@natlibfi/marc-record-merge 5.0.0 → 6.0.0-beta.4

package/src/reducers/select.js

@@ -0,0 +1,107 @@
+import {normalizeSync} from 'normalize-diacritics';
+import createDebugLogger from 'debug';
+
+export function strictEquality(subfieldA, subfieldB) {
+  return subfieldA.code === subfieldB.code &&
+  subfieldA.value === subfieldB.value;
+}
+
+export function subsetEquality(subfieldA, subfieldB) {
+  return subfieldA.code === subfieldB.code &&
+  (subfieldA.value.indexOf(subfieldB.value) !== -1 || subfieldB.value.indexOf(subfieldA.value) !== -1);
+}
+// EqualityFunction can be either strictEquality or subsetEquality
+export default ({tagPattern, equalityFunction = strictEquality}) => (base, source) => {
+  const debug = createDebugLogger('@natlibfi/marc-record-merge:select');
+  const baseFields = base.get(tagPattern);
+  const sourceFields = source.get(tagPattern);
+  const fieldTag = sourceFields.map(field => field.tag);
+  debug(`Comparing field ${fieldTag}`);
+
+  checkFieldType(baseFields);
+  checkFieldType(sourceFields);
+
+  if (baseFields.length > 1 || sourceFields.length > 1) {
+    debug(`Multiple fields in base or source`);
+    debug(`No changes to base`);
+    return base;
+  }
+  const [baseField] = baseFields;
+  const [sourceField] = sourceFields;
+
+  if (baseField.tag === sourceField.tag === false) {
+    debug(`Base tag ${baseField.tag} is not equal to source tag ${sourceField.tag}`);
+    debug(`No changes to base`);
+    return base;
+  }
+  const baseSubs = baseField.subfields;
+  const sourceSubs = sourceField.subfields;
+
+  const baseSubsNormalized = baseSubs
+    .map(({code, value}) => ({code, value: normalizeSubfieldValue(value)}));
+
+  const sourceSubsNormalized = sourceSubs
+    .map(({code, value}) => ({code, value: normalizeSubfieldValue(value)}));
+
+  // Returns the base subfields for which a matching source subfield is found
+  const equalSubfieldsBase = baseSubsNormalized
+    .filter(baseSubfield => sourceSubsNormalized
+      .some(sourceSubfield => equalityFunction(baseSubfield, sourceSubfield)));
+  debug(`equalSubfieldsBase: ${JSON.stringify(equalSubfieldsBase, undefined, 2)}`);
+
+  // Returns the source subfields for which a matching base subfield is found
+  const equalSubfieldsSource = sourceSubsNormalized
+    .filter(sourceSubfield => baseSubsNormalized
+      .some(baseSubfield => equalityFunction(sourceSubfield, baseSubfield)));
+  debug(`equalSubfieldsSource: ${JSON.stringify(equalSubfieldsSource, undefined, 2)}`);
+
+  if (baseSubs.length === sourceSubs.length && equalSubfieldsBase.length < baseSubs.length) {
+    debug(`Base and source subfields are not equal`);
+    debug(`No changes to base`);
+    return base;
+  }
+
+  if (baseSubs.length === sourceSubs.length && equalSubfieldsBase.length === equalSubfieldsSource.length) {
+    debug(`Checking subfield equality`);
+    const totalSubfieldLengthBase = baseSubsNormalized
+      .map(({value}) => value.length)
+      .reduce((acc, value) => acc + value);
+    const totalSubfieldLengthSource = sourceSubsNormalized
+      .map(({value}) => value.length)
+      .reduce((acc, value) => acc + value);
+
+    if (totalSubfieldLengthSource > totalSubfieldLengthBase) {
+      return replaceBasefieldWithSourcefield(base);
+    }
+  }
+
+  if (sourceSubs.length > baseSubs.length && equalSubfieldsBase.length === baseSubs.length) {
+    return replaceBasefieldWithSourcefield(base);
+  }
+
+  debug(`No changes to base`);
+  return base;
+
+  function replaceBasefieldWithSourcefield(base) {
+    const index = base.fields.findIndex(field => field === baseField);
+    base.fields.splice(index, 1, sourceField); // eslint-disable-line functional/immutable-data
+    debug(`Source field is longer, replacing base with source`);
+    return base;
+  }
+
+  function checkFieldType(fields) {
+    const checkedFields = fields.map(field => {
+      if ('value' in field) { // eslint-disable-line functional/no-conditional-statement
+        throw new Error('Invalid control field, expected data field');
+      }
+      return field;
+    });
+    return checkedFields;
+  }
+
+  function normalizeSubfieldValue(value) {
+    // Regexp options: g: global search, u: unicode
+    const punctuation = /[.,\-/#!?$%^&*;:{}=_`~()[\]]/gu;
+    return normalizeSync(value).toLowerCase().replace(punctuation, '', 'u').replace(/\s+/gu, ' ').trim();
+  }
+};

package/src/reducers/select.spec.js

@@ -0,0 +1,45 @@
+import {expect} from 'chai';
+import {MarcRecord} from '@natlibfi/marc-record';
+import createReducer, {subsetEquality} from './select';
+import {READERS} from '@natlibfi/fixura';
+import generateTests from '@natlibfi/fixugen';
+
+MarcRecord.setValidationOptions({subfieldValues: false});
+
+generateTests({
+  callback,
+  path: [__dirname, '..', '..', 'test-fixtures', 'reducers', 'select'],
+  useMetadataFile: true,
+  recurse: false,
+  fixura: {
+    reader: READERS.JSON,
+    failWhenNotFound: false
+  }
+});
+
+function callback({
+  getFixture,
+  disabled = false,
+  tagPatternRegExp = false,
+  expectedError = false,
+  useSubsetEquality = false
+}) {
+  if (disabled) {
+    console.log('TEST DISABLED!'); // eslint-disable-line no-console
+    return;
+  }
+  const base = new MarcRecord(getFixture('base.json'), {subfieldValues: false});
+  const source = new MarcRecord(getFixture('source.json'), {subfieldValues: false});
+  const tagPattern = new RegExp(tagPatternRegExp, 'u');
+  const expectedRecord = getFixture('merged.json');
+  const equalityFunction = useSubsetEquality ? subsetEquality : undefined;
+
+  // Bypass expected error in testing
+  if (expectedError) {
+    expect(() => createReducer.to.throw(Error, 'control field'));
+    return;
+  }
+
+  const mergedRecord = createReducer({tagPattern, equalityFunction})(base, source);
+  expect(mergedRecord.toObject()).to.eql(expectedRecord);
+}

package/test-fixtures/reducers/copy/basic copy/01/base.json

@@ -0,0 +1,9 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/01/merged.json

@@ -0,0 +1,20 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                "code": "a",
+                "value": "##2002021346"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/01/metadata.json

@@ -0,0 +1,5 @@
+{
+    "description": "If base does not contain the field at all, it is copied from source to base",
+    "tagPatternRegExp": "^010$",
+    "disabled": false
+}

package/test-fixtures/reducers/copy/basic copy/01/source.json

@@ -0,0 +1,20 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/02/base.json

@@ -0,0 +1,20 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/02/merged.json

@@ -0,0 +1,20 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                "code": "a",
+                "value": "##2002021346"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/02/metadata.json

@@ -0,0 +1,5 @@
+{
+    "description": "Identical control fields are not copied",
+    "tagPatternRegExp": "^001$",
+    "disabled": false
+}

package/test-fixtures/reducers/copy/basic copy/02/source.json

@@ -0,0 +1,20 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/03/base.json

@@ -0,0 +1,20 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/03/merged.json

@@ -0,0 +1,28 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "001",
+            "value": "007346735"
+        },
+        {
+            "tag": "001",
+            "value": "007346736"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                "code": "a",
+                "value": "##2002021346"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/03/metadata.json

@@ -0,0 +1,5 @@
+{
+    "description": "Add missing control field to base",
+    "tagPatternRegExp": "^001$",
+    "disabled": false
+}

package/test-fixtures/reducers/copy/basic copy/03/source.json

@@ -0,0 +1,24 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346735"
+        },
+        {
+            "tag": "001",
+            "value": "007346736"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/04/base.json

@@ -0,0 +1,39 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                },
+                {
+                    "code": "b",
+                    "value": "##2002021356"
+                }
+            ]
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "c",
+                    "value": "##2002021347"
+                },
+                {
+                    "code": "d",
+                    "value": "##2002021357"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/04/merged.json

@@ -0,0 +1,39 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                },
+                {
+                    "code": "b",
+                    "value": "##2002021356"
+                }
+            ]
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "c",
+                    "value": "##2002021347"
+                },
+                {
+                    "code": "d",
+                    "value": "##2002021357"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/04/metadata.json

@@ -0,0 +1,5 @@
+{
+    "description": "Identical data fields in base and source, not copied",
+    "tagPatternRegExp": "^010$",
+    "disabled": false
+}

package/test-fixtures/reducers/copy/basic copy/04/source.json

@@ -0,0 +1,39 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                },
+                {
+                    "code": "b",
+                    "value": "##2002021356"
+                }
+            ]
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "c",
+                    "value": "##2002021347"
+                },
+                {
+                    "code": "d",
+                    "value": "##2002021357"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/05/base.json

@@ -0,0 +1,39 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                },
+                {
+                    "code": "b",
+                    "value": "##2002021356"
+                }
+            ]
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "c",
+                    "value": "##2002021347"
+                },
+                {
+                    "code": "d",
+                    "value": "##2002021357"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/05/merged.json

@@ -0,0 +1,77 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                },
+                {
+                    "code": "b",
+                    "value": "##2002021356"
+                }
+            ]
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "c",
+                    "value": "##2002021347"
+                },
+                {
+                    "code": "d",
+                    "value": "##2002021357"
+                }
+            ]
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "c",
+                    "value": "##2002021348"
+                },
+                {
+                    "code": "d",
+                    "value": "##2002021358"
+                },
+                {
+                    "code": "e",
+                    "value": "##2002021359"
+                }
+            ]
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "f",
+                    "value": "##2002021311"
+                },
+                {
+                    "code": "g",
+                    "value": "##2002021322"
+                },
+                {
+                    "code": "h",
+                    "value": "##2002021333"
+                }
+            ]
+        }
+    ]
+}

package/test-fixtures/reducers/copy/basic copy/05/metadata.json

@@ -0,0 +1,5 @@
+{
+    "description": "Different data fields are copied from source to base (multiple fields)",
+    "tagPatternRegExp": "^010$",
+    "disabled": false
+}

package/test-fixtures/reducers/copy/basic copy/05/source.json

@@ -0,0 +1,62 @@
+{
+    "leader": "01331cam a22003494i 4500",
+    "fields": [
+        {
+            "tag": "001",
+            "value": "007346734"
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "a",
+                    "value": "##2002021346"
+                },
+                {
+                    "code": "b",
+                    "value": "##2002021356"
+                }
+            ]
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "c",
+                    "value": "##2002021348"
+                },
+                {
+                    "code": "d",
+                    "value": "##2002021358"
+                },
+                {
+                    "code": "e",
+                    "value": "##2002021359"
+                }
+            ]
+        },
+        {
+            "tag": "010",
+            "ind1": " ",
+            "ind2": " ",
+            "subfields": [
+                {
+                    "code": "f",
+                    "value": "##2002021311"
+                },
+                {
+                    "code": "g",
+                    "value": "##2002021322"
+                },
+                {
+                    "code": "h",
+                    "value": "##2002021333"
+                }
+            ]
+        }
+    ]
+}