Add chopPunctuation to Subject

Author: danmichaelo

README.md

@@ -213,11 +213,16 @@ foreach ($record->getSubjects('mesh', Subject::TOPICAL_TERM) as $subject) {
 }
 ```
 
-The string representation of this field makes use of the constant `Subject::glue`
-to glue subject components together. The default value is a space-padded colon,
-making `Physics : History : 20th century` the string representation of
-`650 $aPhysics $xHistory $yHistory`. If you prefer the "LCSH-way" of
-`Physics--History--20th century`, just set `Subject::glue = '--'`.
+Static options:
+
+* `Subject::glue` (default: ` : `) defines what string is used to glue the subfields
+  together in the string representation. For instance, `650 $aPhysics $xHistory $yHistory`
+  becomes `Physics : History : 20th century` when using ` : ` as glue, or
+  `Physics--History--20th century` with `'--'`.
+* `Subject::chopPunctuation` (default: `true`) defines if ending punctuation (.:,;/)
+  is to be chopped off at the end of subjects. Usually, any ending punctuation is an
+  ISBD character that can be safely chopped off, but it might also indicate an abbreviation,
+  and unfortunately there is no way to know.
 
 ## Notes
 

src/Fields/Field.php

@@ -79,9 +79,13 @@ protected function getSubfieldValues($codes)
      * @param string   $glue
      * @return string
      */
-    protected function toString($codes, $glue = ' ')
+    protected function toString($codes, $glue = ' ', $chopPunctuation = true)
     {
-        return trim(implode($glue, $this->getSubfieldValues($codes)));
+        $value = trim(implode($glue, $this->getSubfieldValues($codes)));
+        if ($chopPunctuation) {
+            $value = preg_replace('/[.:,;]$/', '', $value);
+        }
+        return $value;
     }
 
     /**

src/Fields/Subject.php

@@ -12,6 +12,7 @@ class Subject extends Field implements FieldInterface, SubjectInterface
     public $properties = ['type', 'vocabulary', 'term', 'id'];
 
     public static $glue = ' : ';
+    public static $chopPunctuation = true;
     public static $termComponentCodes = ['a', 'b', 'x', 'y', 'z'];
 
     const PERSONAL_NAME = '600';
@@ -92,7 +93,7 @@ public function getParts()
 
     public function getTerm()
     {
-        return $this->toString(self::$termComponentCodes, self::$glue);
+        return $this->toString(self::$termComponentCodes, self::$glue, self::$chopPunctuation);
     }
 
     public function __toString()

tests/SubjectFieldTest.php

@@ -7,16 +7,16 @@
 
 class SubjectFieldTest extends TestCase
 {
-    protected function getNthrecord($n)
+    protected function getNthrecord($filename, $n)
     {
-        $records = Collection::fromFile('tests/data/sru-alma.xml')->toArray();
+        $records = Collection::fromFile('tests/data/' . $filename)->toArray();
 
         return $records[$n - 1];
     }
 
     public function testSubjectString()
     {
-        $record = $this->getNthrecord(1);
+        $record = $this->getNthrecord('sru-alma.xml', 1);
 
         # Vocabulary from indicator2
         $sub = $record->subjects[0];
@@ -25,9 +25,20 @@ public function testSubjectString()
         $this->assertEquals('Eightfold way (Nuclear physics) : Addresses, essays, lectures', strval($sub));
     }
 
+    public function testChopPunctuation()
+    {
+        $record = $this->getNthrecord('sru-loc.xml', 2);
+
+        # Vocabulary from indicator2
+        $sub = $record->subjects[0];
+        $this->assertEquals('lcsh', $sub->vocabulary);
+        $this->assertEquals(Subject::TOPICAL_TERM, $sub->type);
+        $this->assertEquals('Popular music : 1961-1970', strval($sub));
+    }
+
     public function testSubjects()
     {
-        $record = $this->getNthrecord(3);
+        $record = $this->getNthrecord('sru-alma.xml', 3);
 
         # Vocabulary from subfield 2
         $subject = $record->subjects[1];
@@ -40,7 +51,7 @@ public function testSubjects()
 
     public function testRepeated653a()
     {
-        $record = $this->getNthrecord(3);
+        $record = $this->getNthrecord('sru-alma.xml', 3);
 
         $subjects = $record->getSubjects(null, Subject::UNCONTROLLED_INDEX_TERM);
         $this->assertCount(2, $subjects);
@@ -53,7 +64,7 @@ public function testRepeated653a()
 
     public function testGetSubjectsFiltering()
     {
-        $record = $this->getNthrecord(3);
+        $record = $this->getNthrecord('sru-alma.xml', 3);
 
         $lcsh = $record->getSubjects('lcsh');
         $noubomn = $record->getSubjects('noubomn');
@@ -70,7 +81,7 @@ public function testGetSubjectsFiltering()
 
     public function testEdit()
     {
-        $record = $this->getNthrecord(3);
+        $record = $this->getNthrecord('sru-alma.xml', 3);
         $this->assertCount(5, $record->subjects);
 
         $this->assertInstanceOf(Subject::class, $record->subjects[0]);
@@ -94,7 +105,7 @@ public function testEdit()
 
     public function testJsonSerialization()
     {
-        $record = $this->getNthrecord(3);
+        $record = $this->getNthrecord('sru-alma.xml', 3);
         $subject = $record->subjects[1];
 
         $this->assertJsonStringEqualsJsonString(

tests/data/examples/bibliographic4.json

@@ -20,7 +20,7 @@
         {
             "type": "650",
             "vocabulary": "lcsh",
-            "term": "Semiconductors."
+            "term": "Semiconductors"
         },
         {
             "type": "650",
@@ -67,4 +67,4 @@
         "assigning_source": "Provided by publisher."
     },
     "part_of": null
-}
+}