- changed signature of validate() methods

- added checksum class, uses openssl

src/include/bag.hpp

@@ -31,7 +31,7 @@ class Bag {
     Bag();
     Bag(string dfname);
     bool store( string dfname);
-    bool validate();
+    bool validate( list<string> & log );
 };
 #endif
 // vim: set tabstop=4 softtabstop=0 expandtab shiftwidth=4 smarttab

src/include/bagmetadata.hpp

@@ -2,6 +2,7 @@
 #define LIBCBAG_BAGMETADATA
 #include <string>
 #include <map>
+#include <list>
 using namespace std;
 
 enum {
@@ -30,7 +31,7 @@ class Bagmetadata{
     Bagmetadata( string basedir );
     bool has_bagmetadata();
     map<string,string> get_metadata();
-    bool validate();
+    bool validate( list<string> & log );
 };
 
 #endif

src/include/checksum.hpp

@@ -1,5 +1,20 @@
 #ifndef LIBCBAG_CHECKSUM
 #define LIBCBAG_CHECKSUM
+#include <list>
+#include <string>
+#include <openssl/md5.h>
+#include <openssl/sha.h>
+using namespace std;
+
 enum checksum_algorithms {md5=1, sha1};
+
+class Checksum {
+  public:
+    //Checksum();
+    //list<string> get_supported_checksum_algorithms();
+    //bool is_algorithm_supported( string alg );
+    string checksum_of_file(string filepath, checksum_algorithms alg);
+};
+
 #endif
 // vim: set tabstop=4 softtabstop=0 expandtab shiftwidth=4 smarttab

src/include/manifest.hpp

@@ -2,15 +2,17 @@
 #define LIBCBAG_MANIFEST
 #include <string>
 #include <map>
+#include <list>
 #include "checksum.hpp"
 using namespace std;
 
 class Manifest{
-  private:
+  protected:
+    string basedir;
     map<checksum_algorithms,string> manifest_algorithm_files;
   public:
     virtual map<string,string> get_checksum_file_pairs(checksum_algorithms algorithm);
-    virtual bool validate();
+    virtual bool validate( list<string> & log );
 };
 
 #endif

src/include/payload.hpp

@@ -16,7 +16,7 @@ class Payload{
     Payload( string basedir );
     list<string> get_all_relative_paths();
     list<string> get_all_absolute_paths();
-    bool validate();
+    bool validate(list<string> & log);
 };
 #endif
 // vim: set tabstop=4 softtabstop=0 expandtab shiftwidth=4 smarttab

src/include/payloadmanifest.hpp

@@ -8,8 +8,6 @@
 using namespace std;
 
 class Payloadmanifest : public Manifest {
-  private:
-    map<checksum_algorithms,string> manifest_algorithm_files;
   public:
     Payloadmanifest( string basedir );
 };

src/include/tagmanifest.hpp

@@ -9,7 +9,6 @@ using namespace std;
 
 class Tagmanifest : public Manifest {
   private:
-    map<checksum_algorithms,string> manifest_algorithm_files;
     bool exist_manifest_files;
   public:
     Tagmanifest( string basedir );

src/lib/bagmetadata.cpp

@@ -76,8 +76,8 @@ map <string,string> Bagmetadata::get_metadata() {
   return this->metadata;
 }
 
-bool Bagmetadata::validate() {
-	// TODO
-	return true;
+bool Bagmetadata::validate(list<string> & log ) {
+  // TODO
+  return true;
 }
 // vim: set tabstop=4 softtabstop=0 expandtab shiftwidth=4 smarttab

src/lib/checksum.cpp

@@ -0,0 +1,45 @@
+#include <iostream>
+#include <iomanip>
+#include <fstream>
+#include <sstream>
+#include <string>
+#include "checksum.hpp"
+
+
+string Checksum::checksum_of_file(string filepath, checksum_algorithms alg) {
+  ifstream file (filepath, ios::ate);
+  stringstream hex_result;
+  if (file.is_open() ) {
+    ifstream::pos_type fileSize;
+    char * memBlock;
+    fileSize = file.tellg();
+    memBlock = new char[fileSize];
+    file.seekg(0,ios::beg);
+    file.read(memBlock, fileSize);
+    switch ( alg ) {
+      case md5: {
+                  unsigned char result[MD5_DIGEST_LENGTH];
+                  MD5((unsigned char*) memBlock, fileSize, result);
+                  for (int i=0; i<MD5_DIGEST_LENGTH; i++) {
+                    hex_result<< hex << setw(2) << setfill('0') << (int) result[i];
+                  }
+                  break;
+                }
+      case sha1: {
+                   unsigned char result[SHA_DIGEST_LENGTH];
+                   SHA1((unsigned char*) memBlock, fileSize, result);
+                   for (int i=0; i<SHA_DIGEST_LENGTH; i++) {
+                     hex_result<< hex << setw(2) << setfill('0') << (int) result[i];
+                   }
+                   break;
+                 }
+    }
+    // cout << "# # CHECKSUM for filepath '" << filepath << "' " << hex_result.str() << endl;
+    file.close();
+  } else {
+    cout << "file '"<< filepath << "' could not be opened" << endl;
+  }
+  return hex_result.str();
+}
+
+// vim: set tabstop=4 softtabstop=0 expandtab shiftwidth=4 smarttab

src/lib/load_bag.cpp

@@ -6,10 +6,10 @@
 #include <string>
 
 Bag::Bag( string dfname ) {
-  cout << "load constructor (" << dfname << ")" << endl;
+  // cout << "load constructor (" << dfname << ")" << endl;
   // read in file bagit.txt
   string bagit_txt_path =  dfname + "bagit.txt";
-  log << "parse " << bagit_txt_path << endl;
+  //log << "parse " << bagit_txt_path << endl;
   ifstream bagit_txt_file;
   bagit_txt_file.open( bagit_txt_path );
   if (bagit_txt_file.is_open()) {
@@ -26,10 +26,10 @@ Bag::Bag( string dfname ) {
     getline(version_ss, major ,'.');
     getline(version_ss, minor, '.');
     if (0 != vprefix.compare("BagIt-Version:")) { 
-	    log << "wrong vprefix='" << vprefix << "', but 'BagIt-Version:' expected" << endl;
+      // log << "wrong vprefix='" << vprefix << "', but 'BagIt-Version:' expected" << endl;
     }
-    cout << "major:'"<<major<<"'"<<endl;
-    cout << "minor:'"<<minor<<"'"<<endl;
+    //cout << "major:'"<<major<<"'"<<endl;
+    //cout << "minor:'"<<minor<<"'"<<endl;
     Bag::bagit_version_major = stoi(major);
     Bag::bagit_version_minor = stoi(minor);
 
@@ -39,15 +39,10 @@ Bag::Bag( string dfname ) {
     getline(utf8_ss, uprefix, ' ');
     getline(utf8_ss, uvalue, ' ');
     if (0 != uprefix.compare("Tag-File-Character-Encoding:")) { 
-	    log << "wrong uprefix='" << uprefix << "', but 'Tag-File-Character-Encoding:' expected" << endl;
+      //log << "wrong uprefix='" << uprefix << "', but 'Tag-File-Character-Encoding:' expected" << endl;
     }
     Bag::tag_file_character_encoding = uvalue;
-
-
-
-
-
-    cout << "Bagit Version ("<< version_line << ") major=" << Bag::bagit_version_major << " minor=" << Bag::bagit_version_minor << endl;
+    //cout << "Bagit Version ("<< version_line << ") major=" << Bag::bagit_version_major << " minor=" << Bag::bagit_version_minor << endl;
   } else {
     Bag::log << "file " << bagit_txt_path << " could not be opened" <<endl;
   }
@@ -64,19 +59,17 @@ Bag::Bag( string dfname ) {
   map<string,string> md = Bag::bagmetadata_p->get_metadata();
   map<string,string>::iterator m;
   for (m=md.begin(); m!=md.end(); m++) {
-    cout << m->first << " = " << m->second <<endl;
+    // cout << m->first << " = " << m->second <<endl;
   }
 
   list<string>::iterator i;
   for (i=files.begin(); i!=files.end(); i++) {
-    cout << "file/dir (rel):" << (*i) << endl;
+    // cout << "file/dir (rel):" << (*i) << endl;
     //cout << "file/dir (abs):" << (*i) << endl;
   }
   Bag::payloadmanifest_p->get_checksum_file_pairs( md5 );
   Bag::tagmanifest_p->get_checksum_file_pairs( md5 );
 
-  cout << "The bag '"<<dfname<<"' is " << (this->validate() ? "valid" : "invalid") << endl;
-
 }
 
 // vim: set tabstop=4 softtabstop=0 expandtab shiftwidth=4 smarttab

src/lib/manifest.cpp

@@ -12,7 +12,7 @@ namespace fs = boost::filesystem;
 using namespace std;
 
 map<string,string> Manifest::get_checksum_file_pairs(checksum_algorithms alg) {
-  string filename = Manifest::manifest_algorithm_files[ alg ];
+  string filename = this->manifest_algorithm_files[ alg ];
   map<string,string> checksum_file_pairs;
   cout << "using file " << filename << endl;
   ifstream file;
@@ -26,15 +26,31 @@ map<string,string> Manifest::get_checksum_file_pairs(checksum_algorithms alg) {
       line_ss >> checksum;
       line_ss >> subfile;
       checksum_file_pairs[checksum]=subfile;
-      cout << "checksum="<<checksum<<" file="<<subfile<<endl;
+      // cout << "checksum="<<checksum<<" file="<<subfile<<endl;
     }
     file.close();
   }
   return checksum_file_pairs;
 }
 
-bool Manifest::validate() {
-	// TODO
-	return true;
+bool Manifest::validate( list<string> & log ) {
+  Checksum checksum;
+  bool is_valid = true;
+  for (map<checksum_algorithms,string>::iterator it=this->manifest_algorithm_files.begin(); it!=this->manifest_algorithm_files.end(); ++it) {
+    cout << "validate using file "<< (it->second) << " (" << (it->first) << ")" <<endl;
+    map<string,string>checksum_file_pairs = this->get_checksum_file_pairs( it->first );
+    for (map<string,string>::iterator ch=checksum_file_pairs.begin(); ch!=checksum_file_pairs.end(); ++ch) {
+      string expected_checksum = ch->first;
+      string file = this->basedir + ch->second;
+      string calc_checksum = checksum.checksum_of_file( file, it->first );
+      // cout << "\t" << "file="<<file <<" expected:"<<expected_checksum << " found: " << calc_checksum << endl;
+      if (0 != expected_checksum.compare( calc_checksum)) {
+        // cout << ( "Bagit file '" + file + "', checksum '" + expected_checksum + "' is expected, but found: '" + calc_checksum + "'") << endl;
+        log.push_back( "Bagit file '" + file + "', checksum '" + expected_checksum + "' is expected, but found: '" + calc_checksum + "'");
+        is_valid = false;
+      }
+    }
+  }
+  return is_valid;
 }
 // vim: set tabstop=4 softtabstop=0 expandtab shiftwidth=4 smarttab

src/lib/payload.cpp

@@ -11,9 +11,9 @@ using namespace std;
 Payload::Payload( string basedir ) {
   fs::path p{ basedir };
   fs::file_status s = fs::status( p );
-  cout << "basedir "<< p.string() << endl;
-  cout << "is dir: "<< fs::is_directory( s) << endl;
-  cout << "exists: "<< fs::exists(s) << endl;
+  // cout << "basedir "<< p.string() << endl;
+  // cout << "is dir: "<< fs::is_directory( s) << endl;
+  // cout << "exists: "<< fs::exists(s) << endl;
   this->basedir = basedir;
 };
 
@@ -49,8 +49,8 @@ list<string> Payload::get_all_absolute_paths() {
   return strpaths;
 }
 
-bool Payload::validate() {
-	// TODO
-	return true;
+bool Payload::validate( list<string> & log ) {
+  // TODO
+  return true;
 }
 // vim: set tabstop=4

src/lib/payloadmanifest.cpp

@@ -13,20 +13,22 @@ using namespace std;
 
 
 Payloadmanifest::Payloadmanifest( string basedir ) {
+  Payloadmanifest::basedir = basedir;
   map<checksum_algorithms, string> possible_manifest_files;
   possible_manifest_files[md5] = "manifest-md5.txt";
   possible_manifest_files[sha1] = "manifest-sha1.txt";
   for (map<checksum_algorithms, string>::iterator it=possible_manifest_files.begin(); it!=possible_manifest_files.end(); ++it) {
-    // debug
-    cout << it->first << " => " << it->second << endl;
     // test if file exists
     string filename = basedir + it->second;
     fs::path p{ filename };
     fs::file_status s = fs::status( p );
-    cout << "path "<< p.string() << endl;
-    cout << "is file: "<< fs::is_regular_file( s) << endl;
+    // cout << "path "<< p.string() << endl;
+    // cout << "is file: "<< fs::is_regular_file( s) << endl;
     if (fs::is_regular_file( s)) {
       Payloadmanifest::manifest_algorithm_files[it->first] = filename;
+    // debug
+    cout << "PAYLOADMANIFEST: alg="<< it->first << " => filename=" << it->second << endl;
+    cout << "                 alg="<< it->first << " => filename=" << Payloadmanifest::manifest_algorithm_files[it->first] << endl;
     }
   }
 }

src/lib/tagmanifest.cpp

@@ -13,6 +13,7 @@ using namespace std;
 
 
 Tagmanifest::Tagmanifest( string basedir ) {
+  Tagmanifest::basedir = basedir;
   map<checksum_algorithms, string> possible_manifest_files;
   possible_manifest_files[md5] = "tagmanifest-md5.txt";
   possible_manifest_files[sha1] = "tagmanifest-sha1.txt";
@@ -38,7 +39,6 @@ map<string,string> Tagmanifest::get_checksum_file_pairs(checksum_algorithms alg)
   for (map<string,string>::iterator it=checksum_file_pairs.begin(); it!=checksum_file_pairs.end(); ++it) {
     if (it->second.find( "data/" ) == 0) {
       cout << "found 'data/' in the beginning of file '"<<it->second<<"' in tagmanifest file '"<< this->manifest_algorithm_files[ alg] << endl;
-
     }
   }
   return checksum_file_pairs;

src/lib/validate_bag.cpp

@@ -3,67 +3,73 @@
 
 
 
-bool Bag::validate() {
-	bool is_valid = true;
-	if (this->bagit_version_major != 0) {
-		is_valid = false;
-	}
-	if (this->bagit_version_minor != 97) {
-		is_valid = false;
-	}
-	if (0 != tag_file_character_encoding.compare( "UTF-8" )) {
-		is_valid = false;
-	}
-	if (NULL == this->payload_p) {
-		is_valid = false;
-	} else {
-		bool ret = this->payload_p->validate();
-		if (ret == false) {
-			is_valid = false;
-		}
-	}
-	if (NULL == this->payloadmanifest_p) {
-		is_valid = false;
-	} else {
-		bool ret = this->payloadmanifest_p->validate();
-		if (ret == false) {
-			is_valid = false;
-		}
-	}
-	// next elements are optional
-	if (NULL == this->tagmanifest_p) {
-	} else {
-		bool ret = this->tagmanifest_p->validate();
-		if (ret == false) {
-			is_valid = false;
-		}
-	}
-	if (NULL == this->bagmetadata_p) {
-	} else {
-		bool ret = this->bagmetadata_p->validate();
-		if (ret == false) {
-			is_valid = false;
-		}
-	}
-	/*
-	if (NULL == this->fetchfile_p) {
-	} else {
-		bool ret = this->fetchfile_p->validate();
-		if (ret == false) {
-			is_valid = false;
-		}
-	}
+bool Bag::validate( list<string> & log ) {
+  bool is_valid = true;
+  if (this->bagit_version_major != 0) {
+    log.push_back( "Bagit major version 0 is expected, but got: " + to_string(this->bagit_version_major));
+    is_valid = false;
+  }
+  if (this->bagit_version_minor != 97) {
+    log.push_back( "Bagit minor version 97 is expected, but got: " + to_string(this->bagit_version_minor));
+    is_valid = false;
+  }
+  if (0 != tag_file_character_encoding.compare( "UTF-8" )) {
+    log.push_back( "Bagit character encoding UTF-8 is expected, but got: " + this->tag_file_character_encoding);
+    is_valid = false;
+  }
+  if (NULL == this->payload_p) {
+    log.push_back( "Bagit payload directory 'data/' is expected, but could not found");
+    is_valid = false;
+  } else {
+    bool ret = this->payload_p->validate( log );
+    if (ret == false) {
+      is_valid = false;
+    }
+  }
+  if (NULL == this->payloadmanifest_p) {
+    is_valid = false;
+  } else {
+    log.push_back( "Bagit payload manifest");
+    bool ret = this->payloadmanifest_p->validate( log );
+    if (ret == false) {
+      is_valid = false;
+    }
+  }
+  // next elements are optional
+  if (NULL == this->tagmanifest_p) {
+  } else {
+    log.push_back( "Bagit tag manifest");
+    bool ret = this->tagmanifest_p->validate( log );
+    if (ret == false) {
+      is_valid = false;
+    }
+  }
+  if (NULL == this->bagmetadata_p) {
+  } else {
+    bool ret = this->bagmetadata_p->validate( log );
+    if (ret == false) {
+      is_valid = false;
+    }
+  }
+  /*
+     if (NULL == this->fetchfile_p) {
+     } else {
+     bool ret = this->fetchfile_p->validate( log );
+     if (ret == false) {
+     is_valid = false;
+     }
+     }
 
-	if (NULL == this->othertags_p) {
-	} else {
-		bool ret = this->othertags_p->validate();
-		if (ret == false) {
-			is_valid = false;
-		}
-	}
-        */
+     if (NULL == this->othertags_p) {
+     } else {
+     bool ret = this->othertags_p->validate( log);
+     if (ret == false) {
+     is_valid = false;
+     }
+     }
+     */
 
-	return is_valid;
+  return is_valid;
 }
 
 // vim: set tabstop=4 softtabstop=0 expandtab shiftwidth=4 smarttab

src/test/test_bag.cpp

@@ -11,6 +11,13 @@ int main(int argc, const char  *argv[]) {
   }
   const string dir = argv[1];
   Bag bag( dir );
+  list<string> log;
+  bool is_valid = bag.validate(log);
+  cout << "The bag '"<< dir <<"' is " << (is_valid ? "valid" : "invalid") << endl;
+  list<string>::const_iterator iterator;
+  for (iterator = log.begin(); iterator != log.end(); ++iterator) {
+    std::cout << *iterator << endl;
+  }
   return 0;
 }
 // vim: set tabstop=4 softtabstop=0 expandtab shiftwidth=4 smarttab