@natlibfi/marc-record-validators-melinda 10.16.0-alpha.1 → 10.16.0-alpha.2

package/dist/field-008-18-34-character-groups.js

@@ -0,0 +1,189 @@
+"use strict";
+
+Object.defineProperty(exports, "__esModule", {
+  value: true
+});
+exports.default = _default;
+exports.justifyAndSortField008CharacterGroups = justifyAndSortField008CharacterGroups;
+var _clone = _interopRequireDefault(require("clone"));
+var _utils = require("./utils");
+function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
+//import createDebugLogger from 'debug';
+
+// Author(s): Nicholas Volk
+// NB! CR 008/24 vs 008/25-27 is not supported yet!
+
+function _default() {
+  return {
+    description: 'Justify left and sort character groups within 008/18-24',
+    validate,
+    fix
+  };
+  function fix(record) {
+    const typeOfMaterial = record.getTypeOfMaterial();
+    record.fields.forEach(field => {
+      justifyAndSortField008CharacterGroups(field, typeOfMaterial);
+    });
+    // Fix always succeeds (even when it really does not):
+    const res = {
+      message: [],
+      fix: [],
+      valid: true
+    };
+    return res;
+  }
+  function validate(record) {
+    const res = {
+      message: []
+    };
+    const typeOfMaterial = record.getTypeOfMaterial();
+    record.fields?.forEach(field => {
+      validateField(field, res, typeOfMaterial);
+    });
+    res.valid = !(res.message.length >= 1); // eslint-disable-line functional/immutable-data
+    return res;
+  }
+  function validateField(field, res, typeOfMaterial) {
+    if (field.tag !== '008') {
+      // Optimize code a bit...
+      return;
+    }
+    const orig = (0, _utils.fieldToString)(field);
+    const normalizedField = justifyAndSortField008CharacterGroups((0, _clone.default)(field), typeOfMaterial);
+    const mod = (0, _utils.fieldToString)(normalizedField);
+    if (orig !== mod) {
+      // Fail as the input is "broken"/"crap"/sumthing
+      res.message.push(`TODO: '${orig}' => '${mod}'`); // eslint-disable-line functional/immutable-data
+      return;
+    }
+    return;
+  }
+}
+
+// Should we add legal values?
+const characterGroups = [{
+  type: 'BK',
+  start: 18,
+  end: 21,
+  sort: true,
+  name: 'illustrations'
+}, {
+  type: 'BK',
+  start: 24,
+  end: 27,
+  sort: true,
+  name: 'nature of contents'
+},
+// English doc does not explicitly mention alphabetical sorting... Finnish does.
+{
+  type: 'CR',
+  start: 25,
+  end: 27,
+  sort: true,
+  name: 'nature of contents'
+},
+// NB! 24 vs 25-27 logic needs to be implemented separately
+{
+  type: 'MP',
+  start: 18,
+  end: 21,
+  sort: false,
+  name: 'relief'
+},
+// Order of importance!
+{
+  type: 'MP',
+  start: 33,
+  end: 34,
+  sort: false,
+  name: 'special format of characteristics'
+},
+// Order of importance!
+{
+  type: 'MU',
+  start: 24,
+  end: 29,
+  sort: true,
+  name: 'accompanying material'
+}, {
+  type: 'MU',
+  start: 30,
+  end: 31,
+  sort: true,
+  name: 'literary text for sound recordings'
+}];
+const BIG_BAD_VALUE = 999999999;
+function processCharacterGroup(field, group) {
+  const originalContent = field.value.substring(group.start, group.end + 1);
+  const content = removeDuplicateValues(fixBlanks(originalContent));
+  //console.info(`008/${group.start}-${group.end}: '${originalContent}'`); // eslint-disable-line no-console
+  const charArray = content.split('');
+  charArray.sort(function (a, b) {
+    // eslint-disable-line functional/immutable-data, prefer-arrow-callback
+    return scoreChar(a) - scoreChar(b);
+  });
+  const newContent = charArray.join('');
+  if (originalContent === newContent) {
+    return;
+  }
+
+  //console.info(`'${fieldToString(field)}' =>`); // eslint-disable-line no-console
+
+  field.value = `${field.value.substring(0, group.start)}${newContent}${field.value.substring(group.end + 1)}`; // eslint-disable-line functional/immutable-data
+  //console.info(`'${fieldToString(field)}'`); // eslint-disable-line no-console
+
+  function fixBlanks(str) {
+    if (str.includes('|') && str.match(/[^ |]/u)) {
+      return str.replaceAll('|', ' ');
+    }
+    return str;
+  }
+  function scoreChar(c) {
+    if (c === '|' || c === ' ') {
+      return BIG_BAD_VALUE; // Max value, these should code last
+    }
+    if (!group.sort) {
+      // more meaningful comes first: keep the original order
+      return 1;
+    }
+    const asciiCode = c.charCodeAt(0);
+    // a-z get values 1-26:
+    if (asciiCode >= 97 && asciiCode <= 122) {
+      return asciiCode - 96;
+    }
+    // 0-9 get values 100-109
+    if (asciiCode >= 48 && asciiCode <= 57) {
+      return asciiCode + 52;
+    }
+    // Others (=crap) return something between '9' and BIG BAD VALUE
+    return asciiCode + 200;
+  }
+}
+function justifyAndSortField008CharacterGroups(field, typeOfMaterial) {
+  if (field.tag !== '008' || field.subfields) {
+    return field;
+  }
+
+  //console.info(typeOfMaterial); // eslint-disable-line no-console
+
+  const relevantCharacterGroups = characterGroups.filter(gr => gr.type === typeOfMaterial);
+  relevantCharacterGroups.forEach(group => processCharacterGroup(field, group));
+
+  //justifyField008CharacterGroups(field, typeOfMaterial); // Oops: also sorts...
+
+  // NB! add value # and | normalizations
+  //fixBlanks(field, typeOfMaterial);
+
+  return field;
+}
+function removeDuplicateValues(str) {
+  const arr = str.split('');
+  // Take only the first instance of a proper value-carrying character
+  const reducedStr = arr.filter((c, i) => c === ' ' || c === '|' || arr.indexOf(c) === i).join('');
+  //console.info(`I: '${str}'`); // eslint-disable-line no-console
+  //console.info(`M: '${reducedStr}'`); // eslint-disable-line no-console
+  const output = `${reducedStr}${' '.repeat(str.length - reducedStr.length)}`; // Had some weird trouble with str.padEnd(n)
+  //console.info(`M: '${output}'`); // eslint-disable-line no-console
+  return output;
+}
+//# sourceMappingURL=field-008-18-34-character-groups.js.map

package/dist/field-008-18-34-character-groups.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"field-008-18-34-character-groups.js","names":["_clone","_interopRequireDefault","require","_utils","obj","__esModule","default","_default","description","validate","fix","record","typeOfMaterial","getTypeOfMaterial","fields","forEach","field","justifyAndSortField008CharacterGroups","res","message","valid","validateField","length","tag","orig","fieldToString","normalizedField","clone","mod","push","characterGroups","type","start","end","sort","name","BIG_BAD_VALUE","processCharacterGroup","group","originalContent","value","substring","content","removeDuplicateValues","fixBlanks","charArray","split","a","b","scoreChar","newContent","join","str","includes","match","replaceAll","c","asciiCode","charCodeAt","subfields","relevantCharacterGroups","filter","gr","arr","reducedStr","i","indexOf","output","repeat"],"sources":["../src/field-008-18-34-character-groups.js"],"sourcesContent":["//import createDebugLogger from 'debug';\nimport clone from 'clone';\nimport {fieldToString} from './utils';\n// Author(s): Nicholas Volk\n// NB! CR 008/24 vs 008/25-27 is not supported yet!\n\nexport default function () {\n\n  return {\n    description: 'Justify left and sort character groups within 008/18-24',\n    validate, fix\n  };\n\n  function fix(record) {\n    const typeOfMaterial = record.getTypeOfMaterial();\n    record.fields.forEach(field => {\n      justifyAndSortField008CharacterGroups(field, typeOfMaterial);\n    });\n    // Fix always succeeds (even when it really does not):\n    const res = {message: [], fix: [], valid: true};\n    return res;\n  }\n\n  function validate(record) {\n    const res = {message: []};\n\n    const typeOfMaterial = record.getTypeOfMaterial();\n\n    record.fields?.forEach(field => {\n      validateField(field, res, typeOfMaterial);\n    });\n\n    res.valid = !(res.message.length >= 1); // eslint-disable-line functional/immutable-data\n    return res;\n  }\n\n  function validateField(field, res, typeOfMaterial) {\n    if (field.tag !== '008') { // Optimize code a bit...\n      return;\n    }\n    const orig = fieldToString(field);\n\n    const normalizedField = justifyAndSortField008CharacterGroups(clone(field), typeOfMaterial);\n    const mod = fieldToString(normalizedField);\n    if (orig !== mod) { // Fail as the input is \"broken\"/\"crap\"/sumthing\n      res.message.push(`TODO: '${orig}' => '${mod}'`); // eslint-disable-line functional/immutable-data\n      return;\n    }\n    return;\n  }\n}\n\n// Should we add legal values?\nconst characterGroups = [\n  {type: 'BK', start: 18, end: 21, sort: true, name: 'illustrations'},\n  {type: 'BK', start: 24, end: 27, sort: true, name: 'nature of contents'}, // English doc does not explicitly mention alphabetical sorting... Finnish does.\n  {type: 'CR', start: 25, end: 27, sort: true, name: 'nature of contents'}, // NB! 24 vs 25-27 logic needs to be implemented separately\n  {type: 'MP', start: 18, end: 21, sort: false, name: 'relief'}, // Order of importance!\n  {type: 'MP', start: 33, end: 34, sort: false, name: 'special format of characteristics'}, // Order of importance!\n  {type: 'MU', start: 24, end: 29, sort: true, name: 'accompanying material'},\n  {type: 'MU', start: 30, end: 31, sort: true, name: 'literary text for sound recordings'}\n];\n\nconst BIG_BAD_VALUE = 999999999;\n\nfunction processCharacterGroup(field, group) {\n  const originalContent = field.value.substring(group.start, group.end + 1);\n  const content = removeDuplicateValues(fixBlanks(originalContent));\n  //console.info(`008/${group.start}-${group.end}: '${originalContent}'`); // eslint-disable-line no-console\n  const charArray = content.split('');\n\n  charArray.sort(function(a, b) { // eslint-disable-line functional/immutable-data, prefer-arrow-callback\n    return scoreChar(a) - scoreChar(b);\n  });\n\n  const newContent = charArray.join('');\n  if (originalContent === newContent) {\n    return;\n  }\n\n  //console.info(`'${fieldToString(field)}' =>`); // eslint-disable-line no-console\n\n  field.value = `${field.value.substring(0, group.start)}${newContent}${field.value.substring(group.end + 1)}`; // eslint-disable-line functional/immutable-data\n  //console.info(`'${fieldToString(field)}'`); // eslint-disable-line no-console\n\n  function fixBlanks(str) {\n    if (str.includes('|') && str.match(/[^ |]/u)) {\n      return str.replaceAll('|', ' ');\n    }\n    return str;\n  }\n\n  function scoreChar(c) {\n    if (c === '|' || c === ' ') {\n      return BIG_BAD_VALUE; // Max value, these should code last\n    }\n    if (!group.sort) { // more meaningful comes first: keep the original order\n      return 1;\n    }\n    const asciiCode = c.charCodeAt(0);\n    // a-z get values 1-26:\n    if (asciiCode >= 97 && asciiCode <= 122) {\n      return asciiCode - 96;\n    }\n    // 0-9 get values 100-109\n    if (asciiCode >= 48 && asciiCode <= 57) {\n      return asciiCode + 52;\n    }\n    // Others (=crap) return something between '9' and BIG BAD VALUE\n    return asciiCode + 200;\n  }\n}\n\nexport function justifyAndSortField008CharacterGroups(field, typeOfMaterial) {\n  if (field.tag !== '008' || field.subfields) {\n    return field;\n  }\n\n  //console.info(typeOfMaterial); // eslint-disable-line no-console\n\n  const relevantCharacterGroups = characterGroups.filter(gr => gr.type === typeOfMaterial);\n\n  relevantCharacterGroups.forEach(group => processCharacterGroup(field, group));\n\n  //justifyField008CharacterGroups(field, typeOfMaterial); // Oops: also sorts...\n\n  // NB! add value # and | normalizations\n  //fixBlanks(field, typeOfMaterial);\n\n  return field;\n}\n\nfunction removeDuplicateValues(str) {\n  const arr = str.split('');\n  // Take only the first instance of a proper value-carrying character\n  const reducedStr = arr.filter((c, i) => c === ' ' || c === '|' || arr.indexOf(c) === i).join('');\n  //console.info(`I: '${str}'`); // eslint-disable-line no-console\n  //console.info(`M: '${reducedStr}'`); // eslint-disable-line no-console\n  const output = `${reducedStr}${' '.repeat(str.length - reducedStr.length)}`; // Had some weird trouble with str.padEnd(n)\n  //console.info(`M: '${output}'`); // eslint-disable-line no-console\n  return output;\n}\n"],"mappings":";;;;;;;AACA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,MAAA,GAAAD,OAAA;AAAsC,SAAAD,uBAAAG,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AAFtC;;AAGA;AACA;;AAEe,SAAAG,SAAA,EAAY;EAEzB,OAAO;IACLC,WAAW,EAAE,yDAAyD;IACtEC,QAAQ;IAAEC;EACZ,CAAC;EAED,SAASA,GAAGA,CAACC,MAAM,EAAE;IACnB,MAAMC,cAAc,GAAGD,MAAM,CAACE,iBAAiB,CAAC,CAAC;IACjDF,MAAM,CAACG,MAAM,CAACC,OAAO,CAACC,KAAK,IAAI;MAC7BC,qCAAqC,CAACD,KAAK,EAAEJ,cAAc,CAAC;IAC9D,CAAC,CAAC;IACF;IACA,MAAMM,GAAG,GAAG;MAACC,OAAO,EAAE,EAAE;MAAET,GAAG,EAAE,EAAE;MAAEU,KAAK,EAAE;IAAI,CAAC;IAC/C,OAAOF,GAAG;EACZ;EAEA,SAAST,QAAQA,CAACE,MAAM,EAAE;IACxB,MAAMO,GAAG,GAAG;MAACC,OAAO,EAAE;IAAE,CAAC;IAEzB,MAAMP,cAAc,GAAGD,MAAM,CAACE,iBAAiB,CAAC,CAAC;IAEjDF,MAAM,CAACG,MAAM,EAAEC,OAAO,CAACC,KAAK,IAAI;MAC9BK,aAAa,CAACL,KAAK,EAAEE,GAAG,EAAEN,cAAc,CAAC;IAC3C,CAAC,CAAC;IAEFM,GAAG,CAACE,KAAK,GAAG,EAAEF,GAAG,CAACC,OAAO,CAACG,MAAM,IAAI,CAAC,CAAC,CAAC,CAAC;IACxC,OAAOJ,GAAG;EACZ;EAEA,SAASG,aAAaA,CAACL,KAAK,EAAEE,GAAG,EAAEN,cAAc,EAAE;IACjD,IAAII,KAAK,CAACO,GAAG,KAAK,KAAK,EAAE;MAAE;MACzB;IACF;IACA,MAAMC,IAAI,GAAG,IAAAC,oBAAa,EAACT,KAAK,CAAC;IAEjC,MAAMU,eAAe,GAAGT,qCAAqC,CAAC,IAAAU,cAAK,EAACX,KAAK,CAAC,EAAEJ,cAAc,CAAC;IAC3F,MAAMgB,GAAG,GAAG,IAAAH,oBAAa,EAACC,eAAe,CAAC;IAC1C,IAAIF,IAAI,KAAKI,GAAG,EAAE;MAAE;MAClBV,GAAG,CAACC,OAAO,CAACU,IAAI,CAAE,UAASL,IAAK,SAAQI,GAAI,GAAE,CAAC,CAAC,CAAC;MACjD;IACF;IACA;EACF;AACF;;AAEA;AACA,MAAME,eAAe,GAAG,CACtB;EAACC,IAAI,EAAE,IAAI;EAAEC,KAAK,EAAE,EAAE;EAAEC,GAAG,EAAE,EAAE;EAAEC,IAAI,EAAE,IAAI;EAAEC,IAAI,EAAE;AAAe,CAAC,EACnE;EAACJ,IAAI,EAAE,IAAI;EAAEC,KAAK,EAAE,EAAE;EAAEC,GAAG,EAAE,EAAE;EAAEC,IAAI,EAAE,IAAI;EAAEC,IAAI,EAAE;AAAoB,CAAC;AAAE;AAC1E;EAACJ,IAAI,EAAE,IAAI;EAAEC,KAAK,EAAE,EAAE;EAAEC,GAAG,EAAE,EAAE;EAAEC,IAAI,EAAE,IAAI;EAAEC,IAAI,EAAE;AAAoB,CAAC;AAAE;AAC1E;EAACJ,IAAI,EAAE,IAAI;EAAEC,KAAK,EAAE,EAAE;EAAEC,GAAG,EAAE,EAAE;EAAEC,IAAI,EAAE,KAAK;EAAEC,IAAI,EAAE;AAAQ,CAAC;AAAE;AAC/D;EAACJ,IAAI,EAAE,IAAI;EAAEC,KAAK,EAAE,EAAE;EAAEC,GAAG,EAAE,EAAE;EAAEC,IAAI,EAAE,KAAK;EAAEC,IAAI,EAAE;AAAmC,CAAC;AAAE;AAC1F;EAACJ,IAAI,EAAE,IAAI;EAAEC,KAAK,EAAE,EAAE;EAAEC,GAAG,EAAE,EAAE;EAAEC,IAAI,EAAE,IAAI;EAAEC,IAAI,EAAE;AAAuB,CAAC,EAC3E;EAACJ,IAAI,EAAE,IAAI;EAAEC,KAAK,EAAE,EAAE;EAAEC,GAAG,EAAE,EAAE;EAAEC,IAAI,EAAE,IAAI;EAAEC,IAAI,EAAE;AAAoC,CAAC,CACzF;AAED,MAAMC,aAAa,GAAG,SAAS;AAE/B,SAASC,qBAAqBA,CAACrB,KAAK,EAAEsB,KAAK,EAAE;EAC3C,MAAMC,eAAe,GAAGvB,KAAK,CAACwB,KAAK,CAACC,SAAS,CAACH,KAAK,CAACN,KAAK,EAAEM,KAAK,CAACL,GAAG,GAAG,CAAC,CAAC;EACzE,MAAMS,OAAO,GAAGC,qBAAqB,CAACC,SAAS,CAACL,eAAe,CAAC,CAAC;EACjE;EACA,MAAMM,SAAS,GAAGH,OAAO,CAACI,KAAK,CAAC,EAAE,CAAC;EAEnCD,SAAS,CAACX,IAAI,CAAC,UAASa,CAAC,EAAEC,CAAC,EAAE;IAAE;IAC9B,OAAOC,SAAS,CAACF,CAAC,CAAC,GAAGE,SAAS,CAACD,CAAC,CAAC;EACpC,CAAC,CAAC;EAEF,MAAME,UAAU,GAAGL,SAAS,CAACM,IAAI,CAAC,EAAE,CAAC;EACrC,IAAIZ,eAAe,KAAKW,UAAU,EAAE;IAClC;EACF;;EAEA;;EAEAlC,KAAK,CAACwB,KAAK,GAAI,GAAExB,KAAK,CAACwB,KAAK,CAACC,SAAS,CAAC,CAAC,EAAEH,KAAK,CAACN,KAAK,CAAE,GAAEkB,UAAW,GAAElC,KAAK,CAACwB,KAAK,CAACC,SAAS,CAACH,KAAK,CAACL,GAAG,GAAG,CAAC,CAAE,EAAC,CAAC,CAAC;EAC9G;;EAEA,SAASW,SAASA,CAACQ,GAAG,EAAE;IACtB,IAAIA,GAAG,CAACC,QAAQ,CAAC,GAAG,CAAC,IAAID,GAAG,CAACE,KAAK,CAAC,QAAQ,CAAC,EAAE;MAC5C,OAAOF,GAAG,CAACG,UAAU,CAAC,GAAG,EAAE,GAAG,CAAC;IACjC;IACA,OAAOH,GAAG;EACZ;EAEA,SAASH,SAASA,CAACO,CAAC,EAAE;IACpB,IAAIA,CAAC,KAAK,GAAG,IAAIA,CAAC,KAAK,GAAG,EAAE;MAC1B,OAAOpB,aAAa,CAAC,CAAC;IACxB;IACA,IAAI,CAACE,KAAK,CAACJ,IAAI,EAAE;MAAE;MACjB,OAAO,CAAC;IACV;IACA,MAAMuB,SAAS,GAAGD,CAAC,CAACE,UAAU,CAAC,CAAC,CAAC;IACjC;IACA,IAAID,SAAS,IAAI,EAAE,IAAIA,SAAS,IAAI,GAAG,EAAE;MACvC,OAAOA,SAAS,GAAG,EAAE;IACvB;IACA;IACA,IAAIA,SAAS,IAAI,EAAE,IAAIA,SAAS,IAAI,EAAE,EAAE;MACtC,OAAOA,SAAS,GAAG,EAAE;IACvB;IACA;IACA,OAAOA,SAAS,GAAG,GAAG;EACxB;AACF;AAEO,SAASxC,qCAAqCA,CAACD,KAAK,EAAEJ,cAAc,EAAE;EAC3E,IAAII,KAAK,CAACO,GAAG,KAAK,KAAK,IAAIP,KAAK,CAAC2C,SAAS,EAAE;IAC1C,OAAO3C,KAAK;EACd;;EAEA;;EAEA,MAAM4C,uBAAuB,GAAG9B,eAAe,CAAC+B,MAAM,CAACC,EAAE,IAAIA,EAAE,CAAC/B,IAAI,KAAKnB,cAAc,CAAC;EAExFgD,uBAAuB,CAAC7C,OAAO,CAACuB,KAAK,IAAID,qBAAqB,CAACrB,KAAK,EAAEsB,KAAK,CAAC,CAAC;;EAE7E;;EAEA;EACA;;EAEA,OAAOtB,KAAK;AACd;AAEA,SAAS2B,qBAAqBA,CAACS,GAAG,EAAE;EAClC,MAAMW,GAAG,GAAGX,GAAG,CAACN,KAAK,CAAC,EAAE,CAAC;EACzB;EACA,MAAMkB,UAAU,GAAGD,GAAG,CAACF,MAAM,CAAC,CAACL,CAAC,EAAES,CAAC,KAAKT,CAAC,KAAK,GAAG,IAAIA,CAAC,KAAK,GAAG,IAAIO,GAAG,CAACG,OAAO,CAACV,CAAC,CAAC,KAAKS,CAAC,CAAC,CAACd,IAAI,CAAC,EAAE,CAAC;EAChG;EACA;EACA,MAAMgB,MAAM,GAAI,GAAEH,UAAW,GAAE,GAAG,CAACI,MAAM,CAAChB,GAAG,CAAC9B,MAAM,GAAG0C,UAAU,CAAC1C,MAAM,CAAE,EAAC,CAAC,CAAC;EAC7E;EACA,OAAO6C,MAAM;AACf"}

package/dist/field-008-18-34-character-groups.spec.js

@@ -0,0 +1,51 @@
+"use strict";
+
+var _chai = require("chai");
+var _marcRecord = require("@natlibfi/marc-record");
+var _field0081834CharacterGroups = _interopRequireDefault(require("./field-008-18-34-character-groups"));
+var _fixura = require("@natlibfi/fixura");
+var _fixugen = _interopRequireDefault(require("@natlibfi/fixugen"));
+var _debug = _interopRequireDefault(require("debug"));
+function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
+(0, _fixugen.default)({
+  callback,
+  path: [__dirname, '..', 'test-fixtures', 'field-008-18-34-character-groups'],
+  useMetadataFile: true,
+  recurse: false,
+  fixura: {
+    reader: _fixura.READERS.JSON
+  },
+  mocha: {
+    before: () => testValidatorFactory()
+  }
+});
+const debug = (0, _debug.default)('@natlibfi/marc-record-validators-melinda/field-008-18-34-character-groups:test');
+async function testValidatorFactory() {
+  const validator = await (0, _field0081834CharacterGroups.default)();
+  (0, _chai.expect)(validator).to.be.an('object').that.has.any.keys('description', 'validate');
+  (0, _chai.expect)(validator.description).to.be.a('string');
+  (0, _chai.expect)(validator.validate).to.be.a('function');
+}
+async function callback({
+  getFixture,
+  enabled = true,
+  fix = false
+}) {
+  if (enabled === false) {
+    debug('TEST SKIPPED!');
+    return;
+  }
+  const validator = await (0, _field0081834CharacterGroups.default)();
+  const record = new _marcRecord.MarcRecord(getFixture('record.json'));
+  const expectedResult = getFixture('expectedResult.json');
+  // console.log(expectedResult); // eslint-disable-line
+
+  if (!fix) {
+    const result = await validator.validate(record);
+    (0, _chai.expect)(result).to.eql(expectedResult);
+    return;
+  }
+  await validator.fix(record);
+  (0, _chai.expect)(record).to.eql(expectedResult);
+}
+//# sourceMappingURL=field-008-18-34-character-groups.spec.js.map

package/dist/field-008-18-34-character-groups.spec.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"field-008-18-34-character-groups.spec.js","names":["_chai","require","_marcRecord","_field0081834CharacterGroups","_interopRequireDefault","_fixura","_fixugen","_debug","obj","__esModule","default","generateTests","callback","path","__dirname","useMetadataFile","recurse","fixura","reader","READERS","JSON","mocha","before","testValidatorFactory","debug","createDebugLogger","validator","validatorFactory","expect","to","be","an","that","has","any","keys","description","a","validate","getFixture","enabled","fix","record","MarcRecord","expectedResult","result","eql"],"sources":["../src/field-008-18-34-character-groups.spec.js"],"sourcesContent":["import {expect} from 'chai';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport validatorFactory from './field-008-18-34-character-groups';\nimport {READERS} from '@natlibfi/fixura';\nimport generateTests from '@natlibfi/fixugen';\nimport createDebugLogger from 'debug';\n\ngenerateTests({\n  callback,\n  path: [__dirname, '..', 'test-fixtures', 'field-008-18-34-character-groups'],\n  useMetadataFile: true,\n  recurse: false,\n  fixura: {\n    reader: READERS.JSON\n  },\n  mocha: {\n    before: () => testValidatorFactory()\n  }\n});\nconst debug = createDebugLogger('@natlibfi/marc-record-validators-melinda/field-008-18-34-character-groups:test');\n\nasync function testValidatorFactory() {\n  const validator = await validatorFactory();\n\n  expect(validator)\n    .to.be.an('object')\n    .that.has.any.keys('description', 'validate');\n\n  expect(validator.description).to.be.a('string');\n  expect(validator.validate).to.be.a('function');\n}\n\nasync function callback({getFixture, enabled = true, fix = false}) {\n  if (enabled === false) {\n    debug('TEST SKIPPED!');\n    return;\n  }\n\n  const validator = await validatorFactory();\n  const record = new MarcRecord(getFixture('record.json'));\n  const expectedResult = getFixture('expectedResult.json');\n  // console.log(expectedResult); // eslint-disable-line\n\n  if (!fix) {\n    const result = await validator.validate(record);\n    expect(result).to.eql(expectedResult);\n    return;\n  }\n\n  await validator.fix(record);\n  expect(record).to.eql(expectedResult);\n}\n"],"mappings":";;AAAA,IAAAA,KAAA,GAAAC,OAAA;AACA,IAAAC,WAAA,GAAAD,OAAA;AACA,IAAAE,4BAAA,GAAAC,sBAAA,CAAAH,OAAA;AACA,IAAAI,OAAA,GAAAJ,OAAA;AACA,IAAAK,QAAA,GAAAF,sBAAA,CAAAH,OAAA;AACA,IAAAM,MAAA,GAAAH,sBAAA,CAAAH,OAAA;AAAsC,SAAAG,uBAAAI,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AAEtC,IAAAG,gBAAa,EAAC;EACZC,QAAQ;EACRC,IAAI,EAAE,CAACC,SAAS,EAAE,IAAI,EAAE,eAAe,EAAE,kCAAkC,CAAC;EAC5EC,eAAe,EAAE,IAAI;EACrBC,OAAO,EAAE,KAAK;EACdC,MAAM,EAAE;IACNC,MAAM,EAAEC,eAAO,CAACC;EAClB,CAAC;EACDC,KAAK,EAAE;IACLC,MAAM,EAAEA,CAAA,KAAMC,oBAAoB,CAAC;EACrC;AACF,CAAC,CAAC;AACF,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,gFAAgF,CAAC;AAEjH,eAAeF,oBAAoBA,CAAA,EAAG;EACpC,MAAMG,SAAS,GAAG,MAAM,IAAAC,oCAAgB,EAAC,CAAC;EAE1C,IAAAC,YAAM,EAACF,SAAS,CAAC,CACdG,EAAE,CAACC,EAAE,CAACC,EAAE,CAAC,QAAQ,CAAC,CAClBC,IAAI,CAACC,GAAG,CAACC,GAAG,CAACC,IAAI,CAAC,aAAa,EAAE,UAAU,CAAC;EAE/C,IAAAP,YAAM,EAACF,SAAS,CAACU,WAAW,CAAC,CAACP,EAAE,CAACC,EAAE,CAACO,CAAC,CAAC,QAAQ,CAAC;EAC/C,IAAAT,YAAM,EAACF,SAAS,CAACY,QAAQ,CAAC,CAACT,EAAE,CAACC,EAAE,CAACO,CAAC,CAAC,UAAU,CAAC;AAChD;AAEA,eAAezB,QAAQA,CAAC;EAAC2B,UAAU;EAAEC,OAAO,GAAG,IAAI;EAAEC,GAAG,GAAG;AAAK,CAAC,EAAE;EACjE,IAAID,OAAO,KAAK,KAAK,EAAE;IACrBhB,KAAK,CAAC,eAAe,CAAC;IACtB;EACF;EAEA,MAAME,SAAS,GAAG,MAAM,IAAAC,oCAAgB,EAAC,CAAC;EAC1C,MAAMe,MAAM,GAAG,IAAIC,sBAAU,CAACJ,UAAU,CAAC,aAAa,CAAC,CAAC;EACxD,MAAMK,cAAc,GAAGL,UAAU,CAAC,qBAAqB,CAAC;EACxD;;EAEA,IAAI,CAACE,GAAG,EAAE;IACR,MAAMI,MAAM,GAAG,MAAMnB,SAAS,CAACY,QAAQ,CAACI,MAAM,CAAC;IAC/C,IAAAd,YAAM,EAACiB,MAAM,CAAC,CAAChB,EAAE,CAACiB,GAAG,CAACF,cAAc,CAAC;IACrC;EACF;EAEA,MAAMlB,SAAS,CAACe,GAAG,CAACC,MAAM,CAAC;EAC3B,IAAAd,YAAM,EAACc,MAAM,CAAC,CAACb,EAAE,CAACiB,GAAG,CAACF,cAAc,CAAC;AACvC"}

package/package.json

@@ -14,7 +14,7 @@
     "url": "git@github.com:natlibfi/marc-record-validators-melinda.git"
   },
   "license": "MIT",
-  "version": "10.16.0-alpha.1",
+  "version": "10.16.0-alpha.2",
   "main": "./dist/index.js",
   "publishConfig": {
     "access": "public"

package/src/field-008-18-34-character-groups.js

@@ -0,0 +1,142 @@
+//import createDebugLogger from 'debug';
+import clone from 'clone';
+import {fieldToString} from './utils';
+// Author(s): Nicholas Volk
+// NB! CR 008/24 vs 008/25-27 is not supported yet!
+
+export default function () {
+
+  return {
+    description: 'Justify left and sort character groups within 008/18-24',
+    validate, fix
+  };
+
+  function fix(record) {
+    const typeOfMaterial = record.getTypeOfMaterial();
+    record.fields.forEach(field => {
+      justifyAndSortField008CharacterGroups(field, typeOfMaterial);
+    });
+    // Fix always succeeds (even when it really does not):
+    const res = {message: [], fix: [], valid: true};
+    return res;
+  }
+
+  function validate(record) {
+    const res = {message: []};
+
+    const typeOfMaterial = record.getTypeOfMaterial();
+
+    record.fields?.forEach(field => {
+      validateField(field, res, typeOfMaterial);
+    });
+
+    res.valid = !(res.message.length >= 1); // eslint-disable-line functional/immutable-data
+    return res;
+  }
+
+  function validateField(field, res, typeOfMaterial) {
+    if (field.tag !== '008') { // Optimize code a bit...
+      return;
+    }
+    const orig = fieldToString(field);
+
+    const normalizedField = justifyAndSortField008CharacterGroups(clone(field), typeOfMaterial);
+    const mod = fieldToString(normalizedField);
+    if (orig !== mod) { // Fail as the input is "broken"/"crap"/sumthing
+      res.message.push(`TODO: '${orig}' => '${mod}'`); // eslint-disable-line functional/immutable-data
+      return;
+    }
+    return;
+  }
+}
+
+// Should we add legal values?
+const characterGroups = [
+  {type: 'BK', start: 18, end: 21, sort: true, name: 'illustrations'},
+  {type: 'BK', start: 24, end: 27, sort: true, name: 'nature of contents'}, // English doc does not explicitly mention alphabetical sorting... Finnish does.
+  {type: 'CR', start: 25, end: 27, sort: true, name: 'nature of contents'}, // NB! 24 vs 25-27 logic needs to be implemented separately
+  {type: 'MP', start: 18, end: 21, sort: false, name: 'relief'}, // Order of importance!
+  {type: 'MP', start: 33, end: 34, sort: false, name: 'special format of characteristics'}, // Order of importance!
+  {type: 'MU', start: 24, end: 29, sort: true, name: 'accompanying material'},
+  {type: 'MU', start: 30, end: 31, sort: true, name: 'literary text for sound recordings'}
+];
+
+const BIG_BAD_VALUE = 999999999;
+
+function processCharacterGroup(field, group) {
+  const originalContent = field.value.substring(group.start, group.end + 1);
+  const content = removeDuplicateValues(fixBlanks(originalContent));
+  //console.info(`008/${group.start}-${group.end}: '${originalContent}'`); // eslint-disable-line no-console
+  const charArray = content.split('');
+
+  charArray.sort(function(a, b) { // eslint-disable-line functional/immutable-data, prefer-arrow-callback
+    return scoreChar(a) - scoreChar(b);
+  });
+
+  const newContent = charArray.join('');
+  if (originalContent === newContent) {
+    return;
+  }
+
+  //console.info(`'${fieldToString(field)}' =>`); // eslint-disable-line no-console
+
+  field.value = `${field.value.substring(0, group.start)}${newContent}${field.value.substring(group.end + 1)}`; // eslint-disable-line functional/immutable-data
+  //console.info(`'${fieldToString(field)}'`); // eslint-disable-line no-console
+
+  function fixBlanks(str) {
+    if (str.includes('|') && str.match(/[^ |]/u)) {
+      return str.replaceAll('|', ' ');
+    }
+    return str;
+  }
+
+  function scoreChar(c) {
+    if (c === '|' || c === ' ') {
+      return BIG_BAD_VALUE; // Max value, these should code last
+    }
+    if (!group.sort) { // more meaningful comes first: keep the original order
+      return 1;
+    }
+    const asciiCode = c.charCodeAt(0);
+    // a-z get values 1-26:
+    if (asciiCode >= 97 && asciiCode <= 122) {
+      return asciiCode - 96;
+    }
+    // 0-9 get values 100-109
+    if (asciiCode >= 48 && asciiCode <= 57) {
+      return asciiCode + 52;
+    }
+    // Others (=crap) return something between '9' and BIG BAD VALUE
+    return asciiCode + 200;
+  }
+}
+
+export function justifyAndSortField008CharacterGroups(field, typeOfMaterial) {
+  if (field.tag !== '008' || field.subfields) {
+    return field;
+  }
+
+  //console.info(typeOfMaterial); // eslint-disable-line no-console
+
+  const relevantCharacterGroups = characterGroups.filter(gr => gr.type === typeOfMaterial);
+
+  relevantCharacterGroups.forEach(group => processCharacterGroup(field, group));
+
+  //justifyField008CharacterGroups(field, typeOfMaterial); // Oops: also sorts...
+
+  // NB! add value # and | normalizations
+  //fixBlanks(field, typeOfMaterial);
+
+  return field;
+}
+
+function removeDuplicateValues(str) {
+  const arr = str.split('');
+  // Take only the first instance of a proper value-carrying character
+  const reducedStr = arr.filter((c, i) => c === ' ' || c === '|' || arr.indexOf(c) === i).join('');
+  //console.info(`I: '${str}'`); // eslint-disable-line no-console
+  //console.info(`M: '${reducedStr}'`); // eslint-disable-line no-console
+  const output = `${reducedStr}${' '.repeat(str.length - reducedStr.length)}`; // Had some weird trouble with str.padEnd(n)
+  //console.info(`M: '${output}'`); // eslint-disable-line no-console
+  return output;
+}

package/src/field-008-18-34-character-groups.spec.js

@@ -0,0 +1,52 @@
+import {expect} from 'chai';
+import {MarcRecord} from '@natlibfi/marc-record';
+import validatorFactory from './field-008-18-34-character-groups';
+import {READERS} from '@natlibfi/fixura';
+import generateTests from '@natlibfi/fixugen';
+import createDebugLogger from 'debug';
+
+generateTests({
+  callback,
+  path: [__dirname, '..', 'test-fixtures', 'field-008-18-34-character-groups'],
+  useMetadataFile: true,
+  recurse: false,
+  fixura: {
+    reader: READERS.JSON
+  },
+  mocha: {
+    before: () => testValidatorFactory()
+  }
+});
+const debug = createDebugLogger('@natlibfi/marc-record-validators-melinda/field-008-18-34-character-groups:test');
+
+async function testValidatorFactory() {
+  const validator = await validatorFactory();
+
+  expect(validator)
+    .to.be.an('object')
+    .that.has.any.keys('description', 'validate');
+
+  expect(validator.description).to.be.a('string');
+  expect(validator.validate).to.be.a('function');
+}
+
+async function callback({getFixture, enabled = true, fix = false}) {
+  if (enabled === false) {
+    debug('TEST SKIPPED!');
+    return;
+  }
+
+  const validator = await validatorFactory();
+  const record = new MarcRecord(getFixture('record.json'));
+  const expectedResult = getFixture('expectedResult.json');
+  // console.log(expectedResult); // eslint-disable-line
+
+  if (!fix) {
+    const result = await validator.validate(record);
+    expect(result).to.eql(expectedResult);
+    return;
+  }
+
+  await validator.fix(record);
+  expect(record).to.eql(expectedResult);
+}

package/test-fixtures/field-008-18-34-character-groups/01/expectedResult.json

@@ -0,0 +1,6 @@
+{
+  "message": [
+    "TODO: '008    121110s2002    fi  cb| | mk     |fin |' => '008    121110s2002    fi bc   |km      |fin |'"
+  ],
+  "valid": false
+}

package/test-fixtures/field-008-18-34-character-groups/01/metadata.json

@@ -0,0 +1,6 @@
+{
+  "description": "Validate BK 008/19-22 and 008/24-27",
+  "enabled": true,
+  "fix": false,
+  "only": false
+}

package/test-fixtures/field-008-18-34-character-groups/01/record.json

@@ -0,0 +1,6 @@
+{
+  "leader": "01234cam8a22234564i 4500",
+  "fields": [
+    { "tag": "008", "value": "121110s2002    fi  cb| | mk     |fin |" }
+  ]
+}

package/test-fixtures/field-008-18-34-character-groups/02/expectedResult.json

@@ -0,0 +1,7 @@
+{
+  "_validationOptions": {},
+  "leader": "01234cam8a22234564i 4500",
+  "fields": [
+    { "tag": "008", "value": "121110s2002    fi bc   |km      |fin |" }
+  ]
+}

package/test-fixtures/field-008-18-34-character-groups/02/metadata.json

@@ -0,0 +1,6 @@
+{
+  "description": "Fix BK 008/19-22 and 008/24-27",
+  "enabled": true,
+  "fix": true,
+  "only": false
+}

package/test-fixtures/field-008-18-34-character-groups/02/record.json

@@ -0,0 +1,6 @@
+{
+  "leader": "01234cam8a22234564i 4500",
+  "fields": [
+    { "tag": "008", "value": "121110s2002    fi  cb| | mk     |fin |" }
+  ]
+}

package/test-fixtures/field-008-18-34-character-groups/03/expectedResult.json

@@ -0,0 +1,7 @@
+{
+  "_validationOptions": {},
+  "leader": "01234cas8a22234564i 4500",
+  "fields": [
+    { "tag": "008", "value": "121110s2002    fi  cb| | km     |fin |" }
+  ]
+}

package/test-fixtures/field-008-18-34-character-groups/03/metadata.json

@@ -0,0 +1,7 @@
+{
+  "description": "Fix CR 008/19-22 and 008/25-27",
+  "comment": "TODO: handle 008/24 vs 008/25-27 later on",
+  "enabled": true,
+  "fix": true,
+  "only": false
+}

package/test-fixtures/field-008-18-34-character-groups/03/record.json

@@ -0,0 +1,6 @@
+{
+  "leader": "01234cas8a22234564i 4500",
+  "fields": [
+    { "tag": "008", "value": "121110s2002    fi  cb| |  mk    |fin |" }
+  ]
+}

package/test-fixtures/field-008-18-34-character-groups/04/expectedResult.json

@@ -0,0 +1,7 @@
+{
+  "_validationOptions": {},
+  "leader": "01234cjm8a22234564i 4500",
+  "fields": [
+    { "tag": "008", "value": "121110s2002    fi ana   fik   z ||fin |" }
+  ]
+}

package/test-fixtures/field-008-18-34-character-groups/04/metadata.json

@@ -0,0 +1,6 @@
+{
+  "description": "Fix MU 008/24-29 and 008/30-31",
+  "enabled": true,
+  "fix": true,
+  "only": false
+}

package/test-fixtures/field-008-18-34-character-groups/04/record.json

@@ -0,0 +1,6 @@
+{
+  "leader": "01234cjm8a22234564i 4500",
+  "fields": [
+    { "tag": "008", "value": "121110s2002    fi ana   |i f k z||fin |" }
+  ]
+}

package/test-fixtures/field-008-18-34-character-groups/05/expectedResult.json

@@ -0,0 +1,7 @@
+{
+  "_validationOptions": {},
+  "leader": "01234cem8a22234564i 4500",
+  "fields": [
+    { "tag": "008", "value": "121110s2002    fi mc  aa|a  aa ||n fin |" }
+  ]
+}

package/test-fixtures/field-008-18-34-character-groups/05/metadata.json

@@ -0,0 +1,7 @@
+{
+  "description": "Fix MP 008/18-21 and 008/33-34",
+  "comment": "Note that alphanumerical stuff comes in priority order and is not sorted",
+  "enabled": true,
+  "fix": true,
+  "only": false
+}

package/test-fixtures/field-008-18-34-character-groups/05/record.json

@@ -0,0 +1,6 @@
+{
+  "leader": "01234cem8a22234564i 4500",
+  "fields": [
+    { "tag": "008", "value": "121110s2002    fi |m caa|a  aa |||nfin |" }
+  ]
+}