#!/usr/local/bin/gawk -f # @(#) aj.gawk 3.2 97/07/29 # 92/03/09 John DuBois (john@armory.com) # 92/06/29 Ignore children of sdd and mscreen. Now uses awk instead of egrep. # 93/10/17 Ignore children of telnetd & scoterm # 94/02/25 Added IGNOREPROCS and .ajrc file # 94/03/19 Added command line args. Deal with defunct procs. # 94/03/20 Added ability to run on specific users, and i and I options. # 94/03/21 filbo@armory.com: handle "old" processes ("Mar 20" instead of time) # 94/03/28 john@armory.com: Converted to gawk script (use gawk for systime). # 95/08/02 Everest port: make ps input come from /dev/null, to work around # gawk+protlib bugs; added u option; work with Everest style times # (Mmm-dd); added login to default list of ignored parents. # 96/05/25 3.0 Major cleanup. Use getPS(). # 96/05/28 Special meaning of '.' for -a option. More detailed output for -u. # Sort by user name. # 96/11/13 Added q option. # 96/12/06 Added C option. # 97/05/21 Make . turn off -q. Added ntTU options, @ user name option, # make user names given with . or @ be not-users # 97/07/29 Get correct user name if USER not set BEGIN { Name = "aj" Usage = \ "Usage: " Name " [-AcdhHiIntTUuq] [-x] [-a]\n"\ " [-p] [-P] [-C] [user ...]" rcFile = ".ajrc" ARGC = Opts(Name,Usage,"a:cdiIp:P:qw",0, "~/" rcFile ":$UHOME/" rcFile, "ALTUSERS,COUNTDEFUNCT,DISPLAYDEFUNCT,IDLETIME,IDLESINCE,"\ "IGNOREPROCS,PARENTS,QUICK,WIDTH,HEADER",7,"n",0,"","t,T") # If anything is added to this, change the help message too. defParents = "sdd|mscreen|telnetd|scoterm|login" if ("h" in Options) { printf \ "%s: print all interesting jobs being run by a user.\n"\ "%s\n"\ "%s prints all processes that the user is running that are attached to any\n"\ "terminal, with the exception of sdd, mscreen, telnetd, scoterm, login, and\n"\ "their immediate children (these are generally shells, daemons, or other\n"\ "uninteresting processes). %s identifies as owned by the user any\n"\ "processes whose owner is the same as the user ID given by the environment\n"\ "variable USER, or if it not set, the owner of the current process. If\n"\ "any user names are given on the command line (other than with the -a\n"\ "option), %s reports on the named users only. If the user name '.' is\n"\ "given, all users are reported on. If the user name '@' is given, all users\n"\ "who have a login shell listed in /etc/shells are reported on. If either\n"\ "'.' or '@' is given, any other user names given on the command line are\n"\ "taken to be the names of users who should not be reported on.\n"\ "Options:\n"\ "Some of the following options can also be set by assigning values to\n"\ "variables either in the environment or in a configuration file named %s,\n"\ "which is searched for in the invoking user's home directory and in the\n"\ "directory specified by the environment variable UHOME, if it is set. If\n"\ "both files exist, values set in the former take precedence. Variables set\n"\ "in the environment take precedence over both, and options given on the\n"\ "command line have the highest precedence of all. Variables are assigned\n"\ "to with the syntax: varname=value or in the case of flags, by simply\n"\ "putting the indicated variable name in the file without a value. The\n"\ "names of variables that may be set in the environment are followed by\n"\ "\",e\". Variable names are given in parentheses in the option\n"\ "descriptions.\n"\ "-a: is a comma-separated list of users to report on\n"\ " in addition to the user running %s. (ALTUSERS,e)\n"\ "-c: Print a count of matching defunct processes. (COUNTDEFUNCT,e)\n"\ "-d: Show defunct processes. By default, they are ignored. Since TTY\n"\ " information is not available for defunct processes, proceses which\n"\ " were not attached to a TTY may be displayed. (DISPLAYDEFUNCT,e)\n"\ "-h: Print this help.\n"\ "-H: Print a header. (HEADER)\n"\ "-i, -I: In the output, replace the process' start time with the time the\n"\ " user controlling the process has been idle (IDLETIME,e) or the time\n"\ " the user went idle (IDLESINCE,e), respectively. The difference\n"\ " between -i and -I is that if e.g. nothing has been read from the user\n"\ " for 5 minutes, and it is now 10:33:16, -i would show 00:05:00 while -I\n"\ " would show 10:28:16. In both cases, the idle time is determined by\n"\ " the last time any input was read from the process' controlling TTY,\n"\ " which is not a perfect indication of idle time.\n"\ "-p, -P: Ignore any process whose name matches the\n"\ " given pattern. In the case of -P, the children of matching processes\n"\ " are also ignored. There is no default for . The default\n"\ " for is \"%s\".\n"\ " The values given with -p and -P should be egrep(C)-style patterns. \n"\ " The version of the process name that is compared is the last component\n"\ " of the the path that the process' argv[0] is set to (usually, the name\n"\ " the process was invoked by). The pattern is implicitely anchored at\n"\ " the start and end. (IGNOREPROCS,e and PARENTS,e).\n"\ "-u: Show unignored processes along with information about them\n"\ " (particularly the names of their parents) that can be used to ignore\n"\ " them.\n"\ "-T: Do not exclude processes that are not attached to a TTY.\n"\ "-t: Show all processes attached to TTYs. Equivalent to '-p "" -P "" .' \n"\ "-U: Show \"user processes\": equivalent to '-T -p "" -P "" @ root' \n"\ "-q: Quick operation. %s normally gathers information about all processes,\n"\ " not just those owned by the invoking user (and the altusers), because\n"\ " it needs to know the names of the parents of processes, and the parents\n"\ " may not be owned by any of those users. This may take an annoyingly\n"\ " long time on systems that are slow or have a very large number of\n"\ " processes running. If -q is given, information is only gathered for\n"\ " processed owned by the specified users and for root, since root is\n"\ " usually the owner of any other processes of interest. -q is turned off\n"\ " if \".\" or \"@\" is given in the user list. (QUICK)\n"\ "-w: Truncate output lines so that they are no more than \n"\ " characters long. If 0 is given, output lines are not truncated.\n"\ " Normally, output is truncated to a number of columns appropriate to\n"\ " the terminal it is run from so that lines will not wrap. (WIDTH)\n"\ "-n: Do not read values from any configuration files or the environment.\n"\ "-x: Print debugging info. is an integer from\n"\ " 1 to 9. Higher levels produce more detailed debugging information.\n", Name,Usage,Name,Name,Name,rcFile,Name,defParents,Name exit 0 } Debug = ("x" in Options) ? Options["x"] : 0 if ((Err = ExclusiveOptions("i,I,u",Options)) != "") { printf "Error: %s\n",Err > "/dev/stderr" Err = 1 exit(1) } DisplayDefunct = "d" in Options if ("p" in Options) IgnoreProcs = "^(" Options["p"] ")$" if ("P" in Options) ParentPat = Options["P"] else ParentPat = defParents ParentPat = "^(" ParentPat ")$" if ("t" in Options) { ParentPat = IgnoreProcs = "" ARGV[ARGC++] = "." } if ("U" in Options) { ParentPat = IgnoreProcs = "" ARGV[ARGC++] = "@" ARGV[ARGC++] = "root" } # Determine whose processes to pay attention to. if (ARGC > 1) { for (i = 1; i < ARGC; i++) Users[ARGV[i]] if ("." in Users || "@" in Users) { for (user in Users) if (user != "." && user != "@") { notUsers[user] delete Users[user] } } } else { if ("USER" in ENVIRON) Users[ENVIRON["USER"]] else { id(IDs) Users[IDs["user"]] } if ("a" in Options) MakeSet(Users,Options["a"],",") } ttyOnly = !("T" in Options || "U" in Options) if ("q" in Options && !("." in Users) && !("@" in Users)) { UserList = "-uroot," set2list(Users,",") if (Debug) print "User list: " UserList } if ("@" in Users) { split("",Users) # empty Users[] if (!makeShellUser(Users)) { print Name \ ": Error reading /etc/shells or /etc/passwd" > "/dev/stderr" exit 1 } } fWanted = "UID,TTY,CMD,PPID,TIME,ARGS" # Must deal with all processes, since we need to know the names of parents, # which may be owned by a UID that we are not otherwise interested in. if (Debug) { printf "Starting ps at %s\n",strftime("%T") if (Debug > 3) # Only used if debugging is at this level fWanted = fWanted ",LINE" } if (!ShowIdle) # STIME replaced with idle time if showing idle time fWanted = fWanted ",STIME" if ((nProc = getPS(PIDs,Procs,fWanted,Children,Debug > 4,UserList)) < 0) { printf "ps failed. Exiting.\n" > "/dev/stderr" exit 1 } if (Debug) printf "Done with ps at %s\n",strftime("%T") getline PID < "/dev/pid" IgnorePIDs[psPID = PIDs["ps"]] # Ignore the ps IgnorePIDs[PID] # Ignore this program IgnorePPIDs[PID] # Ignore the shell that ran this program IgnorePPIDs[1] # Ignore children of init NumDefunct = Mangle(PIDs,Procs,ParentPat,IgnoreProcs,DisplayDefunct,Users, notUsers,Debug,IgnorePPIDs,GoodPIDs,TTYs,ttyOnly) if (Debug) { printf "Debug=%s\nIgnoreProcs=%s\nParentPat=%s\n", Debug,IgnoreProcs,ParentPat > "/dev/stderr" printf "DisplayDefunct=%s\n",DisplayDefunct > "/dev/stderr" printf "NUser=%s\nNumber of procs: %d\n",NUser,nProc > "/dev/stderr" printf "ps pid: %d\nthis process pid: %d\n",psPID,PID > "/dev/stderr" printf "Number of ttys to stat: %d\n",NumElem(TTYs) > "/dev/stderr" printf "Number of defunct processes: %d\n",NumDefunct > "/dev/stderr" printf "Number of processes to be displayed: %d\n",NumElem(GoodPIDs)\ > "/dev/stderr" } if ("w" in Options) { cols = Options["w"] HeadTailInit(-1,cols ? cols : -1,0,0) } else HeadTailInit(-1) PrintResults("H" in Options,"I" in Options,"i" in Options,"u" in Options, "c" in Options,NumDefunct,TTYs,GoodPIDs,Procs,Debug) if (Debug) printf "Processing completed at %s\n",strftime("%T") } # Generate list of PPIDs to ignore, get list to TTYs to stat, # and record set of processes to be displayed. # Input vars: # PIDs[]: pids of processes in Procs[]. # Procs[]: process data. # ParentPat: process name pattern for processes to ignore along with children. # IgnoreProcs: process name pattern for processes to ignore. # DisplayDefunct: true if defunct processes should be displayed. # Users[]: Set of users whose processes should be scanned for. # Output vars: # IgnorePPIDs[]: process IDs of processes that matched ParentPat. # GoodPIDs[]: set of PIDs of processes to display. The value of each element # is the name of the user who owns the process, for use as a sort key. # TTYs[]: All ttys that procs in GoodPIDs had as controlling TTY. # Return value: number of defunct processes. function Mangle(PIDs,Procs,ParentPat,IgnoreProcs,DisplayDefunct,Users, notUsers,Debug,IgnorePPIDs,GoodPIDs,TTYs,ttyOnly, procname,tty,pid,allUsers,User) { allUsers = "." in Users for (pid in PIDs) { if (pid == "ps") continue if (Debug > 3) printf "Processing: %s\n",Procs[pid,"LINE"] > "/dev/stderr" procname = basename(Procs[pid,"CMD"]) if (ParentPat != "" && procname ~ ParentPat) { IgnorePPIDs[pid] continue } # skip processes not attached to tty, IgnoreProcs, IgnorePIDs tty = Procs[pid,"TTY"] User = Procs[pid,"UID"] if ((allUsers || User in Users) && !(ttyOnly && (tty == "?") || \ pid in IgnorePIDs || IgnoreProcs != "" && procname ~ IgnoreProcs || User in notUsers)) { if (procname == "") { NumDefunct++ if (DisplayDefunct) GoodPIDs[pid] = User } else { GoodPIDs[pid] = User TTYs[Procs[pid,"TTY"] = canonTTY(tty)] } } } return NumDefunct } function PrintResults(PrintHeader,ShowIdleSince,ShowIdleTime,ShowInfo, CountDefunct,NumDefunct,TTYs,GoodPIDs,Procs,Debug, Cmd,CurTime,ttylist,IdleTime,pid,format,Fields,header,ppid,num,k,i) { split("UID,PID,PPID,STIME,TTY,TIME,ARGS",Fields,",") ShowIdle = ShowIdleTime || ShowIdleSince if (ShowIdle && (ttylist = set2list(TTYs," ")) != "") { if (ShowIdleTime) { Cmd = "cd /dev; exec stat -c' ' -nfna " ttylist CurTime = systime() } else Cmd = "cd /dev; exec stat -c' ' -nfnA -t%T " ttylist if (Debug) print "stat command: " Cmd while ((Cmd | getline) == 1) { if (Debug > 3) print "stat line: " $0 if (ShowIdleTime) IdleTime[$1] = sec2hms(max(CurTime - $2,0)) else # IdleSince IdleTime[$1] = $2 } close(Cmd) } if (CountDefunct) printf "%d defunct process(es).\n",NumDefunct if (ShowInfo) { format = "%-8s %5s %-4s %-19s" format = format " " format "\n" printf format,"User","PID","TTY","Process-name","PUser","PPID","PTTY", "Parent-name" } else if (PrintHeader) { header = makePSline(-1,Procs,Fields) if (ShowIdleTime) sub("STIME","IDLE ",header) else if (ShowIdleSince) sub("STIME ","IDLESINC",header) ColPrint(header) } num = qsortArbIndByValue(GoodPIDs,k) for (i = 1; i <= num; i++) if (!(Procs[pid = k[i],"PPID"] in IgnorePPIDs)) { if (ShowInfo) { ppid = Procs[pid,"PPID"] printf format, Procs[pid,"UID"], pid, shortTTY(Procs[pid,"TTY"]), basename(Procs[pid,"CMD"]), Procs[ppid,"UID"], ppid, shortTTY(Procs[ppid,"TTY"]), basename(Procs[ppid,"CMD"]) } else { # tty name in Procs[] has already been canonicalized if (ShowIdle) Procs[pid,"STIME"] = Procs[pid,"TTY"] in IdleTime ? \ IdleTime[Procs[pid,"TTY"]] : "-" ColPrint(makePSline(pid,Procs,Fields)) } } } ### Start of library routines function max(a,b) { if (a > b) return a else return b } function sec2hms(Seconds, Hours,Minutes) { Hours = int(Seconds / 3600) Seconds %= 3600 Minutes = int(Seconds / 60) Seconds %= 60 return sprintf("%02d:%02d:%02d",Hours,Minutes,Seconds) } ### Begin ps lib # getPS 1.1 jhdiii 96/10/09 # 96/02/11 Added Debug flag. # 96/05/09 Added COMM field. # 96/05/23 Added selection args, and saving of "ps" PID. # 96/05/25 Added makePSline() # 96/10/09 Added RUSER field. # Note: makePSline() needs assign() from array lib. # to do: generalize based on -o args to 5.0 ps # Do a ps -f and save the output into an array, indexed by pid and field name. # Input vars: # Fields: Comma-separated list of fields to put in Procs. # If Debug is true, debugging info is output. # selectionArgs may be set to ps options that will report on selected processes # (e.g. -usomeone -ttty01) # The default for selectionArgs is -e, which causes information on all # processes to be recorded. # # Output vars: # PIDs[]: the set of all PIDs seen. # Also, the element with index "ps" is set to the PID for the ps process. # Procs[pid,fieldname]: output by field. # # Possible fields are: # UID: User ID; name if available, else number. # RUSER: Real user ID; name if available, else number. Only available under # 5.0, and cannot be requested along with UID. # PPID: Parent process ID. # C: CPU scheduling. # STIME: Start time. If the start time in the ps output contains a space, # it is replaced with a "-". "-" is returned for a defunct process. # TTY: tty name; may or may not have leading "tty" part. "-" for defunct proc; # "?" for proc with no controlling tty. # TIME: CPU time used. # CMD: First element of arg vector. # ARGS: Entire (truncated) arg vector (command + args). # LINE: Entire ps output line. # COMM: Process accounting name of process: the name of the executable file, # without path. This is only available under 5.0, and cannot be # requested along with CMD or ARGS. # # The header line read is also put in Procs with the index "Header". # The PIDs of the children of each process are put in a comma-separated list # in Children[pid]. # Return value: the number of processes found, or -2 if an invalid field name # is passed, or -1 if an error occurs reading from ps. # Globals: FS is set to " " # # ps -f produces output in these forms, under various conditions & releases: # UID PID PPID C STIME TTY TIME CMD # root 10118 10107 2 Jan-03 ttyp0 00:00:05 -ksh # root 10118 10107 2 Jan 03 ttyp0 00:00:05 -ksh # root 18197 1 0 08:02:56 ttyp0 00:00:03 /usr/bin/X11/scoterm -geo function getPS(PIDs,Procs,Fields,Children,Debug,selectionArgs, stimeI,pidI,ttyI,ppidI,WantLine,psArgs,psSet,newPS, FieldNames,Wanted,Cmd,getI,Field2Ind,i,Name,Lines,WantArgs,Header,CmdIndex) { FS = " " # magic pattern to reset FS to its default special behaviour split("UID,PID,PPID,C,STIME,TTY,TIME,CMD",FieldNames,",") split("user,pid,ppid,c,stime,tty,time,args",psSet,",") Alt["RUSER"] = 1 Alt["COMM"] = 8 FieldNames[0] = "LINE" for (i in FieldNames) Field2Ind[FieldNames[i]] = i split(Fields,Wanted,",") pidI = Field2Ind["PID"] ppidI = Field2Ind["PPID"] stimeI = Field2Ind["STIME"] ttyI = Field2Ind["TTY"] timeI = Field2Ind["TIME"] cmdI = Field2Ind["CMD"] psArgs = "-f" for (i in Wanted) { Name = Wanted[i] if (Debug) printf "Asked for %s\n",Name > "/dev/stderr" if (Name == "ARGS") WantArgs = 1 else if (Name == "LINE") WantLine = 1 else if (Name in Alt) { # New ps fields newPS = 1 psSet[Alt[Name]] = tolower(Name) FieldNames[getI[Field2Ind[Name] = Alt[Name]]] = Name } else if (Name in Field2Ind) getI[Field2Ind[Name]] else return -2 } if (newPS) { psArgs = "" for (i = 1; i in psSet; i++) psArgs = psArgs " -o" psSet[i] } Lines = 0 if (selectionArgs == "") selectionArgs = "-e" Cmd = "echo $$; exec " ( Debug ? "time " : "" ) "/bin/ps " selectionArgs \ " " psArgs " < /dev/null" if ((Cmd | getline PIDs["ps"]) != 1) return -1 if ((Cmd | getline Header) != 1) return -1 Procs["Header"] = Header if (!(CmdIndex = index(Header,"CMD")) && !(CmdIndex = index(Header,"COMMAND"))) return -1 while ((Cmd | getline) == 1) { PIDs[pid = $pidI] if (Debug) printf "Process %d (%d fields): %s\n",pid,NF,$0 > "/dev/stderr" ppid = $ppidI if (ppid in Children) Children[ppid] = Children[ppid] "," pid else Children[ppid] = pid if (WantArgs) Procs[pid,"ARGS"] = substr($0,CmdIndex) # Handle this as a special case so that it can be set before the # line (possibly) modified if (WantLine) Procs[pid,"LINE"] = $0 # Time field with either contain a : (time), a - (new date format), # or neither, in which case it occupies 2 fields (old date format). if (NF == 6) { # old ps defunct proc # Assign new values to fields, from right to left to avoid # overwriting fields before value is moved $cmdI = $ttyI $timeI = $stimeI $ttyI = "-" $stimeI = "-" } if ($stimeI !~ "[-:]") { if (!timePos) timePos = index($0,$stimeI) # Replace space in stime field with "-" $0 = substr($0,1,timePos+2) "-" substr($0,timePos+5) } for (i in getI) { Procs[pid,FieldNames[i]] = $i if (Debug) printf "%s=%s ",FieldNames[i],$i > "/dev/stderr" } if (Debug) print "" > "/dev/stderr" Lines++ } close(Cmd) return Lines } # makePSline: generate a line containing desired fields from ps data. # pid is the ID of the process to generate a line for. # If a pid of -1 is passed, a header line is returned. # Procs[] is the ps data, as generated by getPS(). # Fields[] is the set of fields desired in the output, with indexes starting # at 1. The values are field names as e.g. passed to getPS(). # Sep is the separator to put between fields. If null, a single space is used. # Return value: a line consisting of the fields requested, in the order of # their indices in Fields[]. # Example: # split("UID,PID,PPID,C,STIME,TTY,TIME,CMD",FieldNames,",") # makePSline(pid,psOut,FieldNames) function makePSline(pid,Procs,Fields,Sep, i,fieldName,line,width,value) { if (Sep == "") Sep = " " if (!("PID" in _makePSlineWidths)) # Make TIME before right-adjusted; some versions of ps drop leading # 0 fields from it. Assign(_makePSlineWidths, "UID=-8 PID=5 PPID=5 C=1 STIME=-8 TTY=-4 TIME=8 COMM=-8"," ","=") for (i = 1; i in Fields; i++) { fieldName = Fields[i] if (fieldName in _makePSlineWidths) width = _makePSlineWidths[fieldName] else width = "" if (pid == -1) value = fieldName else if (fieldName == "PID") value = pid else value = Procs[pid,fieldName] if (fieldName == "TTY") sub("^tty","",value) line = line Sep sprintf("%" width "s",value) } return substr(line,length(Sep)+1) } ### End ps lib ### start canonTTY library function nodevTTY(tty) { sub("^/dev/","",tty) return tty } function canonTTY(tty) { if (tty ~ "^/dev/") sub("^/dev/","",tty) # Do not change ?, -, and null names for the sake of ps output, etc. else if (tty !~ /^tty|^([-?]|)$/) tty = "tty" tty return tty } function shortTTY(tty) { sub("^/dev/","",tty) sub("^tty","",tty) return tty } ### end canonTTY library ### Begin utty,id routines # utty: find ttys a user is logged in on. # For each tty User is logged in on, an element is created in TTYs[]. # The index is the name of the tty, with a leading "/dev/". # The value is set to 1 if the user is writable on that tty, 0 if not. # The number of ttys the user is logged in on is returned. function utty(User,TTYs, Cmd,Count) { Cmd = "exec who -T" Count = 0 while ((Cmd | getline) == 1) if ($1 == User) { if ($2 == "+") TTYs[$3] = 1 else TTYs[$3] = 0 Count++ } close(Cmd) return Count } # id returns the numeric user id of the user who owns the current process. # In the array IDs, elements are set as follows: # uid: numeric user id # gid: numeric group id # group: group name, if any # user: user name, if any function id(IDs, Cmd,line,elem) { Cmd = "exec id" Cmd | getline line split(line,elem,"[()=]") close(Cmd) IDs["user"] = elem[3] IDs["gid"] = elem[5] IDs["group"] = elem[6] return IDs["uid"] = elem[2] } ### End utty,id routines ### Begin set library # 96/05/23 added return values jhdiii # Return value: the number of new elements added to Inter function Intersection(A,B,Inter, Elem,Count) { for (Elem in A) if (Elem in B && !(Elem in Inter)) { Inter[Elem] Count++ } return Count } # Return value: the number of new elements added to Both function Union(A,B,Both) { return CopySet(A,Both) + CopySet(B,Both) } # Deletes any elements that are in both Minuend and Subtrahend from Minuend. # Return value: the number of elements deleted. function SubtractSet(Minuend,Subtrahend, Elem,nDel) { for (Elem in Subtrahend) if (Elem in Minuend) { delete Minuend[Elem] nDel++ } return nDel } # Return value: the number of new elements added to To function CopySet(From,To, Elem,n) { for (Elem in From) if (!(Elem in To)) { To[Elem] n++ } return n } # Returns 1 if Set is empty, 0 if not. function IsEmpty(Set, i) { for (i in Set) return 0 return 1 } # MakeSet: make a set from a list. # An index with the name of each element of the list is created in the given # array. # Input variables: # Elements is a string containing the list of elements. # Sep is the character that separates the elements of the list. # Output variables: # Set is the array. # Return value: the number of new elements added to the set. function MakeSet(Set,Elements,Sep, i,Num,Names,nFound,ind) { nFound = 0 Num = split(Elements,Names,Sep) for (i = 1; i <= Num; i++) { ind = Names[i] if (!(ind in Set)) { Set[ind] nFound++ } } return nFound } # Returns the number of elements in set Set function NumElem(Set, elem,Num) { for (elem in Set) Num++ return Num } # Remove all elements from Set function DeleteAll(Set, i) { split("",Set,",") } # Returns a list of all of the elements in Set[], with each pair of elements # separated by Sep. function set2list(Set,Sep, list,elem) { for (elem in Set) list = list Sep elem return substr(list,2) # skip 1st separator } ### End set library function basename(path) { sub(".*/","",path) return path } ### Begin array routines # InitArr: Initialize an array with values. # Ind and Vals are separated into lists on Sep. # For each item in Ind, an index with that name is created in Arr[], # and the value with the same position in Vals is stored in it. # Global variables: none. function InitArr(Arr,Ind,Vals,sep, numind,indnames,values) { split(Ind,indnames,sep) split(Vals,values,sep) for (numind in indnames) Arr[indnames[numind]] = values[numind] } function ClearArr(Arr, Elem) { for (Elem in Arr) delete Arr[Elem] } # Subtract the values in Subtrahend from those in Minuend function SubtractArr(Minuend,Subtrahend, Elem) { for (Elem in Subtrahend) Minuend[Elem] -= Subtrahend[Elem] } # For each element of the array In, an element is created in Out having # an index equal to the value of the element in In and a value equal to # the index of the element in In. function Invert(In,Out, Index) { for (Index in In) Out[In[Index]] = Index } # Assign: make an array from a list of assignments. # An index with the name of each variable in the list is created in the array. # Its value is set to the value given for it. # Input variables: # Elements is a string containing the list of variable-value pairs. # Sep is the string that separates the pairs in the list. # AssignOp is the string that separates variables from values. # Output variables: # Arr is the array. # Return value: the number of elements added to the set. # Example: # Assign(Arr,"foo=blot bar=blat baz=blit"," ","=") function Assign(Arr,Elements,Sep,AssignOp, Num,Names,Elem,Assignments,Assignment,i) { Num = split(Elements,Assignments,Sep) for (i = 1; i <= Num; i++) { Assignment = Assignments[i] Ind = index(Assignment,AssignOp) Arr[substr(Assignment,1,Ind - 1)] = substr(Assignment,Ind + 1) } return Num } # Packs Arr[], which should have integer indices starting at or above n, to # contiguous integer indices starting with n. # If n is not given it defaults to 0. # Num should be the number of elements in Arr. function PackArr(Arr,Num,n, NewInd,OldInd) { NewInd = OldInd = n+0 for (; Num; Num--) { while (!(OldInd in Arr)) OldInd++ if (NewInd != OldInd) { Arr[NewInd] = Arr[OldInd] delete Arr[OldInd] } OldInd++ NewInd++ } } ### End array routines ### Begin head-tail routines # @(#) HeadTail.awk 96/05/09 # 95/04/28 Added tail routines. # 96/05/09 Added all args to HeadTailInit() # Turn on screen-bounded printing. # Current implementation sets global vars LINES, COLUMNS, LINEGAP, and COLGAP. # Sets the number of screen lines and rows to Lines and Rows. # If -1 is passed for either, turns off bounding in that dimension. # If either is not set or 0 is passed for it, its value is taken from the # environment, or if not set there, from terminfo, or if not set there, from # the defaults (24 and 80). # By default, the other functions in this library leave a "grace space" of # 1 column and 1 line. If LineGap or ColGap is passed and is a non-negative # value, the line gap is set to it. function HeadTailInit(Lines,Cols,LineGap,ColGap, Cmd) { # tput will use values in environment, but we want to avoid running # it if possible. if (Cols > 0) COLUMNS = Cols else if (!Cols) if ("COLUMNS" in ENVIRON) COLUMNS = ENVIRON["COLUMNS"] else { Cmd = "exec tput cols" Cmd | getline COLUMNS close(Cmd) if (COLUMNS == "") COLUMNS = 80 } if (Lines > 0) LINES = Lines else if (!Lines) if ("LINES" in ENVIRON) LINES = ENVIRON["LINES"] else { Cmd = "exec tput lines" Cmd | getline LINES close(Cmd) if (LINES == "") LINES = 24 } LINEGAP = (LineGap != "" && LineGap >= 0) ? LineGap : 1 COLGAP = (ColGap != "" && ColGap >= 0) ? ColGap : 1 } # Do screen-bound printing. # If LINES is >0, the last LINES-LINEGAP lines are kept in a circular buffer. # When TailFlush() is called, they are printed. # If LINES = 0, all lines are printed immediately. # If COLUMNS is >0, truncates Line to COLUMNS-COLGAP characters before printing # it. # Global vars: uses LINES & COLUMNS; sets/uses TailPtr; # saves lines in TailLines[] from 1..LINES-LINEGAP # Embedded newlines split the line into multiple lines; trailing newlines are # stripped. Tabs are expanded to spaces. function TailPrint(Line) { if (!LINES) print Line else { if (++TailPtr > (LINES-LINEGAP)) TailPtr = 1 TailLines[TailPtr] = Line } } function TailFlush( NumPrinted,Lines,Line,i,Buffer,PrintLines) { if (!LINES) return NumPrinted = 0 PrintLines = LINES-LINEGAP # Since lines may contain multiple lines, we must create a buffer to be # printed by reading line buffer backwards. # Stop when we have copied enough lines, or if we wrap around to the end # and find that the entire line buffer was not used. while (NumPrinted < PrintLines && TailPtr in TailLines) { # Split line into individual lines, then process them last to first Num = split(TailLines[TailPtr],Lines,"\n") for (i = Num; i >= 1; i--) { Line = Lines[i] if (i == Num && Line == "") # discard trailing newline continue # Put this line at the front of the print buffer if (COLUMNS) Buffer = substr(TabEx(Line),1,COLUMNS - COLGAP) "\n" Buffer else Buffer = Line "\n" Buffer if (++NumPrinted == PrintLines) break } if (!--TailPtr) # Wrap pointer if neccessary TailPtr = PrintLines } printf "%s",Buffer } # Do screen-bound printing. # If LINES >0, returns 0 when LINES-LINEGAP lines have been printed by # HeadPrint(). Otherwise returns 1. # If COLUMNS is >0, truncates Line to COLUMNS-COLGAP characters before printing # it. # Global vars: uses LINES, COLUMNS, LINEGAP, COLGAP; sets/uses LinesPrinted. # Line should not include newlines. function HeadPrint(Line) { # Check first, in case some calls of this function to not check return # value, and in case LINES is 1. if (LINES && LinesPrinted >= (LINES-LINEGAP)) return 0 if (COLUMNS) print substr(Line,1,COLUMNS - COLGAP) else print Line if (LINES && ++LinesPrinted >= (LINES-LINEGAP)) return 0 return 1 } function ColPrint(Line) { if (COLUMNS) print substr(Line,1,COLUMNS - COLGAP) else print Line return 1 } ### End head-tail routines ### Begin qsort routines # Arr[] is an array of values with arbitrary indices. # k[] is returned with numeric indices 1..n. # The values in k[] are the indices of Arr[], # ordered so that if Arr[] is stepped through # in the order Arr[k[1]] .. Arr[k[n]], it will be stepped # through in order of the values of its elements. # The return value is the number of elements in the arrays (n). function qsortArbIndByValue(Arr,k, ArrInd,ElNum) { ElNum = 0 for (ArrInd in Arr) k[++ElNum] = ArrInd qsortSegment(Arr,k,1,ElNum) return ElNum } # Sort a segment of an array. # Arr[] contains data with arbitrary indices. # k[] has indices 1..nelem, with the indices of arr[] as values. # This function sorts the elements of arr that are pointed to by # k[start..end], swapping the values of elements of k[] so that # when this function returns arr[k[start..end]] will be in order. function qsortSegment(Arr,k,start,end, left,right,sepval,tmp,tmpe,tmps) { # handle two-element case explicitly for a tiny speedup if ((end - start) == 1) { if (Arr[tmps = k[start]] > Arr[tmpe = k[end]]) { k[start] = tmpe k[end] = tmps } return } # Make sure comparisons act on these as numbers left = start+0 right = end+0 sepval = Arr[k[int((left + right) / 2)]] # Make every element <= sepval be to the left of every element > sepval while (left < right) { while (Arr[k[left]] < sepval) left++ while (Arr[k[right]] > sepval) right-- if (left < right) { tmp = k[left] k[left++] = k[right] k[right--] = tmp } } if (left == right) if (Arr[k[left]] < sepval) left++ else right-- if (start < right) qsortSegment(Arr,k,start,right) if (left < end) qsortSegment(Arr,k,left,end) } # Arr[] is an array of values with arbitrary indices. # k[] is returned with numeric indices 1..n. # The values in k are the indices of Arr[], # ordered so that if Arr[] is stepped through # in the order Arr[k[1]] .. Arr[k[n]], it will be stepped # through in order of the values of its indices. # The return value is the number of elements in the arrays (n). # If the indexes are numeric, Numeric should be true, so that they can be # compared as such rather than as strings. Numeric indexes do not have to be # contiguous. function qsortByArbIndex(Arr,k,Numeric, ArrInd,ElNum) { ElNum = 0 if (Numeric) # Indexes do not preserve numeric type, so must be forced for (ArrInd in Arr) k[++ElNum] = ArrInd+0 else for (ArrInd in Arr) k[++ElNum] = ArrInd qsortNumIndByValue(k,1,ElNum) return ElNum } # Arr is an array of elements with contiguous numeric indexes to be sorted # by value. # start and end are the starting and ending indexes of the range to be sorted. function qsortNumIndByValue(Arr,start,end, left,right,sepval,tmp,tmpe,tmps) { # handle two-element case explicitly for a tiny speedup if ((start - end) == 1) { if ((tmps = Arr[start]) > (tmpe = Arr[end])) { Arr[start] = tmpe Arr[end] = tmps } return } left = start+0 right = end+0 sepval = Arr[int((left + right) / 2)] while (left < right) { while (Arr[left] < sepval) left++ while (Arr[right] > sepval) right-- if (left <= right) { tmp = Arr[left] Arr[left++] = Arr[right] Arr[right--] = tmp } } if (start < right) qsortNumIndByValue(Arr,start,right) if (left < end) qsortNumIndByValue(Arr,left,end) } ### End qsort routines ### Start of ProcArgs library # @(#) ProcArgs 1.12 97/02/22 # 92/02/29 john h. dubois iii (john@armory.com) # 93/07/18 Added "#" arg type # 93/09/26 Do not count -h against MinArgs # 94/01/01 Stop scanning at first non-option arg. Added ">" option type. # Removed meaning of "+" or "-" by itself. # 94/03/08 Added & option and *()< option types. # 94/04/02 Added NoRCopt to Opts() # 94/06/11 Mark numeric variables as such. # 94/07/08 Opts(): Do not require any args if h option is given. # 95/01/22 Record options given more than once. Record option num in argv. # 95/06/08 Added ExclusiveOptions(). # 96/01/20 Let rcfiles be a colon-separated list of filenames. # Expand $VARNAME at the start of its filenames. # Let varname=0 and -option- turn off an option. # 96/05/05 Changed meaning of 7th arg to Opts; now can specify exactly how many # of the vars should be searched for in the environment. # Check for duplicate rcfiles. # 96/05/13 Return more specific error values. Note: ProcArgs() and InitOpts() # now return various negatives values on error, not just -1, and # Opts() may set Err to various positive values, not just 1. # Added AllowUnrecOpt. # 96/05/23 Check type given for & option # 96/06/15 Re-port to awk # 96/10/01 Moved file-reading code into ReadConfFile(), so that it can be # used by other functions. # 96/10/15 Added OptChars # 96/11/01 Added exOpts arg to Opts() # 96/11/16 Added ; type # 96/12/08 Added Opt2Set() & Opt2Sets() # 96/12/27 Added CmdLineOpt() # 97/02/22 Remove packed elements. # 97/02/28 Make sequence # for rcfiles & environ be "f" and "e". # Replaced CmdLineOpt() with OptsGiven(). # optlist is a string which contains all of the possible command line options. # A character followed by certain characters indicates that the option takes # an argument, with type as follows: # : String argument # ; Non-empty string argument # * Floating point argument # ( Non-negative floating point argument # ) Positive floating point argument # # Integer argument # < Non-negative integer argument # > Positive integer argument # The only difference the type of argument makes is in the runtime argument # error checking that is done. # The & option is a special case used to get numeric options without the # user having to give an option character. It is shorthand for [-+.0-9]. # If & is included in optlist and an option string that begins with one of # these characters is seen, the value given to "&" will include the first # char of the option. & must be followed by a type character other than ":" # or ";". # Note that if e.g. &> is given, an option of -.5 will produce an error. # Strings in argv[] which begin with "-" or "+" are taken to be # strings of options, except that a string which consists solely of "-" # or "+" is taken to be a non-option string; like other non-option strings, # it stops the scanning of argv and is left in argv[]. # An argument of "--" or "++" also stops the scanning of argv[] but is removed. # If an option takes an argument, the argument may either immediately # follow it or be given separately. # "-" and "+" options are treated the same. "+" is allowed because most awks # take any -options to be arguments to themselves. gawk 2.15 was enhanced to # stop scanning when it encounters an unrecognized option, though until 2.15.5 # this feature had a flaw that caused problems in some cases. See the OptChars # parameter to explicitly set the option-specifier characters. # If an option that does not take an argument is given, # an index with its name is created in Options and its value is set to the # number of times it occurs in argv[]. # If an option that does take an argument is given, an index with its name is # created in Options and its value is set to the value of the argument given # for it, and Options[option-name,"count"] is (initially) set to the 1. # If an option that takes an argument is given more than once, # Options[option-name,"count"] is incremented, and the value is assigned to # the index (option-name,instance) where instance is 2 for the second occurance # of the option, etc. # In other words, the first time an option with a value is encountered, the # value is assigned to an index consisting only of its name; for any further # occurances of the option, the value index has an extra (count) dimension. # The sequence number for each option found in argv[] is stored in # Options[option-name,"num",instance], where instance is 1 for the first # occurance of the option, etc. The sequence number starts at 1 and is # incremented for each option, both those that have a value and those that # do not. Options set from a config file get a sequence number of "f", and # options set in the environment get a sequence number of "e". # Options and their arguments are deleted from argv. # Note that this means that there may be gaps left in the indices of argv[]. # If compress is nonzero, argv[] is packed by moving its elements so that # they have contiguous integer indices starting with 0. # Option processing will stop with the first unrecognized option, just as # though -- was given except that unlike -- the unrecognized option will not be # removed from ARGV[]. Normally, an error value is returned in this case. # If AllowUnrecOpt is true, it is not an error for an unrecognized option to # be found, so the number of remaining arguments is returned instead. # If OptChars is not a null string, it is the set of characters that indicate # that an argument is an option string if the string begins with one of the # characters. A string consisting solely of two of the same option-indicator # characters stops the scanning of argv[]. The default is "-+". # argv[0] is not examined. # The number of arguments left in argc is returned. # If an error occurs, the global string OptErr is set to an error message # and a negative value is returned. # Current error values: # -1: option that required an argument did not get it. # -2: argument of incorrect type supplied for an option. # -3: unrecognized (invalid) option. function ProcArgs(argc,argv,OptList,Options,compress,AllowUnrecOpt,OptChars, ArgNum,ArgsLeft,Arg,ArgLen,ArgInd,Option,Pos,NumOpt,Value,HadValue,specGiven, NeedNextOpt,GotValue,OptionNum,Escape,dest,src,count,c,OptTerm,OptCharSet) { # ArgNum is the index of the argument being processed. # ArgsLeft is the number of arguments left in argv. # Arg is the argument being processed. # ArgLen is the length of the argument being processed. # ArgInd is the position of the character in Arg being processed. # Option is the character in Arg being processed. # Pos is the position in OptList of the option being processed. # NumOpt is true if a numeric option may be given. ArgsLeft = argc NumOpt = index(OptList,"&") OptionNum = 0 if (OptChars == "") OptChars = "-+" while (OptChars != "") { c = substr(OptChars,1,1) OptChars = substr(OptChars,2) OptCharSet[c] OptTerm[c c] } for (ArgNum = 1; ArgNum < argc; ArgNum++) { Arg = argv[ArgNum] if (length(Arg) < 2 || !((specGiven = substr(Arg,1,1)) in OptCharSet)) break # Not an option; quit if (Arg in OptTerm) { delete argv[ArgNum] ArgsLeft-- break } ArgLen = length(Arg) for (ArgInd = 2; ArgInd <= ArgLen; ArgInd++) { Option = substr(Arg,ArgInd,1) if (NumOpt && Option ~ /[-+.0-9]/) { # If this option is a numeric option, make its flag be & and # its option string flag position be the position of & in # the option string. Option = "&" Pos = NumOpt # Prefix Arg with a char so that ArgInd will point to the # first char of the numeric option. Arg = "&" Arg ArgLen++ } # Find position of flag in option string, to get its type (if any). # Disallow & as literal flag. else if (!(Pos = index(OptList,Option)) || Option == "&") { if (AllowUnrecOpt) { Escape = 1 break } else { OptErr = "Invalid option: " specGiven Option return -3 } } # Find what the value of the option will be if it takes one. # NeedNextOpt is true if the option specifier is the last char of # this arg, which means that if the option requires a value it is # the next arg. if (NeedNextOpt = (ArgInd >= ArgLen)) { # Value is the next arg if (GotValue = ArgNum + 1 < argc) Value = argv[ArgNum+1] } else { # Value is included with option Value = substr(Arg,ArgInd + 1) GotValue = 1 } if (HadValue = AssignVal(Option,Value,Options, substr(OptList,Pos + 1,1),GotValue,"",++OptionNum,!NeedNextOpt, specGiven)) { if (HadValue < 0) # error occured return HadValue if (HadValue == 2) ArgInd++ # Account for the single-char value we used. else { if (NeedNextOpt) { # option took next arg as value delete argv[++ArgNum] ArgsLeft-- } break # This option has been used up } } } if (Escape) break # Do not delete arg until after processing of it, so that if it is not # recognized it can be left in ARGV[]. delete argv[ArgNum] ArgsLeft-- } if (compress != 0) { dest = 1 src = argc - ArgsLeft + 1 if (src != dest) { for (count = ArgsLeft - 1; count; count--) { ARGV[dest] = ARGV[src] dest++ src++ } for (; dest < src; dest++) delete ARGV[dest] } } return ArgsLeft } # Assignment to values in Options[] occurs only in this function. # Option: Option specifier character. # Value: Value to be assigned to option, if it takes a value. # Options[]: Options array to return values in. # ArgType: Argument type specifier character. # GotValue: Whether any value is available to be assigned to this option. # Name: Name of option being processed. # OptionNum: Number of this option (starting with 1) if set in argv[], # or 0 if it was given in a config file or in the environment. # SingleOpt: true if the value (if any) that is available for this option was # given as part of the same command line arg as the option. Used only for # options from the command line. # specGiven is the option specifier character use, if any (e.g. - or +), # for use in error messages. # Global variables: OptErr # Return value: negative value on error, 0 if option did not require an # argument, 1 if it did & used the whole arg, 2 if it required just one char of # the arg. # Current error values: # -1: Option that required an argument did not get it. # -2: Value of incorrect type supplied for option. # -3: Bad type given for option & function AssignVal(Option,Value,Options,ArgType,GotValue,Name,OptionNum, SingleOpt,specGiven, UsedValue,Err,NumTypes) { # If option takes a value... [ NumTypes = "*()#<>]" if (Option == "&" && ArgType !~ "[" NumTypes) { # ] OptErr = "Bad type given for & option" return -3 } if (UsedValue = (ArgType ~ "[:;" NumTypes)) { # ] if (!GotValue) { if (Name != "") OptErr = "Variable requires a value -- " Name else OptErr = "option requires an argument -- " Option return -1 } if ((Err = CheckType(ArgType,Value,Option,Name,specGiven)) != "") { OptErr = Err return -2 } # Mark this as a numeric variable; will be propogated to Options[] val. if (ArgType != ":" && ArgType != ";") Value += 0 if ((Instance = ++Options[Option,"count"]) > 1) Options[Option,Instance] = Value else Options[Option] = Value } # If this is an environ or rcfile assignment & it was given a value... else if (!OptionNum && Value != "") { UsedValue = 1 # If the value is "0" or "-" and this is the first instance of it, # do not set Options[Option]; this allows an assignment in an rcfile to # turn off an option (for the simple "Option in Options" test) in such # a way that it cannot be turned on in a later file. if (!(Option in Options) && (Value == "0" || Value == "-")) Instance = 1 else Instance = ++Options[Option] # Save the value even though this is a flag Options[Option,Instance] = Value } # If this is a command line flag and has a - following it in the same arg, # it is being turned off. else if (OptionNum && SingleOpt && substr(Value,1,1) == "-") { UsedValue = 2 if (Option in Options) Instance = ++Options[Option] else Instance = 1 Options[Option,Instance] } # If this is a flag assignment without a value, increment the count for the # flag unless it was turned off. The indicator for a flag being turned off # is that the flag index has not been set in Options[] but it has an # instance count. else if (Option in Options || !((Option,1) in Options)) # Increment number of times this flag seen; will inc null value to 1 Instance = ++Options[Option] Options[Option,"num",Instance] = OptionNum return UsedValue } # Option is the option letter # Value is the value being assigned # Name is the var name of the option, if any # ArgType is one of: # : String argument # ; Non-null string argument # * Floating point argument # ( Non-negative floating point argument # ) Positive floating point argument # # Integer argument # < Non-negative integer argument # > Positive integer argument # specGiven is the option specifier character use, if any (e.g. - or +), # for use in error messages. # Returns null on success, err string on error function CheckType(ArgType,Value,Option,Name,specGiven, Err,ErrStr) { if (ArgType == ":") return "" if (ArgType == ";") { if (Value == "") Err = "must be a non-empty string" } # A number begins with optional + or -, and is followed by a string of # digits or a decimal with digits before it, after it, or both else if (Value !~ /^[-+]?([0-9]+|[0-9]*\.[0-9]+|[0-9]+\.)$/) Err = "must be a number" else if (ArgType ~ "[#<>]" && Value ~ /\./) Err = "may not include a fraction" else if (ArgType ~ "[()<>]" && Value < 0) Err = "may not be negative" # ( else if (ArgType ~ "[)>]" && Value == 0) Err = "must be a positive number" if (Err != "") { ErrStr = "Bad value \"" Value "\". Value assigned to " if (Name != "") return ErrStr "variable " substr(Name,1,1) " " Err else { if (Option == "&") Option = Value return ErrStr "option " specGiven substr(Option,1,1) " " Err } } else return "" } # Note: only the above functions are needed by ProcArgs. # The rest of these functions call ProcArgs() and also do other # option-processing stuff. # Opts: Process command line arguments. # Opts processes command line arguments using ProcArgs() # and checks for errors. If an error occurs, a message is printed # and the program is exited. # # Input variables: # Name is the name of the program, for error messages. # Usage is a usage message, for error messages. # OptList the option description string, as used by ProcArgs(). # MinArgs is the minimum number of non-option arguments that this # program should have, non including ARGV[0] and +h. # If the program does not require any non-option arguments, # MinArgs should be omitted or given as 0. # rcFiles, if given, is a colon-seprated list of filenames to read for # variable initialization. If a filename begins with ~/, the ~ is replaced # by the value of the environment variable HOME. If a filename begins with # $, the part from the character after the $ up until (but not including) # the first character not in [a-zA-Z0-9_] will be searched for in the # environment; if found its value will be substituted, if not the filename will # be discarded. # rcfiles are read in the order given. # Values given in them will not override values given on the command line, # and values given in later files will not override those set in earlier # files, because AssignVal() will store each with a different instance index. # The first instance of each variable, either on the command line or in an # rcfile, will be stored with no instance index, and this is the value # normally used by programs that call this function. # VarNames is a comma-separated list of variable names to map to options, # in the same order as the options are given in OptList. # If EnvSearch is given and nonzero, the first EnvSearch variables will also be # searched for in the environment. If set to -1, all values will be searched # for in the environment. Values given in the environment will override # those given in the rcfiles but not those given on the command line. # NoRCopt, if given, is an additional letter option that if given on the # command line prevents the rcfiles and environment from being read. # See ProcArgs() for a description of AllowUnRecOpt and optChars, and # ExclusiveOptions() for a description of exOpts. # Special options: # If x is made an option and is given, some debugging info is output. # h is assumed to be the help option. # Global variables: # The command line arguments are taken from ARGV[]. # The arguments that are option specifiers and values are removed from # ARGV[], leaving only ARGV[0] and the non-option arguments. # The number of elements in ARGV[] should be in ARGC. # After processing, ARGC is set to the number of elements left in ARGV[]. # The option values are put in Options[]. # On error, Err is set to a positive integer value so it can be checked for in # an END block. # Return value: The number of elements left in ARGV is returned. # Must keep OptErr global since it may be set by InitOpts(). function Opts(Name,Usage,OptList,MinArgs,rcFiles,VarNames,EnvSearch,NoRCopt, AllowUnrecOpt,optChars,exOpts, ArgsLeft,e) { if (MinArgs == "") MinArgs = 0 ArgsLeft = ProcArgs(ARGC,ARGV,OptList NoRCopt,Options,1,AllowUnrecOpt, optChars) if (ArgsLeft < (MinArgs+1) && !("h" in Options)) { if (ArgsLeft >= 0) { OptErr = "Not enough arguments" Err = 4 } else Err = -ArgsLeft printf "%s: %s.\nUse -h for help.\n%s\n", Name,OptErr,Usage > "/dev/stderr" exit 1 } if (rcFiles != "" && (NoRCopt == "" || !(NoRCopt in Options)) && (e = InitOpts(rcFiles,Options,OptList,VarNames,EnvSearch)) < 0) { print Name ": " OptErr ".\nUse -h for help." > "/dev/stderr" Err = -e exit 1 } if ((exOpts != "") && ((OptErr = ExclusiveOptions(exOpts,Options)) != "")) { printf "%s: Error: %s\n",Name,OptErr > "/dev/stderr" Err = 1 exit 1 } return ArgsLeft } # ReadConfFile(): Read a file containing var/value assignments, in the form # . # Whitespace (spaces and tabs) around a variable (leading whitespace on the # line and whitespace between the variable name and the assignment character) # is stripped. Lines that do not contain an assignment operator or which # contain a null variable name are ignored, other than possibly being noted in # the return value. If more than one assignment is made to a variable, the # first assignment is used. # Input variables: # File is the file to read. # Comment is the line-comment character. If it is found as the first non- # whitespace character on a line, the line is ignored. # Assign is the assignment string. The first instance of Assign on a line # separates the variable name from its value. # If StripWhite is true, whitespace around the value (whitespace between the # assignment char and trailing whitespace on the line) is stripped. # VarPat is a pattern that variable names must match. # Example: "^[a-zA-Z][a-zA-Z0-9]+$" # If FlagsOK is true, variables are allowed to be "set" by being put alone on # a line; no assignment operator is needed. These variables are set in # the output array with a null value. Lines containing nothing but # whitespace are still ignored. # Output variables: # Values[] contains the assignments, with the indexes being the variable names # and the values being the assigned values. # Lines[] contains the line number that each variable occured on. A flag set # is record by giving it an index in Lines[] but not in Values[]. # Return value: # If any errors occur, a string consisting of descriptions of the errors # separated by newlines is returned. In no case will the string start with a # numeric value. If no errors occur, the number of lines read is returned. function ReadConfigFile(Values,Lines,File,Comment,Assign,StripWhite,VarPat, FlagsOK, Line,Status,Errs,AssignLen,LineNum,Var,Val) { if (Comment != "") Comment = "^" Comment AssignLen = length(Assign) if (VarPat == "") VarPat = "." # null varname not allowed while ((Status = (getline Line < File)) == 1) { LineNum++ sub("^[ \t]+","",Line) if (Line == "") # blank line continue if (Comment != "" && Line ~ Comment) continue if (Pos = index(Line,Assign)) { Var = substr(Line,1,Pos-1) Val = substr(Line,Pos+AssignLen) if (StripWhite) { sub("^[ \t]+","",Val) sub("[ \t]+$","",Val) } } else { Var = Line # If no value, var is entire line Val = "" } if (!FlagsOK && Val == "") { Errs = Errs \ sprintf("\nBad assignment on line %d of file %s: %s", LineNum,File,Line) continue } sub("[ \t]+$","",Var) if (Var !~ VarPat) { Errs = Errs sprintf("\nBad variable name on line %d of file %s: %s", LineNum,File,Var) continue } if (!(Var in Lines)) { Lines[Var] = LineNum if (Pos) Values[Var] = Val } } if (Status) Errs = Errs "\nCould not read file " File close(File) return Errs == "" ? LineNum : substr(Errs,2) # Skip first newline } # Variables: # Data is stored in Options[]. # rcFiles, OptList, VarNames, and EnvSearch are as as described for Opts(). # Global vars: # Sets OptErr. Uses ENVIRON[]. # If anything is read from any of the rcfiles, sets READ_RCFILE to 1. function InitOpts(rcFiles,Options,OptList,VarNames,EnvSearch, Line,Var,Pos,Vars,Map,CharOpt,NumVars,TypesInd,Types,Type,Ret,i,rcFile, fNames,numrcFiles,filesRead,Err,Values,retStr) { split("",filesRead,"") # make awk know this is an array NumVars = split(VarNames,Vars,",") TypesInd = Ret = 0 if (EnvSearch == -1) EnvSearch = NumVars for (i = 1; i <= NumVars; i++) { Var = Vars[i] CharOpt = substr(OptList,++TypesInd,1) if (CharOpt ~ "^[:;*()#<>&]$") CharOpt = substr(OptList,++TypesInd,1) Map[Var] = CharOpt Types[Var] = Type = substr(OptList,TypesInd+1,1) # Do not overwrite entries from environment if (i <= EnvSearch && Var in ENVIRON && (Err = AssignVal(CharOpt,ENVIRON[Var],Options,Type,1,Var,"e")) < 0) return Err } numrcFiles = split(rcFiles,fNames,":") for (i = 1; i <= numrcFiles; i++) { rcFile = fNames[i] if (rcFile ~ "^~/") rcFile = ENVIRON["HOME"] substr(rcFile,2) else if (rcFile ~ /^\$/) { rcFile = substr(rcFile,2) match(rcFile,"^[a-zA-Z0-9_]*") envvar = substr(rcFile,1,RLENGTH) if (envvar in ENVIRON) rcFile = ENVIRON[envvar] substr(rcFile,RLENGTH+1) else continue } if (rcFile in filesRead) continue # rcfiles are liable to be given more than once, e.g. UHOME and HOME # may be the same filesRead[rcFile] if ("x" in Options) printf "Reading configuration file %s\n",rcFile > "/dev/stderr" retStr = ReadConfigFile(Values,Lines,rcFile,"#","=",0,"",1) if (retStr > 0) READ_RCFILE = 1 else if (ret != "") { OptErr = retStr Ret = -1 } for (Var in Lines) if (Var in Map) { if ((Err = AssignVal(Map[Var],Var in Values ? Values[Var] : "", Options,Types[Var],Var in Values,Var,"f")) < 0) return Err } else { OptErr = sprintf(\ "Unknown var \"%s\" assigned to on line %d\nof file %s",Var, Lines[Var],rcFile) Ret = -1 } } if ("x" in Options) for (Var in Map) if (Map[Var] in Options) printf "(%s) %s=%s\n",Map[Var],Var,Options[Map[Var]] > \ "/dev/stderr" else printf "(%s) %s not set\n",Map[Var],Var > "/dev/stderr" return Ret } # OptSets is a semicolon-separated list of sets of option sets. # Within a list of option sets, the option sets are separated by commas. For # each set of sets, if any option in one of the sets is in Options[] AND any # option in one of the other sets is in Options[], an error string is returned. # If no conflicts are found, nothing is returned. # Example: if OptSets = "ab,def,g;i,j", an error will be returned due to # the exclusions presented by the first set of sets (ab,def,g) if: # (a or b is in Options[]) AND (d, e, or f is in Options[]) OR # (a or b is in Options[]) AND (g is in Options) OR # (d, e, or f is in Options[]) AND (g is in Options) # An error will be returned due to the exclusions presented by the second set # of sets (i,j) if: (i is in Options[]) AND (j is in Options[]). # todo: make options given on command line unset options given in config file # todo: that they conflict with. function ExclusiveOptions(OptSets,Options, Sets,SetSet,NumSets,Pos1,Pos2,Len,s1,s2,c1,c2,ErrStr,L1,L2,SetSets,NumSetSets, SetNum,OSetNum) { NumSetSets = split(OptSets,SetSets,";") # For each set of sets... for (SetSet = 1; SetSet <= NumSetSets; SetSet++) { # NumSets is the number of sets in this set of sets. NumSets = split(SetSets[SetSet],Sets,",") # For each set in a set of sets except the last... for (SetNum = 1; SetNum < NumSets; SetNum++) { s1 = Sets[SetNum] L1 = length(s1) for (Pos1 = 1; Pos1 <= L1; Pos1++) # If any of the options in this set was given, check whether # any of the options in the other sets was given. Only check # later sets since earlier sets will have already been checked # against this set. if ((c1 = substr(s1,Pos1,1)) in Options) for (OSetNum = SetNum+1; OSetNum <= NumSets; OSetNum++) { s2 = Sets[OSetNum] L2 = length(s2) for (Pos2 = 1; Pos2 <= L2; Pos2++) if ((c2 = substr(s2,Pos2,1)) in Options) ErrStr = ErrStr "\n"\ sprintf("Cannot give both %s and %s options.", c1,c2) } } } if (ErrStr != "") return substr(ErrStr,2) return "" } # The value of each instance of option Opt that occurs in Options[] is made an # index of Set[]. # The return value is the number of instances of Opt in Options. function Opt2Set(Options,Opt,Set, count) { if (!(Opt in Options)) return 0 Set[Options[Opt]] count = Options[Opt,"count"] for (; count > 1; count--) Set[Options[Opt,count]] return count } # The value of each instance of option Opt that occurs in Options[] that # begins with "!" is made an index of nSet[] (with the ! stripped from it). # Other values are made indexes of Set[]. # The return value is the number of instances of Opt in Options. function Opt2Sets(Options,Opt,Set,nSet, count,aSet,ret) { ret = Opt2Set(Options,Opt,aSet) for (value in aSet) if (substr(value,1,1) == "!") nSet[substr(value,2)] else Set[value] return ret } # Returns true if any option in the string Opts was given, as indicated by the # data in Options[]. If any of Arg, Env, or File are true, the given opts are # only considered to have been set if they were set in the command line # arguments, environment, or in a configuration file, respectively. function OptsGiven(Options,Opts,Arg,Env,File, l,i,Opt,j,c) { if (!Arg && !Env && !File) Arg = Env = File = 1 l = length(Opts) for (i = 1; i <= l; i++) { Opt = substr(Opts,i,1) for (j = 1; (Opt,"num",j) in Options; j++) { c = Options[Opt,"num",j] if (Arg && c+0 > 0 || File && c == "f" || Env && c == "e") return 1 } } return 0 } ### End of ProcArgs library # Put a list of login shells (from /etc/shells) into set LoginShells[]. # Returns -1 if /etc/shells could not be read, else the number of shells found. function ReadShells(LoginShells, ret,Num,Line) { while (ret = ((getline Line < "/etc/shells") == 1)) if (Line ~ "^/") { Num++ sub(/[ \t]+/,"",Line) LoginShells[Line] } close("/etc/shells") _DidReadShells = 1 return ret ? -1 : Num } # Makes array shellUser[] have an index for each user who has a shell in # /etc/shells. # Returns 1 on success, 0 if there is a problem reading /etc/shells or # /etc/passwd. function makeShellUser(shellUser, LoginShells,ret,oFS) { if (ReadShells(LoginShells) < 0) return 0 oFS = FS FS = ":" while (ret = ((getline < "/etc/passwd") == 1)) if ($7 in LoginShells) shellUser[$1] close("/etc/passwd") FS = oFS return !ret }