/* >From: EVERHART%ARISIA.decnet@GE-CRD.ARPA Newsgroups: comp.os.vms Subject: Small upgrade to TAR reader Message-ID: <8901280458.AA21037@ucbvax.Berkeley.EDU> Date: 27 Jan 89 21:23:00 GMT Sender: daemon@ucbvax.BERKELEY.EDU Organization: The Internet Lines: 489 Folks - I noticed that some TAR tape directory trees are too deep for VMS when trying to use Sid Penstone's TAR reader here. To fix things up, I added a small edit. Its effect is to add an s option to the reader to effectively partly flatten directory trees. This is done by replacing the / between levels of directories after the first with _. As a result, one can arrange that a directory that might have been created as, for example [x11r3.lib.clx.debug] gets created instead as [x11r3.lib_clx.debug] and so on. Using the s option more than once allows multiple such replacements to occur. Thus, basically no information is lost so long as the total directory length is short enough to fit in a VMS file specification. The trees remain shallow enough for VMS to handle them and you can read them in. I send it out in thanks to Sid Penstone for the original TAR reader, which works very nicely. Glenn Everhart Everhart%Arisia.decnet@ge-crd.arpa ----------- cut here ---------- */ /* Read a TAR format tape or file , move files into VMS directories */ /* Copyright 1986, Sid Penstone, * Department of Electrical Engineering, * Queen's University, * Kingston, Ontario, Canada K7L3N6 * (613)-545-5925 * BITNET: PENSTONE@QUCDNEE1 (Preferred) * or PENSTONE@QUCDN * * Version 2.3a, Jan. 17, 1989 * mods: - Added a few more chars to VMS forbidden list and allowed - * - in filenames. Added "s" option to skip a directory spec, * - done by changing the . for subdirs into a _. Multiple s's * - cause multiple dirs to be so "skipped". gce. * - Use ss to pass one dir, sss for two and so on. * Version 2.3, Jan.9,1987 * mods: - corrected header size (thanks to Eric Gisin, U .of Waterloo) * - No more of the dreaded QIO's ( " " " ) * - tried to sort out link flag format * - uses a tape or a file as input * - NOTE: default is NO conversion to vms standard text format (cr) * 2.1 - trapped commas in file names, converted to '_' * 2.2 - reported translations of names * - continued after error in opening output file * - exit correctly after error opening input file * 2.3 - fixed bug in make_new on top level file (thanks to Ursula Natrup, * natrup@vax.hmi.dfn ) * - reject "@" in filenames */ /* The input data is in record format, length 512, blocks of 10240 bytes; */ #include stdio #include time #include ssdef #include iodef #include descrip #include ctype #define ERROR1 -1 #define BUFFSIZE 512 #define ISDIRE 1 #define ISFILE 0 #define NAMSIZE 100 #define SIZE 20480 /* Block size */ #define DSIZE 512 /* Data block size */ struct /* A Tar header */ { char title[NAMSIZE]; char protection[8]; char field1[8]; /* this is the user id */ char field2[8]; /* this is the group id */ char count[12]; /* was 11 in error */ char time[12]; /* UNIX format date */ char chksum[8]; /* Header Checksum (ignored) */ char linkcount; /* hope this is right */ char linkname[NAMSIZE]; /* Space for the name of the link */ char dummy[255]; /* and the rest */ } header; static char buffer[DSIZE]; /* BUFFER for a record */ /* Function flags, options:*/ int extract, /* x option (default) */ list, /* t option : list tape contents */ verbose, /* v option, report actions */ wait; /* Miscellaneous globals, etc. */ char *tarfile = "tape", /* Input file name */ pathname[NAMSIZE], /* File name as found on tape (UNIX) */ directory[NAMSIZE], /* Current directory */ new_directory[NAMSIZE], /* Directory of current file */ top[NAMSIZE], /* Top level or root */ newfile[NAMSIZE], /* VMS format of file name */ outfile[NAMSIZE], /* Complete output file specification */ temp[256], /* Scratch */ creation[NAMSIZE], /* Date as extracted from the TAR file */ *ctime(), /* System function */ linkname[NAMSIZE]; /* Linked file name */ int bytecount, mode, uic1, uic2, linktype;/* Data from header */ int tarfd; /* The input file descriptor */ int dirskp; /* directories to skip in input */ main(argc,argv) int argc; char *argv[]; { int isterm,status,file_type,j,c, flag; char *make_directory(), *cp; /* Decode the options and parameters: */ dirskp=0; if(argc ==1) { extract = 1; /* Default for now */ verbose = 1; wait = 0; /* Don't wait for prompt */ } while(--argc > 0) { cp = argv[1]; while(c = *cp++) { switch(c) { case 't': list=1; break; case 'x': extract=1; break; case 'v': verbose=1; break; case 'w': wait=1; break; case 's': dirskp++; break; default: printf("Option '%c' not recognized.\n",c); } } } /* Find if this is a terminal */ isterm = isatty(0); /* Set up directory names */ strcpy(top,getenv("PATH")); /* Start with the default as the top */ strcpy(directory,top); /* open the file for reading */ if((tarfd = opentar()) <= 0) { printf("Error opening the Tar tape\n"); exit(2); } /* Now keep reading headers from this file, and decode the names, etc. */ while((status=hdr_read(&header))==DSIZE) /* 0 on end of file */ { if(strlen(header.title)!=0) /* Valid header */ { decode_header(); if(extract) { file_type=scan_title(pathname,new_directory,newfile); if( make_new(new_directory)!=0) printf("Error creating %s\n",new_directory); if(file_type == ISDIRE) {} if(file_type == ISFILE) /* Now move the data into the output file */ if(bytecount>0) { strcpy(outfile,new_directory); strcat(outfile,newfile); if((j=copyfile(outfile,bytecount))<0) printf("Error writing file %s\n",outfile); } } else /* listing only */ { printf("%o %6d %s %s\n", mode,bytecount,creation+4,pathname); if(linktype == 0) tarskip(bytecount); else printf(" *****( Linked to file: %s)\n",linkname); } } else /* Empty header means the end!!! */ { status = 1; printf("End of Tar file found.\n"); break; } } /* end while */ if(status == 1) /* Empty header */ { printf("Do you wish to move past the EOF mark ? y/n\n"); gets(temp); if(tolower(temp[0]) == 'y') while((status=hdr_read(&header)) >0); else exit(SS$_NORMAL); } if(status==0) /* End of tar file */ { printf("End of file encountered\n"); exit(SS$_NORMAL); } if(status<0) /* An error */ { printf("Error reading input.\n"); exit(2); } } /* This function simply copies the file to the output, no conversion */ int copyfile(outfile,nbytes) char outfile[]; /* name of output version */ int nbytes; { int inbytes, fil; /* Open the output file */ if((fil=creat(outfile,0)) == ERROR1) { printf(" Creation error in opening %s \n",outfile); tarskip(bytecount); return(-2); } if(linktype !=0) { sprintf(buffer,"This file is linked to %s\n",linkname); write(fil,buffer,strlen(temp)); } else { while(nbytes>0) { if((inbytes=read(tarfd,buffer,DSIZE)) > 0) { write(fil,buffer,(nbytes > DSIZE)? DSIZE:nbytes); nbytes -= inbytes; } else { printf("End of input file detected\n"); close(fil); return(-1); } } } /* Close the file */ close(fil); if(verbose) { printf("CREATED: %s\n",outfile); if(linktype!=0) printf(" *** REAL DATA IS IN: %s\n",linkname); } return(0); } /* Decode a file name into the directory, and the name, return * a value to indicate if this is a directory name, or another file * We return the extracted directory string in "dire", and the * filename (if it exists) in "fname". The full title is in "line" * at input. */ int scan_title(line,dire,fname) char line[],dire[],fname[]; { char temp[NAMSIZE],*end1; int len,len2,i,ind; int dlvl; /* The format will be UNIX at input, so we have to scan for the * UNIX directory separator '/' * If the name ends with '/' then it is actually a directory name. * If the directory consists only of '.', then don't add a subdirectory * The output directory will be a complete file spec, based on the default * directory. */ dlvl=0; strcpy(dire,top); /* Start with the top level */ if(strncmp(line,"./",2)==0) strcpy(line,line+2); /* ignore "./" */ strcpy(temp,line); /* Start in local buffer */ ind=vms_cleanup(temp); /* Remove illegal vms characters */ if((end1=strrchr(temp,'/'))==0) /* No directory at all ? */ strcpy(fname,temp); /* Only a file name */ else { /* End of directory name is '/' */ *end1 = 0; /* Terminate directory name */ dlvl=0; strcpy(fname,end1+1); /* File name without directory */ for (i=0;temp[i];i++) /* Change '/' to '.' in directory */ if(temp[i]=='/'){ temp[i]='.'; /* Change first "n" . chars to _ if s option given */ if(dlvl1||ind)&& verbose ) /* Any translations ? */ printf("****RENAMED: %s \n TO: %s\n",line,fname); return(ISFILE); } /* Create a new directory, finding out any higher levels that are missing */ /* We will parse the directory name into the next higher directory, and the * desired directory as "desired.dir". * Thus: "DEV:[top.sub1.sub2]" is made into "DEV:[top.sub1]sub2.dir" . If * the directory does not exist , then create the original directory. There * may be higher levels missing, so we can recurse until we reach the top * level directory, then work our way back, creating directories at each * successive level. * Bug fix: if the input file was at top level, we will not find a '.' * and 'name' will be garbage. */ int make_new(want) char want[]; { int i,len; char a[NAMSIZE],parent[NAMSIZE],*end,name[NAMSIZE]; strcpy(parent,want); len = strlen(parent); parent[len-1] =0 ; /* Get rid of the "]" */ end = strrchr(parent,'.'); /* Find the last '.' */ if(end != NULL) { strcpy(a,end+1); /* Get the last parent */ strcat(a,".dir"); /* Add the extension */ *end++ = ']' ; /* Reduce the directory parent */ *end = 0; /* Terminate the directory */ strcpy(name,parent); strcat(name,a); if(access(name,0) <0) /* Does the directory exist ? */ { if(strcmp(parent,top)!=0) /* No, are we at the top? */ if(make_new(parent)) /* No, look again */ return(-1); /* recurse */ if(mkdir(want,0755,0,0,0)) /* make it */ return(-1); /* Leave on error */ else if(verbose) printf("CREATED: %s\n",want); return(0); } } return(0); } /* Function to open and get data from the blocked input file */ int opentar() { int fd; fd = open(tarfile, 0, "rfm = fix","mrs = 512"); if(fd < 0) { printf("Can't open input file \n"); return(0); } return(fd); } /* Get the next file header from the input file buffer. We will always * move to the next 512 byte boundary. */ int hdr_read(buffer) char *buffer; { int stat; stat = read(tarfd,buffer,DSIZE); /* read the header */ return(stat); /* Catch them next read ? */ } /* This is supposed to skip over data to get to the desired position */ /* Position is the number of bytes to skip. We should never have to use * this during data transfers; just during listings. */ int tarskip(bytes) int bytes; { int i=0; while(bytes > 0) { if((i=read(tarfd,buffer,DSIZE)) == 0) { printf("End of file encountered while skipping.\n"); return(-1); } bytes -= i; } return(0); } /* Decode the fields of the header */ int decode_header() { int idate, *bintim; char ll; bintim = &idate; linktype=0; strcpy(linkname,""); strcpy(pathname,header.title); sscanf(header.time,"%o",bintim); strcpy(creation,ctime(bintim)); /* Work on this! */ creation[24]=0; sscanf(header.count,"%o",&bytecount); sscanf(header.protection,"%o",&mode); sscanf(header.field1,"%o",&uic1); sscanf(header.field2,"%o",&uic2); /* We may have the link written as binary or as character: */ linktype = isdigit(header.linkcount)? (header.linkcount - '0'):header.linkcount; if(linktype != 0) sscanf(header.linkname,"%s",linkname); return(0); } /* remove illegal characters from directory and file names; replace * hyphens and commas with underscores.Returns number of translations * that were made. */ vms_cleanup(string) char string[]; { int i,flag=0; char c; for(i=0;c=string[i];i++) { switch (c) { /* allow hyphens; ok sinve vms 4.4 */ /* case '-': */ /* No hyphens in file names */ case ',': /* No commas in file names */ case '+': /* No plusses in file names */ case '@': /* No '@' allowed in a name */ case '^': /* No '^' allowed in a name */ case '&': /* No '&' allowed in a name */ case '?': /* No '?' allowed in a name */ case '*': /* No '*' allowed in a name */ string[i]= '_'; flag++; /* Record if any changes were made */ break; default: break; } } return(flag); }