Auto-detect filetype

src/ifcconvert/IfcConvert.cpp

@@ -206,7 +206,7 @@ size_t read_filters_from_file(const std::string&, inclusion_filter&, inclusion_t
 void parse_filter(geom_filter &, const std::vector<std::string>&);
 std::vector<IfcGeom::filter_t> setup_filters(const std::vector<geom_filter>&, const std::string&);
 
-bool init_input_file(const std::string& filename, IfcParse::IfcFile*& ifc_file, bool no_progress, bool mmap, bool rocksdb);
+bool init_input_file(const std::string& filename, IfcParse::IfcFile*& ifc_file, bool no_progress, bool mmap);
 
 // from https://stackoverflow.com/questions/31696328/boost-program-options-using-zero-parameter-options-multiple-times
 struct verbosity_counter {
@@ -239,7 +239,6 @@ int main(int argc, char** argv) {
 	path_t cache_file;
 	std::string log_format;
 	std::string geometry_kernel;
-	std::string input_format;
 
     po::options_description generic_options("Command line options");
 	verbosity_counter vcounter;
@@ -255,7 +254,6 @@ int main(int argc, char** argv) {
 		("yes,y", "answer 'yes' automatically to possible confirmation queries (e.g. overwriting an existing output file)")
 		("no-progress", "suppress possible progress bar type of prints that use carriage return")
 		("log-format", po::value<std::string>(&log_format), "log format: plain or json")
-		("input-format", po::value<std::string>(&input_format), "input format: ifcspf, ifcxml, rocksdb")
 		("log-file", new po::typed_value<path_t, char_t>(&log_file), "redirect log output to file");
 
     po::options_description fileio_options;
@@ -662,7 +660,7 @@ int main(int argc, char** argv) {
 	if (output_extension == XML) {
 		int exit_code = EXIT_FAILURE;
 		try {
-			if (init_input_file(IfcUtil::path::to_utf8(input_filename), ifc_file, no_progress || quiet, mmap, input_format=="rocksdb")) {
+			if (init_input_file(IfcUtil::path::to_utf8(input_filename), ifc_file, no_progress || quiet, mmap)) {
 				time_t start, end;
 				time(&start);
 				XmlSerializer s(ifc_file, IfcUtil::path::to_utf8(output_temp_filename));
@@ -682,7 +680,7 @@ int main(int argc, char** argv) {
 	} else if (output_extension == IFC) {
 		int exit_code = EXIT_FAILURE;
 		try {
-			if (init_input_file(IfcUtil::path::to_utf8(input_filename), ifc_file, no_progress || quiet, mmap, input_format == "rocksdb")) {
+			if (init_input_file(IfcUtil::path::to_utf8(input_filename), ifc_file, no_progress || quiet, mmap)) {
                 time_t start, end;
 				time(&start);
 				std::ofstream fs(output_filename.c_str());
@@ -708,7 +706,7 @@ int main(int argc, char** argv) {
 	else if (output_extension == RDB) {
 		int exit_code = EXIT_FAILURE;
 		try {
-			if (init_input_file(IfcUtil::path::to_utf8(input_filename), ifc_file, no_progress || quiet, mmap, input_format == "rocksdb")) {
+			if (init_input_file(IfcUtil::path::to_utf8(input_filename), ifc_file, no_progress || quiet, mmap)) {
 				time_t start, end;
 				time(&start);
 				RocksDbSerializer s(ifc_file, IfcUtil::path::to_utf8(output_filename));
@@ -941,7 +939,7 @@ int main(int argc, char** argv) {
 	time_t start,end;
 	time(&start);
 	
-    if (!init_input_file(IfcUtil::path::to_utf8(input_filename), ifc_file, no_progress || quiet, mmap, input_format == "rocksdb")) {
+    if (!init_input_file(IfcUtil::path::to_utf8(input_filename), ifc_file, no_progress || quiet, mmap)) {
         write_log(!quiet);
 		serializer.reset();
         IfcUtil::path::delete_file(IfcUtil::path::to_utf8(output_temp_filename)); /**< @todo Windows Unicode support */
@@ -1274,7 +1272,7 @@ void write_log(bool header) {
 
 #include <boost/algorithm/string/predicate.hpp>
 
-bool init_input_file(const std::string& filename, IfcParse::IfcFile*& ifc_file, bool no_progress, bool mmap, bool rocksdb) {
+bool init_input_file(const std::string& filename, IfcParse::IfcFile*& ifc_file, bool no_progress, bool mmap) {
     time_t start, end;
 
     // Prevent IfcFile::Init() prints by setting output to null temporarily
@@ -1293,7 +1291,7 @@ bool init_input_file(const std::string& filename, IfcParse::IfcFile*& ifc_file,
 		ifc_file = new IfcParse::IfcFile(filename, mmap);
 #else
 		(void)mmap;
-		ifc_file = new IfcParse::IfcFile(filename, rocksdb ? IfcParse::rocksdb : IfcParse::ifcspf);
+		ifc_file = new IfcParse::IfcFile(filename);
 #endif
 	}
 

src/ifcparse/IfcFile.cpp

@@ -499,3 +499,43 @@ IfcParse::IfcFile::~IfcFile() {
         delete p.second;
     }
 }
+
+#include <filesystem>
+#include <fstream>
+
+IfcParse::filetype IfcParse::guess_file_type(const std::string& fn) {
+    namespace fs = std::filesystem;
+
+    if (!fs::exists(fn)) {
+        // @todo this is just weird, but for consistency with earlier behaviour
+        // for now the only intent for this function is to auto-detect RocksDB
+        return FT_IFCSPF;
+    }
+
+    if (fs::is_directory(fn)) {
+        // Typical RocksDB file to look for
+        auto currentFile = fs::path(fn) / "CURRENT";
+
+        if (!fs::exists(currentFile) || !fs::is_regular_file(currentFile)) {
+            return FT_UNKNOWN;
+        }
+
+        std::ifstream infile(currentFile);
+        if (!infile) {
+            return FT_UNKNOWN;
+        }
+
+        std::string line;
+        if (!std::getline(infile, line)) {
+            return FT_UNKNOWN;
+        }
+
+        // RocksDB's CURRENT file typically contains a line like "MANIFEST-000001".
+        if (line.find("MANIFEST-") == 0) {
+            return FT_ROCKSDB;
+        }
+    } else {
+        // @todo just return SPF for now, but ideally this will be augmented with all other options
+        return FT_IFCSPF;
+    }
+}

src/ifcparse/IfcFile.h

@@ -572,12 +572,16 @@ namespace impl {
 }
 
 enum filetype {
-    ifcspf,
-    ifcxml,
-    rocksdb,
-    autodetect
+    FT_IFCSPF,
+    FT_IFCXML,
+    FT_IFCZIP,
+    FT_ROCKSDB,
+    FT_UNKNOWN,
+    FT_AUTODETECT
 };
 
+filetype guess_file_type(const std::string& fn);
+
 /// This class provides access to the entity instances in an IFC file
 /// The file takes ownership of instances added to this file and deletes them when the file is deleted.
 class IFC_PARSE_API IfcFile {
@@ -623,13 +627,13 @@ private:
 #ifdef USE_MMAP
     IfcFile(const std::string& path, bool mmap = false);
 #else
-    IfcFile(const std::string& path, filetype ty=ifcspf);
+    IfcFile(const std::string& path, filetype ty=FT_AUTODETECT);
 #endif
     IfcFile(std::istream& stream, int length);
     IfcFile(void* data, int length);
     IfcFile(IfcParse::IfcSpfStream* stream);
     // @nb path is only used in rocksdb mode, for spf file is in-memory only until write() is called
-    IfcFile(const IfcParse::schema_definition* schema = IfcParse::schema_by_name("IFC4"), filetype ty = ifcspf, const std::string& path = "");
+    IfcFile(const IfcParse::schema_definition* schema = IfcParse::schema_by_name("IFC4"), filetype ty = FT_AUTODETECT, const std::string& path = "");
 
     ~IfcFile();
 

src/ifcparse/IfcParse.cpp

@@ -1296,7 +1296,10 @@ IfcFile::IfcFile(const std::string& fn, bool mmap) {
 #else
 IfcFile::IfcFile(const std::string& path, filetype ty) {
     // @todo allow for rocksdb from path
-    if (ty == ifcspf) {
+    if (ty == FT_AUTODETECT) {
+        ty = guess_file_type(path);
+    }
+    if (ty == FT_IFCSPF) {
         IfcSpfStream s(path);
         storage_.emplace<1>(); // impl::in_memory_file_storage{};
         // @todo assign in constructor
@@ -1307,7 +1310,7 @@ IfcFile::IfcFile(const std::string& path, filetype ty) {
         byid_ = decltype(byid_)(&std::get<impl::in_memory_file_storage>(storage_).byid_);
         byref_excl_ = decltype(byref_excl_)(&std::get<impl::in_memory_file_storage>(storage_).byref_excl_);
         // byidentity_ = decltype(byidentity_)(&std::get<impl::in_memory_file_storage>(storage_).byidentity_);
-    } else {
+    } else if (ty == FT_ROCKSDB) {
         // @todo this can only be used for databases that already exist, because otherwise there is no way to specify the schema
         storage_.emplace<2>(path, this);
         std::get<impl::rocks_db_file_storage>(storage_).read_schema(schema_);
@@ -1315,6 +1318,8 @@ IfcFile::IfcFile(const std::string& path, filetype ty) {
         byid_ = decltype(byid_)(&std::get<impl::rocks_db_file_storage>(storage_).instance_by_name_);
         byref_excl_ = decltype(byref_excl_)(&std::get<impl::rocks_db_file_storage>(storage_).byref_excl_);
         // byidentity_ = decltype(byidentity_)(&std::get<impl::rocks_db_file_storage>(storage_).instance_cache_);
+    } else {
+        throw std::runtime_error("Unsupported file format");
     }
     ifcroot_type_ = schema_->declaration_by_name("IfcRoot");
 }
@@ -1344,21 +1349,26 @@ IfcFile::IfcFile(const IfcParse::schema_definition* schema, filetype ty, const s
     : schema_(schema)
     , ifcroot_type_(schema_->declaration_by_name("IfcRoot"))
     , max_id_(0)
-    , _header(ty == rocksdb ? this : nullptr)
+    , _header(ty == FT_ROCKSDB ? this : nullptr)
 {
-    if (ty == ifcspf) {
+    if (ty == FT_AUTODETECT) {
+        ty = guess_file_type(path);
+    }
+    if (ty == FT_IFCSPF) {
         storage_.emplace<1>();
         std::get<impl::in_memory_file_storage>(storage_).file = this;
 
         byid_ = decltype(byid_)(&std::get<impl::in_memory_file_storage>(storage_).byid_);
         byref_excl_ = decltype(byref_excl_)(&std::get<impl::in_memory_file_storage>(storage_).byref_excl_);
         // byidentity_ = decltype(byidentity_)(&std::get<impl::in_memory_file_storage>(storage_).byidentity_);
-    } else {
+    } else if (ty == FT_ROCKSDB) {
         storage_.emplace<2>(path, this);
 
         byid_ = decltype(byid_)(&std::get<impl::rocks_db_file_storage>(storage_).instance_by_name_);
         byref_excl_ = decltype(byref_excl_)(&std::get<impl::rocks_db_file_storage>(storage_).byref_excl_);
         // byidentity_ = decltype(byidentity_)(&std::get<impl::rocks_db_file_storage>(storage_).instance_cache_);
+    } else {
+        throw std::runtime_error("Unsupported file format");
     }
     setDefaultHeaderValues();
 }