Contains all utility functions used by FOSSology. More...

#include "ununpack.h"
#include "externs.h"
#include "regex.h"

Include dependency graph for utils.c:

Enumerations
enum	BITS { BITS_PROJECT = 27 , BITS_ARTIFACT = 28 , BITS_CONTAINER = 29 }
	File mode BITS.

Functions
int	IsInflatedFile (char *FileName, int InflateSize)
	Test if the file is a compression bomb. More...

void	SafeExit (int rc)
	Close scheduler and database connections, then exit. More...

void	RemovePostfix (char *Name)
	get rid of the postfix More...

void	InitCmd ()
	Initialize the metahandler CMD table. More...

int	TaintString (char Dest, int DestLen, char Src, int ProtectQuotes, char *Replace)
	Protect strings intelligently. More...

int	Prune (char *Fname, struct stat Stat)
	Given a filename and its stat, prune it. More...

int	MkDirs (char *Fname)
	Same as command-line "mkdir -p". More...

int	MkDir (char *Fname)
	Smart mkdir. More...

int	IsDir (char *Fname)
	Given a filename, is it a directory? More...

int	IsFile (char *Fname, int Link)
	Given a filename, is it a file? More...

int	ReadLine (FILE Fin, char Line, int MaxLine)
	Read a command from a stream. More...

int	IsExe (char *Exe, int Quiet)
	Check if the executable exists. More...

int	CopyFile (char Src, char Dst)
	Copy a file. For speed: mmap and save. More...

int	ParentWait ()
	Wait for a child. Sets child status. More...

void	CheckCommands (int Show)
	Make sure all commands are usable. More...

int	RunCommand (char Cmd, char CmdPre, char File, char CmdPost, char Out, char Where)
	Try a command and return command code. More...

int	InitMagic ()
	Open and load Magic file Initializes global MagicCookie. More...

int	IsDebianSourceFile (char *Filename)
	Read file to see if it is a Debian source file. More...

void	OctetType (char Filename, char TypeBuf)
	Figure out the real type of "octet" files in case we can unarchive them. More...

int	FindCmd (char *Filename)
	Given a file name, determine the type of extraction command. This uses Magic. More...

void	FreeDirList (dirlist *DL)
	Free a list of files in a directory list. More...

dirlist *	MakeDirList (char *Fullname)
	Create a list of files in a directory. More...

void	SetDir (char Dest, int DestLen, char Smain, char *Sfile)
	Set a destination directory name. More...

void	DebugContainerInfo (ContainerInfo *CI)
	Print a ContainerInfo structure. More...

int	DBInsertPfile (ContainerInfo CI, char Fuid)
	Insert a Pfile record. Sets the pfile_pk in CI. More...

int	TestSCMData (char *sourcefilename)
	Search for SCM data in the filename. More...

int	DBInsertUploadTree (ContainerInfo *CI, int Mask)
	Insert an UploadTree record. More...

int	AddToRepository (ContainerInfo CI, char Fuid, int Mask)
	Add a ContainerInfo record to the repository AND to the database. More...

int	DisplayContainerInfo (ContainerInfo *CI, int Cmd)
	Print what can be printed in XML. More...

int	RemoveDir (char *dirpath)
	Remove all files under dirpath (rm -rf) More...

char *	PathCheck (char *DirPath)
	Check if path contains a "%U" or "%H". If so, substitute a unique ID for U. More...

void	deleteTmpFiles (char *dir)

void	Usage (char Name, char Version)
	Display program usage. More...

void	SQLNoticeProcessor (void arg, const char message)
	Dummy postgresql notice processor. This prevents Notices from being written to stderr. More...

int	ShouldExclude (char Filename, const char ExcludePatterns)
	Determines if a file or folder should be excluded. More...

Variables
const char *	SCM_REGEX = "/\\.git\|\\.hg\|\\.bzr\|CVS/ROOT\|\\.svn/"

Detailed Description

Contains all utility functions used by FOSSology.

Definition in file utils.c.

Function Documentation

◆ AddToRepository()

int AddToRepository	(	ContainerInfo *	CI,
		char *	Fuid,
		int	Mask
	)

Add a ContainerInfo record to the repository AND to the database.

This modifies the CI record's pfile and ufile indexes!

Parameters

CI
Fuid	sha1.md5.sha256.size
Mask	file mode mask

Returns: 1 if added, 0 if already exists!

Definition at line 1403 of file utils.c.

◆ CheckCommands()

void CheckCommands ( int Show )

Make sure all commands are usable.

Parameters

Show Unused

Returns: void but updates global CMD Status

Definition at line 568 of file utils.c.

◆ CopyFile()

int CopyFile	(	char *	Src,
		char *	Dst
	)

Copy a file. For speed: mmap and save.

Parameters

	Src	Source file path
[out]	Dst	Destination file path

Returns: 0 if copy worked, 1 if failed.

Definition at line 441 of file utils.c.

◆ DBInsertPfile()

int DBInsertPfile	(	ContainerInfo *	CI,
		char *	Fuid
	)

Insert a Pfile record. Sets the pfile_pk in CI.

Parameters

CI
Fuid	string of sha1.md5.size

Returns: 1 if record exists, 0 if record does not exist.

< Temporary storage for mimetype fk from DB

< Temporary storage for pfile_sha256 from DB

Definition at line 1115 of file utils.c.

◆ DBInsertUploadTree()

int DBInsertUploadTree	(	ContainerInfo *	CI,
		int	Mask
	)

Insert an UploadTree record.

If the tree is a duplicate, then we need to replicate all of the uploadtree records for the tree. This uses Upload_Pk.

Parameters

CI
Mask	mask file mode for ufile_mode

Returns: 1 if tree exists for some other project (duplicate) and 0 if tree does not exist.

Definition at line 1287 of file utils.c.

◆ DebugContainerInfo()

void DebugContainerInfo ( ContainerInfo * CI )

Print a ContainerInfo structure.

Parameters

CI	ContainerInfo struct to print

Definition at line 1086 of file utils.c.

◆ DisplayContainerInfo()

int DisplayContainerInfo	(	ContainerInfo *	CI,
		int	Cmd
	)

Print what can be printed in XML.

Parameters

CI
Cmd	Command used to create this file (parent) CI->Cmd = command to be used ON this file (child)

Returns: 1 if item is unique, 0 if duplicate.

Definition at line 1458 of file utils.c.

◆ FindCmd()

int FindCmd ( char * Filename )

Given a file name, determine the type of extraction command. This uses Magic.

Returns: index to command-type, or -1 on error.

Definition at line 851 of file utils.c.

◆ FreeDirList()

void FreeDirList ( dirlist * DL )

Free a list of files in a directory list.

Parameters

DL	directory list

Definition at line 952 of file utils.c.

◆ InitCmd()

void InitCmd ( )

Initialize the metahandler CMD table.

This ensures that:

Every mimetype is loaded
Every mimetype has an DBindex.

Definition at line 109 of file utils.c.

◆ InitMagic()

int InitMagic ( )

Open and load Magic file Initializes global MagicCookie.

Returns: 0 on success

Definition at line 716 of file utils.c.

◆ IsDebianSourceFile()

int IsDebianSourceFile ( char * Filename )

Read file to see if it is a Debian source file.

Assumes that all Debian source files have a .dsc filename extension.

Parameters

Filename File to open

Returns: 1 if Filename is a Debian source file, else 0

Definition at line 734 of file utils.c.

◆ IsDir()

int IsDir ( char * Fname )

Given a filename, is it a directory?

Parameters

Fname file name

Returns: 1=yes, 0=no.

Definition at line 320 of file utils.c.

◆ IsExe()

int IsExe	(	char *	Exe,
		int	Quiet
	)

Check if the executable exists.

(Like the command-line "which" but without returning the path.)

Note: This should only be used on relative path executables.

Parameters

Exe	Executable file name
Quiet	If true, do not write warning on file not found

Returns: 1 if exists, 0 if does not exist.

Definition at line 394 of file utils.c.

◆ IsFile()

int IsFile	(	char *	Fname,
		int	Link
	)

Given a filename, is it a file?

Parameters

Fname	Path of file to check
Link	True if should it follow symbolic links

Returns: 1=yes, 0=no.

Definition at line 336 of file utils.c.

◆ IsInflatedFile()

int IsInflatedFile	(	char *	FileName,
		int	InflateSize
	)

Test if the file is a compression bomb.

If the size of FileName is a factor of InflateSize more than the size of the directory containing it, then it is a bomb.

Parameters

FileName	pathname to file
InflateSize	Inflation factor.

Returns: 1 on is one inflated file, 0 on is not

Definition at line 41 of file utils.c.

◆ MakeDirList()

dirlist* MakeDirList ( char * Fullname )

Create a list of files in a directory.

Parameters

Fullname Path to top level directory.

Returns: the directory list

Definition at line 971 of file utils.c.

◆ MkDir()

int MkDir ( char * Fname )

Smart mkdir.

If mkdir fails, then try running MkDirs.

Parameters

Fname file name

Returns: 0 on success, 1 on failure.

Definition at line 304 of file utils.c.

◆ MkDirs()

int MkDirs ( char * Fname )

Same as command-line "mkdir -p".

Parameters

Fname file name

Returns: 0 on success, 1 on failure.

Definition at line 249 of file utils.c.

◆ OctetType()

void OctetType	(	char *	Filename,
		char *	TypeBuf
	)

Figure out the real type of "octet" files in case we can unarchive them.

Parameters

Filename
Static	buffer to return with new Type

Definition at line 773 of file utils.c.

◆ ParentWait()

int ParentWait ( )

Wait for a child. Sets child status.

Returns: the queue record, or -1 if no more children.

Definition at line 510 of file utils.c.

◆ PathCheck()

char* PathCheck ( char * DirPath )

Check if path contains a "%U" or "%H". If so, substitute a unique ID for U.

This substitution parameter must be at the end of the DirPath. Substitute hostname for H. @parm DirPath Directory path.

Returns: new directory path

Definition at line 1663 of file utils.c.

◆ Prune()

int Prune	(	char *	Fname,
		struct stat	Stat
	)

Given a filename and its stat, prune it.

Remove anything that is not a regular file or directory
Remove files when hard-link count > 1 (duplicate search)
Remove zero-length files
Returns
1=pruned, 0=no change.

Definition at line 218 of file utils.c.

◆ ReadLine()

int ReadLine	(	FILE *	Fin,
		char *	Line,
		int	MaxLine
	)

Read a command from a stream.

Read a line each time from one file.

If the line is empty, then try again.

Parameters

	Fin	Input file pointer
[out]	Line	Output line buffer
	MaxLine	Max line length

Returns: line length, or -1 of EOF.

Definition at line 357 of file utils.c.

◆ RemoveDir()

int RemoveDir ( char * dirpath )

Remove all files under dirpath (rm -rf)

Parameters

dirpath

Returns: shell exit code of rm -rf

Definition at line 1642 of file utils.c.

◆ RemovePostfix()

void RemovePostfix ( char * Name )

get rid of the postfix

For example: test.gz --> test

Parameters

[in,out] Name input file name

Definition at line 91 of file utils.c.

◆ RunCommand()

int RunCommand	(	char *	Cmd,
		char *	CmdPre,
		char *	File,
		char *	CmdPost,
		char *	Out,
		char *	Where
	)

Try a command and return command code.

Command becomes:

‘Cmd CmdPre 'File’ CmdPost Out`
If there is a s, then that becomes Where.
Parameters

Cmd

CmdPre

File

CmdPost

Out

Where

Returns
-1 if command could not run.

Definition at line 622 of file utils.c.

◆ SafeExit()

void SafeExit ( int rc )

Close scheduler and database connections, then exit.

Parameters

rc exit code

Returns: no return, calls exit()

Definition at line 78 of file utils.c.

◆ SetDir()

void SetDir	(	char *	Dest,
		int	DestLen,
		char *	Smain,
		char *	Sfile
	)

Set a destination directory name.

This will concatenate Smain and Sfile, but remove and terminating filename.

Parameters

[in,out]	Dest	returned directory name
	DestLen	size of Dest
	Smain	main extraction directory (may be null)
	Sfile	filename

NOTE: Someone that embeds "../" within the path can still climb out!

Definition at line 1048 of file utils.c.

◆ ShouldExclude()

int ShouldExclude	(	char *	Filename,
		const char *	ExcludePatterns
	)

Determines if a file or folder should be excluded.

This function checks whether the supplied file name, Filename, contains any of the substrings listed in the comma-separated string ExcludePatterns. Each pattern is matched directly as a substring; no wildcard or directory-specific matching is performed.

Parameters

Filename	The name of the file or folder to be examined.
ExcludePatterns	A comma-separated list of substrings used for determining exclusion.

Returns: 1 if a substring match is found (folder is to be excluded), or 0 otherwise.

Definition at line 1788 of file utils.c.

◆ SQLNoticeProcessor()

void SQLNoticeProcessor	(	void *	arg,
		const char *	message
	)

Dummy postgresql notice processor. This prevents Notices from being written to stderr.

Parameters

arg	unused
message	unused

Definition at line 1773 of file utils.c.

◆ TaintString()

int TaintString	(	char *	Dest,
		int	DestLen,
		char *	Src,
		int	ProtectQuotes,
		char *	Replace
	)

Protect strings intelligently.

Prevents filenames containing ' or % or \ from screwing up system() and snprintf(). Even supports a "%s".

Note: s is assumed to be in single quotes!

Parameters

[in,out]	Dest	Destination to store tainted string
	DestLen	Length of Dest
	Src	Source string
	ProtectQuotes	Set to protect quotes for shell
	Replace	String to replace with

Returns: 0 on success, 1 on overflow.

Definition at line 166 of file utils.c.

◆ TestSCMData()

int TestSCMData ( char * sourcefilename )

Search for SCM data in the filename.

SCM data is one of these: Git (.git)Data(char *FileName) Mercurial (.hg) Bazaar (.bzr) CVS (CVS/Root) Subversion (.svn)

Parameters

sourcefilename

Returns: 1 if SCM data is found

Definition at line 1225 of file utils.c.

◆ Usage()

void Usage	(	char *	Name,
		char *	Version
	)

Display program usage.

Parameters

Name	program name
Version	program version

Definition at line 1724 of file utils.c.

Variable Documentation

◆ SCM_REGEX

const char* SCM_REGEX = "/\\.git|\\.hg|\\.bzr|CVS/ROOT|\\.svn/"

regular expression to detect SCM data

Definition at line 28 of file utils.c.

Enumerations

Functions

Variables

Detailed Description

Function Documentation

◆ AddToRepository()

◆ CheckCommands()

◆ CopyFile()

◆ DBInsertPfile()

◆ DBInsertUploadTree()

◆ DebugContainerInfo()

◆ DisplayContainerInfo()

◆ FindCmd()

◆ FreeDirList()

◆ InitCmd()

◆ InitMagic()

◆ IsDebianSourceFile()

◆ IsDir()

◆ IsExe()

◆ IsFile()

◆ IsInflatedFile()

◆ MakeDirList()

◆ MkDir()

◆ MkDirs()

◆ OctetType()

◆ ParentWait()

◆ PathCheck()

◆ Prune()

◆ ReadLine()

◆ RemoveDir()

◆ RemovePostfix()

◆ RunCommand()

◆ SafeExit()

◆ SetDir()

◆ ShouldExclude()

◆ SQLNoticeProcessor()

◆ TaintString()

◆ TestSCMData()

◆ Usage()

Variable Documentation

◆ SCM_REGEX