Make openSinglePieceOrSplitZimFile better handling single/multi part …

…opening.

Now, it tries to open as single piece only if not ending by `.zimaa`.
If it cannot find the file, or if ending with `.zimaa`, it tries multi zim part.

In any case, FileImpl constructor check for size and throw a
ZimFileFormatError is file's size doesn't correspond to what is the header.

src/file_compound.cpp

@@ -41,12 +41,20 @@ void FileCompound::addPart(FilePart* fpart)
   _fsize += fpart->size();
 }
 
-std::shared_ptr<FileCompound> FileCompound::openSinglePieceOrSplitZimFile(const std::string& filename) {
+std::shared_ptr<FileCompound> FileCompound::openSinglePieceOrSplitZimFile(std::string filename) {
+  std::shared_ptr<FileCompound> fileCompound;
+  if (filename.size() > 6 && filename.substr(filename.size()-6) == ".zimaa") {
+    filename.resize(filename.size()-2);
+  } else {
   try {
-    return std::make_shared<FileCompound>(filename);
-  } catch (...) {
-    return std::make_shared<FileCompound>(filename, FileCompound::MultiPartToken::Multi);
+      fileCompound = std::make_shared<FileCompound>(filename);
+    } catch(...) { }
   }
+
+  if ( !fileCompound ) {
+    fileCompound = std::make_shared<FileCompound>(filename, FileCompound::MultiPartToken::Multi);
+  }
+  return fileCompound;
 }
 
 FileCompound::FileCompound(const std::string& filename):

src/file_compound.h

@@ -57,7 +57,7 @@ class FileCompound : private std::map<Range, FilePart*, less_range> {
     enum class MultiPartToken { Multi };
 
   public: // functions
-    static std::shared_ptr<FileCompound> openSinglePieceOrSplitZimFile(const std::string& filename);
+    static std::shared_ptr<FileCompound> openSinglePieceOrSplitZimFile(std::string filename);
     explicit FileCompound(const std::string& filename);
     explicit FileCompound(const std::string& filename, MultiPartToken token);
 

src/fileimpl.cpp

@@ -207,6 +207,15 @@ class Grouping
       throw ZimFileFormatError("error reading zim-file header.");
     }
 
+    // This can happen for several reasons:
+    // - Zim file is corrupted (corrupted header)
+    // - Zim file is too small (ongoing download, truncated file...)
+    // - Zim file is embedded at beginning of another file (and we try to open the file as a zim file)
+    //   If open through a FdInput, size should be set in FdInput.
+    if (header.hasChecksum() && (header.getChecksumPos() + 16) != size_type(zimReader->size())) {
+      throw ZimFileFormatError("Zim file(s) is of bad size or corrupted.");
+    }
+
     auto pathPtrReader = sectionSubReader(*zimReader,
                                           "Dirent pointer table",
                                           offset_t(header.getPathPtrPos()),
@@ -297,10 +306,6 @@ class Grouping
         throw ZimFileFormatError("last cluster offset larger than file size; file corrupt");
       }
     }
-
-    if (header.hasChecksum() && header.getChecksumPos() != (getFilesize().v-16) ) {
-      throw ZimFileFormatError("Checksum position is not valid");
-    }
   }
 
   offset_type FileImpl::getMimeListEndUpperLimit() const

test/archive.cpp

@@ -264,6 +264,31 @@ TEST(ZimArchive, openRealZimArchive)
   }
 }
 
+TEST(ZimArchive, openSplitZimArchive)
+{
+  const char* fname = "wikibooks_be_all_nopic_2017-02_splitted.zim";
+
+  for (auto& testfile: getDataFilePath(fname)) {
+    const TestContext ctx{ {"path", testfile.path+"aa" } };
+    std::unique_ptr<zim::Archive> archive;
+    EXPECT_NO_THROW( archive.reset(new zim::Archive(testfile.path+"aa")) ) << ctx;
+    if ( archive ) {
+      EXPECT_TRUE( archive->check() ) << ctx;
+    }
+  }
+}
+
+TEST(ZimArchive, openDontFallbackOnNonSplitZimArchive)
+{
+  const char* fname = "wikibooks_be_all_nopic_2017-02.zim";
+
+  for (auto& testfile: getDataFilePath(fname)) {
+    const TestContext ctx{ {"path", testfile.path+"aa" } };
+    std::unique_ptr<zim::Archive> archive;
+    EXPECT_THROW( archive.reset(new zim::Archive(testfile.path+"aa")), std::runtime_error) << ctx;
+  }
+}
+
 TEST(ZimArchive, randomEntry)
 {
   const char* const zimfiles[] = {
@@ -434,7 +459,7 @@ TEST(ZimArchive, validate)
 
   TEST_BROKEN_ZIM_NAME(
     "invalid.invalid_checksumpos.zim",
-    "Checksum position is not valid\n"
+    "Zim file(s) is of bad size or corrupted.\n"
   );
 
   TEST_BROKEN_ZIM_NAME(