tool/dbtotxt.c

   1 /*
   2 ** Copyright 2008 D. Richard Hipp and Hipp, Wyrick & Company, Inc.
   3 ** All Rights Reserved
   4 **
   5 ******************************************************************************
   6 **
   7 ** This file implements a stand-alone utility program that converts
   8 ** a binary file (usually an SQLite database) into a text format that
   9 ** is compact and friendly to human-readers.
  10 **
  11 ** Usage:
  12 **
  13 **         dbtotxt [OPTIONS] FILENAME
  14 **
  15 ** where OPTIONS are zero or more of:
  16 **
  17 **    --for-cli          prepending '.open --hexdb' to the output
  18 **
  19 **    --script           The input file is expected to start with a
  20 **                       zero-terminated SQL string.  Output the
  21 **                       ".open --hexdb" header, then the database
  22 **                       then the SQL.
  23 **
  24 **    --pagesize N       set the database page size for later reading
  25 **
  26 ** The translation of the database appears on standard output.  If the
  27 ** --pagesize command-line option is omitted, then the page size is taken
  28 ** from the database header.
  29 **
  30 ** Compactness is achieved by suppressing lines of all zero bytes.  This
  31 ** works well at compressing test databases that are mostly empty.  But
  32 ** the output will probably be lengthy for a real database containing lots
  33 ** of real content.  For maximum compactness, it is suggested that test
  34 ** databases be constructed with "zeroblob()" rather than "randomblob()"
  35 ** used for filler content and with "PRAGMA secure_delete=ON" selected to
  36 ** zero-out deleted content.
  37 */
  38 #include <stdio.h>
  39 #include <string.h>
  40 #include <stdlib.h>
  41 #include <ctype.h>
  42
  43 /* Return true if the line is all zeros */
  44 static int allZero(unsigned char *aLine){
  45   int i;
  46   for(i=0; i<16 && aLine[i]==0; i++){}
  47   return i==16;
  48 }
  49
  50 int main(int argc, char **argv){
  51   int pgsz = 0;               /* page size */
  52   int forCli = 0;             /* whether to prepend with .open */
  53   int bSQL = 0;               /* Expect and SQL prefix */
  54   long szFile;                /* Size of the input file in bytes */
  55   FILE *in;                   /* Input file */
  56   int nSQL;                   /* Number of bytes of script */
  57   int i, j;                   /* Loop counters */
  58   int nErr = 0;               /* Number of errors */
  59   const char *zInputFile = 0; /* Name of the input file */
  60   const char *zBaseName = 0;  /* Base name of the file */
  61   int lastPage = 0;           /* Last page number shown */
  62   int iPage;                  /* Current page number */
  63   unsigned char *aData = 0;   /* All data */
  64   unsigned char *aLine;       /* A single line of the file */
  65   unsigned char *aHdr;        /* File header */
  66   unsigned char bShow[256];   /* Characters ok to display */
  67   memset(bShow, '.', sizeof(bShow));
  68   for(i=' '; i<='~'; i++){
  69     if( i!='{' && i!='}' && i!='"' && i!='\\' ) bShow[i] = (unsigned char)i;
  70   }
  71   for(i=1; i<argc; i++){
  72     if( argv[i][0]=='-' ){
  73       const char *z = argv[i];
  74       z++;
  75       if( z[0]=='-' ) z++;
  76       if( strcmp(z,"pagesize")==0 ){
  77         i++;
  78         pgsz = atoi(argv[i]);
  79         if( pgsz<512 || pgsz>65536 || (pgsz&(pgsz-1))!=0 ){
  80           fprintf(stderr, "Page size must be a power of two between"
  81                           " 512 and 65536.\n");
  82           nErr++;
  83         }
  84         continue;
  85       }else if( strcmp(z,"for-cli")==0 ){
  86         forCli = 1;
  87         continue;
  88       }else if( strcmp(z,"script")==0 ){
  89         forCli = 1;
  90         bSQL = 1;
  91         continue;
  92       }
  93       fprintf(stderr, "Unknown option: %s\n", argv[i]);
  94       nErr++;
  95     }else if( zInputFile ){
  96       fprintf(stderr, "Already using a different input file: [%s]\n", argv[i]);
  97       nErr++;
  98     }else{
  99       zInputFile = argv[i];
 100     }
 101   }
 102   if( zInputFile==0 ){
 103     fprintf(stderr, "No input file specified.\n");
 104     nErr++;
 105   }
 106   if( nErr ){
 107     fprintf(stderr,
 108        "Usage: %s [--pagesize N] [--script] [--for-cli] FILENAME\n", argv[0]);
 109     exit(1);
 110   }
 111   in = fopen(zInputFile, "rb");
 112   if( in==0 ){
 113     fprintf(stderr, "Cannot open input file [%s]\n", zInputFile);
 114     exit(1);
 115   }
 116   fseek(in, 0, SEEK_END);
 117   szFile = ftell(in);
 118   rewind(in);
 119   if( szFile<100 ){
 120     fprintf(stderr, "File too short. Minimum size is 100 bytes.\n");
 121     exit(1);
 122   }
 123   aData = malloc( szFile+16 );
 124   if( aData==0 ){
 125     fprintf(stderr, "Failed to allocate %ld bytes\n", szFile);
 126     exit(1);
 127   }
 128   if( fread(aData, szFile, 1, in)!=1 ){
 129     fprintf(stderr, "Cannot read file info memory\n");
 130     exit(1);
 131   }
 132   memset(aData+szFile, 0, 16);
 133   fclose(in);
 134   if( bSQL ){
 135     for(i=0; i<szFile && aData[i]!=0; i++){}
 136     if( i==szFile ){
 137       fprintf(stderr, "No zero terminator on SQL script\n");
 138       exit(1);
 139     }
 140     nSQL = i+1;
 141     if( szFile - nSQL<100 ){
 142       fprintf(stderr, "Less than 100 bytes in the database\n");
 143       exit(1);
 144     }
 145   }else{
 146     nSQL = 0;
 147   }
 148   aHdr = aData + nSQL;
 149   if( pgsz==0 ){
 150     pgsz = (aHdr[16]<<8) | aHdr[17];
 151     if( pgsz==1 ) pgsz = 65536;
 152     if( pgsz<512 || (pgsz&(pgsz-1))!=0 ){
 153       fprintf(stderr, "Invalid page size in header: %d\n", pgsz);
 154       exit(1);
 155     }
 156   }
 157   zBaseName = zInputFile;
 158   for(i=0; zInputFile[i]; i++){
 159     if( zInputFile[i]=='/' && zInputFile[i+1]!=0 ) zBaseName = zInputFile+i+1;
 160   }
 161   if( forCli ){
 162     printf(".open --hexdb\n");
 163   }
 164   printf("| size %d pagesize %d filename %s\n",(int)szFile,pgsz,zBaseName);
 165   for(i=nSQL; i<szFile; i+=16){
 166     aLine = aData+i;
 167     if( allZero(aLine) ) continue;
 168     iPage = i/pgsz + 1;
 169     if( lastPage!=iPage ){
 170       printf("| page %d offset %d\n", iPage, (iPage-1)*pgsz);
 171       lastPage = iPage;
 172     }
 173     printf("|  %5d:", i-(iPage-1)*pgsz);
 174     for(j=0; j<16; j++) printf(" %02x", aLine[j]);
 175     printf("   ");
 176     for(j=0; j<16; j++){
 177       unsigned char c = (unsigned char)aLine[j];
 178       fputc( bShow[c], stdout);
 179     }
 180     fputc('\n', stdout);
 181   }
 182   printf("| end %s\n", zBaseName);
 183   if( nSQL>0 ){
 184     printf("%s\n", aData);
 185   }
 186   free( aData );
 187   return 0;
 188 }