Add support for table cells spanning multiple rows

2 files changed, 85 insertions, 18 deletions

diff --git a/lib/documents.js b/lib/documents.js
index 8b979c2..5665198 100644
--- a/lib/documents.js
+++ b/lib/documents.js
@@ -1,6 +1,6 @@
 var _ = require("underscore");
 
-var types = {
+var types = exports.types = {
     document: "document",
     paragraph: "paragraph",
     run: "run",
@@ -144,7 +144,8 @@ function TableCell(children, options) {
     return {
         type: types.tableCell,
         children: children,
-        colSpan: options.colSpan == null ? 1 : options.colSpan
+        colSpan: options.colSpan == null ? 1 : options.colSpan,
+        rowSpan: options.rowSpan == null ? 1 : options.rowSpan
     };
 }
 
diff --git a/lib/docx/body-reader.js b/lib/docx/body-reader.js
index 126896c..70d5402 100644
--- a/lib/docx/body-reader.js
+++ b/lib/docx/body-reader.js
@@ -157,22 +157,9 @@ function BodyReader(options) {
                 }
             });
         },
-        "w:tbl": function(element) {
-            return readXmlElements(element.children).map(documents.Table);
-        },
-        "w:tr": function(element) {
-            return readXmlElements(element.children).map(documents.TableRow);
-        },
-        "w:tc": function(element) {
-            return readXmlElements(element.children).map(function(children) {
-                var gridSpan = element
-                    .firstOrEmpty("w:tcPr")
-                    .firstOrEmpty("w:gridSpan")
-                    .attributes["w:val"];
-                var colSpan = gridSpan ? parseInt(gridSpan) : 1;
-                return documents.TableCell(children, {colSpan: colSpan});
-            });
-        },
+        "w:tbl": readTable,
+        "w:tr": readTableRow,
+        "w:tc": readTableCell,
         "w:footnoteReference": noteReferenceReader("footnote"),
         "w:endnoteReference": noteReferenceReader("endnote"),
         "w:br": function(element) {
@@ -225,6 +212,85 @@ function BodyReader(options) {
             return numbering.findLevel(numId, level);
         }
     }
+    
+    function readTable(element) {
+        return readXmlElements(element.children)
+            .flatMap(calculateRowSpans)
+            .map(documents.Table);
+    }
+    
+    function readTableRow(element) {
+        return readXmlElements(element.children).map(documents.TableRow);
+    }
+    
+    function readTableCell(element) {
+        return readXmlElements(element.children).map(function(children) {
+            var properties = element.firstOrEmpty("w:tcPr");
+            
+            var gridSpan = properties.firstOrEmpty("w:gridSpan").attributes["w:val"];
+            var colSpan = gridSpan ? parseInt(gridSpan) : 1;
+            
+            var cell = documents.TableCell(children, {colSpan: colSpan});
+            cell._vMerge = readVMerge(properties);
+            return cell;
+        });
+    }
+    
+    function readVMerge(properties) {
+        var element = properties.first("w:vMerge");
+        if (element) {
+            var val = element.attributes["w:val"];
+            return val ? val : "continue";
+        } else {
+            return null;
+        }
+    }
+    
+    function calculateRowSpans(rows) {
+        var unexpectedNonRows = _.any(rows, function(row) {
+            return row.type !== documents.types.tableRow;
+        });
+        if (unexpectedNonRows) {
+            return elementResultWithMessages(rows, [warning(
+                "unexpected non-row element in table, cell merging may be incorrect"
+            )]);
+        }
+        var unexpectedNonCells = _.any(rows, function(row) {
+            return _.any(row.children, function(cell) {
+                return cell.type !== documents.types.tableCell;
+            });
+        });
+        if (unexpectedNonCells) {
+            return elementResultWithMessages(rows, [warning(
+                "unexpected non-cell element in table row, cell merging may be incorrect"
+            )]);
+        }
+        
+        var columns = {};
+        
+        rows.forEach(function(row, rowIndex) {
+            var cellIndex = 0;
+            row.children.forEach(function(cell) {
+                if (cell._vMerge === "continue" && columns[cellIndex]) {
+                    columns[cellIndex].rowSpan++;
+                } else {
+                    columns[cellIndex] = cell;
+                }
+                cellIndex += cell.colSpan;
+            });
+        });
+        
+        rows.forEach(function(row, rowIndex) {
+            row.children = row.children.filter(function(cell) {
+                return cell._vMerge !== "continue";
+            });
+            row.children.forEach(function(cell) {
+                delete cell._vMerge;
+            });
+        });
+        
+        return elementResult(rows);
+    }
 
     function readDrawingElement(element) {
         var blips = element