pub2k2+: improve handling of default paragraph styles (fixes spacing issues in TS030002173.pub)

4 files changed, 80 insertions, 57 deletions

diff --git a/src/lib/MSPUBCollector.cpp b/src/lib/MSPUBCollector.cpp
index ab7441b..8c8d28b 100644
--- a/src/lib/MSPUBCollector.cpp
+++ b/src/lib/MSPUBCollector.cpp
@@ -739,14 +739,14 @@ boost::function<void(void)> libmspub::MSPUBCollector::paintShape(const ShapeInfo
     m_painter->startTextObject(props, WPXPropertyListVector());
     for (unsigned i_lines = 0; i_lines < text.size(); ++i_lines)
     {
-      WPXPropertyList paraProps = getParaStyleProps(text[i_lines].style, text[i_lines].style.defaultCharStyleIndex);
+      WPXPropertyList paraProps = getParaStyleProps(text[i_lines].style, text[i_lines].style.m_defaultCharStyleIndex);
       m_painter->startTextLine(paraProps);
       for (unsigned i_spans = 0; i_spans < text[i_lines].spans.size(); ++i_spans)
       {
         WPXString textString;
         appendCharacters(textString, text[i_lines].spans[i_spans].chars,
                          m_encoding.get_value_or(UTF_16));
-        WPXPropertyList charProps = getCharStyleProps(text[i_lines].spans[i_spans].style, text[i_lines].style.defaultCharStyleIndex);
+        WPXPropertyList charProps = getCharStyleProps(text[i_lines].spans[i_spans].style, text[i_lines].style.m_defaultCharStyleIndex);
         m_painter->startTextSpan(charProps);
         m_painter->insertText(textString);
         m_painter->endTextSpan();
@@ -989,12 +989,13 @@ void libmspub::MSPUBCollector::addFont(std::vector<unsigned char> name)
   m_fonts.push_back(name);
 }
 
-WPXPropertyList libmspub::MSPUBCollector::getParaStyleProps(const ParagraphStyle &style, unsigned defaultParaStyleIndex) const
+WPXPropertyList libmspub::MSPUBCollector::getParaStyleProps(const ParagraphStyle &style, boost::optional<unsigned> defaultParaStyleIndex) const
 {
   ParagraphStyle _nothing;
-  const ParagraphStyle &defaultParaStyle = defaultParaStyleIndex < m_defaultParaStyles.size() ? m_defaultParaStyles[defaultParaStyleIndex] : _nothing;
+  const ParagraphStyle &defaultStyle = defaultParaStyleIndex.is_initialized() && defaultParaStyleIndex.get() < m_defaultParaStyles.size() ? m_defaultParaStyles[defaultParaStyleIndex.get()] : _nothing;
   WPXPropertyList ret;
-  Alignment align = style.align >= 0 ? style.align : defaultParaStyle.align;
+  Alignment align = style.m_align.get_value_or(
+      defaultStyle.m_align.get_value_or(LEFT));
   switch (align)
   {
   case RIGHT:
@@ -1011,8 +1012,10 @@ WPXPropertyList libmspub::MSPUBCollector::getParaStyleProps(const ParagraphStyle
     ret.insert("fo:text-align", "left");
     break;
   }
-  double lineSpacing = style.lineSpacing;
-  LineSpacingType lineSpacingType = style.lineSpacingType;
+  LineSpacingInfo info = style.m_lineSpacing.get_value_or(
+      defaultStyle.m_lineSpacing.get_value_or(LineSpacingInfo()));
+  LineSpacingType lineSpacingType = info.m_type;
+  double lineSpacing = info.m_amount;
   if (!(lineSpacingType == LINE_SPACING_SP && lineSpacing == 1))
   {
     if (lineSpacingType == LINE_SPACING_SP)
@@ -1024,33 +1027,43 @@ WPXPropertyList libmspub::MSPUBCollector::getParaStyleProps(const ParagraphStyle
       ret.insert("fo:line-height", lineSpacing, WPX_POINT);
     }
   }
-  if (style.spaceAfterEmu != 0)
+  unsigned spaceAfterEmu = style.m_spaceAfterEmu.get_value_or(
+      defaultStyle.m_spaceAfterEmu.get_value_or(0));
+  unsigned spaceBeforeEmu = style.m_spaceBeforeEmu.get_value_or(
+      defaultStyle.m_spaceBeforeEmu.get_value_or(0));
+  unsigned firstLineIndentEmu = style.m_firstLineIndentEmu.get_value_or(
+      defaultStyle.m_firstLineIndentEmu.get_value_or(0));
+  unsigned leftIndentEmu = style.m_leftIndentEmu.get_value_or(
+      defaultStyle.m_leftIndentEmu.get_value_or(0));
+  unsigned rightIndentEmu = style.m_rightIndentEmu.get_value_or(
+      defaultStyle.m_rightIndentEmu.get_value_or(0));
+  if (spaceAfterEmu != 0)
   {
-    ret.insert("fo:margin-bottom", (double)style.spaceAfterEmu / EMUS_IN_INCH);
+    ret.insert("fo:margin-bottom", (double)spaceAfterEmu / EMUS_IN_INCH);
   }
-  if (style.spaceBeforeEmu != 0)
+  if (spaceBeforeEmu != 0)
   {
-    ret.insert("fo:margin-top", (double)style.spaceBeforeEmu / EMUS_IN_INCH);
+    ret.insert("fo:margin-top", (double)spaceBeforeEmu / EMUS_IN_INCH);
   }
-  if (style.firstLineIndentEmu != 0)
+  if (firstLineIndentEmu != 0)
   {
-    ret.insert("fo:text-indent", (double)style.firstLineIndentEmu / EMUS_IN_INCH);
+    ret.insert("fo:text-indent", (double)firstLineIndentEmu / EMUS_IN_INCH);
   }
-  if (style.leftIndentEmu != 0)
+  if (leftIndentEmu != 0)
   {
-    ret.insert("fo:margin-left", (double)style.leftIndentEmu / EMUS_IN_INCH);
+    ret.insert("fo:margin-left", (double)leftIndentEmu / EMUS_IN_INCH);
   }
-  if (style.rightIndentEmu != 0)
+  if (rightIndentEmu != 0)
   {
-    ret.insert("fo:margin-right", (double)style.rightIndentEmu / EMUS_IN_INCH);
+    ret.insert("fo:margin-right", (double)rightIndentEmu / EMUS_IN_INCH);
   }
   return ret;
 }
 
-WPXPropertyList libmspub::MSPUBCollector::getCharStyleProps(const CharacterStyle &style, unsigned defaultCharStyleIndex) const
+WPXPropertyList libmspub::MSPUBCollector::getCharStyleProps(const CharacterStyle &style, boost::optional<unsigned> defaultCharStyleIndex) const
 {
   CharacterStyle _nothing = CharacterStyle(false, false, false);
-  const CharacterStyle &defaultCharStyle = defaultCharStyleIndex < m_defaultCharStyles.size() ? m_defaultCharStyles[defaultCharStyleIndex] : _nothing;
+  const CharacterStyle &defaultCharStyle = defaultCharStyleIndex.is_initialized() && defaultCharStyleIndex.get() < m_defaultCharStyles.size() ? m_defaultCharStyles[defaultCharStyleIndex.get()] : _nothing;
   WPXPropertyList ret;
   if (style.italic ^ defaultCharStyle.italic)
   {
diff --git a/src/lib/MSPUBCollector.h b/src/lib/MSPUBCollector.h
index 93edcb1..9b90905 100644
--- a/src/lib/MSPUBCollector.h
+++ b/src/lib/MSPUBCollector.h
@@ -197,8 +197,8 @@ private:
   boost::function<void(void)> paintShape(const ShapeInfo &info, const Coordinate &relativeTo, const VectorTransformation2D &foldedTransform, bool isGroup, const VectorTransformation2D &thisTransform) const;
   double getCalculationValue(const ShapeInfo &info, unsigned index, bool recursiveEntry, const std::vector<int> &adjustValues) const;
 
-  WPXPropertyList getCharStyleProps(const CharacterStyle &, unsigned defaultCharStyleIndex) const;
-  WPXPropertyList getParaStyleProps(const ParagraphStyle &, unsigned defaultParaStyleIndex) const;
+  WPXPropertyList getCharStyleProps(const CharacterStyle &, boost::optional<unsigned> defaultCharStyleIndex) const;
+  WPXPropertyList getParaStyleProps(const ParagraphStyle &, boost::optional<unsigned> defaultParaStyleIndex) const;
   double getSpecialValue(const ShapeInfo &info, const CustomShape &shape, int arg, const std::vector<int> &adjustValues) const;
 public:
   static WPXString getColorString(const Color &);
diff --git a/src/lib/MSPUBParser.cpp b/src/lib/MSPUBParser.cpp
index 7f5953f..b3281cb 100644
--- a/src/lib/MSPUBParser.cpp
+++ b/src/lib/MSPUBParser.cpp
@@ -1083,20 +1083,14 @@ std::vector<libmspub::MSPUBParser::TextSpanReference> libmspub::MSPUBParser::par
 }
 libmspub::ParagraphStyle libmspub::MSPUBParser::getParagraphStyle(WPXInputStream *input)
 {
-  Alignment align = (Alignment)-1;
-  double lineSpacing = 1;
-  LineSpacingType lineSpacingType = LINE_SPACING_SP;
-  unsigned defaultCharStyleIndex = 0;
-  unsigned spaceBeforeEmu = 0;
-  unsigned spaceAfterEmu = 0;
-  int firstLineIndentEmu = 0;
-  unsigned leftIndentEmu = 0;
-  unsigned rightIndentEmu = 0;
+  ParagraphStyle ret;
+
   bool isList = false;
   uint32_t bulletChar = '\u0000';
   NumberingType numberingType = STANDARD_WESTERN;
   NumberingDelimiter numberingDelimiter = NO_DELIMITER;
   boost::optional<unsigned> numberIfRestarted;
+
   unsigned offset = input->tell();
   unsigned len = readU32(input);
   while (stillReading(input, offset + len))
@@ -1105,10 +1099,10 @@ libmspub::ParagraphStyle libmspub::MSPUBParser::getParagraphStyle(WPXInputStream
     switch(info.id)
     {
     case PARAGRAPH_ALIGNMENT:
-      align = (Alignment)(info.data & 0xFF); // Is this correct?
+      ret.m_align = (Alignment)(info.data & 0xFF); // Is this correct?
       break;
     case PARAGRAPH_DEFAULT_CHAR_STYLE:
-      defaultCharStyleIndex = info.data;
+      ret.m_defaultCharStyleIndex = info.data;
       break;
     case PARAGRAPH_LINE_SPACING:
       if (info.data & 1)
@@ -1116,32 +1110,32 @@ libmspub::ParagraphStyle libmspub::MSPUBParser::getParagraphStyle(WPXInputStream
         // line spacing expressed in points in the UI,
         // in eighths of an emu in the file format.
         // (WTF??)
-        lineSpacing = static_cast<double>(info.data - 1) / 8 * 72 / EMUS_IN_INCH;
-        lineSpacingType = LINE_SPACING_PT;
+        ret.m_lineSpacing = LineSpacingInfo(LINE_SPACING_PT, 
+            static_cast<double>(info.data - 1) / 8 * 72 / EMUS_IN_INCH);
       }
       else if (info.data & 2)
       {
         // line spacing expressed in SP in the UI,
         // in what would be EMUs if font size were 96pt in the file format
         // (WTF??)
-        lineSpacing = static_cast<double>(info.data - 2) / EMUS_IN_INCH * 72 / 96;
-        lineSpacingType = LINE_SPACING_SP;
+        ret.m_lineSpacing = LineSpacingInfo(LINE_SPACING_SP,
+            static_cast<double>(info.data - 2) / EMUS_IN_INCH * 72 / 96);
       }
       break;
     case PARAGRAPH_SPACE_BEFORE:
-      spaceBeforeEmu = info.data;
+      ret.m_spaceBeforeEmu = info.data;
       break;
     case PARAGRAPH_SPACE_AFTER:
-      spaceAfterEmu = info.data;
+      ret.m_spaceAfterEmu = info.data;
       break;
     case PARAGRAPH_FIRST_LINE_INDENT:
-      firstLineIndentEmu = (int)info.data;
+      ret.m_firstLineIndentEmu = (int)info.data;
       break;
     case PARAGRAPH_LEFT_INDENT:
-      leftIndentEmu = info.data;
+      ret.m_leftIndentEmu = info.data;
       break;
     case PARAGRAPH_RIGHT_INDENT:
-      rightIndentEmu = info.data;
+      ret.m_rightIndentEmu = info.data;
       break;
     case PARAGRAPH_LIST_INFO:
     {
@@ -1171,21 +1165,20 @@ libmspub::ParagraphStyle libmspub::MSPUBParser::getParagraphStyle(WPXInputStream
       break;
     }
   }
-  boost::optional<ListInfo> listInfo;
   if (isList)
   {
     if (bulletChar != '\u0000')
     {
-      listInfo = ListInfo(bulletChar);
+      ret.m_listInfo = ListInfo(bulletChar);
     }
     else
     {
-      listInfo = ListInfo(numberIfRestarted, numberingType,
+      ret.m_listInfo = ListInfo(numberIfRestarted, numberingType,
                           numberingDelimiter);
     }
   }
-  return ParagraphStyle(align, defaultCharStyleIndex, lineSpacing, lineSpacingType, spaceBeforeEmu, spaceAfterEmu,
-                        firstLineIndentEmu, leftIndentEmu, rightIndentEmu, listInfo);
+
+  return ret;
 }
 
 libmspub::CharacterStyle libmspub::MSPUBParser::getCharacterStyle(WPXInputStream *input)
diff --git a/src/lib/MSPUBTypes.h b/src/lib/MSPUBTypes.h
index 4fffd1b..61e5b56 100644
--- a/src/lib/MSPUBTypes.h
+++ b/src/lib/MSPUBTypes.h
@@ -133,19 +133,36 @@ enum LineSpacingType
   LINE_SPACING_PT
 };
 
+struct LineSpacingInfo
+{
+  LineSpacingType m_type;
+  double m_amount;
+  LineSpacingInfo() : m_type(LINE_SPACING_SP), m_amount(1)
+  {
+  }
+  LineSpacingInfo(LineSpacingType type, double amount) :
+    m_type(type), m_amount(amount)
+  {
+  }
+};
+
 struct ParagraphStyle
 {
-  ParagraphStyle(Alignment a = (Alignment)-1, unsigned dCSI = 0, double ls = 1, LineSpacingType lst = LINE_SPACING_SP, unsigned sb = 0, unsigned sa = 0, int fli = 0, unsigned li = 0, unsigned ri = 0, boost::optional<ListInfo> lInfo = boost::optional<ListInfo>()) : align(a), defaultCharStyleIndex(dCSI), lineSpacing(ls), lineSpacingType(lst), spaceBeforeEmu(sb), spaceAfterEmu(sa), firstLineIndentEmu(fli), leftIndentEmu(li), rightIndentEmu(ri), listInfo(lInfo) { }
-  Alignment align;
-  unsigned defaultCharStyleIndex;
-  double lineSpacing;
-  LineSpacingType lineSpacingType;
-  unsigned spaceBeforeEmu;
-  unsigned spaceAfterEmu;
-  int firstLineIndentEmu;
-  unsigned leftIndentEmu;
-  unsigned rightIndentEmu;
-  boost::optional<ListInfo> listInfo;
+  boost::optional<Alignment> m_align;
+  boost::optional<unsigned> m_defaultCharStyleIndex;
+  boost::optional<LineSpacingInfo> m_lineSpacing;
+  boost::optional<unsigned> m_spaceBeforeEmu;
+  boost::optional<unsigned> m_spaceAfterEmu;
+  boost::optional<int> m_firstLineIndentEmu;
+  boost::optional<unsigned> m_leftIndentEmu;
+  boost::optional<unsigned> m_rightIndentEmu;
+  boost::optional<ListInfo> m_listInfo;
+  ParagraphStyle() :
+    m_align(), m_defaultCharStyleIndex(), m_lineSpacing(), m_spaceBeforeEmu(),
+    m_spaceAfterEmu(), m_firstLineIndentEmu(), m_leftIndentEmu(),
+    m_rightIndentEmu(), m_listInfo()
+  {
+  }
 };
 
 struct TextSpan