Add f245 title splitting to title+subtitle if some of most common pat…

…terns match to title

src/transform/convert/common/generate2xx.js

@@ -5,18 +5,22 @@
  */
 export function generate245({getFields}) {
   const isAddedEntry = generateIsAddedEntry();
+  const ind1 = isAddedEntry ? '1' : '0';
+  const ind2 = ' '; // NB: generated in validation phase by marc-record-validators-melinda:IndicatorFixes
 
   const fields = getFields('dc.title');
-  const title = fields.length > 0 ? fields[0].$.value : null;
+  if (fields.length === 0) {
+    return [];
+  }
+
+  const titleText = fields.length > 0 ? fields[0].$.value : null;
+
+  const {title, alternativeSubtitle} = getTitle(titleText);
+
+  return alternativeSubtitle
+    ? [{tag: '245', ind1, ind2, subfields: [{code: 'a', value: `${title} :`}, {code: 'b', value: `${alternativeSubtitle}.`}]}]
+    : [{tag: '245', ind1, ind2, subfields: [{code: 'a', value: `${title}.`}]}];
 
-  return title ? [
-    {
-      tag: '245',
-      ind1: isAddedEntry ? '1' : '0',
-      ind2: '0',
-      subfields: [{code: 'a', value: `${title}.`}]
-    }
-  ] : [];
 
   function generateIsAddedEntry() {
     const fields = getFields(p => [
@@ -26,6 +30,42 @@ export function generate245({getFields}) {
 
     return fields.length > 0;
   }
+
+  // Splits title to title+subtitle if title contains any of patterns that require this type of processing.
+  // Note: this getter is same as one defined within ONIX-transformer
+  function getTitle(titleText) {
+    const regexObj = findRegex(titleText);
+    const result = regexObj ? regexObj.regex.exec(titleText) : undefined;
+
+    if (!result) {
+      return {title: titleText.trimEnd(), alternativeSubtitle: undefined};
+    }
+
+    const titleResult = regexObj.keepResult === true ? {
+      title: (titleText.slice(0, result.index + regexObj.keepCharactersFromStart) + result).trimEnd(),
+      alternativeSubtitle: titleText.slice(result.index + result[0].length - regexObj.keepCharactersFromEnd).trimEnd().trimStart()
+    }
+      : {
+        title: titleText.slice(0, result.index + regexObj.keepCharactersFromStart).trimEnd(),
+        alternativeSubtitle: titleText.slice(result.index + result[0].length - regexObj.keepCharactersFromEnd).trimEnd().trimStart()
+      };
+
+    return titleResult;
+
+    function findRegex(titleText) {
+      // Note: order defines priority
+      const pluralOfRegex = [
+        // split title to mainTitle and subtitle at first ':', do not keep ':'
+        {keepCharactersFromStart: 0, keepCharactersFromEnd: 0, regex: /:\s+/u},
+        // split title to mainTitle and subtitle at first ' - ', do not keep the separator
+        {keepCharactersFromStart: 1, keepCharactersFromEnd: 1, regex: /[^0-9]\s+[\u2013\u2014-]\s+[^0-9]/u},
+        // split title to mainTitle and subtitle at '! ' or '? ', keep question and exclamation marks, they are part of the title
+        {keepCharactersFromStart: 0, keepCharactersFromEnd: 0, keepResult: true, regex: /!+|\?+/u}
+      ];
+
+      return pluralOfRegex.find(({regex}) => regex.test(titleText));
+    }
+  }
 }
 
 /**

test-fixtures/transform/convert/common/generate2xx/generate245/01/input.json

@@ -3,7 +3,7 @@
     "$": {
       "schema": "dc",
       "element": "title",
-      "value": "Dublin Core - MARC21 -konversio"
+      "value": "Dublin Core"
     }
   }
 ]

test-fixtures/transform/convert/common/generate2xx/generate245/01/output.json

@@ -1,8 +1,8 @@
 [
   {
-    "tag": "245", "ind1": "0", "ind2": "0",
+    "tag": "245", "ind1": "0", "ind2": " ",
     "subfields": [
-      {"code": "a", "value": "Dublin Core - MARC21 -konversio."}
+      {"code": "a", "value": "Dublin Core."}
     ]
   }
 ]

test-fixtures/transform/convert/common/generate2xx/generate245/02/input.json

@@ -10,7 +10,7 @@
     "$": {
       "schema": "dc",
       "element": "title",
-      "value": "Dublin Core - MARC21 -konversio"
+      "value": "Dublin Core"
     }
   }
 ]

test-fixtures/transform/convert/common/generate2xx/generate245/02/metadata.json

@@ -1,4 +1,4 @@
 {
-  "description": "Ind1=1 when do.contibutor.author or dc.creator information can be found.",
+  "description": "Ind1=1 when dc.contibutor.author or dc.creator information can be found.",
   "only": false
 }

test-fixtures/transform/convert/common/generate2xx/generate245/02/output.json

@@ -1,8 +1,8 @@
 [
   {
-    "tag": "245", "ind1": "1", "ind2": "0",
+    "tag": "245", "ind1": "1", "ind2": " ",
     "subfields": [
-      {"code": "a", "value": "Dublin Core - MARC21 -konversio."}
+      {"code": "a", "value": "Dublin Core."}
     ]
   }
 ]

test-fixtures/transform/convert/common/generate2xx/generate245/05/input.json

@@ -0,0 +1,9 @@
+[
+  {
+    "$": {
+      "schema": "dc",
+      "element": "title",
+      "value": "Dublin Core - MARC21 -conversion"
+    }
+  }
+]

test-fixtures/transform/convert/common/generate2xx/generate245/05/metadata.json

@@ -0,0 +1,4 @@
+{
+  "description": "Splits title to title+subtitle from character \"-\"",
+  "only": false
+}

test-fixtures/transform/convert/common/generate2xx/generate245/05/output.json

@@ -0,0 +1,9 @@
+[
+  {
+    "tag": "245", "ind1": "0", "ind2": " ",
+    "subfields": [
+      {"code": "a", "value": "Dublin Core :"},
+      {"code": "b", "value": "MARC21 -conversion."}
+    ]
+  }
+]

test-fixtures/transform/convert/common/generate2xx/generate245/06/input.json

@@ -0,0 +1,9 @@
+[
+  {
+    "$": {
+      "schema": "dc",
+      "element": "title",
+      "value": "Dublin Core: MARC21 -conversion"
+    }
+  }
+]

test-fixtures/transform/convert/common/generate2xx/generate245/06/metadata.json

@@ -0,0 +1,4 @@
+{
+  "description": "Splits title to title+subtitle from character \":\"",
+  "only": false
+}

test-fixtures/transform/convert/common/generate2xx/generate245/06/output.json

@@ -0,0 +1,9 @@
+[
+  {
+    "tag": "245", "ind1": "0", "ind2": " ",
+    "subfields": [
+      {"code": "a", "value": "Dublin Core :"},
+      {"code": "b", "value": "MARC21 -conversion."}
+    ]
+  }
+]

test-fixtures/transform/convert/common/generate2xx/generate245/07/input.json

@@ -0,0 +1,9 @@
+[
+  {
+    "$": {
+      "schema": "dc",
+      "element": "title",
+      "value": "Dublin Core! MARC21 -conversion"
+    }
+  }
+]

test-fixtures/transform/convert/common/generate2xx/generate245/07/metadata.json

@@ -0,0 +1,4 @@
+{
+  "description": "Splits title to title+subtitle from character \"!\"",
+  "only": false
+}

test-fixtures/transform/convert/common/generate2xx/generate245/07/output.json

@@ -0,0 +1,9 @@
+[
+  {
+    "tag": "245", "ind1": "0", "ind2": " ",
+    "subfields": [
+      {"code": "a", "value": "Dublin Core! :"},
+      {"code": "b", "value": "MARC21 -conversion."}
+    ]
+  }
+]

test-fixtures/transform/convert/common/generate2xx/generate245/08/input.json

@@ -0,0 +1,9 @@
+[
+  {
+    "$": {
+      "schema": "dc",
+      "element": "title",
+      "value": "Dublin Core? MARC21 -conversion"
+    }
+  }
+]

test-fixtures/transform/convert/common/generate2xx/generate245/08/metadata.json

@@ -0,0 +1,4 @@
+{
+  "description": "Splits title to title+subtitle from character \"?\"",
+  "only": false
+}

test-fixtures/transform/convert/common/generate2xx/generate245/08/output.json

@@ -0,0 +1,9 @@
+[
+  {
+    "tag": "245", "ind1": "0", "ind2": " ",
+    "subfields": [
+      {"code": "a", "value": "Dublin Core? :"},
+      {"code": "b", "value": "MARC21 -conversion."}
+    ]
+  }
+]

test-fixtures/transform/integration/01/output.json

@@ -103,7 +103,11 @@
         "subfields": [
           {
             "code": "a",
-            "value": "<\"Cool Stuff\", like 'this': Dublin Core & MARC21 ♥ >."
+            "value": "<\"Cool Stuff\", like 'this' :"
+          },
+          {
+            "code": "b",
+            "value": "Dublin Core & MARC21 ♥ >."
           }
         ]
       },
@@ -409,7 +413,7 @@
           },
           {
             "code": "k",
-            "value": "MELINDA_RECORD_IMPORT_REPO:FOOBAR:3f39198fe332d48b27b8d18cfab4356d286ce43f3afd2799c21a302556bd04a8"
+            "value": "MELINDA_RECORD_IMPORT_REPO:FOOBAR:00eb8bec1dddd268cf2306e35693e6889d402f30ccb4a882d4978c78cfcaccaf"
           },
           {
             "code": "q",
@@ -461,7 +465,7 @@
     },
     {
       "description": "Normalizes indicator values",
-      "state": "valid"
+      "state": "fixed"
     },
     {
       "description": "Adds URN for record, to 856-field (if not existing). If isLegalDeposit is active, adds legal deposit subfields to the f856s with URN.",