#!/usr/local/bin/gawk -f # @(#) ftpi2html.gawk 1.5 96/07/23 # 94/03/25 john h. dubois iii (john@armory.com) # 94/07/22 Updated for new index file format. # 94/08/14 Include a 'name' attribute in anchors. Deal with leading blanks. # Link dir descriptions in preamble to dir section in body. # Print horizontal rule between dir sections. # 95/08/04 Make hostnames be fully qualified. # 95/12/31 Allow an archive description to be added for each directory. # 96/01/11 Generalized; added rcfile, help, all options. # 96/03/29 Fixed incorrect creation of file URLs. # 96/07/06 Also read rcfiles in home dirs. # 96/07/23 Exit cleanly after printing help message. # Buffer output so that an index of directories can be put at the top. # Let lines be commented out. # Print a description line with appropriate anchor # Uses global ArchDir # InBody is false if we are in the global preamble, else true function PrintDesc(InBody, Blanks,Text,Filename) { if ($0 ~ "^[ \t]") { # Line with leading whitespace # Split line into leading whitespace (if any) and everything after it. match($0,"^[ \t]+") Blanks = substr($0,1,RLENGTH) Text = substr($0,RLENGTH+1) Filename = $2 } else { Text = $0 Filename = $1 Blanks = "" } # Find where the filename ends. match(Text,"[ \t]") if (InBody) { # File description line directory section # Make the 'name' be just the filename with the dir, # to make it easier to find & because in general the only instances of # multiple identical names will be multiple links to the same file. # Exception: include the dir name in index files since they will be # different. OutLines[++OutInd] = \ sprintf("%s%s%s", Blanks,Filename == "index" ? ArchDir Filename : Filename,ftpHost, ArchDir,Filename,substr(Text,1,RSTART-1),substr(Text,RSTART)) } else { # Directory description line in preamble OutLines[++OutInd] = \ sprintf("%s%s%s", Blanks,Filename,substr(Text,1,RSTART-1),substr(Text,RSTART)) noDirIndex = 1 } } function PrintDirectoryIndex( i,n,link,maxLinkLen,l,dots) { if (noDirIndex) return if (0) { # Experimental code print "" for (i = 1; i <= IndInd; i++) { n = link = IndexLinks[i] sub("/$","",n) # Don't bother with trailing / in dir index printf \ "\n", link,n } print "
FTP Directory Index
%s%s
" } else { printf "
" maxLinkLen = 0 for (i = 1; i <= IndInd; i++) { l = length(IndexLinks[i]) - 1 if (l > maxLinkLen) maxLinkLen = l } dots = "" for (i = 1; i <= maxLinkLen; i++) dots = dots "." for (i = 1; i <= IndInd; i++) { n = link = IndexLinks[i] sub("/$","",n) # Don't bother with trailing / in dir index printf "%s %s %s\n",link,n, substr(dots,1,maxLinkLen - length(n)), IndexDesc[i] } printf "
" } } BEGIN { Name = "ftpi2html" Usage = \ "Usage: " Name " [-chn] [-t] [-p<FTP-host>] [-m<maintainer>]\n"\ " [index-file ...]" rcFile = ".ftpi2html" globrcFile = "/etc/default/ftpi2html" defTitle = "The Anonymous FTP Archives" defIndex = "/usr/local/ftp/pub/index" ARGC = Opts(Name,Usage,"ct:f:u:M:p:hx",0, "~/" rcFile ":$UHOME/" rcFile ":" globrcFile, "NOCTYPE,TITLE,INDEX,UPLINK,MAINTAINER,FTPHOST",0,"n") if ("h" in Options) { printf \ "%s: Read FTP index files and convert them to an HTML listing.\n"\ "%s\n"\ "This program reads FTP index files and builds an HTML index from them.\n"\ "The file(s) must be in this format:\n"\ "\n"\ "Preamble\n"\ "Directory-Index\n"\ "Directory-Index\n"\ "...\n"\ "\n"\ "Preamble consists of a ordinary text lines and directory descriptions.\n"\ "The directory description lines consist of a directory name, one or more\n"\ "tabs, and a description of the files to be found in the directory. The\n"\ "directory name may contain embedded / characters but should not be\n"\ "followed by a trailing /. The directory name will be linked to the\n"\ "matching Directory-Name line in a Directory-Index section.\n"\ "Preamble example:\n"\ "\n"\ "deepthought's archives are in the following directories:\n"\ "dirname1/<tab>What's in directory 1\n"\ "dirname2/<tab>What's in directory 2\n"\ "\n"\ "If no directory name lines are found in the preamble, an index of all\n"\ "directories found is built and put (by default) before the preamble.\n"\ "A Directory-Index section has this format:\n"\ "\n"\ "Directory-Name/\n"\ "Description ...\n"\ "\n"\ "Header\n"\ "Filename File-Description\n"\ "Filename File-Description\n"\ "...\n"\ "\n"\ "Directory-Name is the name of the directory whose files will be described.\n"\ "It must be followed by a /. It will be given an anonymous FTP link to the\n"\ "named directory, and its anchor name will be the directory name as it\n"\ "appears (with a trailing /).\n"\ "Description is an option section describing what is in this directory.\n"\ "If an index of directories is built, the description attached to each\n"\ "directory is the first line of its description section, up through the\n"\ "first period (if any) that is followed by whitespace.\n"\ "A blank line must come before the header line.\n"\ "Header is a line that describes what follows.\n"\ "Filename is the name of the file in the directory. It will be given an\n"\ "anonymous FTP link to the file, and its anchorname will be the filename\n"\ "without having the directory name attached; this means that a URL that\n"\ "points to the anchor will find only the first instance of multiple\n"\ "identical filenames.\n"\ "A line that contains /* at the start of the line and has only whitespace\n"\ "after the /* opens a comment section. Everything starting with that line\n"\ "and continuing up through a similar */ line is discarded.\n"\ "A line that begins with # is a comment line or a directive line.\n"\ "Neither is included in the output. The only directive is #index. A line\n"\ "that begins with the word #index causes the index of directories that is\n"\ "generated (if any) to be placed at the position of the #index line in the\n"\ "output, instead of just before the preamble. #index would typically be\n"\ "put somewhere in the preamble.\n"\ "If no index-files are given on the command line, the default file read\n"\ "is %s. The default file read can be changed by assigning\n"\ "a value to the variable INDEX in one of the files described below.\n"\ "Options:\n"\ "Some of the following options can also be set by assigning values to\n"\ "variables in a configuration file. Three configuration files are read, in\n"\ "order: a file named %s in the invoking user's home directory; a file\n"\ "named %s in the directory specified by the environment variable UHOME\n"\ "(if it is set); and the file %s. Variables are\n"\ "assigned to with the syntax: varname=value or in the case of flags, by\n"\ "simply putting the indicated variable name in the file without a value.\n"\ "A variable assigned to in one of these files will override values assigned\n"\ "to the same variable in one of the files read after it. To turn off a flag\n"\ "option and prevent it from being set in a file read later, assign it a\n"\ "value of 0. e.g. if NOCTYPE is set in %s, NOCTYPE=0\n"\ "in a %s file will override it. Flag options can be turned off on the\n"\ "command line by following them immediately with '-', e.g. -c- to turn off\n"\ "the c option in such a way that it cannot be turned on in a config file.\n"\ "Variable names appear in parentheses in the option descriptions.\n"\ "-c: Do not print a content-type header. Use this if the output is to a\n"\ " cache file. By default, a Content-type header is output, which is\n"\ " appropriate if run as a cgi-bin script. (NOCTYPE)\n"\ "-p<FTP-host>: Set the FTP hostname. By default, the local hostname is \n"\ " used. (FTPHOST)\n"\ "-u<uplink>: If given, this is printed near the bottom of the page; it can\n"\ " be used to provide a link back to another page. (UPLINK)\n"\ "-M<maintainer>: If given, this should be the name of the maintainer (and\n"\ " perhaps a mailto: or other link to reach the maintainer). It will be\n"\ " printed at the bottom of the page and identified as the page\n"\ " maintainer. (MAINTAINER)\n"\ "-h: Print this help.\n"\ "-n: Do not read the configuration file.\n"\ "-t<title>: Set the title. The title is used for the <title> in the HTML\n"\ " head section, and also as a <h1> heading in the body. The default\n"\ " title is \"%s\". (TITLE)\n",Name,Usage,defIndex,rcFile,rcFile, globrcFile,globrcFile,rcFile,defTitle ERR = 0 exit 0 } Debug = "x" in Options # Initially, make FS be just a tab, since that is how description lines are # distinguished in the preamble FS = "\t" if (ARGC < 2) { ARGC = 2 if ("f" in Options) # f pseudo-option; actually INDEX var ARGV[1] = Options["f"] else ARGV[1] = defIndex } if (!("c" in Options)) print "Content-type: text/html\n" if ("t" in Options) Title = Options["t"] else Title = defTitle if ("u" in Options) upLink = Options["u"] if ("M" in Options) Maintainer = Options["M"] if ("p" in Options) ftpHost = Options["p"] else { if (CmdReadLine("hostname") != 1) { printf "Could not get hostname. Exiting.\n" > "/dev/stderr" exit 1 } ftpHost = $0 } if (Debug) printf "ftp hostname: %s\n",ftpHost > "/dev/stderr" print "<html><head><title>" Title "" print "

" Title "

" print "
"
    OutInd = IndInd = InBody = 0
}

# comment start
$0 ~ /^\/\*[ \t]*$/ {
    inComment = 1
    next
}

$0 ~ /^\*\/[ \t]*$/ {
    inComment = 0
    next
}

# commented-out line
inComment {
    next
}

# commented-out line or directive
$0 ~ /^#/ {
    if ($1 == "#index")
	indexPosition = OutInd
    next
}

# A directory name; start of a directory index
NF == 1 && $1 ~ "/$" {
    InBody = 1	# No longer in global preamble
    ArchDir = $1
    # Leave the / attached in the name since we know it's a directory name
    # if we want to go to it, and it might have the same name as a file.
    OutLines[++OutInd] = \
    sprintf("
%s",$1,ftpHost,$1,$1) IndexLinks[++IndInd] = $1 # No longer in preamble InArchDesc = 1 FirstDescLine = 1 FS = "[ \t]+" HeaderNext = 1 next } # In the per-directory archive description section. InArchDesc { if (!NF) { InArchDesc = 0 OutInd++ next } # Attach the first sentence of the description to the directory index if (FirstDescLine) { line = $0 gsub("^[ \t]+|[ \t]+$","",line) # discard leading/trailing whitespace # Get rid of everything after first period followed by whitespace if (!sub(/\.[ \t].*/,"",line)) # if that doesn't exist, remove trailing period, if any sub(/\.$/,"",line) if (line != "") IndexDesc[IndInd] = line FirstDescLine = 0 } line = $0 count = 0 # Make references to directories be links to the appropriate section of # the output document while (match(line,"(^| )[^ ]+/( |$)") && ++count < 40) { lastC = RSTART+RLENGTH-1 if (substr(line,RSTART,1) == " ") { RSTART++ RLENGTH-- } if (substr(line,lastC,1) == " ") RLENGTH-- word = substr(line,RSTART,RLENGTH) sub(/^\.\//,ArchDir,word) word = "" word "" line = InsertStr(DelStr(line,RSTART,RLENGTH),RSTART,word) } OutLines[++OutInd] = line next } # A blank or header line, or a non-description line in the preamble HeaderNext || !NF || !InBody && NF != 2 { if (NF) HeaderNext = 0 OutLines[++OutInd] = $0 next } # A description line { PrintDesc(InBody) } END { if (ERR != "") exit ERR # Only bother with the directory index if there is more than one directory if (!indexPosition && IndInd > 1) PrintDirectoryIndex() if (IndInd <= 1) indexPosition = 0 for (i = 1; i <= OutInd; i++) { print OutLines[i] if (i == indexPosition) PrintDirectoryIndex() } print "
" if (upLink != "") printf "
%s\n",upLink if (Maintainer != "") printf "
This web page maintained by %s\n",Maintainer print "" } ### End of main program. ### Begin Strings routines # Delete the string starting at Start and having length Num from the middle # of string S, and return the remaining part. function DelStr(S,Start,Num) { return substr(S,1,Start - 1) substr(S,Start+Num) } # Insert NewStr into S at position Pos (between the Pos-1'th and the Pos'th # characters). S is padded with spaces if neccessary. function InsertStr(S,Pos,NewStr, e) { e = length(S)+1 # The position after the end of S if (e >= Pos) return substr(S,1,Pos-1) NewStr substr(S,Pos) for (; e < Pos; e++) S = S " " return S NewStr } # Search for char C in string S starting at position Pos, in the direction # specified by Dir (1 = forward, -1 = backward). # Return position char found at for success, 0 if not found before start or end # of string. function FindC(S,Pos,C,Dir, FoundC) { while (Pos > 0 && (FoundC = substr(S,Pos,1)) != C && FoundC != "") Pos += Dir if (FoundC == C) return Pos else return 0 } # Split string S into array Arr, one character per index, starting with 1. # The number of characters in the string is returned. function SplitS(S,Arr, len,i) { len = length(S) for (i = 1; i <= len; i++) Arr[i] = substr(S,i,1) return len } # Paste NewStr onto S at position Pos, overwriting what was there # S is padded with spaces if neccessary. function PasteStr(S,Pos,NewStr, e) { e = length(S)+1 # The position after the end of S if (e >= Pos) return substr(S,1,Pos-1) NewStr substr(S,Pos+length(NewStr)) for (; e < Pos; e++) S = S " " return S NewStr } ### End Strings routines # @(#) CmdReadLine 95/09/04 # Run Command, read a single line of output from it, then close it. # If Verbose is true, a complaint is issued if the read fails. # Output is returned in $* # The return value from getline is returned. It will be 1 on a successful # read; 0 if no lines were read due because the command produced no output # or could not be run. ERRNO is never set since pipes are run by a shell. function CmdReadLine(Command,Verbose, ret) { if (Debug) { print "* Issuing command: " Command "\n"\ "* Waiting for single line of output..." > "/dev/stderr" } ret = Command | getline if (Verbose && ret != 1) printf "Read from pipe '%s' failed\n",Command # close doesn't return a value under awk, only gawk close(Command) if (Debug) print "* Output: " $0 > "/dev/stderr" return ret } ### Start of ProcArgs library # @(#) ProcArgs 1.11 96/12/08 # 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() # 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 have a value of 0 assigned to this. # 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 for (count = ArgsLeft - 1; count; count--) { ARGV[dest] = ARGV[src] dest++ src++ } } 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 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,0)) < 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,0)) < 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 option Opt was given on the command line. function CmdLineOpt(Options,Opt, i) { for (i = 1; (Opt,"num",i) in Options; i++) if (Options[Opt,"num",i] != 0) return 1 return 0 } ### End of ProcArgs library