Replace boost::filesystem by std::filesystem

.github/workflows/test.yml

@@ -11,7 +11,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v2
-    - uses: cachix/install-nix-action@v7
+    - uses: cachix/install-nix-action@v10
     - run: nix-channel --add ${{ matrix.channel }} nixpkgs
     - run: nix-channel --update
-    - run: nix-build
+    - run: nix-build

CMakeLists.txt

@@ -5,7 +5,7 @@ enable_testing()
 
 set (ALPINOCORPUS_VERSION "2.6.0")
 
-set (CMAKE_CXX_STANDARD 11)
+set (CMAKE_CXX_STANDARD 17)
 
 if (NOT CMAKE_CONFIGURATION_TYPES AND NOT CMAKE_BUILD_TYPE)
   set (CMAKE_BUILD_TYPE Release)
@@ -41,7 +41,7 @@ endif()
 
 include(GNUInstallDirs REQUIRED)
 
-find_package(Boost 1.50 COMPONENTS system filesystem REQUIRED)
+find_package(Boost 1.50 COMPONENTS system REQUIRED)
 
 if(Boost_FOUND)
   include_directories(${Boost_INCLUDE_DIRS})

src/CompactCorpusReaderPrivate.cpp

@@ -1,3 +1,4 @@
+#include <filesystem>
 #include <fstream>
 #include <memory>
 #include <mutex>
@@ -7,8 +8,6 @@
 #include <typeinfo>
 #include <vector>
 
-#include <boost/filesystem.hpp>
-
 #include <AlpinoCorpus/Error.hh>
 
 #include "DzIstream.hh"
@@ -20,7 +19,7 @@ namespace {
     char const * const INDEX_EXT = ".index";
 }
 
-namespace bf = boost::filesystem;
+namespace fs = std::filesystem;
 
 namespace alpinocorpus {
 
@@ -51,12 +50,12 @@ void CompactCorpusReaderPrivate::construct(std::string const &canonical,
     std::string const &indexPath)
 {
     // XXX race condition up ahead
-    bf::path dataP(dataPath);
-    if (!bf::is_regular_file(dataP))
+    fs::path dataP(dataPath);
+    if (!fs::is_regular_file(dataP))
         throw OpenError(dataPath, "not a regular file");
 
-    bf::path indexP(indexPath);
-    if (!bf::is_regular_file(indexP))
+    fs::path indexP(indexPath);
+    if (!fs::is_regular_file(indexP))
         throw OpenError(indexPath, "not a regular file");
 
     open(dataPath, indexPath);

src/DbCorpusWriter.cpp

@@ -1,19 +1,18 @@
 #include <cerrno>
 #include <cstdio>
 #include <cstring>
+#include <filesystem>
 #include <sstream>
 #include <string>
 
-#include <boost/filesystem.hpp>
-
 #include <dbxml/DbXml.hpp>
 
 #include <AlpinoCorpus/DbCorpusWriter.hh>
 #include <AlpinoCorpus/Entry.hh>
 #include <AlpinoCorpus/Error.hh>
 #include <AlpinoCorpus/util/NonCopyable.hh>
 
-namespace bf = boost::filesystem;
+namespace fs = std::filesystem;
 namespace db = DbXml;
 
 namespace alpinocorpus {
@@ -76,7 +75,7 @@ namespace alpinocorpus {
                                                     db::XmlContainer
                                                     ::NodeContainer);
             } else {
-                if (bf::exists(path))
+                if (fs::exists(path))
                     d_container = d_mgr.openContainer(path, config);
                 else
                     d_container = d_mgr.createContainer(path, config,
@@ -149,7 +148,7 @@ namespace alpinocorpus {
                                db::XmlUpdateContext &ctx)
     {
         try {
-            std::string canonical(bf::path(name).generic_string());
+            std::string canonical(fs::path(name).generic_string());
             d_container.putDocument(canonical, content, ctx,
                                   db::DBXML_WELL_FORMED_ONLY);
         } catch (db::XmlException const &e) {

src/DirectoryCorpusReaderPrivate.cpp

@@ -1,4 +1,5 @@
 #include <algorithm>
+#include <filesystem>
 #include <fstream>
 #include <iterator>
 #include <limits>
@@ -7,8 +8,6 @@
 #include <typeinfo>
 
 #include <boost/algorithm/string/predicate.hpp>
-#include <boost/assert.hpp>
-#include <boost/filesystem.hpp>
 
 #include <AlpinoCorpus/Error.hh>
 #include <AlpinoCorpus/Entry.hh>
@@ -18,17 +17,17 @@
 #include "util/NameCompare.hh"
 #include "util/textfile.hh"
 
-namespace bf = boost::filesystem;
+namespace fs = std::filesystem;
 
 namespace {
     class DirIter : public alpinocorpus::IterImpl
     {
-        boost::filesystem::recursive_directory_iterator iter;
-        boost::filesystem::path d_directory;
+        fs::recursive_directory_iterator iter;
+        fs::path d_directory;
 
       public:
-        DirIter(boost::filesystem::path const &path,
-            boost::filesystem::recursive_directory_iterator i);
+        DirIter(fs::path const &path,
+            fs::recursive_directory_iterator i);
         alpinocorpus::IterImpl *copy() const;
         bool hasNext();
         alpinocorpus::Entry next(alpinocorpus::CorpusReader const &rdr);
@@ -37,7 +36,7 @@ namespace {
     };
 
     DirIter::DirIter(
-        bf::path const &path, bf::recursive_directory_iterator i) :
+        fs::path const &path, fs::recursive_directory_iterator i) :
         d_directory(path), iter(i)
     {
     }
@@ -51,7 +50,7 @@ namespace {
     bool DirIter::isValid()
     {
         // End is a correct iterator state.
-        if (iter == bf::recursive_directory_iterator())
+        if (iter == fs::recursive_directory_iterator())
             return true;
 
         return iter->path().extension() == ".xml";
@@ -64,7 +63,7 @@ namespace {
           ++iter;
         }
 
-        return iter != bf::recursive_directory_iterator();
+        return iter != fs::recursive_directory_iterator();
     }
 
     alpinocorpus::Entry DirIter::next(alpinocorpus::CorpusReader const &rdr)
@@ -78,7 +77,7 @@ namespace {
         if (entryPathStr[0] == '/')
             entryPathStr.erase(0, 1);
 
-        bf::path entryPath(entryPathStr);
+        fs::path entryPath(entryPathStr);
 
         // Move the iterator.
         ++iter;
@@ -94,8 +93,8 @@ namespace {
         std::vector<std::string>::const_iterator d_iter;
 
       public:
-        SortedDirIter(boost::filesystem::path const &path,
-            boost::filesystem::recursive_directory_iterator i);
+        SortedDirIter(fs::path const &path,
+            fs::recursive_directory_iterator i);
         alpinocorpus::IterImpl *copy() const;
         bool hasNext();
         alpinocorpus::Entry next(alpinocorpus::CorpusReader const &rdr);
@@ -106,9 +105,9 @@ namespace {
     };
 
     SortedDirIter::SortedDirIter(
-        bf::path const &path, bf::recursive_directory_iterator i)
+        fs::path const &path, fs::recursive_directory_iterator i)
     {
-        for (; i != bf::recursive_directory_iterator(); i++)
+        for (; i != fs::recursive_directory_iterator(); i++)
         {
             std::string entryPathStr = i->path().string();
             entryPathStr.erase(0, path.string().size());
@@ -161,7 +160,7 @@ namespace {
     {
         // We assume the iterator is valid, since hasNext() should be called
 
-        bf::path entryPath(*d_iter);
+        fs::path entryPath(*d_iter);
 
         // Move the iterator.
         ++d_iter;
@@ -180,12 +179,12 @@ DirectoryCorpusReaderPrivate::DirectoryCorpusReaderPrivate(
     d_nEntries(std::numeric_limits<size_t>::max())
 {
     if (directory[directory.size() - 1] == '/')
-        d_directory = bf::path(directory).parent_path();
+        d_directory = fs::path(directory).parent_path();
     else
-        d_directory = bf::path(directory);
+        d_directory = fs::path(directory);
 
-    if (!bf::exists(d_directory) ||
-        !bf::is_directory(d_directory))
+    if (!fs::exists(d_directory) ||
+        !fs::is_directory(d_directory))
         throw OpenError(directory, "non-existent or not a directory");
 }
 
@@ -197,10 +196,12 @@ CorpusReader::EntryIterator DirectoryCorpusReaderPrivate::getEntries(SortOrder s
     switch (sortOrder) {
         case NaturalOrder:
             return EntryIterator(new DirIter(d_directory,
-                bf::recursive_directory_iterator(d_directory, bf::symlink_option::recurse)));
+                fs::recursive_directory_iterator(d_directory,
+		  fs::directory_options::follow_directory_symlink)));
         case NumericalOrder:
             return EntryIterator(new SortedDirIter(d_directory,
-                bf::recursive_directory_iterator(d_directory, bf::symlink_option::recurse)));
+                fs::recursive_directory_iterator(d_directory,
+		  fs::directory_options::follow_directory_symlink)));
         default:
             throw NotImplemented("Unexpected sort order.");
     }
@@ -232,12 +233,12 @@ size_t DirectoryCorpusReaderPrivate::getSize() const
 
 std::string DirectoryCorpusReaderPrivate::readEntry(std::string const &entry) const
 {
-    bf::path p(d_directory);
+    fs::path p(d_directory);
     p /= entry;
     return util::readFile(p.string());
 }
 
-bf::path DirectoryCorpusReaderPrivate::cachePath() const
+fs::path DirectoryCorpusReaderPrivate::cachePath() const
 {
     return d_directory.parent_path() / d_directory.filename().replace_extension(".dir_index");
 }

src/DirectoryCorpusReaderPrivate.hh

@@ -1,11 +1,10 @@
 #ifndef ALPINO_DIRECTORYCORPUSREADER_PRIVATE_HH
 #define ALPINO_DIRECTORYCORPUSREADER_PRIVATE_HH
 
+#include <filesystem>
 #include <string>
 #include <vector>
 
-#include <boost/filesystem.hpp>
-
 #include <AlpinoCorpus/CorpusReader.hh>
 #include <AlpinoCorpus/Entry.hh>
 #include <AlpinoCorpus/IterImpl.hh>
@@ -32,9 +31,9 @@ public:
     virtual size_t getSize() const;
 
 private:
-    boost::filesystem::path cachePath() const;
+    std::filesystem::path cachePath() const;
 
-    boost::filesystem::path d_directory;
+    std::filesystem::path d_directory;
     mutable size_t d_nEntries;
     bool d_entriesRead;
 };

src/DzOstreamBuf.cpp

@@ -2,7 +2,9 @@
 #include <cmath>
 #include <cstdint>
 #include <cstdio>
+#include <cstdlib>
 #include <cstring>
+#include <filesystem>
 #include <iostream>
 #include <limits>
 #include <stdexcept>
@@ -11,14 +13,12 @@
 
 #include <zlib.h>
 
-#include <boost/filesystem.hpp>
-
 #include "DzOstreamBuf.hh"
 #include "gzip.hh"
 #include "util/bufutil.hh"
 
 
-namespace bf = boost::filesystem;
+namespace fs = std::filesystem;
 
 namespace {
 
@@ -44,13 +44,18 @@ DzOstreamBuf::DzOstreamBuf(char const *filename) : d_size(0), d_crc32(crc32(0L,
 		return;
 	}
 
-  // XXX - There is a race condition here, but Boost does not seem to
-  // provide a variant that returns a file descriptor. We used mkstemp
-  // previously, but it is not portable.
-  std::string tmpFilename =
-    bf::unique_path(std::string(filename) + "-%%%%-%%%%-%%%%-%%%%").string();
-  d_tmpFilename = tmpFilename;
-  d_zDataStream = fopen(tmpFilename.c_str(), "w");
+	// XXX - Seems like C++17 does not provide an alternative to mkstemp
+	// yet. But since we haven't recently built any Windows versions and
+	// Windows has WSL anyway, this should be fine.
+	std::string tmpFilename =
+	  std::string(filename) + "-XXXX-XXXX-XXXX-XXXX";
+	int fd;
+	if ((fd = mkstemp(tmpFilename.data())) == -1)
+		throw std::runtime_error(std::string("DzOstreamBuf::DzOstreamBuf: Could not create temporary file") +
+			d_tmpFilename);
+
+	d_tmpFilename = tmpFilename;
+	d_zDataStream = fdopen(fd, "w");
 
 	if (d_zDataStream == NULL)
 		throw std::runtime_error(std::string("DzOstreamBuf::DzOstreamBuf: Could not open ") +
@@ -114,7 +119,7 @@ DzOstreamBuf::~DzOstreamBuf()
 
 	fclose(d_dzStream);
 
-  bf::remove(d_tmpFilename);
+	fs::remove(d_tmpFilename);
 }
 
 void DzOstreamBuf::flushBuffer()
@@ -202,14 +207,14 @@ void DzOstreamBuf::writeHeader()
   auto secsSinceEpoch = std::chrono::duration_cast<std::chrono::seconds>(
     clock.time_since_epoch()).count();
 
-  if (secsSinceEpoch > std::numeric_limits<boost::int32_t>::max())
+  if (secsSinceEpoch > std::numeric_limits<int32_t>::max())
       secsSinceEpoch = 0;
 
 	header[GZ_HEADER_ID1] = gzipId1;
 	header[GZ_HEADER_ID2] = gzipId2;
 	header[GZ_HEADER_CM] = GZ_CM_DEFLATE;
 	header[GZ_HEADER_FLG] = GZ_FLG_EXTRA;
-	util::writeToBuf<boost::uint32_t>(&header[0] + GZ_HEADER_MTIME, secsSinceEpoch);
+	util::writeToBuf<uint32_t>(&header[0] + GZ_HEADER_MTIME, secsSinceEpoch);
 	header[GZ_HEADER_XFL] = GZ_XFL_MAX;
 	header[GZ_HEADER_OS] = GZ_OS_UNIX;
 
@@ -220,8 +225,8 @@ void DzOstreamBuf::writeTrailer()
 {
 	std::vector<unsigned char> trailer(GZ_TRAILER_SIZE);
 
-	util::writeToBuf<boost::uint32_t>(&trailer[0] + GZ_TRAILER_CRC32, d_crc32);
-	util::writeToBuf<boost::uint32_t>(&trailer[0] + GZ_TRAILER_ISIZE, d_size);
+	util::writeToBuf<uint32_t>(&trailer[0] + GZ_TRAILER_CRC32, d_crc32);
+	util::writeToBuf<uint32_t>(&trailer[0] + GZ_TRAILER_ISIZE, d_size);
 
 	fwrite(&trailer[0], 1, GZ_TRAILER_SIZE, d_dzStream);
 }