xls-dump.py was tested using libreoffice's set of test files; several small issues were fixed

6 files changed, 114 insertions, 24 deletions

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..0d20b64
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1 @@
+*.pyc
diff --git a/misc/test-files.sh b/misc/test-files.sh
new file mode 100755
index 0000000..164952d
--- /dev/null
+++ b/misc/test-files.sh
@@ -0,0 +1,15 @@
+#! /bin/sh
+
+test_dir=$1
+
+if [ ! -d $test_dir ]
+then
+    echo "Usage: test-files.sh TEST_DIR"
+    exit 1
+fi
+
+for x in `find $test_dir -name \*.xls`; do
+    (python xls-dump.py $x | grep -v "rror inter" > /dev/null) || echo "Flat dump failed for" $x
+    python xls-dump.py --dump-mode=xml $x > /dev/null || echo "Xml dump failed for" $x
+    python xls-dump.py --dump-mode=cxml $x > /dev/null || echo "CXml dump failed for" $x
+done
diff --git a/src/xlsmodel.py b/src/xlsmodel.py
index 5c10cad..5094ee5 100644
--- a/src/xlsmodel.py
+++ b/src/xlsmodel.py
@@ -79,7 +79,7 @@ class Workbook(ModelBase):
         sheets = filter(lambda x: isinstance(x, Worksheet), self.__sheets)
         n = len(sheets)
         if n == 0:
-            return
+            return nd
 
         wbglobal = self.getWorkbookGlobal()
         nd.appendChild(wbglobal.createDOM(self))
diff --git a/src/xlsparser.py b/src/xlsparser.py
index b9d5204..d105f24 100644
--- a/src/xlsparser.py
+++ b/src/xlsparser.py
@@ -119,8 +119,11 @@ class Req(BaseParser):
     def parse(self, stream):
         parsed = safeParse(self.__parser, stream)
         if parsed is None:
+            currentToken = "<<<End Of Token Stream>>>"
+            if stream.currentIndex < len(stream.tokens):
+                currentToken = stream.tokens[stream.currentIndex]
             raise ParseException("%s failed but it is required, next token is [%s]" % 
-                                 (str(self.__parser), str(stream.tokens[stream.currentIndex])))
+                                 (str(self.__parser), str(currentToken)))
         return parsed
     
     def __str__(self):
@@ -334,8 +337,14 @@ class PlotGrowth(BaseParser):
 class Series(BaseParser):
     PARSER = Term(xlsrecord.Series)
 
+class SeriesText(BaseParser):
+    PARSER = Term(xlsrecord.SeriesText)
+
+class BRAI(BaseParser):
+    PARSER = Term(xlsrecord.Brai)
+
 class AI(BaseParser):
-    PARSER = Term(xlsrecord.Brai) # TODO: we use Brai instead of AI now, fix it
+    PARSER = Req(BRAI()) << SeriesText()
 
 class SerParent(BaseParser): pass        
 class SerAuxTrend(BaseParser): pass        
@@ -351,9 +360,13 @@ class DataFormat(BaseParser):
 class Chart3DBarShape(BaseParser):
     PARSER = Term(xlsrecord.Chart3DBarShape)
 
-class PieFormat(BaseParser): pass        
-class SerFmt(BaseParser): pass        
-class MarkerFormat(BaseParser): pass
+class PieFormat(BaseParser): 
+    PARSER = Term(xlsrecord.PieFormat)
+
+class SerFmt(BaseParser): pass
+
+class MarkerFormat(BaseParser): 
+    PARSER = Term(xlsrecord.MarkerFormat)
 
 class Text(BaseParser):
     PARSER = Term(xlsrecord.Text)
@@ -487,7 +500,9 @@ class BobPopCustom(BaseParser): pass
 class Bar(BaseParser):
     PARSER = Term(xlsrecord.CHBar)
 
-class Line(BaseParser): pass
+class Line(BaseParser): 
+    PARSER = Term(xlsrecord.CHLine)
+    
 class Pie(BaseParser): pass
 class Area(BaseParser): pass
 class Scatter(BaseParser): pass
@@ -538,15 +553,15 @@ class CHARTFORMATS(BaseParser):
     #*2DFTTEXT AxesUsed 1*2AXISPARENT [CrtLayout12A] [DAT] *ATTACHEDLABEL [CRTMLFRT]
     #*([DataLabExt StartObject] ATTACHEDLABEL [EndObject]) [TEXTPROPS] *2CRTMLFRT End
     PARSER = Group('chart-fmt', Req(Chart()) << Req(Begin()) << Many('font-lists', FONTLIST(), max=2) <<
-                Req(Scl()) << Req(PlotGrowth()) << FRAME() << Many('series-fmt-list', SERIESFORMAT()) <<
+                Req(Scl()) << Req(PlotGrowth()) << Opt(FRAME()) << Many('series-fmt-list', SERIESFORMAT()) <<
                 Many('ss-list', SS()) << Req(ShtProps()) << Many('dft-texts', DFTTEXT(), max=2) <<
                 Req(AxesUsed()) << Many('axis-roots', AXISPARENT(), min=1, max=2) <<
-                CrtLayout12A() << DAT() << Many('attached-labels', ATTACHEDLABEL()) <<
-                CRTMLFRT() << Many('datalab-exts', Seq(Seq(Req(DataLabExt()), 
-                                                           Req(StartObject())), 
-                                                     Req(ATTACHEDLABEL()), 
-                                                     EndObject())) <<
-                TEXTPROPS() << Many('crtmlfrt-list', CRTMLFRT()) << Req(End()))
+                CrtLayout12A() << Opt(DAT()) << Many('attached-labels', ATTACHEDLABEL()) <<
+                Opt(CRTMLFRT()) << Many('datalab-exts', Seq(Opt(Seq(Req(DataLabExt()), 
+                                                                    Req(StartObject()))), 
+                                                            Req(ATTACHEDLABEL()), 
+                                                            EndObject())) <<
+                Opt(TEXTPROPS()) << Many('crtmlfrt-list', CRTMLFRT()) << Req(End()))
 
 class Dimensions(BaseParser):
     PARSER = Term(xlsrecord.Dimensions)
@@ -560,7 +575,9 @@ class Number(BaseParser):
 
 class BoolErr(BaseParser): pass
 class Blank(BaseParser): pass
-class Label(BaseParser): pass
+
+class Label(BaseParser):
+    PARSER = Term(xlsrecord.Label)
 
 class SERIESDATA(BaseParser):
     #SERIESDATA = Dimensions 3(SIIndex *(Number / BoolErr / Blank / Label))
diff --git a/src/xlsrecord.py b/src/xlsrecord.py
index 1e5b0e6..21f0beb 100644
--- a/src/xlsrecord.py
+++ b/src/xlsrecord.py
@@ -1127,6 +1127,12 @@ class Label(BaseRecordHandler):
         self.appendLine("XF record ID: %d"%self.xfIdx)
         self.appendLine("label text: %s"%self.text)
 
+    def dumpData(self):
+        self.__parseBytes()
+        return ('label', {'col': self.col, 
+                          'row': self.row,
+                          'xf-idx': self.xfIdx,
+                          'text': self.text})
 
 class LabelSST(BaseRecordHandler):
 
@@ -3531,6 +3537,48 @@ class AreaFormat(BaseRecordHandler):
                                  ('icv-fore', dumpIcv(self.icvFore)),
                                  ('icv-back', dumpIcv(self.icvBack))])
 
+class PieFormat(BaseRecordHandler):
+    def __parseBytes(self):
+        self.pcExplode = self.readSignedInt(2)
+        
+    def parseBytes (self):
+        self.__parseBytes()
+        # TODO: dump all data
+
+    def dumpData(self):
+        self.__parseBytes()
+        return ('pie-format', {'pc-explode': self.pcExplode})
+
+class MarkerFormat(BaseRecordHandler):
+    def __parseBytes(self):
+        self.rgbFore = self.readLongRGB()
+        self.rgbBack = self.readLongRGB()
+        self.imk = self.readUnsignedInt(2)
+        flags = self.readUnsignedInt(2)
+        self.auto = (flags & 0x001) != 0 # A
+        # next 3 bits are reserved
+        self.notShowInt = (flags & 0x010) != 0 # C 
+        self.notShowBrd = (flags & 0x020) != 0 # D 
+        self.icvFore = self.readICV()
+        self.icvBack = self.readICV()
+        self.miSize = self.readUnsignedInt(4)
+        
+    def parseBytes (self):
+        self.__parseBytes()
+        # TODO: dump all data
+
+    def dumpData(self):
+        self.__parseBytes()
+        return ('marker-format', {'imk': self.imk,
+                                  'auto': self.auto,
+                                  'not-show-int': self.notShowInt,
+                                  'not-show-brd': self.notShowBrd,
+                                  'mi-size': self.miSize},
+                                [('rgb-fore', dumpRgb(self.rgbFore)),
+                                 ('rgb-back', dumpRgb(self.rgbBack)),
+                                 ('icv-fore', dumpIcv(self.icvFore)),
+                                 ('icv-back', dumpIcv(self.icvBack))])
+
 class DataFormat(BaseRecordHandler):
     def __parseBytes(self):
         self.xi = self.readUnsignedInt(2)
@@ -4104,15 +4152,24 @@ class CHBar(BaseRecordHandler):
 
 class CHLine(BaseRecordHandler):
 
-    def parseBytes (self):
+    def __parseBytes (self):
         flags   = globals.getUnsignedInt(self.readBytes(2))
-        stacked = (flags & 0x0001)
-        percent = (flags & 0x0002)
-        shadow  = (flags & 0x0004)
+        self.stacked = (flags & 0x0001)
+        self.percent = (flags & 0x0002)
+        self.shadow  = (flags & 0x0004)
 
-        self.appendLine("stacked: %s"%self.getYesNo(stacked))
-        self.appendLine("percent: %s"%self.getYesNo(percent))
-        self.appendLine("shadow: %s"%self.getYesNo(shadow))
+    def parseBytes (self):
+        self.__parseBytes()
+        self.appendLine("stacked: %s"%self.getYesNo(self.stacked))
+        self.appendLine("percent: %s"%self.getYesNo(self.percent))
+        self.appendLine("shadow: %s"%self.getYesNo(self.shadow))
+    
+    def dumpData(self):
+        self.__parseBytes()
+        return ('line', {'stacked': self.stacked,
+                         'percent': self.percent,
+                         'shadow': self.shadow})
+        
 
 
 class Brai(BaseRecordHandler):
diff --git a/src/xlsstream.py b/src/xlsstream.py
index 5c569ad..ceb6eeb 100644
--- a/src/xlsstream.py
+++ b/src/xlsstream.py
@@ -243,9 +243,9 @@ recData = {
     0x1003: ["SERIES", "Data Properties for Series, Trendlines or Error Bars", xlsrecord.Series],
     0x1006: ["CHDATAFORMAT", "Data point or series that the formatting information that follows applies to (2.4.74)", xlsrecord.DataFormat],
     0x1007: ["LINEFORMAT", "Appearance of A Line", xlsrecord.LineFormat],
-    0x1009: ["CHMARKERFORMAT", "?"],
+    0x1009: ["CHMARKERFORMAT", "Color, size, and shape of the markers", xlsrecord.MarkerFormat],
     0x100A: ["AREAFORMAT", "Patterns and Colors in Filled Region of Chart", xlsrecord.AreaFormat],
-    0x100B: ["CHPIEFORMAT", "?"],
+    0x100B: ["CHPIEFORMAT", "Distance of a data point from the center", xlsrecord.PieFormat],
     0x100C: ["CHATTACHEDLABEL", "?"],
     0x100D: ["SERIESTEXT", "Series Category Name or Title Text in Chart", xlsrecord.SeriesText],
     0x1014: ["CHTYPEGROUP", "Properties of a chart group", xlsrecord.ChartFormat],