[ooo-build-commit] scratch/mso-dumper

[Kohei Yoshida]

 scratch/mso-dumper/src/xlsmodel.py  |   30 ++++++++++++++++++++++--------
 scratch/mso-dumper/src/xlsrecord.py |   29 ++++++++++++++++++++++-------
 2 files changed, 44 insertions(+), 15 deletions(-)

New commits:
commit 160a4dd8003f4ad4cf7b55cc1a369b9c29446a4e
Author: Kohei Yoshida <kyoshida at novell.com>
Date:   Wed Dec 30 17:35:12 2009 -0500

    [xls-dump] Fetch string values from SST IDs in canonical XML mode.
    
    * scratch/mso-dumper/src/xlsmodel.py:
    * scratch/mso-dumper/src/xlsrecord.py:

diff --git a/scratch/mso-dumper/src/xlsmodel.py b/scratch/mso-dumper/src/xlsmodel.py
index 045a40d..b48c576 100644
--- a/scratch/mso-dumper/src/xlsmodel.py
+++ b/scratch/mso-dumper/src/xlsmodel.py
@@ -45,10 +45,10 @@ class Workbook(ModelBase):
             return
 
         wbglobal = self.__sheets[0]
-        nd.appendChild(wbglobal.createDOM())
+        nd.appendChild(wbglobal.createDOM(self))
         for i in xrange(1, n):
             sheet = self.__sheets[i]
-            sheetNode = sheet.createDOM()
+            sheetNode = sheet.createDOM(self)
             nd.appendChild(sheetNode)
             if i > 0:
                 data = wbglobal.getSheetData(i-1)
@@ -68,7 +68,7 @@ class SheetBase(object):
         self.modelType = modelType
         self.version = None
 
-    def createDOM (self):
+    def createDOM (self, wb):
         nd = node.Element('sheet')
         return nd
 
@@ -83,14 +83,23 @@ class WorkbookGlobal(SheetBase):
         SheetBase.__init__(self, SheetModelType.WorkbookGlobal)
 
         self.__sheetData = []
+        self.__sharedStrings = []
 
-    def createDOM (self):
+    def createDOM (self, wb):
         nd = node.Element('workbook-global')
         return nd
 
     def appendSheetData (self, data):
         self.__sheetData.append(data)
 
+    def appendSharedString (self, sst):
+        self.__sharedStrings.append(sst)
+
+    def getSharedString (self, strID):
+        if len(self.__sharedStrings) <= strID:
+            return None
+        return self.__sharedStrings[strID]
+
     def getSheetData (self, i):
         return self.__sheetData[i]
 
@@ -107,7 +116,7 @@ class Worksheet(SheetBase):
 
         self.rows[row][col] = cell
 
-    def createDOM (self):
+    def createDOM (self, wb):
         nd = node.Element('worksheet')
         nd.setAttr('version', self.version)
         rows = self.rows.keys()
@@ -118,7 +127,7 @@ class Worksheet(SheetBase):
             cols = self.rows[row].keys()
             for col in cols:
                 cell = self.rows[row][col]
-                cellNode = cell.createDOM()
+                cellNode = cell.createDOM(wb)
                 rowNode.appendChild(cellNode)
                 cellNode.setAttr('col', col)
         return nd
@@ -138,9 +147,14 @@ class CellBase(object):
 class LabelCell(CellBase):
     def __init__ (self):
         CellBase.__init__(self, CellModelType.Label)
+        self.strID = None
 
-    def createDOM (self):
+    def createDOM (self, wb):
         nd = node.Element('label-cell')
+        if self.strID != None:
+            sst = wb.getWorkbookGlobal().getSharedString(self.strID)
+            if sst != None:
+                nd.setAttr('value', sst.baseText)
         return nd
 
 class NumberCell(CellBase):
@@ -148,7 +162,7 @@ class NumberCell(CellBase):
         CellBase.__init__(self, CellModelType.Number)
         self.value = value
 
-    def createDOM (self):
+    def createDOM (self, wb):
         nd = node.Element('number-cell')
         nd.setAttr('value', self.value)
         return nd
diff --git a/scratch/mso-dumper/src/xlsrecord.py b/scratch/mso-dumper/src/xlsrecord.py
index 2abc69a..beb7985 100644
--- a/scratch/mso-dumper/src/xlsrecord.py
+++ b/scratch/mso-dumper/src/xlsrecord.py
@@ -404,6 +404,7 @@ class LabelSST(BaseRecordHandler):
         self.__parseBytes()
         sheet = model.getCurrentSheet()
         cell = xlsmodel.LabelCell()
+        cell.strID = self.strId
         sheet.setCell(self.col, self.row, cell)
 
 
@@ -604,15 +605,29 @@ class String(BaseRecordHandler):
 
 class SST(BaseRecordHandler):
 
-    def parseBytes (self):
-        refCount = self.readSignedInt(4) # total number of references in workbook
-        strCount = self.readSignedInt(4) # total number of unique strings.
-        self.appendLine("total number of references: %d"%refCount)
-        self.appendLine("total number of unique strings: %d"%strCount)
-        for i in xrange(0, strCount):
+    def __parseBytes (self):
+        self.refCount = self.readSignedInt(4) # total number of references in workbook
+        self.strCount = self.readSignedInt(4) # total number of unique strings.
+        self.sharedStrings = []
+        for i in xrange(0, self.strCount):
             extText, bytesRead = globals.getUnicodeRichExtText(self.bytes[self.getCurrentPos():])
             self.readBytes(bytesRead) # advance current position.
-        return
+            self.sharedStrings.append(extText)
+
+    def parseBytes (self):
+        self.__parseBytes()
+        self.appendLine("total number of references: %d"%self.refCount)
+        self.appendLine("total number of unique strings: %d"%self.strCount)
+        i = 0
+        for s in self.sharedStrings:
+            self.appendLine("s%d: %s"%(i, s.baseText))
+            i += 1
+
+    def fillModel (self, model):
+        self.__parseBytes()
+        wbg = model.getWorkbookGlobal()
+        for sst in self.sharedStrings:
+            wbg.appendSharedString(sst)
 
 
 class Blank(BaseRecordHandler):