48 # include <winsock2.h>
60 const std::string S_ToolName =
"urldbsample_extended";
61 const std::string S_ToolVersion =
"1.3";
68 " <redist-folder> <ticket> <product> <url-list-file>\n"
69 " redist-folder - the folder where the DCA is installed to\n"
70 " ticket - a valid ticket\n"
71 " product - the product associated with your ticket\n"
72 " url-list-file - file that includes the URLs to classify\n\n"
80 # define DCA_BINDIR "bin/Win32"
82 # define DCA_BINDIR "bin/linux"
89 #define DCA_INITDIR "init"
94 #define DCA_LOGDIR "./logs"
122 static bool StartupLibraries()
127 WORD wVersionRequested = MAKEWORD( 2, 2 );
129 int err = WSAStartup( wVersionRequested, &wsaData );
131 std::cout <<
"Error on WSAStartup (" << err <<
") occured, aborting" <<
149 static void ShutdownLibraries()
168 static void SetupLicense(
const std::string& ticket,
const std::string& product,
171 licenseData.
ticket = ticket;
199 if( numOfCats == 0 ) {
200 std::cout <<
"No categories found." << std::endl;
209 const Category myCategory = myCategories.
byId( result );
210 const std::string catname = myCategory.
name(
"en_US" );
213 std::cout << (i+1) <<
".\t Category '" << catname <<
214 "' (id=" << myCategory.
id() <<
215 ", groupid=" << myCategory.
groupId() <<
227 std::cout <<
"IBM DCA Sample: " << S_ToolName <<
" (" << S_ToolVersion <<
")" << std::endl;
237 std::cout <<
"usage: " << name <<
S_UsageString << std::endl;
246 static void LoadUrlFile(
const std::string& fileName, std::vector<std::string>& urlList )
248 std::ifstream fstream( fileName.c_str(), std::ios::in );
249 if (!fstream.is_open())
return;
253 while ( std::getline(fstream, line) )
255 if( !line.empty() && line[line.length()-1] ==
'\r')
256 line.erase( line.length() - 1 );
259 urlList.push_back( line );
281 void TestUrlClassification(
const std::string& aUrlListFile,
const DcaInstance& myDca,
284 std::cout <<
"Starting URL db classification routine..." << std::endl;
288 std::vector<std::string> myUrlList;
289 LoadUrlFile( aUrlListFile, myUrlList );
291 for( std::vector<std::string>::const_iterator U = myUrlList.begin(),
292 UEnd = myUrlList.end(); U != UEnd; ++U ) {
294 const std::string& myUrlString = *U;
296 std::cout <<
" Starting URL db classification for URL '" <<
297 myUrlString <<
"'" << std::endl;
307 std::cerr <<
"Error from uURLrl db classification. Details: " << myFR.
getDescription() <<
308 " (" << myFR.
getReturnCode() <<
"). Continuing with next URL." << std::endl;
314 std::cout <<
"Results: URL '" << myUrlString <<
"' is not known in the database. " <<
315 "Continuing with next URL." << std::endl;
322 std::cout <<
"Results: URL '" << myUrlString <<
"' is not categorized (but known in database). "
323 <<
"Continuing with next URL" << std::endl;
328 std::cout <<
"Classification Results for URL '" << myUrlString <<
"'" << std::endl;
329 PrintResults( myCategoriesInfo, myUrlClassificationResults );
332 std::cout <<
"Leaving URL db classification routine." << std::endl;
342 int main(
int argc,
char *argv[] )
355 std::string myRedistFolder = argv[ 1 ];
356 const std::string myTicket = argv[ 2 ];
357 const std::string myProduct = argv[ 3 ];
358 const std::string myUrlList = argv[ 4 ];
360 if( myRedistFolder.empty() || myTicket.empty() ||
361 myProduct.empty() || myUrlList.empty() ) {
367 const char c = myRedistFolder[ myRedistFolder.length() - 1 ];
368 if( c !=
'/' && c !=
'\\' )
369 myRedistFolder +=
"/";
372 if( !StartupLibraries() )
390 std::cout <<
"DCA is not licensed!" << std::endl;
396 SetupConnectionData( myDbConnectionData );
422 TestUrlClassification( myUrlList, myDca, myUrlDbClassifier, myCategoriesInfo );
432 catch(
const ExDca& ex ) {
433 std::cerr <<
"DCA Exception occured. Details: " << ex.
getDescription() <<
437 catch(
const std::exception& s ) {
438 std::cerr <<
"std::exception occured. Details: " << s.what() <<
"." << std::endl;
442 std::cerr <<
"Unknown exception caught." << std::endl;
static DCA_MODULE_ID_TYPE ID
The unique ID of the URL classification module.
Example implementation of a schedule event subscriber.
Is used to create a License object. A license first must be created with DcaInstance::createLicense t...
Exception class used in the DCA.
bool isCategorized() const
Returns whether or not the URL matched one or more categories.
static Url create(const DcaInstance &aDcaInstance, const std::string &urlString)
Standard Url creation function.
bool cancelUpdate() const
Cancels a currently running update process. If there is currently no update running,...
const DCA_CATEGORIES_INFO_TYPE DCA_CAT_INFO_TYPE_URL
Refers to the internal categories info for URL classification.
void InitCUrl()
Initializes libcurl. Do not use any DCA function before initializing libcurl.
std::string initDir
the directory in which the DCA init files are stored
bool signal(unsigned int signal) const
Cancels functions which could potentially take a long time to return, such as DcaInstance::schedule()...
static UpdateModule create(const DcaInstance &aDcaInstance, const License &aLicense, const ProxySettings &proxySettings=ProxySettings())
std::string getDescription() const
Returns a description of the error.
Definition of a container class for Category objects.
CategoriesInfo getCategoriesInfo(DCA_CATEGORIES_INFO_TYPE categoryType) const
Returns the DCA internal categories, groups and locales.
DCA_RESULT_TYPE getReturnCode() const
Returns the last error code (if any).
FunctionResult classify(const Url &aUrl, UrlClassificationResults &urlResults) const
Performs the URL classification and returns the results.
static void SetupInitData(const std::string &redist_folder, InitData &initData)
Sets up the given initData by substituting the given redist_folder with DCA subdirectories.
The update module is used to download and install DCA content and engine updates.
std::string binDir
the directory in which the DCA binary (*.dca) files are stored
This header includes initialization/deinitialization support functions for the 3rd party libraries us...
A container class that allows access to the contained Categories, Groups and Locales.
Sets up options for embedded URL detection and provided Feedback mechanism.
Header file for the schedule event subscriber.
void SetOpenSslCallbacks()
Initializes the required callbacks for OpenSSL when using HTTPS or SSL connections in a multi-threade...
const unsigned int DCA_SIG_ABORT
Currently the only supported signal parameter for the DcaInstance::signal() function.
Stores the connection data for a database.
void UnsetOpenSslCallbacks()
Unsets the openssl callbacks. Do not call any DCA function after you have called this function.
void StartupThreads(const dca::DcaInstance &aDcaInstance, const dca::UpdateModule &aUpdateModule, volatile bool *schedulerErrorSignal, volatile bool *performUpdateErrorSignal)
Starts up the update and schedule threads and supplies the given DcaInstance and UpdateModule.
DCA_CATEGORY_ID_TYPE id() const
The category id.
void ShutdownThreads()
Shuts down the previously started update and schedule threads.
void DeinitCUrl()
Deinitializes libcurl. Do not call any DCA function after you have called this function.
bool isUnknownUrl() const
Returns whether a URL is known or unknown. A URL is unknown if it is not contained in the database.
Database connection class for a local or remote database.
This header includes all header files of the URL Classification Package.
#define DCA_LOGDIR
Relative directory for logfile(s).
Main class for the URL classification.
#define DCA_INITDIR
DCA subdirectory of the DCA initialization data.
bool enable_Feedback
This switches the Feedback feature on or off. This is switched off by default.
DCA_RESULT_TYPE getReturnCode() const
Gets the code of the error.
DCA_CATEGORY_ID_TYPE UrlClassificationResult
The item of an URL classification result is typedef'd as DCA_CATEGORY_ID_TYPE.
void DumpEventMessages() const
Dumps all collected messages to screen.
Results of an URL classification.
bool detect_EmbeddedUrlsInUrlPath
If set to true, you can specify that embedded URL detection is additionally performed in the path par...
URL database classifier class.
static void PrintToolHeader()
Prints out the name and the version of this sample.
Use a License to initialize a classification package or a toolbox package.
Category byId(DCA_CATEGORY_ID_TYPE id) const
Returns the category with the given category id.
bool isLicensed(DCA_MODULE_ID_TYPE id=0, bool force=false) const
Checks whether the given License is valid for the given module id.
std::string ticket
The ticket as provided in the license.
This header includes all header files of the DCA Base Package.
bool useLocalDatabase
Set to true to connect to a local or custom database, set to false to use a remote database.
Encapsulates the init and deinit of the DCA API.
Category NullCategory
Defines a constant unassigned Category you can use for checks. if( myCat == NullCategory ) --> myCat ...
std::string product
The product code used with the license.
const DbType DBT_Url
Used for DbConnection classes for URL classification.
static UrlClassification create(const DcaInstance &aDcaInstance, const License &aLicense)
Creates the URL classification module by using the given DcaInstance and License.
std::string logDir
the directory in which the DCA log file should be created
static void PrintUsage(const char *name)
Prints out the syntax of the sample.
size_t DCA_INDEX_TYPE
Type for index access (used for arrays and collections).
DbType dbType
The type of the database.
Categories getCategories() const
Returns the contained Categories.
Header file for functions related to start and stop the update and schedule threads.
std::string getDescription() const
Returns the description for the error or warning.
size_t DCA_SIZE_TYPE
Type for size (used for size of array and collections).
std::string name(const std::string &localeString=std::string()) const
Returns the localized (display) name of the category.
DCA_GROUP_ID_TYPE groupId() const
If the category is associated with a group, this is the group id.
bool enable_EmbeddedUrlDetection
If set to true embedded URL detection will be used in URL classification. This is switched on by defa...
#define DCA_BINDIR
DCA subdirectory of the DCA binaries.
Standard function result.
Encapsulates a URL object.
UrlDbClassifier createDbClassifier(const DbConnection &aDbConnection, const UrlDbClassifierOptions &options=UrlDbClassifierOptions()) const
Create a URL database classifier. The classifier is created by using the provided database connection...
const std::string S_UsageString
Usage string, displayed if a parameter is missing.
DbConnection createDbConnection(const License &aLicense, const DbConnectionData &dbcData, const ProxySettings &proxySettings=ProxySettings(), LogLevel aLogLevel=LOG_Initial) const
Creates a DbConnection object using the given DbConnectionData.
This structure is used to initialize the DcaInstance.
Encapsulates a category as defined in the categories XML schema (see Categories XML: Categories).
static DcaInstance create(const InitData &initData)
Creates a DcaInstance, starts up the DCA API and initializes the required main module.
static void SetupLicense(const std::string &ticket, const std::string &product, LicenseData &licenseData)
Sets up the given licenseData by copying the given ticket and product strings.
License createLicense(const LicenseData &licData, const ProxySettings &proxySettings=ProxySettings(), LogLevel aLogLevel=LOG_Initial) const
Creates a License object using the given LicenseData.
DCA_SIZE_TYPE size() const
Returns the number of results in the container.
int main(int argc, char *argv[])
The main routine.