@natlibfi/marc-record-validators-melinda 3.2.0 → 3.3.14-alpha.3

package/src/punctuation/index.js

@@ -0,0 +1,292 @@
+import {MarcRecord} from '@natlibfi/marc-record';
+import createDebug from 'debug';
+import {autRules, bibRules} from './rules';
+
+const debug = createDebug('@natlibfi/marc-record-validator-melinda/punctuation');
+
+function cloneDeep(field) {
+  const r = new MarcRecord();
+  r.appendField(field);
+  return r.get(field.tag)?.[0];
+}
+
+export default function () {
+  function readPunctuationRulesFromJSON(recordType) {
+    const json = getRules(recordType);
+    const rules = json.filter(row => row.selector !== '').map(row => {
+      const {selector, namePortion, description, portion, preceedingPunctuation, exceptions} = row;
+      return {
+        selector: new RegExp(selector.replace(/X/ug, '.'), 'u'),
+        namePortion: namePortion.replace(/\$/ug, '').trim(),
+        description, portion, preceedingPunctuation, exceptions
+      };
+    });
+
+    return rules;
+  }
+
+  function getRules(recordType) {
+    if (recordType === 'z') {
+      return autRules;
+    }
+
+    return bibRules;
+  }
+
+  function fieldToString(field) {
+    if (field && field.subfields) {
+      const ind1 = field.ind1 || ' ';
+      const ind2 = field.ind2 || ' ';
+      const subfields = field.subfields.map(subfield => `‡${subfield.code}${subfield.value}`).join('');
+      return `${field.tag} ${ind1}${ind2} ${subfields}`;
+    }
+
+    return `${field.tag}    ${field.value}`;
+  }
+
+  function validateField(recordType = 'a') {
+    return function (element) {
+      const testField = cloneDeep(element);
+      debug(`Original field: ${JSON.stringify(element)}`);
+      const punctuated = punctuateField(testField, recordType);
+      debug(`Punctuation result: ${JSON.stringify(punctuated)}`);
+      if (!punctuated) {
+        debug('No punctuation result -> true');
+        return true;
+      }
+
+      if (MarcRecord.isEqual(punctuated, element)) {
+        debug(`Original field (element): ${JSON.stringify(element)}`);
+        debug('Punctuation result equals original field');
+        return true;
+      }
+
+      return false;
+    };
+  }
+
+  function punctuateField(field, recordType) {
+    const rules = readPunctuationRulesFromJSON(recordType);
+    debug(`Handling field ${field.tag}`);
+    debug(`Field contents: ${fieldToString(field)}`);
+    const rulesForField = getRulesForField(field.tag, rules);
+    if (rulesForField.length === 0) {
+      debug(`No matching rules for field ${field.tag}`);
+      return;
+    }
+
+    let currentPortion; // eslint-disable-line functional/no-let
+    let preceedingField; // eslint-disable-line functional/no-let
+    let inNamePortion = true; // eslint-disable-line functional/no-let
+
+    debug(`Field subfields: ${field.subfields.map(sub => sub.code)}`);
+    debug(`Field portions: ${field.subfields.map(sub => getPortion(sub, rulesForField))}`);
+
+    field.subfields.forEach(subfield => {
+      debug(`Handling subfield ${subfield.code}`);
+      let portion = getPortion(subfield, rulesForField); // eslint-disable-line functional/no-let
+
+      if (portion === false) {
+        return;
+      }
+
+      if (portion === 'CF' || portion === 'NC') {
+        return;
+      }
+
+      if (inNamePortion && portion.includes('T', 'S')) { // eslint-disable-line functional/no-conditional-statements
+        debug(`Portion changed to ${portion}. Not in name portion anymore`);
+        inNamePortion = false;
+      }
+
+      if (inNamePortion && portion === 'NT') { // eslint-disable-line functional/no-conditional-statements
+        portion = 'N';
+      }
+
+      if (!inNamePortion && portion === 'NT') { // eslint-disable-line functional/no-conditional-statements
+        portion = 'T';
+      }
+
+      debug(`Current portion is ${portion}.`);
+
+      if (currentPortion) {
+        if (currentPortion === portion) { // eslint-disable-line functional/no-conditional-statements
+          debug(`Current stayed as ${portion}. Adding punctuation for subfield.`);
+          addSubfieldPunctuation(preceedingField, subfield, rulesForField);
+        } else {
+          debug(`Current portion changed to ${portion}.`);
+          if (portion !== 'S') { // eslint-disable-line functional/no-conditional-statements
+            debug('Adding punctuation for portion.');
+            addNamePortionPunctuation(preceedingField);
+          }
+        }
+      }
+
+      currentPortion = portion;
+      preceedingField = subfield;
+    });
+
+    if (recordType !== 'z') { // eslint-disable-line functional/no-conditional-statements
+      addNamePortionPunctuation(preceedingField);
+    }
+
+    debug(`After punctuation: ${fieldToString(field)}`);
+
+    return field;
+  }
+
+  function getRulesForField(tag, rules) {
+    return rules.filter(rule => rule.selector.test(tag));
+  }
+
+  function getPortion(subfield, rules) {
+    debug(`Looking for namePortion for ${subfield.code}`);
+    const [portion] = rules.filter(rule => rule.namePortion === subfield.code).map(rule => rule.portion);
+
+    if (portion === undefined) {
+      return false;
+    }
+
+    return portion.toUpperCase();
+  }
+
+  function addNamePortionPunctuation(preceedingSubfield) {
+    const subfieldContainsPunctuation = (/[?")\].\-!,]$/u).test(preceedingSubfield.value);
+    if (!subfieldContainsPunctuation) { // eslint-disable-line functional/no-conditional-statements
+      const nextValue = `${preceedingSubfield.value}.`;
+      debug(`Updated subfield ${preceedingSubfield.code} from '${preceedingSubfield.value}' to '${nextValue}'`);
+      preceedingSubfield.value = nextValue; // eslint-disable-line functional/immutable-data
+    }
+  }
+
+  function addSubfieldPunctuation(preceedingSubfield, currentSubfield, rules) {
+    const punctType = getPrecedingPunctuation(currentSubfield, rules);
+    const exceptionsFunctions = getExceptions(currentSubfield, rules);
+
+    const isExceptionCase = exceptionsFunctions.some(fn => fn(preceedingSubfield));
+
+    if (isExceptionCase) {
+      return;
+    }
+
+    const endsInPunctuation = (/[?")\]\-!,]$/u).test(preceedingSubfield.value);
+    debug(`addSubfieldPunctuation -- punctType: ${punctType} endsInPunctuation: ${endsInPunctuation}`);
+
+    if (!endsInPunctuation) {
+      if (punctType === 'PERIOD' && !(/\.$/u).test(preceedingSubfield.value)) { // eslint-disable-line functional/no-conditional-statements
+        const nextValue = `${preceedingSubfield.value}.`;
+        debug(`Updated subfield ${preceedingSubfield.code} from '${preceedingSubfield.value}' to '${nextValue}'`);
+        preceedingSubfield.value = nextValue; // eslint-disable-line functional/immutable-data
+      }
+    }
+
+    if (punctType === 'COMMA') {
+      if (!(/,$/u).test(preceedingSubfield.value)) {
+        if (!(/^[[(]/u).test(currentSubfield.value)) { // eslint-disable-line functional/no-conditional-statements
+          const nextValue = `${preceedingSubfield.value},`;
+          debug(`Updated subfield ${preceedingSubfield.code} from '${preceedingSubfield.value}' to '${nextValue}'`);
+          preceedingSubfield.value = nextValue; // eslint-disable-line functional/immutable-data
+        }
+      }
+    }
+
+    if (punctType === 'COND_COMMA') {
+      if (!(/[-,]$/u).test(preceedingSubfield.value)) { // eslint-disable-line functional/no-conditional-statements
+        const nextValue = `${preceedingSubfield.value},`;
+        debug(`Updated subfield ${preceedingSubfield.code} from '${preceedingSubfield.value}' to '${nextValue}'`);
+        preceedingSubfield.value = nextValue; // eslint-disable-line functional/immutable-data
+      }
+    }
+
+    if (punctType === 'SPACECOLON') {
+      if (!(/:$/u).test(preceedingSubfield.value)) { // eslint-disable-line functional/no-conditional-statements
+        const nextValue = `${preceedingSubfield.value} :`;
+        debug(`Updated subfield ${preceedingSubfield.code} from '${preceedingSubfield.value}' to '${nextValue}'`);
+        preceedingSubfield.value = nextValue; // eslint-disable-line functional/immutable-data
+      }
+      if ((/[^ ]:$/u).test(preceedingSubfield.value)) { // eslint-disable-line functional/no-conditional-statements
+        const nextValue = `${preceedingSubfield.value.slice(0, -1)} :`;
+        debug(`Updated subfield ${preceedingSubfield.code} from '${preceedingSubfield.value}' to '${nextValue}'`);
+        preceedingSubfield.value = nextValue; // eslint-disable-line functional/immutable-data
+      }
+
+    }
+
+    debug('addSubfieldPunctuation -- end');
+  }
+
+  function getPrecedingPunctuation(subfield, rules) {
+    const [punct] = rules.filter(rule => rule.namePortion === subfield.code).map(rule => rule.preceedingPunctuation);
+
+    if (punct === undefined) {
+      throw new Error(`Unknown subfield code ${subfield.code}`);
+    }
+
+    return punct.toUpperCase();
+  }
+
+  function getExceptions(subfield, rules) {
+    const [exception] = rules.filter(rule => rule.namePortion === subfield.code).map(rule => parseExceptions(rule.exceptions));
+
+    if (exception === undefined) {
+      throw new Error(`Unknown subfield code ${subfield.code}`);
+    }
+
+    return exception;
+  }
+
+  function parseExceptions(expectionsString) {
+    const exceptionRules = expectionsString.split('\n');
+    const exceptionFuncs = [];
+
+    exceptionRules.forEach(exceptionRule => {
+      const match = (/- (.*) if preceded by (.*)/u).exec(exceptionRule); // eslint-disable-line prefer-named-capture-group
+      if (match) { // eslint-disable-line functional/no-conditional-statements
+        const [, type, preceededCode] = match;
+        const normalizedType = type.trim().toUpperCase().trim();
+        const normalizedCode = preceededCode.replace(/\$/ug, '').trim();
+        exceptionFuncs.push(ifPrecededByException(normalizedCode, normalizedType)); // eslint-disable-line functional/immutable-data
+      }
+    });
+
+    return exceptionFuncs;
+  }
+
+  function ifPrecededByException(code, type) {
+    return preceedingSubfield => {
+      if (code === preceedingSubfield.code) {
+        debug(`Adding ${type} to ${preceedingSubfield.code}`);
+        if (type === 'SEMICOLON' && !(/;$/u).test(preceedingSubfield.value)) { // eslint-disable-line functional/no-conditional-statements
+          const nextValue = `${preceedingSubfield.value} ;`;
+          debug(`Updated subfield ${preceedingSubfield.code} from '${preceedingSubfield.value}' to '${nextValue}'`);
+          preceedingSubfield.value = nextValue; // eslint-disable-line functional/immutable-data
+        }
+
+        if (type === 'COLON' && !(/:$/u).test(preceedingSubfield.value)) { // eslint-disable-line functional/no-conditional-statements
+          const nextValue = `${preceedingSubfield.value} :`;
+          debug(`Updated subfield ${preceedingSubfield.code} from '${preceedingSubfield.value}' to '${nextValue}'`);
+          preceedingSubfield.value = nextValue; // eslint-disable-line functional/immutable-data
+        }
+
+        return true;
+      }
+
+      return false;
+    };
+  }
+
+  function validate(record) {
+    return {valid: record.fields.every(validateField(record.leader[6]))};
+  }
+
+  function fix(record) {
+    record.fields.map(field => punctuateField(field, record.leader[6]));
+    return true;
+  }
+
+  return {
+    description: 'Fixes punctuation of fields',
+    validate,
+    fix
+  };
+}

package/src/punctuation/rules/aut.js

@@ -0,0 +1,372 @@
+const autRules =
+[
+  {
+    selector: '[1457]00',
+    namePortion: '$a',
+    description: 'Personal name (NR)',
+    portion: 'N',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]10',
+    namePortion: '$a',
+    description: 'Corporate name or jurisdiction name as entry element (NR)',
+    portion: 'N',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]11',
+    namePortion: '$a',
+    description: 'Meeting name or jurisdiction name as entry element (NR)',
+    portion: 'N',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]00',
+    namePortion: '$b',
+    description: 'Numeration (NR)',
+    portion: 'N',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]10',
+    namePortion: '$b',
+    description: 'Subordinate unit (R)',
+    portion: 'N',
+    preceedingPunctuation: 'period',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]00',
+    namePortion: '$c',
+    description: 'Titles and words associated with a name (R)',
+    portion: 'N',
+    preceedingPunctuation: 'comma',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](10|11)',
+    namePortion: '$c',
+    description: 'Location of meeting (R)',
+    portion: 'N',
+    preceedingPunctuation: 'comma',
+    exceptions: '- colon if preceded by $d\n- semicolon if preceded by $c'
+  },
+  {
+    selector: '[1457]00',
+    namePortion: '$d',
+    description: 'Dates associated with a name (NR)',
+    portion: 'N',
+    preceedingPunctuation: 'comma',
+    exceptions: '- colon if preceded by $n'
+  },
+  {
+    selector: '[1457]11',
+    namePortion: '$d',
+    description: 'Date of meeting (NR)',
+    portion: 'N',
+    preceedingPunctuation: '',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10)',
+    namePortion: '$e',
+    description: 'Relator term (R)',
+    portion: 'N',
+    preceedingPunctuation: 'cond_comma',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]11',
+    namePortion: '$e',
+    description: 'Subordinate unit (R)',
+    portion: 'N',
+    preceedingPunctuation: 'period',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]11',
+    namePortion: '$j',
+    description: 'Relator term (R)',
+    portion: 'N',
+    preceedingPunctuation: 'comma',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]00',
+    namePortion: '$j',
+    description: 'Attribution qualifier (R)',
+    portion: 'N',
+    preceedingPunctuation: 'comma',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]00',
+    namePortion: '$q',
+    description: 'Fuller form of name (NR)',
+    portion: 'N',
+    preceedingPunctuation: 'comma',
+    exceptions: ''
+  },
+  {
+    selector: '[1457]11',
+    namePortion: '$q',
+    description: 'Name of meeting following jurisdiction name entry element (NR)',
+    portion: 'N',
+    preceedingPunctuation: '',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|11)',
+    namePortion: '$u',
+    description: 'Affiliation (NR)',
+    portion: 'N',
+    preceedingPunctuation: 'period',
+    exceptions: '?'
+  },
+  {
+    selector: '[1457]10',
+    namePortion: '$u',
+    description: 'Affiliation (NR) or Address',
+    portion: 'N',
+    preceedingPunctuation: 'period',
+    exceptions: '- period, if address'
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$4',
+    description: 'Relator code (R)',
+    portion: 'NC',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$f',
+    description: 'Date of a work (NR)',
+    portion: 'T',
+    preceedingPunctuation: 'period',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$h',
+    description: 'Medium (NR)',
+    portion: 'T',
+    preceedingPunctuation: 'period',
+    exceptions: ''
+  },
+  {
+    selector: '[457](00|10|11)',
+    namePortion: '$i',
+    description: 'Relationship information (R)',
+    portion: 'cf',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$k',
+    description: 'Form subheading (R)',
+    portion: 'T',
+    preceedingPunctuation: 'period',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$l',
+    description: 'Language of a work (NR)',
+    portion: 'T',
+    preceedingPunctuation: 'comma / period',
+    exceptions: '- Finnish MARC21 comma,\n- MARC21 period'
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$m',
+    description: 'Medium of performance for music',
+    portion: 'T',
+    preceedingPunctuation: 'comma',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$n',
+    description: 'Number of part/section of a work (R)',
+    portion: 'T',
+    preceedingPunctuation: 'comma, period',
+    exceptions: ' - depends on data?\n- period, if preceded by $k, $m, $t\n- comma, if preceded by $b, $m'
+  },
+  {
+    selector: '[1457](00|10)',
+    namePortion: '$o',
+    description: 'Arranged statement for music (NR)',
+    portion: 'T',
+    preceedingPunctuation: 'semicolon',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$p',
+    description: 'Name of part/section of a work (R)',
+    portion: 'T',
+    preceedingPunctuation: 'comma, period, none',
+    exceptions: '- depends on data?\n- period if preceded by $t, $n, $p\n- comma, if preceded by $n\n- none, if preceded by $k'
+  },
+  {
+    selector: '[1457](00|10)',
+    namePortion: '$r',
+    description: 'Key for music (NR)',
+    portion: 'T',
+    preceedingPunctuation: 'comma',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$s',
+    description: 'Version (NR)',
+    portion: 'T',
+    preceedingPunctuation: 'period',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$t',
+    description: 'Title of a work (NR)',
+    portion: 'T',
+    preceedingPunctuation: 'none',
+    exceptions: '- usually first in section -> period'
+  },
+  {
+    selector: '[1457](10|11)',
+    namePortion: '$d',
+    description: 'Date of meeting or treaty signing (R)',
+    portion: 'NT',
+    preceedingPunctuation: '',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$g',
+    description: 'Miscellaneous information (NR)',
+    portion: 'NT',
+    preceedingPunctuation: 'comma',
+    exceptions: '- colon, if preceded by $d'
+  },
+  {
+    selector: '[1457]10',
+    namePortion: '$g',
+    description: 'Miscellaneous information (NR), other party to treaties.',
+    portion: 'NT',
+    preceedingPunctuation: 'comma',
+    exceptions: '- colon, if preceded by $d\n- period, if preceded by $t'
+  },
+  {
+    selector: '[1457](10|11)',
+    namePortion: '$n',
+    description: 'Number of part/section/meeting (R)',
+    portion: 'NT',
+    preceedingPunctuation: '',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$v',
+    description: 'Form subdivision (R)   [600]',
+    portion: 'S',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$x',
+    description: 'General subdivision (R)   [600]',
+    portion: 'S',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$y',
+    description: 'Chronological subdivision (R)   [600]',
+    portion: 'S',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$z',
+    description: 'Geographic subdivision (R)   [600]',
+    portion: 'S',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[457](00|10|11)',
+    namePortion: '$w',
+    description: '',
+    portion: 'cf',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$0',
+    description: 'Authority record control number (R)',
+    portion: 'cf',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '7(00|10|11)',
+    namePortion: '$2',
+    description: 'Source of heading or term (NR)',
+    portion: 'cf',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$5',
+    description: 'Institution to which field applies (NR)',
+    portion: 'cf',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$6',
+    description: 'Linkage (NR)',
+    portion: 'cf',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$7',
+    description: 'Control subfield (NR)',
+    portion: 'cf',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$8',
+    description: 'Field link and sequence number (R)',
+    portion: 'cf',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  },
+  {
+    selector: '[1457](00|10|11)',
+    namePortion: '$9',
+    description: 'Local control subfield',
+    portion: 'cf',
+    preceedingPunctuation: 'none',
+    exceptions: ''
+  }
+];
+export default autRules;