == 'object') { return validateThing(parentFieldName, typeFieldValidators, doc); } else { setError(`"${parentFieldName}" must be a ${acceptedTypesString} object or URI, not ${JSON.stringify(doc)}`); return false; } } // Validates subtypes of Thing // // typeFieldValidators is a map: {type => {fieldName => fieldValidator}} function validateThing(parentFieldName, typeFieldValidators, doc) { // TODO: check there is either id or @id but not both // TODO: check there is either type or @type but not both var acceptedTypesString = Object.keys(typeFieldValidators).join('/'); var documentType = getDocumentType(doc); - var id = doc["id"] || doc["@id"]; + var id = getDocumentId(doc); if (id !== undefined && !isUrl(id)) { setError(`"${fieldName}" has an invalid URI as id: ${JSON.stringify(id)}"`); return false; } if (documentType === undefined) { if (id === undefined) { setError(`"${parentFieldName}" must be a (list of) ${acceptedTypesString} object(s) or an URI, but is missing a type/@type.`); return false; } else { // FIXME: we have an @id but no @type, what should we do? return true; } } for (expectedType in typeFieldValidators) { if (isCompactTypeEqual(documentType, expectedType)) { var fieldValidators = typeFieldValidators[expectedType]; return Object.entries(doc).every((entry) => { var fieldName = entry[0]; var subdoc = entry[1]; if (fieldName == "type" || fieldName == "@type") { // Was checked before return true; } else { var validator = fieldValidators[fieldName]; if (validator === undefined) { // TODO: find if it's a field that belongs to another type, // and suggest that to the user setError(`Unknown field "${fieldName}" in "${parentFieldName}".`) return false; } else { return validator(fieldName, subdoc); } } }); } } setError(`"${parentFieldName}" type must be a (list of) ${acceptedTypesString} object(s), not ${JSON.stringify(documentType)}`); return false; } // Helper function to validate a field is either X or a list of X. function validateListOrSingle(fieldName, doc, validator) { if (Array.isArray(doc)) { return doc.every((subdoc) => validator(subdoc, true)); } else { return validator(doc, false); } } // Validates a CreativeWork or an array of CreativeWork function validateCreativeWorks(fieldName, doc) { return validateListOrSingle(fieldName, doc, (subdoc, inList) => { return validateCreativeWork(fieldName, subdoc); }); } // Validates a single CreativeWork function validateCreativeWork(fieldName, doc) { return validateThingOrId(fieldName, { "CreativeWork": creativeWorkFieldValidators, "SoftwareSourceCode": softwareFieldValidators, "SoftwareApplication": softwareFieldValidators, }, doc); } // Validates a Person, Organization or an array of these function validateActors(fieldName, doc) { return validateListOrSingle(fieldName, doc, (subdoc, inList) => { return validateActor(fieldName, subdoc); }); } // Validates a Person or an array of Person function validatePersons(fieldName, doc) { return validateListOrSingle(fieldName, doc, (subdoc, inList) => { return validatePerson(fieldName, subdoc); }); } // Validates an Organization or an array of Organization function validateOrganizations(fieldName, doc) { return validateListOrSingle(fieldName, doc, (subdoc, inList) => { return validateOrganization(fieldName, subdoc); }); } // Validates a single Person or Organization function validateActor(fieldName, doc) { return validateThingOrId(fieldName, { "Person": personFieldValidators, "Organization": organizationFieldValidators, }, doc); } // Validates a single Person object function validatePerson(fieldName, doc) { return validateThingOrId(fieldName, {"Person": personFieldValidators}, doc); } // Validates a single Organization object function validateOrganization(fieldName, doc) { return validateThingOrId(fieldName, {"Organization": organizationFieldValidators}, doc); } var softwareFieldValidators = { "@id": validateUrl, "id": validateUrl, "codeRepository": validateUrls, "programmingLanguage": noValidation, "runtimePlatform": validateTexts, "targetProduct": noValidation, // TODO: validate SoftwareApplication "applicationCategory": validateTextsOrUrls, "applicationSubCategory": validateTextsOrUrls, "downloadUrl": validateUrls, "fileSize": validateText, // TODO "installUrl": validateUrls, "memoryRequirements": validateTextsOrUrls, "operatingSystem": validateTexts, "permissions": validateTexts, "processorRequirements": validateTexts, "releaseNotes": validateTextsOrUrls, "softwareHelp": validateCreativeWorks, "softwareRequirements": noValidation, // TODO: validate SoftwareSourceCode "softwareVersion": validateText, // TODO? "storageRequirements": validateTextsOrUrls, "supportingData": noValidation, // TODO "author": validateActors, "citation": validateCreativeWorks, // TODO "contributor": validateActors, "copyrightHolder": validateActors, "copyrightYear": validateNumbers, "creator": validateActors, // TODO: still in codemeta 2.0, but removed from master "dateCreated": validateDate, "dateModified": validateDate, "datePublished": validateDate, "editor": validatePersons, "encoding": noValidation, "fileFormat": validateTextsOrUrls, "funder": validateActors, // TODO: may be other types "keywords": validateTexts, "license": validateCreativeWorks, "producer": validateActors, "provider": validateActors, "publisher": validateActors, "sponsor": validateActors, "version": validateNumberOrText, "isAccessibleForFree": validateBoolean, "isPartOf": validateCreativeWorks, "hasPart": validateCreativeWorks, "position": noValidation, "identifier": noValidation, // TODO "description": validateText, "name": validateText, "sameAs": validateUrls, "url": validateUrls, "relatedLink": validateUrls, "softwareSuggestions": noValidation, // TODO: validate SoftwareSourceCode "maintainer": validateActors, "contIntegration": validateUrls, "buildInstructions": validateUrls, "developmentStatus": validateText, // TODO: use only repostatus strings? "embargoDate": validateDate, "funding": validateText, "issueTracker": validateUrls, "referencePublication": noValidation, // TODO? "readme": validateUrls, }; var creativeWorkFieldValidators = { "@id": validateUrl, "id": validateUrl, "author": validateActors, "citation": validateCreativeWorks, // TODO "contributor": validateActors, "copyrightHolder": validateActors, "copyrightYear": validateNumbers, "creator": validateActors, // TODO: still in codemeta 2.0, but removed from master "dateCreated": validateDate, "dateModified": validateDate, "datePublished": validateDate, "editor": validatePersons, "encoding": noValidation, "funder": validateActors, // TODO: may be other types "keywords": validateTexts, "license": validateCreativeWorks, "producer": validateActors, "provider": validateActors, "publisher": validateActors, "sponsor": validateActors, "version": validateNumberOrText, "isAccessibleForFree": validateBoolean, "isPartOf": validateCreativeWorks, "hasPart": validateCreativeWorks, "position": noValidation, "identifier": noValidation, // TODO "description": validateText, "name": validateText, "sameAs": validateUrls, "url": validateUrls, }; var personFieldValidators = { "@id": validateUrl, "id": validateUrl, "givenName": validateText, "familyName": validateText, "email": validateText, "affiliation": validateOrganizations, "identifier": validateUrls, "name": validateText, // TODO: this is technically valid, but should be allowed here? "url": validateUrls, }; var organizationFieldValidators = { "@id": validateUrl, "id": validateUrl, "email": validateText, "identifier": validateUrls, "name": validateText, "address": validateText, "sponsor": validateActors, "funder": validateActors, // TODO: may be other types "isPartOf": validateOrganizations, "url": validateUrls, // TODO: add more? };