Added SQLite header parsing functionality and associated tests (SRombauts#249)

* Added SQLite header parsing functionality and associated tests

* Removed unused header file.

* Removed an accidental copy pasted remove() statement

* Replaced stdint with plain old C types for now. Will apply fixed with datatypes to cpp11 branch

* Added test scenarios to simulate blank file name, non existant file and a corrupt header

* Refactored exception flow to match latest tidying, brought casts out of function calls and cleared up invalid header exception message

Author: ptrks

include/SQLiteCpp/Database.h

@@ -12,7 +12,6 @@
 
 #include <SQLiteCpp/Column.h>
 #include <SQLiteCpp/Utils.h>    // definition of nullptr for C++98/C++03 compilers
-
 #include <string.h>
 
 // Forward declarations to avoid inclusion of <sqlite3.h> in a header
@@ -53,6 +52,32 @@ const char* getLibVersion() noexcept; // nothrow
 /// Return SQLite version number using runtime call to the compiled library
 int   getLibVersionNumber() noexcept; // nothrow
 
+// Public structure for representing all fields contained within the SQLite header.
+// Official documentation for fields: https://www.sqlite.org/fileformat.html#the_database_header
+struct Header {
+    unsigned char headerStr[16];
+    unsigned int pageSizeBytes;
+    unsigned char fileFormatWriteVersion;
+    unsigned char fileFormatReadVersion;
+    unsigned char reservedSpaceBytes;
+    unsigned char maxEmbeddedPayloadFrac;
+    unsigned char minEmbeddedPayloadFrac;
+    unsigned char leafPayloadFrac;
+    unsigned long fileChangeCounter;
+    unsigned long  databaseSizePages;
+    unsigned long firstFreelistTrunkPage;
+    unsigned long totalFreelistPages;
+    unsigned long schemaCookie;
+    unsigned long schemaFormatNumber;
+    unsigned long defaultPageCacheSizeBytes;
+    unsigned long largestBTreePageNumber;
+    unsigned long databaseTextEncoding;
+    unsigned long userVersion;
+    unsigned long incrementalVaccumMode;
+    unsigned long applicationId;
+    unsigned long versionValidFor;
+    unsigned long sqliteVersion;
+};
 
 /**
  * @brief RAII management of a SQLite Database Connection.
@@ -434,6 +459,21 @@ class Database
     */
     static bool isUnencrypted(const std::string& aFilename);
 
+    /**
+    * @brief Parse SQLite header data from a database file.
+    *
+    *  This function reads the first 100 bytes of a SQLite database file
+    *  and reconstructs groups of individual bytes into the associated fields
+    *  in a Header object.
+    *  
+    * @param[in] aFilename path/uri to a file
+    *
+    * @return Header object containing file data
+    *
+    * @throw SQLite::Exception in case of error
+    */
+    static Header getHeaderInfo(const std::string& aFilename);
+
     /**
      * @brief BackupType for the backup() method
      */

src/Database.cpp

@@ -298,6 +298,136 @@ bool Database::isUnencrypted(const std::string& aFilename)
     return strncmp(header, "SQLite format 3\000", 16) == 0;
 }
 
+// Parse header data from a database.
+Header Database::getHeaderInfo(const std::string& aFilename)
+{
+    Header h;
+    unsigned char buf[100];
+    char* pBuf = reinterpret_cast<char*>(&buf[0]);
+    char* pHeaderStr = reinterpret_cast<char*>(&h.headerStr[0]);
+
+    if (aFilename.empty())
+    {
+        throw SQLite::Exception("Could not open database, the aFilename parameter was empty.");
+    }
+
+    std::ifstream fileBuffer(aFilename.c_str(), std::ios::in | std::ios::binary);
+
+    if (fileBuffer.is_open())
+    {
+        fileBuffer.seekg(0, std::ios::beg);
+        fileBuffer.read(pBuf, 100);
+        fileBuffer.close();
+        strncpy(pHeaderStr, pBuf, 16);
+    }
+
+    else
+    {
+        throw SQLite::Exception("Error opening file: " + aFilename);
+    }
+
+    // If the "magic string" can't be found then header is invalid, corrupt or unreadable
+    if (!strncmp(pHeaderStr, "SQLite format 3", 15) == 0)
+    {
+        throw SQLite::Exception("Invalid or encrypted SQLite header");
+    }
+
+    h.pageSizeBytes = (buf[16] << 8) | buf[17];
+    h.fileFormatWriteVersion = buf[18];
+    h.fileFormatReadVersion = buf[19];
+    h.reservedSpaceBytes = buf[20];
+    h.maxEmbeddedPayloadFrac = buf[21];
+    h.minEmbeddedPayloadFrac = buf[22];
+    h.leafPayloadFrac = buf[23];
+
+    h.fileChangeCounter =
+        (buf[24] << 24) |
+        (buf[25] << 16) |
+        (buf[26] << 8)  |
+        (buf[27] << 0);
+
+    h.databaseSizePages =
+        (buf[28] << 24) |
+        (buf[29] << 16) |
+        (buf[30] << 8)  |
+        (buf[31] << 0);
+
+    h.firstFreelistTrunkPage =
+        (buf[32] << 24) |
+        (buf[33] << 16) |
+        (buf[34] << 8)  |
+        (buf[35] << 0);
+
+    h.totalFreelistPages =
+        (buf[36] << 24) |
+        (buf[37] << 16) |
+        (buf[38] << 8)  |
+        (buf[39] << 0);
+
+    h.schemaCookie =
+        (buf[40] << 24) |
+        (buf[41] << 16) |
+        (buf[42] << 8)  |
+        (buf[43] << 0);
+
+    h.schemaFormatNumber =
+        (buf[44] << 24) |
+        (buf[45] << 16) |
+        (buf[46] << 8)  |
+        (buf[47] << 0);
+
+    h.defaultPageCacheSizeBytes =
+        (buf[48] << 24) |
+        (buf[49] << 16) |
+        (buf[50] << 8)  |
+        (buf[51] << 0);
+
+    h.largestBTreePageNumber =
+        (buf[52] << 24) |
+        (buf[53] << 16) |
+        (buf[54] << 8)  |
+        (buf[55] << 0);
+
+    h.databaseTextEncoding =
+        (buf[56] << 24) |
+        (buf[57] << 16) |
+        (buf[58] << 8)  |
+        (buf[59] << 0);
+
+    h.userVersion =
+        (buf[60] << 24) |
+        (buf[61] << 16) |
+        (buf[62] << 8)  |
+        (buf[63] << 0);
+
+    h.incrementalVaccumMode =
+        (buf[64] << 24) |
+        (buf[65] << 16) |
+        (buf[66] << 8)  |
+        (buf[67] << 0);
+
+    h.applicationId =
+        (buf[68] << 24) |
+        (buf[69] << 16) |
+        (buf[70] << 8)  |
+        (buf[71] << 0);
+
+    h.versionValidFor =
+        (buf[92] << 24) |
+        (buf[93] << 16) |
+        (buf[94] << 8)  |
+        (buf[95] << 0);
+
+    h.sqliteVersion =
+        (buf[96] << 24) |
+        (buf[97] << 16) |
+        (buf[98] << 8)  |
+        (buf[99] << 0);
+
+    return h;
+}
+
+
 // This is a reference implementation of live backup taken from the official sit:
 // https://www.sqlite.org/backup.html
 

tests/Database_test.cpp

@@ -16,6 +16,7 @@
 #include <gtest/gtest.h>
 
 #include <cstdio>
+#include <fstream>
 
 #ifdef SQLITECPP_ENABLE_ASSERT_HANDLER
 namespace SQLite
@@ -354,6 +355,67 @@ TEST(Database, loadExtension)
     // TODO: test a proper extension
 }
 
+TEST(Database, getHeaderInfo)
+{
+    remove("test.db3");
+    {
+        //Call without passing a database file name
+        EXPECT_THROW(SQLite::Database::getHeaderInfo(""),SQLite::Exception);
+
+        //Call with a non existant database
+        EXPECT_THROW(SQLite::Database::getHeaderInfo("test.db3"), SQLite::Exception);
+
+        //Simulate a corrupt header by writing garbage to a file
+        unsigned char badData[100];
+        char* pBadData = reinterpret_cast<char*>(&badData[0]);
+
+        std::ofstream corruptDb;
+        corruptDb.open("corrupt.db3", std::ios::app | std::ios::binary);
+        corruptDb.write(pBadData, 100);
+
+        EXPECT_THROW(SQLite::Database::getHeaderInfo("corrupt.db3"), SQLite::Exception);
+        
+        remove("corrupt.db3");
+
+        // Create a new database
+        SQLite::Database db("test.db3", SQLite::OPEN_READWRITE | SQLite::OPEN_CREATE);
+        db.exec("CREATE TABLE test (id INTEGER PRIMARY KEY, value TEXT)");
+        
+        // Set assorted SQLite header values using associated PRAGMA
+        db.exec("PRAGMA main.user_version = 12345");
+        db.exec("PRAGMA main.application_id = 2468");
+
+        // Parse header fields from test database
+        SQLite::Header h = SQLite::Database::getHeaderInfo("test.db3");
+
+        //Test header values expliticly set via PRAGMA statements
+        EXPECT_EQ(h.userVersion, 12345);
+        EXPECT_EQ(h.applicationId, 2468);
+
+        //Test header values with expected default values
+        EXPECT_EQ(h.pageSizeBytes, 4096);
+        EXPECT_EQ(h.fileFormatWriteVersion,1);
+        EXPECT_EQ(h.fileFormatReadVersion,1);
+        EXPECT_EQ(h.reservedSpaceBytes,0);
+        EXPECT_EQ(h.maxEmbeddedPayloadFrac, 64);
+        EXPECT_EQ(h.minEmbeddedPayloadFrac, 32);
+        EXPECT_EQ(h.leafPayloadFrac, 32);
+        EXPECT_EQ(h.fileChangeCounter, 3);
+        EXPECT_EQ(h.databaseSizePages, 2);
+        EXPECT_EQ(h.firstFreelistTrunkPage, 0);
+        EXPECT_EQ(h.totalFreelistPages, 0);
+        EXPECT_EQ(h.schemaCookie, 1);
+        EXPECT_EQ(h.schemaFormatNumber, 4);
+        EXPECT_EQ(h.defaultPageCacheSizeBytes, 0);
+        EXPECT_EQ(h.largestBTreePageNumber, 0);
+        EXPECT_EQ(h.databaseTextEncoding, 1);
+        EXPECT_EQ(h.incrementalVaccumMode, 0);
+        EXPECT_EQ(h.versionValidFor, 3);
+        EXPECT_EQ(h.sqliteVersion, SQLITE_VERSION_NUMBER);
+    }
+    remove("test.db3");
+}
+
 #ifdef SQLITE_HAS_CODEC
 TEST(Database, encryptAndDecrypt)
 {