refactor format detection

5 files changed, 90 insertions, 32 deletions

diff --git a/configure.ac b/configure.ac
index f9054a3..49e9b52 100644
--- a/configure.ac
+++ b/configure.ac
@@ -52,6 +52,7 @@ AC_SUBST(REVENGE_LIBS)
 # ==================
 AC_CHECK_HEADERS(
         boost/optional.hpp \
+        boost/scoped_ptr.hpp \
         boost/shared_ptr.hpp \
         ,
     [],
diff --git a/src/lib/SW602Document.cpp b/src/lib/SW602Document.cpp
index 2be0e20..038bd1f 100644
--- a/src/lib/SW602Document.cpp
+++ b/src/lib/SW602Document.cpp
@@ -9,6 +9,9 @@
 
 #include <libsw602/SW602Document.h>
 
+#include <cassert>
+
+#include <boost/scoped_ptr.hpp>
 #include <boost/shared_ptr.hpp>
 
 #include <librevenge-stream/librevenge-stream.h>
@@ -16,8 +19,6 @@
 #include "WinText602Header.h"
 #include "WinText602Parser.h"
 
-using boost::shared_ptr;
-
 using librevenge::RVNGInputStream;
 
 using std::string;
@@ -28,15 +29,63 @@ namespace libsw602
 namespace
 {
 
-shared_ptr<RVNGInputStream> getContent(librevenge::RVNGInputStream *const ip)
+struct DetectionInfo
+{
+  DetectionInfo();
+
+  SW602Document::Kind m_kind;
+  SW602Document::Type m_type;
+  SW602Document::Confidence m_confidence;
+  bool m_needsEncoding;
+};
+
+DetectionInfo::DetectionInfo()
+  : m_kind(SW602Document::KIND_UNKNOWN)
+  , m_type(SW602Document::TYPE_UNKNOWN)
+  , m_confidence(SW602Document::CONFIDENCE_NONE)
+  , m_needsEncoding(false)
 {
-  const shared_ptr<RVNGInputStream> input(ip, SW602_shared_ptr_noop_deleter<librevenge::RVNGInputStream>());
+}
 
-  shared_ptr<RVNGInputStream> strm;
-  if (input->isStructured())
-    strm.reset(input->getSubStreamByName("CONTENTS"));
+template<class Header>
+void fillDetectionInfo(const Header &header, DetectionInfo &info);
+
+void fillDetectionInfo(const WinText602Header &/*header*/, DetectionInfo &info)
+{
+  info.m_kind = SW602Document::KIND_TEXT;
+  info.m_type = SW602Document::TYPE_WINTEXT602;
+  info.m_needsEncoding = false;
+  // TODO: handle encryption
+  info.m_confidence = SW602Document::CONFIDENCE_EXCELLENT;
+}
+
+template<class Header>
+bool probe(RVNGInputStream &input, DetectionInfo &info) try
+{
+  input.seek(0, librevenge::RVNG_SEEK_SET);
+  Header header;
+  header.construct(input);
+  if (header.isValid())
+  {
+    fillDetectionInfo(header, info);
+    assert(m_kind != SW602Document::KIND_UNKNOWN);
+    assert(m_type != SW602Document::TYPE_UNKNOWN);
+  }
+  return info.m_confidence != SW602Document::CONFIDENCE_NONE;
+}
+catch (...)
+{
+  return false;
+}
 
-  return strm;
+SW602Document::Confidence pass(const DetectionInfo &info, SW602Document::Kind &kind, SW602Document::Type *const type, bool *const needsEncoding)
+{
+  kind = info.m_kind;
+  if (type)
+    *type = info.m_type;
+  if (needsEncoding)
+    *needsEncoding = info.m_needsEncoding;
+  return info.m_confidence;
 }
 
 }
@@ -46,20 +95,18 @@ SW602Document::Confidence SW602Document::isSupported(librevenge::RVNGInputStream
   if (!input)
     return CONFIDENCE_NONE;
 
-  shared_ptr<RVNGInputStream> content = getContent(input);
-  if (bool(content))
+  DetectionInfo info;
+
+  if (input->isStructured())
   {
-    content->seek(0, librevenge::RVNG_SEEK_SET);
-    WinText602Header header(*content);
-    // TODO: handle encryption
-    if (header.isValid())
+    if (input->existsSubStream("CONTENTS"))
     {
-      kind = KIND_TEXT;
-      if (type)
-        *type = TYPE_WINTEXT602;
-      if (needsEncoding)
-        *needsEncoding = false;
-      return CONFIDENCE_EXCELLENT;
+      const boost::scoped_ptr<RVNGInputStream> content(input->getSubStreamByName("CONTENTS"));
+      if (bool(content))
+      {
+        if (probe<WinText602Header>(*content, info))
+          return pass(info, kind, type, needsEncoding);
+      }
     }
   }
 
@@ -104,15 +151,14 @@ SW602Document::Result SW602Document::parse(librevenge::RVNGInputStream *const in
   {
   case TYPE_WINTEXT602:
   {
-    shared_ptr<RVNGInputStream> content = getContent(input);
-    if (bool(content))
-    {
-      content->seek(0, librevenge::RVNG_SEEK_SET);
-      WinText602Parser parser(content);
-      parser.parse(document);
-      return RESULT_OK;
-    }
-    break;
+    if (!input->existsSubStream("CONTENTS"))
+      return RESULT_UNSUPPORTED_FORMAT;
+    const boost::shared_ptr<RVNGInputStream> content(input->getSubStreamByName("CONTENTS"));
+    if (!content)
+      return RESULT_UNSUPPORTED_FORMAT;
+    WinText602Parser parser(content);
+    parser.parse(document);
+    return RESULT_OK;
   }
   default:
     return RESULT_UNSUPPORTED_FORMAT;
diff --git a/src/lib/WinText602Header.cpp b/src/lib/WinText602Header.cpp
index 8f88a50..75237d2 100644
--- a/src/lib/WinText602Header.cpp
+++ b/src/lib/WinText602Header.cpp
@@ -18,10 +18,14 @@
 namespace libsw602
 {
 
-WinText602Header::WinText602Header(librevenge::RVNGInputStream &input)
+WinText602Header::WinText602Header()
   : m_length(0)
   , m_valid(false)
 {
+}
+
+void WinText602Header::construct(librevenge::RVNGInputStream &input)
+{
   m_length = readU32(input);
   skip(input, 2);
   const unsigned char *const sig = readNBytes(input, 5);
diff --git a/src/lib/WinText602Header.h b/src/lib/WinText602Header.h
index 7dd038d..9ed2656 100644
--- a/src/lib/WinText602Header.h
+++ b/src/lib/WinText602Header.h
@@ -18,7 +18,9 @@ namespace libsw602
 class WinText602Header
 {
 public:
-  explicit WinText602Header(librevenge::RVNGInputStream &input);
+  WinText602Header();
+
+  void construct(librevenge::RVNGInputStream &input);
 
   bool isValid() const;
 
diff --git a/src/lib/WinText602Parser.cpp b/src/lib/WinText602Parser.cpp
index 96d371b..1bcd768 100644
--- a/src/lib/WinText602Parser.cpp
+++ b/src/lib/WinText602Parser.cpp
@@ -17,6 +17,7 @@
 #include <librevenge/librevenge.h>
 
 #include "SW602MemoryStream.h"
+#include "libsw602_utils.h"
 
 using boost::shared_ptr;
 
@@ -111,7 +112,7 @@ namespace libsw602
 
 WinText602Parser::WinText602Parser(const boost::shared_ptr<librevenge::RVNGInputStream> input)
   : SW602TextParser(input)
-  , m_header(*input)
+  , m_header()
   , m_input(input)
   , m_document(0)
   , m_sectionMap()
@@ -122,6 +123,10 @@ void WinText602Parser::parse(librevenge::RVNGTextInterface *documentInterface)
 {
   m_document = documentInterface;
 
+  m_header.construct(*m_input);
+  if (!m_header.isValid())
+    throw ParseException();
+
   m_sectionMap.reset(new WinText602SectionMap(*m_input));
 
   m_document->startDocument(librevenge::RVNGPropertyList());