From 8bac924d48dd9faa3cc4d8b75d841ccf2be0c11a Mon Sep 17 00:00:00 2001
From: Adrien Loison <aloison@box.com>
Date: Wed, 3 Jun 2015 10:54:53 -0700
Subject: [PATCH] Add support for more cell types

Added proper support for booleans, dates, numbers, errors.
Added unescaping of the read string.
Fixed a bug when cells did not have any values => now returns empty string.
---
 src/Spout/Reader/XLSX.php                     |  77 +++++++++++++-----
 tests/Spout/Reader/XLSXTest.php               |  21 +++++
 .../xlsx/sheet_with_all_cell_types.xlsx       | Bin 0 -> 3715 bytes
 3 files changed, 78 insertions(+), 20 deletions(-)
 create mode 100644 tests/resources/xlsx/sheet_with_all_cell_types.xlsx

diff --git a/src/Spout/Reader/XLSX.php b/src/Spout/Reader/XLSX.php
index adb9544..83bee9d 100644
--- a/src/Spout/Reader/XLSX.php
+++ b/src/Spout/Reader/XLSX.php
@@ -177,6 +177,7 @@ class XLSX extends AbstractReader
             throw new BadUsageException('You must call nextSheet() before calling hasNextRow() or nextRow()');
         }
 
+        $escaper = new \Box\Spout\Common\Escaper\XLSX();
         $isInsideRowTag = false;
         $rowData = [];
 
@@ -188,6 +189,7 @@ class XLSX extends AbstractReader
                     $lastCellIndex = $matches[1];
                     $this->numberOfColumns = CellHelper::getColumnIndexFromCellIndex($lastCellIndex) + 1;
                 }
+
             } else if ($this->xmlReader->nodeType == \XMLReader::ELEMENT && $this->xmlReader->name === 'row') {
                 // Start of the row description
                 $isInsideRowTag = true;
@@ -200,32 +202,15 @@ class XLSX extends AbstractReader
                     $numberOfColumnsForRow = intval($numberOfColumnsForRow);
                 }
                 $rowData = ($numberOfColumnsForRow !== 0) ? array_fill(0, $numberOfColumnsForRow, '') : [];
+
             } else if ($isInsideRowTag && $this->xmlReader->nodeType == \XMLReader::ELEMENT && $this->xmlReader->name === 'c') {
                 // Start of a cell description
                 $currentCellIndex = $this->xmlReader->getAttribute('r');
                 $currentColumnIndex = CellHelper::getColumnIndexFromCellIndex($currentCellIndex);
+
                 $node = $this->xmlReader->expand();
+                $rowData[$currentColumnIndex] = $this->getCellValue($node, $escaper);
 
-                $hasInlineString = ($this->xmlReader->getAttribute('t') === 'inlineStr');
-                $hasSharedString = ($this->xmlReader->getAttribute('t') === 's');
-
-                if ($hasInlineString) {
-                    // inline strings are formatted this way:
-                    // <c r="A1" t="inlineStr"><is><t>[INLINE_STRING]</t></is></c>
-                    $tNode = $node->getElementsByTagName('t')->item(0);
-                    $rowData[$currentColumnIndex] = trim($tNode->nodeValue);
-                } else if ($hasSharedString) {
-                    // shared strings are formatted this way:
-                    // <c r="A1" t="s"><v>[SHARED_STRING_INDEX]</v></c>
-                    $vNode = $node->getElementsByTagName('v')->item(0);
-                    $sharedStringIndex = intval($vNode->nodeValue);
-                    $rowData[$currentColumnIndex] = $this->sharedStringsHelper->getStringAtIndex($sharedStringIndex);
-                } else {
-                    // other values are formatted this way:
-                    // <c r="A1"><v>[VALUE]</v></c>
-                    $vNode = $node->getElementsByTagName('v')->item(0);
-                    $rowData[$currentColumnIndex] = intval($vNode->nodeValue);
-                }
             } else if ($this->xmlReader->nodeType == \XMLReader::END_ELEMENT && $this->xmlReader->name === 'row') {
                 // End of the row description
                 // If needed, we fill the empty cells
@@ -238,6 +223,58 @@ class XLSX extends AbstractReader
         return ($rowData !== []) ? $rowData : null;
     }
 
+    /**
+     * Returns the (unescaped) cell value associated to the given XML node.
+     *
+     * @param \DOMNode $node
+     * @param \Box\Spout\Common\Escaper\XLSX $escaper
+     * @return string|int|float|bool|null The value associated with the cell (null when the cell has an error)
+     */
+    protected function getCellValue($node, $escaper)
+    {
+        $cellValue = '';
+
+        // Default cell type is "n"
+        $cellType = $node->getAttribute('t') ?: 'n';
+
+        if ($cellType === 'inlineStr') {
+            // inline strings are formatted this way:
+            // <c r="A1" t="inlineStr"><is><t>[INLINE_STRING]</t></is></c>
+            $tNode = $node->getElementsByTagName('t')->item(0);
+            $escapedCellValue = trim($tNode->nodeValue);
+            $cellValue = $escaper->unescape($escapedCellValue);
+        } else {
+            // all other cell types should have a "v" tag containing the value.
+            // if not, the returned value should be empty string.
+            $vNode = $node->getElementsByTagName('v')->item(0);
+
+            if ($vNode !== null) {
+                if ($cellType === 's') {
+                    // shared strings are formatted this way:
+                    // <c r="A1" t="s"><v>[SHARED_STRING_INDEX]</v></c>
+                    $sharedStringIndex = intval($vNode->nodeValue);
+                    $escapedCellValue = $this->sharedStringsHelper->getStringAtIndex($sharedStringIndex);
+                    $cellValue = $escaper->unescape($escapedCellValue);
+                } else if ($cellType === 'b') {
+                    // !! is similar to boolval()
+                    $cellValue = !!$vNode->nodeValue;
+                } else if ($cellType === 'n') {
+                    $nodeValue = $vNode->nodeValue;
+                    $cellValue = is_int($nodeValue) ? intval($nodeValue) : floatval($nodeValue);
+                } else if ($cellType === 'd') {
+                    $cellValue = new \DateTime($vNode->nodeValue);
+                } else if ($cellType === 'e') {
+                    $cellValue = null;
+                } else if ($cellType === 'str') {
+                    $escapedCellValue = trim($vNode->nodeValue);
+                    $cellValue = $escaper->unescape($escapedCellValue);
+                }
+            }
+        }
+
+        return $cellValue;
+    }
+
     /**
      * Closes the reader. To be used after reading the file.
      *
diff --git a/tests/Spout/Reader/XLSXTest.php b/tests/Spout/Reader/XLSXTest.php
index 24aa89c..f611db4 100644
--- a/tests/Spout/Reader/XLSXTest.php
+++ b/tests/Spout/Reader/XLSXTest.php
@@ -114,6 +114,27 @@ class XLSXTest extends \PHPUnit_Framework_TestCase
         $this->assertEquals($expectedRows, $allRows);
     }
 
+    /**
+     * @return void
+     */
+    public function testReadShouldSupportAllCellTypes()
+    {
+        $allRows = $this->getAllRowsForFile('sheet_with_all_cell_types.xlsx');
+
+        $expectedRows = [
+            [
+                's1--A1', 's1--A2',
+                false, true,
+                \DateTime::createFromFormat('Y-m-d H:i:s', '2015-06-03 13:21:58'),
+                \DateTime::createFromFormat('Y-m-d H:i:s', '2015-06-01 00:00:00'),
+                10, 10.43,
+                null,
+            ],
+            ['', '', '', '', '', '', '', '', ''],
+        ];
+        $this->assertEquals($expectedRows, $allRows);
+    }
+
     /**
      * @return void
      */
diff --git a/tests/resources/xlsx/sheet_with_all_cell_types.xlsx b/tests/resources/xlsx/sheet_with_all_cell_types.xlsx
new file mode 100644
index 0000000000000000000000000000000000000000..17b730308375b7b8415829ea2acaeb2691428570
GIT binary patch
literal 3715
zcmai12{e>#8y-W3FqUlDC2Pf$DPt-7l3_-cY*B{E7|h7dpeQ7XBq>W-QXgN+AcRUI
zU)JnP$x_Ll&z>#+ix|=W|Iaz^IdkT{ujhI0>t3$=HA2$Tae_b~29RN*xi&p+vL8+Z
z0!7e)Kzo3Htu=`RPb|Untf`+H*26~D*VUygP8~^sasD$N(v$jC#`PHMz9s^sJRO?k
z7*RFC*Cx?dhA1SCBR^?r<zp#Ad>TC~D~km&@l&qt=lRvh-DVgo11)+K$^KIo2fC(Y
z$*R_Iz_H(*=h6`&Z_@~gN;fg_A$le4P}bfbC&zM4ZWb2Bst*-d1(zln<YdgshD7=q
z2!?z-^0FF|nRYbh5N9Q+RbBk6yatl2=X3K@Pz6UZhs>2WJeQ>jA!)v{cCIvT)bVj{
zu&l4zyB^<<-q8V5ML`omHf~?fPB0p<sO|5RYj%J^zmI&t!)#x93rI_eNHyP?C<&Nl
zeXbY~wnu;3xlwwA-qNV6K<$zDAToS{KlGfC*rX_;xU&8GE7u}y@QE0QZ;cYq`ZH*5
zL2~G>b64`>qG7guZbyt4Bni@ssx_m6M!!e~f!fR{IsmVmiyruO)*b8OA-7?(9ta;z
z;oJe>9R>^nF>Tt&u3A2cHAGgx*uus`Rzzz96NlBcKG<ln&q|O#`x#`q;-<>?<Co{4
z`-ZQX?(#rR)=!<d0#EEj58p(T@LD0V@J5^l`Lff8y1wU}e)P;zSO~5Fx4UT|bf#KA
zOoL!2C7$e*T&E`F!L?^s@z6ffL2|#XBwTDZBYI>(Ys!(H!K~_oSoRaw;KEtiNnhbF
z4EVXEK?UvHk)+fboRCWBq+!pUD={}oGKXq!1^H)Z)9v;@IlAx2vRSpPo}UaZ_rWo+
ztvELWkM6uCI`e5i;*~@Yw+eblRXFOKMtQ8wX-Zso9beg;S=%NbJ2_(n_;4WF8@UtR
zw$sLWs;#FCc&!BFFxyr$J2$u0<V}t#zXm?crZvv=Q}dmT{C9P&BVr-Gwr*`CJ%2cC
z2zv6s!rb$s#{*;5UghaoqYL+Ryn>Cz?6L>)bPlMYML*}ZzFH={Xvrc|dUZ5#Gb~xQ
zKDXo+O{6sY7|a;K&e`Z%Kdixwq+U*ah0X<fnX24(?r=DBBPD)Gk%l}^xEC6HfR>)2
zc?mo$dBLl1MT2%;_oe?J(b6~8ys9&$BP!6K;b>dw+xsO|e%HHm4>pd8?z5PhkFEb^
z^P+_I_Y46wof_l@+#CZQ_N}hk6Wy_^zPjF2M^i+ALX`*#4<j+<pEqgphSZ_%C0L}J
zgWun<kfJ$LNR)fxY~$Hw+Z`<&JtwCC`6Nzqvz^J*fQw1di246zV4d22KCwHk0z%k7
z?oZ356r!8Bu#m+WAJJW&H*%=x!E4Dz2bDd~HA>hHQaU09ao_Xw5;~R!lfQ|^=e&)N
zM0bPh87=Ktf{{W9q0ql<t`BC!VNQLJYc~W-!#p39(?vV!Ovq)MA*_mg?FlK3G)c2O
zeShV}K%R2rdBa?!BxLK+6N}AXYNlIF?_Mwm8+93Rp#A(WF+x->8Ki|RUsv$yzP1>C
zK5{szAu%o%F30Ep&m>n(yR@aehW?W7krI0_bQW)rbECy??^O8;IO`S728lb<iWVTz
z*X94@9boXMa+P88wrvCO`arqzt?#pL=0kLM#t@0ls{#2{y>)mBjQP@dXxZh)zzbx}
zz2XEpY{N*~@k<xt5b)fnuglY8*@qVtbUtc&9gZDE3+v2$CJD4ia?Bm(eSD3Blb$E>
zgeQG>yIl19X9=nX!{!1m@yer%75YNEj`ez=<ezl<2^``%Y!`##cgnXk9`+;??3G*4
zSjAS)YEY_7ggn{mC_S#Vw14apGrr8ww<e+Ec^L)kDJ&MY^d4SEN_z;+3+^dpgAPVY
zbaPx|tqSilA7u3%_p*AKWB$ox=s;80am<Nf0e#gs?xv83R`;J&1VA=eWdL1AQFPLz
zxa#i%EVKa6s<r^6dEo5au?{Al?oI?pkG1SXsqcORV=HC)DLPtaIHIl<%gG)Pd_pAO
znICuTXbIPH*~I6nVrF>?%>IRWQtKZ1a4qf+FK9wr9*H*K`n=0bAw~yN=v5^(n1~L8
zUEcNy+P7tMLr*#=;_H^LG<qa+XT)}y+_1e?oBmBaUD#J64l!cWJnA(atTy8$E?x5A
z^zLY`_o~Z0*dNHS%krO#o@ALg=OfZ1O;AIV)qx<YL#wiAp`SIN2-I0W5MGvF4te^y
zVAnkHGBzYqM3~FaA1<nOnUjSwo%F3{IGhH&nsVTW898|A2(6>-(eNR2viszed=TrG
zfWZ7PK@qL7XT47&ugkx#Zski1xk#+J<5R!y<lhf3Fzgx^nVcD(myT}P`vXsN^&vMe
zQ#Z#vt0<;5<o1O=m2;RlrqjkDsqBrA7_U(|OR;#%S&6?}-kGSzqjCg{9d3UV6PFv9
z4t-Jn=*$up;$)1id>g0fWOq(M9Vsto)P!$l5wB5J;!(iRUgZh@kf00QbChc$mH`|e
zfw3<g3*ZyWv_ok>orAEMKk^;WNd7r+u`xKQn$HMMa%(VDejhuP*HzE1sElm4ItG1M
zTo^de%d9)X#XFTQWT7e(MzXeA_?s!EE$9A}qQO&PN@@?o4LozkT;>6rp6P|TV&0jB
z8P4|QGs~N$xUP7t=i?~<0Nfr0cxGMacx&-QMXRGIYRuYmdR=y3k0)m^T}jLMWU6OV
zC~)o_qVDltWpeoQE}=Wd9iO9?U5DmZyaTIlDW~Ev!w+mtdt>}>L_h^g$IjljeeJ}^
zNvXIZ7?AeHf96L+isg;*T3ts_6SUI!TP0~sw^p4;MXVEht|~Pl?sB^&OiPd*nX*LA
zN(qEU^y=Wgsxk|OwR{O%jJ|G4U@MpA%zJ8okE=E@5C3XqFnZx5Vs|HIG{?iWF(;7!
zkRrk<mW3R3^oR1>6^zwIbm;<kBd>P0A1Kvl4;&WjxmDWU15<+!i-kQ3;KmZ@Q@lHD
z*C2;i3oGi@DfAl{XK0V%5-YMtw_Q~04uNi0h2)+Jo}OWEyD00De6HypQfCXCCX+NW
zq}WsDI{xEpdQONS?|}k+Mt-gRpxYxFS+4pi?UDc9qibU{yQ+UcuS(M5NR_PI58Ffq
zYj6>Wu{txuGv}mKfgfYRGuls=-XupID{w1<TjTj@({P$z+FzvjnyvJ-l{7FvDGiBH
zwRV|Pb9`|lBJ?tS&n{Y5S8}+>zK$;RI{zkun93E^4Y3GkN4qFYURJ^ScEt4x7R8A~
zjFYc6+f+IQHC-H8e3+FHBIyW|bgi^GZ^?_YM+t(V)qAUx(SFM%!=oMV7nVGF&gTC_
zVA5fH;f-!}VJB;28)uNhn8wS4PXfmwX6?@7eetTn^-4v2v&F&zD*1Qk@iRqA^={YD
z7cm?dI{0VUtAoXg$GK_}8szt*mZiT;&pmqg66gaoK+XU60T~Fymi32mnYzKwkgpDZ
z)P)6tyP++=AG&=Qq+%jAF>B)TzYbfSnyBk8;MJBH>i6@uPf=9Z>#eX~GZppF^+Mi0
zm{8H#K<aKDx-q6u;X4Uhm1OGr7x*Yy6X=cG?gaeqB>S#4z@JIoF4@#`5L;ned(bwH
z|15UYeRP2M{z>|%9<LL&U0tX<@Icmg*wmWs(Ctl=3MDbDLAPF?3f@WFYWJZoBETbX
p^HuO>FWMlT0raPCM8ziVz*2jY5pvgB&O<>^5DEyc@am@x=s&)cY$^Z%

literal 0
HcmV?d00001