Utilities used by copyright and ecc agent. More...

#include "copyrightUtils.hpp"
#include <boost/program_options.hpp>
#include <iostream>
#include <sstream>

Include dependency graph for copyrightUtils.cc:

Macros
#define	RGX_FMT_SEPARATOR "@@"

Functions
int	queryAgentId (PGconn *dbConn)
	Get agent id, exit if agent id is incorrect. More...

int	writeARS (int agentId, int arsId, int uploadId, int success, const fo::DbManager &dbManager)
	Call C function fo_WriteARS() and translate the arguments. More...

void	bail (int exitval)
	Disconnect with scheduler returning an error code and exit. More...

bool	parseCliOptions (int argc, char **argv, CliOptions &dest, std::vector< std::string > &fileNames, std::string &directoryToScan)
	Parse the options sent by CLI to CliOptions object. More...

static void	addDefaultScanners (CopyrightState &state)
	Add default scanners to the agent state. More...

scanner *	makeRegexScanner (const std::string &regexDesc, const std::string &defaultType)
	Make a boost regex scanner object based on regex desc and type. More...

CopyrightState	getState (CliOptions &&cliOptions)
	Create a new state for the current agent based on CliOptions. More...

bool	saveToDatabase (const string &s, const list< match > &matches, unsigned long pFileId, int agentId, const CopyrightDatabaseHandler &copyrightDatabaseHandler)
	Save findings to the database if agent was called by scheduler. More...

void	matchFileWithLicenses (const string &sContent, unsigned long pFileId, CopyrightState const &state, int agentId, CopyrightDatabaseHandler &databaseHandler)
	Scan a given file with all available scanners and save findings to database. More...

void	matchPFileWithLicenses (CopyrightState const &state, int agentId, unsigned long pFileId, CopyrightDatabaseHandler &databaseHandler)
	Get the file contents, scan for statements and save findings to database. More...

bool	processUploadId (const CopyrightState &state, int agentId, int uploadId, CopyrightDatabaseHandler &databaseHandler, bool ignoreFilesWithMimeType)
	Process a given upload id, scan from statements and add to database. More...

pair< string, list< match > >	processSingleFile (const CopyrightState &state, const string fileName)

void	appendToJson (const std::string fileName, const std::pair< string, list< match >> resultPair, bool &printComma)

void	printResultToStdout (const std::string fileName, const std::pair< string, list< match >> resultPair)

Detailed Description

Utilities used by copyright and ecc agent.

Definition in file copyrightUtils.cc.

Function Documentation

◆ addDefaultScanners()

static void addDefaultScanners ( CopyrightState & state )

static

Add default scanners to the agent state.

Parameters

state State to which scanners are being added

Definition at line 208 of file copyrightUtils.cc.

◆ appendToJson()

void appendToJson	(	const std::string	fileName,
		const std::pair< string, list< match >>	resultPair,
		bool &	printComma
	)

Append a new result from scanner to main output json object

Parameters

fileName	File which was scanned
resultPair	The result pair from scanSingleFile()
printComma	Set true to print comma. Will be set true after first data is printed

Definition at line 469 of file copyrightUtils.cc.

◆ bail()

void bail ( int exitval )

Disconnect with scheduler returning an error code and exit.

Parameters

exitval Error code

Definition at line 62 of file copyrightUtils.cc.

◆ getState()

CopyrightState getState ( CliOptions && cliOptions )

Create a new state for the current agent based on CliOptions.

Called during instantiation of agent.

Parameters

cliOptions CliOptions passed to the agent

Returns: New CopyrightState object for the agent

Definition at line 279 of file copyrightUtils.cc.

◆ makeRegexScanner()

scanner* makeRegexScanner	(	const std::string &	regexDesc,
		const std::string &	defaultType
	)

Make a boost regex scanner object based on regex desc and type.

Parameters

regexDesc	Regex format for scanner
defaultType	Type of scanner

Returns: scanner object or NULL in case of error

Definition at line 248 of file copyrightUtils.cc.

◆ matchFileWithLicenses()

void matchFileWithLicenses	(	const string &	sContent,
		unsigned long	pFileId,
		CopyrightState const &	state,
		int	agentId,
		CopyrightDatabaseHandler &	databaseHandler
	)

Scan a given file with all available scanners and save findings to database.

Parameters

sContent	Content of file
pFileId	id of the pfile sent for scan
state	State of the agent
agentId	Agent id
databaseHandler	Database handler used by agent

Definition at line 337 of file copyrightUtils.cc.

◆ matchPFileWithLicenses()

void matchPFileWithLicenses	(	CopyrightState const &	state,
		int	agentId,
		unsigned long	pFileId,
		CopyrightDatabaseHandler &	databaseHandler
	)

Get the file contents, scan for statements and save findings to database.

Reads the file contents of the pFileId and send it for scanning to matchFileWithLicenses().

If the pfile is not found for pFileId, bails with error code 8.

If the pfile is not found in repository, bails with error code 7.

Parameters

state	State of the agent
agentId	Agent id
pFileId	pFile to be scanned
databaseHandler	Database handler used by agent

Definition at line 361 of file copyrightUtils.cc.

◆ parseCliOptions()

bool parseCliOptions	(	int	argc,
		char **	argv,
		CliOptions &	dest,
		std::vector< std::string > &	fileNames,
		std::string &	directoryToScan
	)

Parse the options sent by CLI to CliOptions object.

Parameters

[in]	argc
[in]	argv
[out]	dest	The parsed CliOptions object
[out]	fileNames	List of files to be scanned
[out]	directoryToScan	Directory to be scanned

Returns: True if success, false otherwise

Todo:: Change and add help based on IDENTITY

Definition at line 78 of file copyrightUtils.cc.

◆ printResultToStdout()

void printResultToStdout	(	const std::string	fileName,
		const std::pair< string, list< match >>	resultPair
	)

Print the result of current scan to stdout

Parameters

fileName	File which was scanned
resultPair	Result pair from scanSingleFile()

Definition at line 535 of file copyrightUtils.cc.

◆ processSingleFile()

pair<string, list<match> > processSingleFile	(	const CopyrightState &	state,
		const string	fileName
	)

Read a single file and run all scanners on it based of CopyrightState.

Parameters

state	Copyright state
fileName	Location of the file to be scanned

Returns: A pair of file scanned and list of matches found.

Definition at line 439 of file copyrightUtils.cc.

◆ processUploadId()

bool processUploadId	(	const CopyrightState &	state,
		int	agentId,
		int	uploadId,
		CopyrightDatabaseHandler &	databaseHandler,
		bool	ignoreFilesWithMimeType
	)

Process a given upload id, scan from statements and add to database.

The agent runs in parallel with the help of omp. A new thread is created for every pfile.

Parameters

state	State of the agent
agentId	Agent id
uploadId	Upload id to be processed
databaseHandler	Database handler object
ignoreFilesWithMimeType	To ignore files with particular mimetype

Returns: True when upload is processed

Definition at line 405 of file copyrightUtils.cc.

◆ queryAgentId()

int queryAgentId ( PGconn * dbConn )

Get agent id, exit if agent id is incorrect.

Parameters

[in] dbConn Database connection object

Returns: ID of the agent

Definition at line 25 of file copyrightUtils.cc.

◆ saveToDatabase()

bool saveToDatabase	(	const string &	s,
		const list< match > &	matches,
		unsigned long	pFileId,
		int	agentId,
		const CopyrightDatabaseHandler &	copyrightDatabaseHandler
	)

Save findings to the database if agent was called by scheduler.

Parameters

s	Statement found
matches	List of regex matches for highlight
pFileId	Id of pfile on which the statement was found
agentId	Id of agent who discovered the statements
copyrightDatabaseHandler	Database handler object

Returns: True of successful insertion, false otherwise

Definition at line 296 of file copyrightUtils.cc.

◆ writeARS()

int writeARS	(	int	agentId,
		int	arsId,
		int	uploadId,
		int	success,
		const fo::DbManager &	dbManager
	)

Call C function fo_WriteARS() and translate the arguments.

See also: fo_WriteARS()

Definition at line 53 of file copyrightUtils.cc.

Macros

Functions

Detailed Description

Function Documentation

◆ addDefaultScanners()

◆ appendToJson()

◆ bail()

◆ getState()

◆ makeRegexScanner()

◆ matchFileWithLicenses()

◆ matchPFileWithLicenses()

◆ parseCliOptions()

◆ printResultToStdout()

◆ processSingleFile()

◆ processUploadId()

◆ queryAgentId()

◆ saveToDatabase()

◆ writeARS()