Handle the issue with FNB's multiple elements on a single line which …

lib/OfxParser/Ofx.php

@@ -147,32 +147,47 @@ private function buildBankAccounts(SimpleXMLElement $xml)
         // Loop through the bank accounts
         $bankAccounts = [];
         foreach ($xml->BANKMSGSRSV1->STMTTRNRS as $accountStatement) {
-            $bankAccounts[] = $this->buildBankAccount($accountStatement);
+            foreach ($accountStatement->STMTRS as $statementResponse) {
+                $bankAccounts[] = $this->buildBankAccount($accountStatement->TRNUID, $statementResponse);
+            }
         }
         return $bankAccounts;
     }
 
     /**
-     * @param SimpleXMLElement $xml
+     * @param string $transactionUid
+     * @param SimpleXMLElement $statementResponse
      * @return BankAccount
      * @throws \Exception
      */
-    private function buildBankAccount(SimpleXMLElement $xml)
+    private function buildBankAccount($transactionUid, SimpleXMLElement $statementResponse)
     {
         $bankAccount = new BankAccount();
-        $bankAccount->transactionUid = $xml->TRNUID;
-        $bankAccount->agencyNumber = $xml->STMTRS->BANKACCTFROM->BRANCHID;
-        $bankAccount->accountNumber = $xml->STMTRS->BANKACCTFROM->ACCTID;
-        $bankAccount->routingNumber = $xml->STMTRS->BANKACCTFROM->BANKID;
-        $bankAccount->accountType = $xml->STMTRS->BANKACCTFROM->ACCTTYPE;
-        $bankAccount->balance = $xml->STMTRS->LEDGERBAL->BALAMT;
-        $bankAccount->balanceDate = $this->createDateTimeFromStr($xml->STMTRS->LEDGERBAL->DTASOF, true);
+        $bankAccount->transactionUid = $transactionUid;
+        $bankAccount->agencyNumber = $statementResponse->BANKACCTFROM->BRANCHID;
+        $bankAccount->accountNumber = $statementResponse->BANKACCTFROM->ACCTID;
+        $bankAccount->routingNumber = $statementResponse->BANKACCTFROM->BANKID;
+        $bankAccount->accountType = $statementResponse->BANKACCTFROM->ACCTTYPE;
+        $bankAccount->balance = $statementResponse->LEDGERBAL->BALAMT;
+        $bankAccount->balanceDate = $this->createDateTimeFromStr(
+            $statementResponse->LEDGERBAL->DTASOF,
+            true
+        );
 
         $bankAccount->statement = new Statement();
-        $bankAccount->statement->currency = $xml->STMTRS->CURDEF;
-        $bankAccount->statement->startDate = $this->createDateTimeFromStr($xml->STMTRS->BANKTRANLIST->DTSTART);
-        $bankAccount->statement->endDate = $this->createDateTimeFromStr($xml->STMTRS->BANKTRANLIST->DTEND);
-        $bankAccount->statement->transactions = $this->buildTransactions($xml->STMTRS->BANKTRANLIST->STMTTRN);
+        $bankAccount->statement->currency = $statementResponse->CURDEF;
+
+        $bankAccount->statement->startDate = $this->createDateTimeFromStr(
+            $statementResponse->BANKTRANLIST->DTSTART
+        );
+
+        $bankAccount->statement->endDate = $this->createDateTimeFromStr(
+            $statementResponse->BANKTRANLIST->DTEND
+        );
+
+        $bankAccount->statement->transactions = $this->buildTransactions(
+            $statementResponse->BANKTRANLIST->STMTTRN
+        );
 
         return $bankAccount;
     }
@@ -264,6 +279,8 @@ private function buildStatus(SimpleXMLElement $xml)
      */
     private function createDateTimeFromStr($dateString, $ignoreErrors = false)
     {
+        if((!isset($dateString) || trim($dateString) === '')) return null;
+
         $regex = '/'
             . "(\d{4})(\d{2})(\d{2})?"     // YYYYMMDD             1,2,3
             . "(?:(\d{2})(\d{2})(\d{2}))?" // HHMMSS   - optional  4,5,6
@@ -310,7 +327,7 @@ private function createDateTimeFromStr($dateString, $ignoreErrors = false)
     private function createAmountFromStr($amountString)
     {
         // Decimal mark style (UK/US): 000.00 or 0,000.00
-        if (preg_match('/^-?([\d,]+)(\.?)([\d]{2})$/', $amountString) === 1) {
+        if (preg_match('/^(-|\+)?([\d,]+)(\.?)([\d]{2})$/', $amountString) === 1) {
             return (float)preg_replace(
                 ['/([,]+)/', '/\.?([\d]{2})$/'],
                 ['', '.$1'],
@@ -319,7 +336,7 @@ private function createAmountFromStr($amountString)
         }
 
         // European style: 000,00 or 0.000,00
-        if (preg_match('/^-?([\d\.]+,?[\d]{2})$/', $amountString) === 1) {
+        if (preg_match('/^(-|\+)?([\d\.]+,?[\d]{2})$/', $amountString) === 1) {
             return (float)preg_replace(
                 ['/([\.]+)/', '/,?([\d]{2})$/'],
                 ['', '.$1'],

lib/OfxParser/Parser.php

@@ -63,6 +63,10 @@ private function conditionallyAddNewlines($ofxContent)
             return str_replace('<', "\n<", $ofxContent); // add line breaks to allow XML to parse
         }
 
+        if (preg_match('/<.+>.*<.+>/', $ofxContent) === 1) {
+            return str_replace('<', "\n<", $ofxContent); // add line breaks to allow XML to parse
+        }
+
         return $ofxContent;
     }
 
@@ -98,7 +102,7 @@ private function closeUnclosedXmlTags($line)
         // Does not match: <SOMETHING>
         // Does not match: <SOMETHING>blah</SOMETHING>
         if (preg_match(
-            "/<([A-Za-z0-9.]+)>([\wà-úÀ-Ú0-9\.\-\_\+\, ;:\[\]\'\&\/\\\*\(\)\+\{\|\}\!\£\$\?=@€£#%±§~`]+)$/",
+            "/<([A-Za-z0-9.]+)>([\wà-úÀ-Ú0-9\.\-\_\+\, ;:\[\]\'\&\/\\\*\(\)\+\{\|\}\!\£\$\?=@€£#%±§~`\"]+)$/",
             trim($line),
             $matches
         )) {
@@ -117,6 +121,8 @@ private function convertSgmlToXml($sgml)
     {
         $sgml = str_replace(["\r\n", "\r"], "\n", $sgml);
 
+        $sgml = preg_replace('/&(?!#?[a-z0-9]+;)/', '&amp;', $sgml);
+
         $lines = explode("\n", $sgml);
 
         $xml = '';

lib/OfxParser/Parserv2.php

@@ -0,0 +1,126 @@
+<?php
+
+namespace OfxParser;
+
+/**
+ * An OFX parser library
+ *
+ * Heavily refactored from Guillaume Bailleul's grimfor/ofxparser
+ *
+ * @author Guillaume BAILLEUL <[email protected]>
+ * @author James Titcumb <[email protected]>
+ * @author Oliver Lowe <[email protected]>
+ */
+class Parserv2
+{
+    /**
+     * Load an OFX file into this parser by way of a filename
+     *
+     * @param string $ofxFile A path that can be loaded with file_get_contents
+     * @return Ofx
+     * @throws \Exception
+     */
+    public function loadFromFile($ofxFile)
+    {
+        if (!file_exists($ofxFile)) {
+            throw new \InvalidArgumentException("File '{$ofxFile}' could not be found");
+        }
+
+        return $this->loadFromString(file_get_contents($ofxFile));
+    }
+
+    /**
+     * Load an OFX by directly using the text content
+     *
+     * @param string $ofxContent
+     * @return  Ofx
+     * @throws \Exception
+     */
+    public function loadFromString($ofxContent)
+    {
+        $ofxContent = utf8_encode($ofxContent);
+        $xml = $this->xmlLoadString($ofxContent);
+
+        return new Ofx($xml);
+    }
+
+    /**
+     * Detect if the OFX file is on one line. If it is, add newlines automatically.
+     *
+     * @param string $ofxContent
+     * @return string
+     */
+    private function conditionallyAddNewlines($ofxContent)
+    {
+        if (preg_match('/<OFX>.*<\/OFX>/', $ofxContent) === 1) {
+            return str_replace('<', "\n<", $ofxContent); // add line breaks to allow XML to parse
+        }
+
+        if (preg_match('/<.+>.*<.+>/', $ofxContent) === 1) {
+            return str_replace('<', "\n<", $ofxContent); // add line breaks to allow XML to parse
+        }
+
+        return $ofxContent;
+    }
+
+    /**
+     * Load an XML string without PHP errors - throws exception instead
+     *
+     * @param string $xmlString
+     * @throws \Exception
+     * @return \SimpleXMLElement
+     */
+    private function xmlLoadString($xmlString)
+    {
+        libxml_clear_errors();
+        libxml_use_internal_errors(true);
+        $xml = simplexml_load_string($xmlString);
+
+        if ($errors = libxml_get_errors()) {
+            throw new \RuntimeException('Failed to parse OFX: ' . var_export($errors, true));
+        }
+
+        return $xml;
+    }
+
+    /**
+     * Detect any unclosed XML tags - if they exist, close them
+     *
+     * @param string $line
+     * @return string
+     */
+    private function closeUnclosedXmlTags($line)
+    {
+        // Matches: <SOMETHING>blah
+        // Does not match: <SOMETHING>
+        // Does not match: <SOMETHING>blah</SOMETHING>
+        if (preg_match(
+            "/<([A-Za-z0-9.]+)>([\wà-úÀ-Ú0-9\.\-\_\+\, ;:\[\]\'\&\/\\\*\(\)\+\{\|\}\!\£\$\?=@€£#%±§~`]+)$/",
+            trim($line),
+            $matches
+        )) {
+            return "<{$matches[1]}>{$matches[2]}</{$matches[1]}>";
+        }
+        return $line;
+    }
+
+    /**
+     * Convert an SGML to an XML string
+     *
+     * @param string $sgml
+     * @return string
+     */
+    private function convertSgmlToXml($sgml)
+    {
+        $sgml = str_replace(["\r\n", "\r"], "\n", $sgml);
+
+        $lines = explode("\n", $sgml);
+
+        $xml = '';
+        foreach ($lines as $line) {
+            $xml .= trim($this->closeUnclosedXmlTags($line)) . "\n";
+        }
+
+        return trim($xml);
+    }
+}

tests/OfxParser/OfxTest.php

@@ -38,6 +38,12 @@ public function amountConversionProvider()
             '1' => ['1', 1.0],
             '10' => ['10', 10.0],
             '100' => ['100', 1.0], // @todo this is weird behaviour, should not really expect this
+            '+1' => ['+1', 1.0],
+            '+10' => ['+10', 10.0],
+            '+1000.00' => ['+1000.00', 1000.0],
+            '+1000,00' => ['+1000,00', 1000.0],
+            '+1,000.00' => ['+1,000.00', 1000.0],
+            '+1.000,00' => ['+1.000,00', 1000.0],
         ];
     }
 
@@ -81,6 +87,11 @@ public function testCreateDateTimeFromOFXDateFormats()
         // Test YYYYMMDDHHMMSS.XXX
         $DateTimeFour = $method->invoke($Ofx, '20081005132200.124');
         self::assertEquals($expectedDateTime->getTimestamp(), $DateTimeFour->getTimestamp());
+
+        // Test empty datetime
+        $DateTimeFour = $method->invoke($Ofx, '');
+        self::assertEquals(null, $DateTimeFour);
+
     }
 
     public function testBuildsSignOn()

tests/OfxParser/ParserTest.php

@@ -112,6 +112,18 @@ public function convertSgmlToXmlProvider()
 <ACCTID>XXXXXXXXXXX</ACCTID>
 <ACCTTYPE>CHECKING</ACCTTYPE>
 </BANKACCTFROM>
+HERE
+            ],[<<<HERE
+<SOMETHING>
+    <FOO>bar & restaurant
+    <BAZ>bat</BAZ>
+</SOMETHING>
+HERE
+        , <<<HERE
+<SOMETHING>
+<FOO>bar &amp; restaurant</FOO>
+<BAZ>bat</BAZ>
+</SOMETHING>
 HERE
             ],
         ];
@@ -163,8 +175,12 @@ public function loadFromStringProvider()
             'ofxdata-oneline.ofx' => [dirname(__DIR__).'/fixtures/ofxdata-oneline.ofx'],
             'ofxdata-cmfr.ofx' => [dirname(__DIR__).'/fixtures/ofxdata-cmfr.ofx'],
             'ofxdata-bb.ofx' => [dirname(__DIR__).'/fixtures/ofxdata-bb.ofx'],
+            'ofxdata-bb-two-stmtrs.ofx' => [dirname(__DIR__).'/fixtures/ofxdata-bb-two-stmtrs.ofx'],
             'ofxdata-credit-card.ofx' => [dirname(__DIR__).'/fixtures/ofxdata-credit-card.ofx'],
             'ofxdata-bpbfc.ofx' => [dirname(__DIR__).'/fixtures/ofxdata-bpbfc.ofx'],
+            'ofxdata-memoWithQuotes.ofx' => [dirname(__DIR__).'/fixtures/ofxdata-memoWithQuotes.ofx'],
+            'ofxdata-emptyDateTime.ofx' => [dirname(__DIR__).'/fixtures/ofxdata-emptyDateTime.ofx'],
+            'ofxdata-memoWithAmpersand.ofx' => [dirname(__DIR__).'/fixtures/ofxdata-memoWithAmpersand.ofx'],
         ];
     }