Initial xentral_oss_20.3.c9ffacf

classes/Modules/HocrParser/Finder/BoundingBoxFinder.php

@@ -0,0 +1,63 @@
+<?php
+
+namespace Xentral\Modules\HocrParser\Finder;
+
+use Xentral\Modules\HocrParser\Data\BoundingBox;
+use Xentral\Modules\HocrParser\Data\BoundingBoxCollection;
+
+class BoundingBoxFinder
+{
+    /** @var BoundingBoxCollection $boxes */
+    private $boxes;
+
+    /** @var array|RelativePositionFinderFacet[] $criteria */
+    private $criteria;
+
+    /** @var array $currencies */
+    private $currencies;
+
+    /**
+     * @param BoundingBoxCollection               $boxes
+     * @param array|RelativePositionFinderFacet[] $criteria
+     * @param array                               $validCurrencies
+     */
+    public function __construct(BoundingBoxCollection $boxes, array $criteria, array $validCurrencies = [])
+    {
+        $this->boxes = $boxes;
+        $this->criteria = $criteria;
+        $this->currencies = $validCurrencies;
+    }
+
+    /**
+     * @return array
+     */
+    public function Find()
+    {
+        $search = [];
+        $result = [];
+
+        // Alle Boxen finden die den jeweiligen Suchbegriff enthalten
+        /** @var BoundingBox $box */
+        foreach ($this->boxes->GetBoxes() as $box) {
+            foreach ($this->criteria as $searchKey => $criteria) {
+                $result[$searchKey] = null;
+
+                // Alle Boxen sammeln die den Vorbedingungen entsprechen
+                if ($criteria->MatchPreCondition($box->GetData('text'))) {
+                    $search[$searchKey]['boxes'][] = $box;
+                    $search[$searchKey]['criteria'] = $criteria;
+                }
+            }
+        }
+
+        // Schauen welche der Kandidaten die genauen Vorgaben erfüllt; der erste Treffer gewinnt
+        foreach ($search as $searchKey => $searchInfo) {
+            $criteria = $searchInfo['criteria'];
+            $candidates = $searchInfo['boxes'];
+            $result[$searchKey] = $criteria->Select($candidates, $this->boxes);
+        }
+
+        return $result;
+    }
+}
+

classes/Modules/HocrParser/Finder/CurrencyCodeFinderFacet.php

@@ -0,0 +1,75 @@
+<?php
+
+namespace Xentral\Modules\HocrParser\Finder;
+
+use Xentral\Modules\HocrParser\Data\BoundingBox;
+use Xentral\Modules\HocrParser\Data\BoundingBoxCollection;
+
+class CurrencyCodeFinderFacet implements FinderFacetInterface
+{
+    /** @var array $validCodes */
+    private $validCodes;
+
+    /**
+     * @param array $validCodes Array mit gültigen Währungscodes (drei-stelliger ISO-Code)
+     */
+    public function __construct(array $validCodes = ['EUR'])
+    {
+        $this->validCodes = $validCodes;
+    }
+
+    /**
+     * @param string $text
+     *
+     * @return bool
+     */
+    public function MatchPreCondition($text)
+    {
+        if (!$this->IsCurrencyLikeValue($text)) {
+            return false;
+        }
+
+        if (!$this->IsValidCurrency($text)) {
+            return false;
+        }
+
+        return true;
+    }
+
+    /**
+     * @param string $value
+     *
+     * @return bool
+     */
+    private function IsCurrencyLikeValue($value)
+    {
+        return (bool)preg_match('/^[A-Z]{3}$/', $value);
+    }
+
+    /**
+     * @param string $value
+     *
+     * @return bool
+     */
+    private function IsValidCurrency($value)
+    {
+        return in_array($value, $this->validCodes, true);
+    }
+
+    /**
+     * @param array|BoundingBox[]   $candidates
+     * @param BoundingBoxCollection $boxes
+     *
+     * @return string|false
+     */
+    public function Select(array $candidates, BoundingBoxCollection $boxes)
+    {
+        if (empty($candidates)) {
+            return false;
+        }
+
+        // Einfach das erste Ergebnis zurückliefern;
+        // In den PreConditions wurde schon sichergestellt dass es eine gültige Währung ist
+        return $candidates[0]->GetData('text');
+    }
+}

classes/Modules/HocrParser/Finder/FinderFacetInterface.php

@@ -0,0 +1,12 @@
+<?php
+
+namespace Xentral\Modules\HocrParser\Finder;
+
+use Xentral\Modules\HocrParser\Data\BoundingBoxCollection;
+
+interface FinderFacetInterface
+{
+    public function MatchPreCondition($text);
+
+    public function Select(array $candidates, BoundingBoxCollection $boxes);
+}

classes/Modules/HocrParser/Finder/PatternMatcher.php

@@ -0,0 +1,123 @@
+<?php
+
+namespace Xentral\Modules\HocrParser\Finder;
+
+use Xentral\Modules\HocrParser\Exception\InvalidArgumentException;
+
+class PatternMatcher
+{
+    public const PATTERN_DOCUMENT_NUMBER = 'documentnumber';
+    public const PATTERN_MONEY = 'money';
+    public const PATTERN_DATE = 'date';
+    public const PATTERN_DEFAULT = 'default';
+
+    /** @var array $validPatterns */
+    private static $validPatterns = [
+        self::PATTERN_DOCUMENT_NUMBER,
+        self::PATTERN_MONEY,
+        self::PATTERN_DATE,
+        self::PATTERN_DEFAULT,
+    ];
+
+    /** @var string $pattern */
+    private $pattern;
+
+    /**
+     * @param string $pattern
+     */
+    public function __construct($pattern = self::PATTERN_DEFAULT)
+    {
+        if (!in_array($pattern, self::$validPatterns, true)) {
+            throw new InvalidArgumentException(sprintf('Pattern "%s" is not allowed.', $pattern));
+        }
+
+        $this->pattern = $pattern;
+    }
+
+    /**
+     * @param string $value
+     *
+     * @return bool
+     */
+    public function Match($value)
+    {
+        $value = trim((string)$value);
+
+        if (empty($value)) {
+            return false;
+        }
+
+        switch ($this->pattern) {
+            case self::PATTERN_DATE:
+                return $this->IsDateLikeValue($value);
+                break;
+            case self::PATTERN_MONEY:
+                return $this->IsMoneyLikeValue($value);
+                break;
+            case self::PATTERN_DOCUMENT_NUMBER:
+                return $this->IsDocumentNumberLikeValue($value);
+                break;
+            case self::PATTERN_DEFAULT:
+                return $this->IsCandidateValue($value);
+                break;
+        }
+
+        return false;
+    }
+
+    /**
+     * @param string $value
+     *
+     * @return bool
+     */
+    private function IsDateLikeValue($value)
+    {
+        return (bool)preg_match('/\d{1,2}\.\d{1,2}\.\d{2,4}/', $value);
+    }
+
+    /**
+     * @param string $value
+     *
+     * @return bool
+     */
+    private function IsMoneyLikeValue($value)
+    {
+        // Mit Tausendertrenner: z.B.: 11.111,11 oder 11,111.11
+        $withThousands = (bool)preg_match('/\d+[\.,]\d{3}[\.,]{1}\d{2}$/', $value);
+        if ($withThousands) {
+            return true;
+        }
+
+        // Ohne Tausendertrenner: z.B.: 1111111,11 oder 1111111.11
+        return (bool)preg_match('/^\d+[\.,]{1}\d{2}$/', $value);
+    }
+
+    /**
+     * @param $value
+     *
+     * @return bool
+     */
+    private function IsDocumentNumberLikeValue($value)
+    {
+        // Nur Grossbuchstaben, Zahlen, Minus und Unterstrich sind erlaubt
+        $containsInvalidChars = (bool)preg_match('/[^A-Z0-9\-_]+/', $value);
+        if ($containsInvalidChars) {
+            return false;
+        }
+
+        return (bool)preg_match('/\d{4,}/', $value);
+    }
+
+    /**
+     * @param string $value
+     *
+     * @return bool
+     */
+    private function IsCandidateValue($value)
+    {
+        return $this->IsDateLikeValue($value)
+            || $this->IsDocumentNumberLikeValue($value)
+            || $this->IsMoneyLikeValue($value);
+    }
+}
+

classes/Modules/HocrParser/Finder/RelativePositionFinderFacet.php

@@ -0,0 +1,111 @@
+<?php
+
+namespace Xentral\Modules\HocrParser\Finder;
+
+use Xentral\Modules\HocrParser\Data\BoundingBox;
+use Xentral\Modules\HocrParser\Data\BoundingBoxCollection;
+use Xentral\Modules\HocrParser\Exception\InvalidArgumentException;
+
+class RelativePositionFinderFacet implements FinderFacetInterface
+{
+    const DIRECTION_LEFT = 'left';
+    const DIRECTION_RIGHT = 'right';
+    const DIRECTION_ABOVE = 'above';
+    const DIRECTION_BELOW = 'below';
+
+    /** @var array $validDirections */
+    private static $validDirections = [
+        self::DIRECTION_ABOVE,
+        self::DIRECTION_RIGHT,
+        self::DIRECTION_BELOW,
+        self::DIRECTION_LEFT,
+    ];
+
+    /** @var string $text */
+    private $text;
+
+    /** @var PatternMatcher $matcher */
+    private $matcher;
+
+    /** @var string $direction */
+    private $direction;
+
+    /**
+     * @param string $searchText
+     * @param string $direction
+     * @param string $pattern
+     */
+    public function __construct($searchText, $direction, $pattern)
+    {
+        if (!in_array($direction, self::$validDirections, true)) {
+            throw new InvalidArgumentException(sprintf('Direction "%s" is not allowed.', $direction));
+        }
+
+        $this->matcher = new PatternMatcher($pattern);
+        $this->text = trim($searchText);
+        $this->direction = $direction;
+    }
+
+    /**
+     * @param string $text
+     *
+     * @return bool
+     */
+    public function MatchPreCondition($text)
+    {
+        return $text === $this->text;
+    }
+
+    /**
+     * @param array|BoundingBox[]   $candidates
+     * @param BoundingBoxCollection $boxes
+     *
+     * @return string|false
+     */
+    public function Select(array $candidates, BoundingBoxCollection $boxes)
+    {
+        foreach ($candidates as $candidate) {
+            $coords = $candidate->GetCenterPoint();
+            /** @var BoundingBox $nearestBox */
+            $nearestBox = false;
+
+            switch ($this->direction) {
+                case self::DIRECTION_RIGHT:
+                    $nearestBox = $boxes->GetNearestBoxRightFromPoint($coords['x'], $coords['y']);
+                    break;
+                case self::DIRECTION_LEFT:
+                    $nearestBox = $boxes->GetNearestBoxLeftFromPoint($coords['x'], $coords['y']);
+                    break;
+                case self::DIRECTION_ABOVE:
+                    $nearestBox = $boxes->GetNearestBoxAboveFromPoint($coords['x'], $coords['y']);
+                    break;
+                case self::DIRECTION_BELOW:
+                    $nearestBox = $boxes->GetNearestBoxBelowFromPoint($coords['x'], $coords['y']);
+                    break;
+            }
+
+            if ($nearestBox === false) {
+                continue;
+            }
+
+            // Prüfen ob Wert einem bestimmten Muster folgt
+            $patternMatching = $this->IsPatternMatching($nearestBox->GetData('text'));
+            if ($patternMatching) {
+                return $nearestBox->GetData('text');
+            }
+        }
+
+        return false;
+    }
+
+    /**
+     * @param string $value
+     *
+     * @return bool
+     */
+    private function IsPatternMatching($value)
+    {
+        return $this->matcher->Match($value);
+    }
+}
+