nomos.h

Go to the documentation of this file.

/*
 SPDX-FileCopyrightText: © 2006-2014 Hewlett-Packard Development Company, L.P.
 SPDX-FileCopyrightText: © 2014 Siemens AG
 
 SPDX-License-Identifier: GPL-2.0-only
*/
 
#ifndef _NOMOS_H
#define _NOMOS_H 1
#ifndef _GNU_SOURCE
#define _GNU_SOURCE
#endif  /* not defined _GNU_SOURCE */
#include <glib.h>
#include <stdio.h>
#include <assert.h>
#include <stdlib.h>
#include <search.h>
#include <unistd.h>
#include <fcntl.h>
#include <string.h>
#include <strings.h>
#include <ctype.h>
#include <dirent.h>
#include <ftw.h>
#include <regex.h>
#include <getopt.h>
#include <time.h>
#include <libgen.h>
#include <errno.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/wait.h>
#include <sys/mman.h>
#include <sys/time.h>
#include "nomos_gap.h"
#include <stdbool.h>
#include <semaphore.h>
#include <stdbool.h>
#include "json_writer.h"
 
#ifdef STANDALONE
#include "standalone.h"
#else
#include <libfossology.h>
#include <libfossdbmanager.h>
#endif
 
#ifdef  MEMORY_TRACING
#include "DMalloc.h"
#endif  /* MEMORY_TRACING */
 
#define PRECHECK
#define GPLV2_BEATS_GPLV3
#define SAVE_UNCLASSIFIED_LICENSES
/*#define FLAG_NO_COPYRIGHT*/
 
#ifdef  PROC_TRACE_SWITCH
#define PROC_TRACE
#endif  /* PROC_TRACE_SWITCH */
 
#define myBUFSIZ  4096      
#define MAX_RENAME  1000    
#define TEMP_FILE_LEN 100   
 
#define MAX_SCANBYTES 1024*1024
 
#define OPTS_DEBUG    0x1
#define OPTS_TRACE_SWITCH 0x2
#define OPTS_LONG_CMD_OUTPUT 0x4
#define OPTS_HIGHLIGHT_STDOUT 0x8
#define OPTS_NO_HIGHLIGHTINFO 0x10
#define OPTS_JSON_OUTPUT 0x20
#define OPTS_SCANNING_DIRECTORY 0x40
 
extern char debugStr[myBUFSIZ];        
extern char dbErrString[myBUFSIZ];     
extern char saveLics[myBUFSIZ];        
 
extern size_t hashEntries;             
 
#define FL_SAVEBASE 0x20
#define FL_FRAGMENT 0x40
#define FL_SHOWMATCH  0x80
#define FL_NOCOPYRIGHT  0x100
 
#define FILE_FOUND  "Found.txt"
#define FILE_SCORES "_scores"
#define DEBUGLOG    "/tmp/NomosDebugLog"
 
 
#define NO  0
#define YES 1
 
/* List-sorting flags */
#define UNSORTED    0
#define SORT_BY_NAME    1
#define SORT_BY_NAME_ICASE  2
#define SORT_BY_COUNT_DSC 3
#define SORT_BY_COUNT_ASC 4
#define SORT_BY_ALIAS   5
#define SORT_BY_BASENAME  6
 
/* Interest level (licenses) */
#define IL_HIGH   3
#define IL_MED    2
#define IL_LOW    1
#define IL_NONE   0
#define IL_INIT   -1
 
#define LTSR_RMASK  ((char) 1)  
#define LTSR_SMASK  ((char) 2)  
#define LTSR_YES  ((char) 3)    
#define LTSR_NO   LTSR_SMASK    
#define STR_NOTPKG      "None (not an rpm-format package)"
 
/*
 * License-scanning limits
 */
#define _scCOMFORT  9 
#define _scINVALID  4 
#define LS_NONE   "None"
#define LS_UNLIKELY "LikelyNot"
#define LS_NOSUM  "No_license_found"
#define LS_UNCL   "UnclassifiedLicense"
#define LS_NOT_PD "NOT-public-domain"
#define LS_PD_CLM "Public-domain"
#define LS_PD_CPRT  "Public-domain(C)"
#define LS_PD_ONLY  "Public-domain-ref"
#define LS_CPRTONLY "Misc-Copyright"
#define LS_TDMKONLY "Trademark-ref"
#define LS_LICRONLY "License-ref"
#define LS_PATRONLY "Patent-ref"
 
/*
 * NULL values
 */
#define NULL_ITEM (item_t *) NULL 
#define NULL_LIST (list_t *) NULL 
#define NULL_FH   (fh_t *) NULL   
#define NULL_CHAR '\0'            
#define NULL_STR  (char *) NULL   
 
#define isEOL(x)  (((x == '\n') || (x == '\r') || (x == '\v'))) 
#define IS_HUGE(x)  (x >= gl.blkUpperLimit)
 
 
 
#define NOMOS_TEMP  "/tmp/nomos.tempdir"
#define NOMOS_TLOCK "/tmp/nomos.tempdir/.lock.tmp," 
struct mm_cache {
    int inUse;            
    int fd;               
    unsigned long size;   
    void *mmPtr;          
    char label[myBUFSIZ]; 
};
 
 
/*
  CDB - This is kind of tricky, the way it uses the same fields for
  different meanings. If we had objects, we could subclass. It works
  okay, but is just a PITA for debugging.
 */
 
struct listitem {
  int val;
  int val2;
  int val3;
  char *str;    
  void *buf;    
};
typedef struct listitem item_t;
 
 
#define seqNo   val
#define foundTool val
#define refCount  val
#define num   val
#define bStart    val
#define iFlag   val
#define ssComp    val2
#define isProcessed val2
#define iLevel    val2
#define nMatch    val2
#define bLen    val2
#define bDocLen   val3
#define bIndex    val3
#define bList   buf
 
struct list {
    char name[64]; 
    int used; 
    int size; 
    int ix; 
    int sorted; 
    int desc; 
    item_t *items; 
};
typedef struct list list_t;
 
struct searchString {
    int csLen;      
    char *csData;   
};
typedef struct searchString searchString_t;
 
struct licenseSpec {
    searchString_t seed;  
    searchString_t text;  
};
typedef struct licenseSpec licSpec_t;
 
 
struct globals {
    char initwd[myBUFSIZ];  
    char progName[64];      
    int progOpts;           
    int flags;              
    int uPsize;             
#ifdef  GLOBAL_DEBUG
  int DEEBUG;
  int MEM_DEEBUG;
#endif  /* GLOBAL_DEBUG */
#ifdef  PROC_TRACE_SWITCH
  int ptswitch;
#endif  /* PROC_TRACE_SWITCH */
    list_t sHash;           
    /* Agent-specific Things */
    int agentPk;            
    long uploadFk;          
    int arsPk;              
    PGconn *pgConn;         
    fo_dbManager *dbManager;  
};
 
typedef struct  {
    int start;    
    int end;      
    int index;    
} MatchPositionAndType;
 
typedef struct  {
    GArray* matchPositions;   
    GArray* indexList;        
    char* licenceName;        
    int licenseFileId;        
} LicenceAndMatchPositions;
 
 
 
struct curScan {
  char cwd[myBUFSIZ];      
  char targetDir[myBUFSIZ];    /* check */
  char targetFile[myBUFSIZ];   /* check */
  char filePath[myBUFSIZ];    
  long pFileFk;            
  char pFile[myBUFSIZ];       
  char *licPara;
  char *matchBase;
  size_t targetLen;
  size_t cwdLen;
  struct stat stbuf;
  regmatch_t regm;
  list_t regfList;
  list_t fLicFoundMap;
  list_t parseList;
  list_t offList;
  list_t lList;
  char compLic[myBUFSIZ];   
  int nLines;
  int cliMode;                
  char *tmpLics;              
  char *licenseList[512];     
  GArray* indexList; 
  GArray* theMatches; 
  GArray* keywordPositions; 
  GArray* docBufferPositionsAndOffsets;
  int currentLicenceIndex;
};
 
struct license {
    int len;      
    char *patt;   
};
 
struct licensetext {
    char *regex;  
    char *tseed;  
  int nAbove;
  int nBelow;
  int compiled;
  int plain;
};
typedef struct licensetext licText_t;
 
#define _REGEX(x) licText[x].regex
#define _SEED(x)  licText[x].tseed
 
struct scanResults {
  int score;        
  int kwbm;
  int size;
  int flag;         
  int dataOffset;
  char fullpath[myBUFSIZ];
  char linkname[16];
  char *licenses;
  char *relpath;
  size_t nameOffset;
};
typedef struct scanResults scanres_t;
 
#define MTAG_UNSORTKEY  "list/str (initially-UNsorted key)"
#define MTAG_SORTKEY  "list/str (initially-sorted key)"
#define MTAG_LISTKEY  "list/str (sorted/unsorted key)"
#define MTAG_REPLKEY  "list/str (replaced primary key)"
#define MTAG_LISTBUF  "list/buf (any data)"
#define MTAG_PATHBASE "list/buf (path basename)"
#define MTAG_PKGINFO  "list/buf (pkg rname/type/name/vers/lic)"
#define MTAG_PKG_NV "list/buf (pkg name/vers)"
#define MTAG_MD5SUM "list/buf (distro-arch MD5SUM)"
#define MTAG_COUNTER  "list/buf integer (counter)"
#define MTAG_PKGNAME  "list/buf (package-name)"
#define MTAG_PKGVERS  "list/buf (package-vers)"
#define MTAG_CLAIMLIC "list/buf (claimlic copy)"
#define MTAG_COMPLIC  "list/buf (pkg compLic copy)"
#define MTAG_URLCOPY  "list/buf (pkg URL copy)"
#define MTAG_FILELIC  "list/buf (file-license copy)"
#define MTAG_FIXNAME  "list/buf (fixed-package name)"
#define MTAG_SEEDTEXT "search-seed text"
#define MTAG_SRCHTEXT "license-search text"
#define MTAG_MMAPFILE "mmap-file data"
#define MTAG_MAGICDATA  "file magic description"
#define MTAG_PATTRS "pkg-attr buffer"
#define MTAG_DOUBLED  "doubled (reallocated) data"
#define MTAG_SEARCHBUF  "initial search-data buffer"
#define MTAG_TOOSMALL "too-small half-size buffer"
#define MTAG_TEXTPARA "paragraph text"
#define MTAG_LIST "dynamically-allocated list"
#define MTAG_ENV  "environment variable"
#define MTAG_SCANRES  "scan-results list"
 
 
/*
   Functions defined in nomos.c, used in other files
 */
void Bail(int exitval);
int optionIsSet(int val);
 
/*
  Global Declarations
 */
extern struct globals gl;
extern struct curScan cur;
extern licText_t licText[];
extern licSpec_t licSpec[];
extern int schedulerMode; /* Non-zero if being run by scheduler */
 
#ifdef  MEMORY_TRACING
char *memAllocTagged();
void memFreeTagged();
#define memFree(x,y)    memFreeTagged(x, y)
#define memAlloc(x,y)   memAllocTagged(x, y)
#else /* NOT MEMORY_TRACING */
#define memFree(x,/*notused*/y) free(x)
#define memAlloc(x,y)   calloc(x, 1)
#endif  /* NOT MEMORY_TRACING */
 
/*
 * Macros for timing - refer to findPhrase() for usage examples
 */
/* need TIMING_DECL in the declarations section of function */
#define DECL_TIMER  struct timeval bTV, eTV; float proctime
#define ZERO_TIMER  memcpy((void *) &bTV, (void *) &eTV, sizeof(eTV))
#define RESET_TIMER END_TIMER; ZERO_TIMER
#define START_TIMER RECORD_TIMER(bTV)
#define END_TIMER   RECORD_TIMER(eTV) ; \
    proctime = (float) (eTV.tv_sec - bTV.tv_sec) + \
    ((float) (eTV.tv_usec - bTV.tv_usec) * 0.000001)
#define RECORD_TIMER(x) (void) gettimeofday(&x, (struct timezone *) NULL)
#define PRINT_TIMER(x,y)  printf("%11.6f seconds: %s\n", proctime, x); \
    if (y) { DUMP_TIMERS; }
#define DUMP_TIMERS printf("[1]: %d.%06d\n", bTV.tv_sec, bTV.tv_usec); \
    printf("[2]: %d.%06d\n", eTV.tv_sec, eTV.tv_usec)
 
 
/*
 * Cut-and-paste this stuff to turn on timing
 */
#if 0
#ifdef  TIMING
DECL_TIMER; /* timer declaration */
#endif
/* */
#ifdef  TIMING
START_TIMER;  /* turn on the timer */
#endif  /* TIMING */
/* */
#ifdef  TIMING
END_TIMER;  /* stop the timer */
PRINT_TIMER("unpack", 0); /* ... and report */
START_TIMER;  /* optionally re-start timer */
#endif  /* TIMING */
#endif
 
#endif /* _NOMOS_H */