bankRef & splitTransactions (#39)

added function body to method: bankRef & refactored splitTransactions() in AbstractParser.php
sandermarechal · Jun 3, 2021 · 9309715 · 9309715
1 parent 8b3af41
commit 9309715
Show file tree

Hide file tree

Showing 17 changed files with 359 additions and 124 deletions.
diff --git a/composer.json b/composer.json
@@ -12,8 +12,8 @@
         },
         {
             "name": "powercloud GmbH / Dominic Richter",
-            "email": "d.richter@powercloud.de",
-            "homepage": "https://www.powercloud.de"
+            "email": "dominic.richter@power.cloud",
+            "homepage": "https://www.power.cloud"
         }
     ],
     "require": {

diff --git a/lib/Jejik/MT940/Parser/AbstractParser.php b/lib/Jejik/MT940/Parser/AbstractParser.php
@@ -29,17 +29,13 @@
  */
 abstract class AbstractParser
 {
-    // Properties {{{
-
     /**
      * Reference to the MT940 reader
      *
      * @var Reader
      */
     protected $reader;
 
-    // }}}
-
     /**
      * Constructor
      *
@@ -130,13 +126,14 @@ public function parse(string $text): array
      *
      * The contents may be several lines long (e.g. :86: descriptions)
      *
-     * @param string $id       The line ID (e.g. "20"). Can be a regular
+     * @param string $id The line ID (e.g. "20"). Can be a regular
      *                         expression (e.g. "60F|60M")
-     * @param string $text     The text to search
-     * @param int    $offset   The offset to start looking
-     * @param int    $position Starting position of the found line
-     * @param int    $length   Length of the found line (before trimming),
+     * @param string $text The text to search
+     * @param int $offset The offset to start looking
+     * @param int|null $position Starting position of the found line
+     * @param int|null $length Length of the found line (before trimming),
      *                         including EOL
+     * @return string|null
      */
     protected function getLine(
         string $id,
@@ -164,6 +161,40 @@ protected function getLine(
         return null;
     }
 
+    /**
+     * Get the contents of an MT940 line
+     *
+     * The contents may be several lines long (e.g. :86: descriptions)
+     *
+     * @param string $text The text to search
+     * @return string|null
+     */
+    protected function getTransactionLines($text): ?array {
+
+        $amountLine = [];
+        $pcre = '/(?:^|\r\n)\:(?:61)\:(.+)(?::?$|\r\n\:[[:alnum:]]{2,3}\:)/Us';
+
+        if (preg_match_all($pcre, $text, $match)) {
+            $amountLine = $match;
+        }
+
+        $multiPurposeField = [];
+        $pcre = '/(?:^|\r\n)\:(?:86)\:(.+)(?:[\r\n])(?:\:(?:6[0-9]{1}[a-zA-Z]?)\:|(?:[\r\n]-))/Us';
+
+        if (preg_match_all($pcre, $text, $match)) {
+            $multiPurposeField = $match;
+        }
+
+        $count = count($amountLine[1]);
+        $result = [];
+        for ($i = 0; $i < $count; $i++) {
+            $result[$i][] = trim($amountLine[1][$i]);
+            $result[$i][] = trim(str_replace(':86:', '', $multiPurposeField[1][$i]));
+        }
+
+        return $result;
+    }
+
     /**
      * Split the text into separate statement chunks
      *
@@ -195,35 +226,8 @@ protected function splitStatements(string $text): array
      */
     protected function splitTransactions(string $text): array
     {
-        $offset = 0;
-        $length = 0;
-        $position = 0;
-        $transactions = [];
-
-        while ($line = $this->getLine('61', $text, $offset, $offset, $length)) {
-            $offset += 4 + $length + 2;
-            $transaction = [$line];
-
-            // See if the next description line belongs to this transaction line.
-            // The description line should immediately follow the transaction line.
-            $description = [];
-            while ($line = $this->getLine('86', $text, $offset, $position, $length)) {
-                if ($position == $offset) {
-                    $offset += 4 + $length + 2;
-                    $description[] = $line;
-                } else {
-                    break;
-                }
-            }
-
-            if ($description) {
-                $transaction[] = implode("\r\n", $description);
-            }
-
-            $transactions[] = $transaction;
-        }
-
-        return $transactions;
+        $transactionLines = $this->getTransactionLines($text);
+        return $transactionLines ?? [];
     }
 
     /**
@@ -329,7 +333,7 @@ protected function accountNumber(string $text): ?string
     protected function accountCurrency($text): ?string
     {
         $accountNumber = $this->accountNumber($text);
-        if ($accountNumber == null) {
+        if ($accountNumber === null) {
             return null;
         }
         // last 3 characters comprises its ISO currency code
@@ -534,7 +538,11 @@ protected function contraAccountName(array $lines): ?string
      */
     protected function description(?string $description): ?string
     {
-        return $description;
+        if ($description === null) {
+            return null;
+        }
+        //return implode('', array_map('trim', explode("\r\n", $description)));
+        return implode("\r\n", explode("\r\n", $description));
     }
 
     /**

diff --git a/lib/Jejik/MT940/Parser/GermanBank.php b/lib/Jejik/MT940/Parser/GermanBank.php
@@ -20,7 +20,6 @@
  */
 abstract class GermanBank extends AbstractParser
 {
-
     protected const IDENTIFIER_EREF = 'EREF';
     protected const IDENTIFIER_KREF = 'KREF';
     protected const IDENTIFIER_MREF = 'MREF';
@@ -51,6 +50,7 @@ protected function gvc(array $lines): ?string
 
     /**
      * Parse code for provided transaction lines
+     * @todo: return code with leading N
      */
     protected function code(array $lines): ?string
     {
@@ -79,7 +79,6 @@ protected function code(array $lines): ?string
      */
     protected function ref(array $lines): ?string
     {
-        // get :61: line -- it is first in provided array [:61:,:86:,....]
         $refLine = isset($lines[0]) ? $lines[0] : null;
 
         // assure ref line
@@ -88,24 +87,37 @@ protected function ref(array $lines): ?string
         }
 
         // match it
-        preg_match('#(\d{6})(\d{4})?(R?(?:C|D))([0-9,]{1,15})N([a-zA-Z0-9]+)#', $refLine, $match);
+        preg_match("/(?'valuta'\d{6})(?'bookingdate'\d{4})?(?'debitcreditid'R?(?:C|D))(?'amount'[0-9,]{1,15})(?:\s*)(?'bookingkey'N[a-zA-Z0-9]{3})(?'reference'[a-zA-Z0-9+]+)(?:\/\/)*(?'bankref'[0-9a-zA-Z]{1,16})*/", $refLine, $match);
 
         // assure match
-        if (!isset($match[5])) {
+        if (!isset($match['reference'])) {
             return null;
         }
 
-        // return
-        return substr($match[5], 3);
+        return $match['reference'];
     }
 
     /**
      * Parse bankRef for provided transaction lines
      */
     protected function bankRef(array $lines): ?string
     {
-        // TODO search its proper explanation and implement it
-        return $this->ref($lines);
+        $refLine = isset($lines[0]) ? $lines[0] : null;
+
+        // assure ref line
+        if ($refLine == null) {
+            return null;
+        }
+
+        // match it
+        preg_match("/(?'valuta'\d{6})(?'bookingdate'\d{4})?(?'debitcreditid'R?(?:C|D))(?'amount'[0-9,]{1,15})(?:\s*)(?'bookingkey'N[a-zA-Z0-9]{3})(?'reference'[a-zA-Z0-9+]+)(?:\/\/)*(?'bankref'[0-9a-zA-Z]{1,16})*/", $refLine, $match);
+
+        // assure match
+        if (!isset($match['bankref'])) {
+            return null;
+        }
+
+        return $match['bankref'];
     }
 
     /**
@@ -227,7 +239,7 @@ protected function getSubfield(string $multiUseLine, string $identifier): ?strin
         $subfields = [];
 
         // check if leading value is an separator: '?20<text>' '?2X' '<identifier>' '<content>'
-        if (current($splitReferenceLine) != '?20') {
+        if (current($splitReferenceLine) !== '?20') {
             // remove first element if no separator found
             next($splitReferenceLine);
         }
@@ -321,7 +333,7 @@ protected function accountHolder(array $lines): ?string
 
         // TODO try to match names containing ? character
         preg_match(
-            '#\?32((?:[a-zA-ZöäüÖÄÜß0-9\(\)\s,\-\./\+]+(?:\?[^\?33|34])?)+)#',
+            '#\?32((?:[a-zA-ZöäüÖÄÜß0-9\(\)\s,\-\./\+]+(?:\?[^\?33|34])?)+)#u',
             $this->removeNewLinesFromLine($accHolderLine),
             $match
         );
@@ -333,7 +345,7 @@ protected function accountHolder(array $lines): ?string
         // additional field ?33
         /** @var string $accHolderLine */
         preg_match(
-            '#\?33([a-zA-ZöäüÖÄÜß0-9\(\)\s,\-\./\+]+)#',
+            '#\?33([a-zA-ZöäüÖÄÜß0-9\(\)\s,\-\./\+]+)#u',
             $this->removeNewLinesFromLine($accHolderLine),
             $matchAdd
         );
@@ -385,7 +397,6 @@ protected function mref(array $lines): ?string
         // get :86: line -- it is second in provided array [:61:,:86:,....]
         $mrefLine = isset($lines[1]) ? $lines[1] : null;
 
-        // pattern
         $pattern = 'M(?:\?2[1-9])?R(?:\?2[1-9])?E(?:\?2[1-9])?F(?:\?2[1-9])?\+([a-zA-ZöäüÖÄÜß0-9\./?\+\-\s,]+)(C(?:\?2[1-9])?R(?:\?2[1-9])?E(?:\?2[1-9])?D|S(?:\?2[1-9])?V(?:\?2[1-9])?W(?:\?2[1-9])?Z)';
 
         // match it
@@ -396,7 +407,6 @@ protected function mref(array $lines): ?string
             return null;
         }
 
-        // return
         return preg_replace('#(\?\d{0,2})#', '', $match[1]);
     }
 
@@ -408,7 +418,6 @@ protected function cred(array $lines): ?string
         // get :86: line -- it is second in provided array [:61:,:86:,....]
         $credLine = isset($lines[1]) ? $lines[1] : null;
 
-        // pattern
         $pattern = 'C(?:\?2[1-9])?R(?:\?2[1-9])?E(?:\?2[1-9])?D(?:\?2[1-9])?\+([a-zA-ZöäüÖÄÜß0-9\./?\+\-\s,]+)S(?:\?2[1-9])?V(?:\?2[1-9])?W(?:\?2[1-9])?Z';
 
         // match it
@@ -430,10 +439,8 @@ protected function svwz(array $lines): ?string
         // get :86: line -- it is second in provided array [:61:,:86:,....]
         $svwzLine = isset($lines[1]) ? $lines[1] : null;
 
-        // pattern
         $pattern = "(S(?:\?2[1-9])?V(?:\?2[1-9])?W(?:\?2[1-9])?Z(?:\?2[1-9])?\+)(?:\?(?:2[1-9]))?(?'SVWZ'.*)(?:\?30)";
 
-        // match it
         /** @var string $svwzLine */
         preg_match("/{$pattern}/", $this->removeNewLinesFromLine($svwzLine), $match);
 

diff --git a/tests/Jejik/Tests/MT940/Fixture/document/K4262927_20200905-080000-952.txt b/tests/Jejik/Tests/MT940/Fixture/document/K4262927_20200905-080000-952.txt
@@ -1,10 +1,10 @@
-:20:STARTUMS TA
-:25:28020050/1426292700
-:28C:172/1
-:60F:C200904EUR0,
-:61:2009040904C230,00N051NONREF
-:86:166?00GUTSCHRIFT?100004772?20EREF+SCP 100 / 0082002528?21SVWZ+D 8
-03020001000145464 +?221000145463 Manthey,August?30OLBODEH2XXX?31D
-E59280200501422112100?32EWE VERTRIEB GmbH?35EWE VERTRIEB powerclo
-ud
-:65F:Dnullnull7414,3
+:20:STARTUMS TA
+:25:28020050/1426292700
+:28C:172/1
+:60F:C200904EUR0,
+:61:2009040904C230,00N051NONREF
+:86:166?00GUTSCHRIFT?100004772?20EREF+SCP 100 / 0082002528?21SVWZ+D 8
+03020001000145464 +?221000145463 XXXXXXX,XXXXXX?30OLBODEH2XXX?31D
+E59280200501000000000?32EWE VERTRIEB GmbH?35EWE VERTRIEB powerclo
+ud
+:65F:Dnullnull7414,3
diff --git a/tests/Jejik/Tests/MT940/Fixture/document/bug-core-5401.txt b/tests/Jejik/Tests/MT940/Fixture/document/bug-core-5401.txt
@@ -0,0 +1,10 @@
+:20:DEUTDEMMXXX
+:25:DE19662800530622160900
+:28C:00111/1
+:60F:C210330EUR0,00
+:61:2103310331C50,00 NTRFNONREF//2109025460313532
+:86:166?00SEPA-GUTSCHRIFT?109075/611?20EREF+NOTPROVIDED SVWZ+VERTR?21
+:KO:NR:401113172APRIL 2021?30DEUTDEDB237?31DE93230707000621787100
+?32XXXXXXA MXXXXXX
+:62F:C210331EUR0,00
+-
diff --git a/tests/Jejik/Tests/MT940/Fixture/document/deutschebank.txt b/tests/Jejik/Tests/MT940/Fixture/document/deutschebank.txt
@@ -1,14 +1,14 @@
-:20:DEUTDEMMXXX
-:25:20070000/0123456601
-:28C:00143/1
-:60F:C200605EUR0,00
-:61:2006080608D12,35 NRTIKREF+//2016021783252833
-/OCMT/EUR11,85//CHGS/EUR0,50/
-:86:109?00SEPA-LASTSCHR. RETOURE CORE?109075/629?20EREF+A1.200080779.
-400143254?21.4961336 KREF+SEPA-DA202006?2201221740-34972000-P1 MR
-EF+2?230852HW2723821 CRED+DE41EON0?240000129793 OAMT+11,85 SVWZ+?
-25SONSTIGE GRUENDE ENDABRECHN?26UNG NR. 500106875 ZU VERTRA?2740
-0143254, KUNDENNUM MER 2?2802227779?30CSDBDE71XXX?31DE50712345600
-200691329?32TESTER?33EL
-:62F:C200608EUR0,00
--
+:20:DEUTDEMMXXX
+:25:20070000/0123456601
+:28C:00143/1
+:60F:C200605EUR0,00
+:61:2006080608D12,35 NRTIKREF+//2016021783252833
+/OCMT/EUR11,85//CHGS/EUR0,50/
+:86:109?00SEPA-LASTSCHR. RETOURE CORE?109075/629?20EREF+A1.200080779.
+400143254?21.4961336 KREF+SEPA-DA202006?2201221740-34972000-P1 MR
+EF+2?230852HW2723821 CRED+DE41EON0?240000129793 OAMT+11,85 SVWZ+?
+25SONSTIGE GRUENDE ENDABRECHN?26UNG NR. 500106875 ZU VERTRA?2740
+0143254, KUNDENNUM MER 2?2802227779?30CSDBDE71XXX?31DE50712345600
+200691329?32TESTER?33EL
+:62F:C200608EUR0,00
+-
diff --git a/tests/Jejik/Tests/MT940/Fixture/document/lbbw.txt b/tests/Jejik/Tests/MT940/Fixture/document/lbbw.txt
@@ -1,15 +1,15 @@
-:20:LBBW
-:21:NONREF
-:25:12345678/1324357
-:28C:1
-:60F:C210120EUR0,00
-:61:2102080208D0,01NCMZNONREF
-:86:834?00KONTENPOOL?102?20BUCHUNG AUF KTO 7402050699?21BANKLEITZAHL
-60050101?3060050101?3111111111
-:61:2102080208C0,01NCOLKREF+//202102040007693
-:86:171?00SEPA EINZUGSAUFTRAG?101?20KREF+SEPA-20210203175805-00?21154
-800-P1?22SVWZ+E-MOBILITY ABRECHNUNG?23NR. 28 ZU VERTRAG 9433, KUN
-?24DENNUMMER 11111?30SOLADEST600?31DE59600501010007907986?32NIC R
-ICHTER
-:62F:C210208EUR0,00
--
+:20:LBBW
+:21:NONREF
+:25:12345678/1324357
+:28C:1
+:60F:C210120EUR0,00
+:61:2102080208D0,01NCMZNONREF
+:86:834?00KONTENPOOL?102?20BUCHUNG AUF KTO 7402050699?21BANKLEITZAHL
+60050101?3060050101?3111111111
+:61:2102080208C0,01NCOLKREF+//202102040007693
+:86:171?00SEPA EINZUGSAUFTRAG?101?20KREF+SEPA-20210203175805-00?21154
+800-P1?22SVWZ+E-MOBILITY ABRECHNUNG?23NR. 28 ZU VERTRAG 9433, KUN
+?24DENNUMMER 11111?30SOLADEST600?31DE59600501010007907986?32NIC R
+ICHTER
+:62F:C210208EUR0,00
+-
diff --git a/tests/Jejik/Tests/MT940/Fixture/document/oldenburgischelandesbank.txt b/tests/Jejik/Tests/MT940/Fixture/document/oldenburgischelandesbank.txt
@@ -1,8 +1,8 @@
-:20:STARTUMS MC
-:25:28020050/1324354687
-:28C:135/1
-:60F:C180713EUR0,
-:61:1807160716C104,50N051NONREF
-:86:166?00GUTSCHRIFT?100004770?20WOHNBAU DIEPHOLZ GMBH?21EREF+NOTPROVIDED?22SVWZ+EWE,ENERGIEAUSWEIS RG.?23-NR. 1234567890 KD.-NR. 123?2477777?30AARBDE5W250?31DE99123456771234567888?32WOHNBAU DIEPHOLZ GMBH?35EWE VERTRIEB GMBH
-:62F:C180716EUR0,
--
+:20:STARTUMS MC
+:25:28020050/1324354687
+:28C:135/1
+:60F:C180713EUR0,
+:61:1807160716C104,50N051NONREF
+:86:166?00GUTSCHRIFT?100004770?20WOHNBAU DIEPHOLZ GMBH?21EREF+NOTPROVIDED?22SVWZ+EWE,ENERGIEAUSWEIS RG.?23-NR. 1234567890 KD.-NR. 123?2477777?30AARBDE5W250?31DE99123456771234567888?32WOHNBAU DIEPHOLZ GMBH?35EWE VERTRIEB GMBH
+:62F:C180716EUR0,
+-
diff --git a/tests/Jejik/Tests/MT940/Fixture/document/oldenburgischelandesbank2.txt b/tests/Jejik/Tests/MT940/Fixture/document/oldenburgischelandesbank2.txt
@@ -0,0 +1,10 @@
+:20:STARTUMS TA
+:25:DE19662800530622160900
+:28C:172/1
+:60F:C200904EUR0,
+:61:2009040904C230,00N051NONREF
+:86:166?00GUTSCHRIFT?100004772?20EREF+SCP 100 / 0082002528?21SVWZ+D 8
+03020001000145464 +?221000145463 XXXXXXX,XXXXXX?30OLBODEH2XXX?31D
+E59280200501000000000?32EWE VERTRIEB GmbH?35EWE VERTRIEB powerclo
+ud
+:65F:Dnullnull7414,3