#!/usr/local/bin/gawk -f
# determine file type and output information in plain2html.awk format
# $Id: magic.awk,v 1.4 1999/12/22 10:55:47 aonuma Exp aonuma $
# $Log: magic.awk,v $
# Revision 1.4 1999/12/22 10:55:47 aonuma
# Added NOHEADER, HEADERONLY, FOOTERONLY (for xargs)
#
# If resource is temporarily unavailable, try `unlimit'.
BEGIN {
CURRENTFILE="unknownmagic.xxx"
READLINES=10
IGNORECASE=1
MONTHS="00JanFebMarAprMayJunJulAugSepOctNovDec"
# TOP
# NOHEADER
# HEADERONLY
# FOOTERONLY
if (FOOTERONLY) NOHEADER=1
#LIST or TABLE
TABLE=1 # default is TABLE
if (LIST) TABLE=0 # set LIST=non-zero to output list format.
if (!NOHEADER) {
print ".LANG en"
print ".TI directory listing"
if (TABLE) {
print ".TTAG 1"
print ".TS ///"
print "#name /// file type and description ///>size /// YYYY/MM/DD"
if (!TOP) print "../ /// UP /// . /// ."
}
}
if (HEADERONLY || FOOTERONLY) exit
}
END {
if (!SKIP && CURRENTFILE!="unknownmagic.xxx") {
if (TABLE) PrintOutTable()
else PrintOutList()
}
if (!HEADERONLY && (!NOHEADER || FOOTERONLY) ) PrintFooter()
}
(FILENAME != CURRENTFILE) {
if (!SKIP && CURRENTFILE!="unknownmagic.xxx") {
if (TABLE) PrintOutTable()
else PrintOutList()
}
CLINE=1
SKIP=0
FRCS=""
CURRENTFILE=BASE=FILENAME
if (FILENAME ~ /~$/) SKIP=1
else {
"ls -LTl "FILENAME | getline tmp
split(tmp, FINFO)
# 1 2 3 4 5 6 7 8 9 10
# -rwxr-xr-x 1 aonuma zoo 4424 Nov 21 22:29:23 1999 magic.awk
FSIZE=FINFO[5]
if (FSIZE<1024) FSIZE=sprintf("%i bytes", FSIZE)
else {
FSIZE=FSIZE/1024
if (FSIZE<1024) FSIZE=sprintf("%i Kbytes", FSIZE)
else {
FSIZE=FSIZE/1024
FSIZE=sprintf("%i Mbytes", FSIZE)
}
}
FDATE=sprintf("%04i/%02i/%02i", FINFO[9], index(MONTHS, FINFO[6])/3, FINFO[7])
}
while (index(BASE, "/")) BASE=substr(BASE, index(BASE, "/")+1)
EXT=BASE
while (index(EXT, ".")) EXT=substr(EXT, index(EXT, ".")+1)
}
($0 && CLINE<=READLINES && !SKIP) {
LINE[CLINE++]=$0
}
($0 ~ /\$Id: / && !SKIP) {
n=split(substr($0, index($0, "$Id: ")), RCS)
# if (RCS[4]) FDATE=RCS[4]
if (n>6 && RCS[3]) FRCS=RCS[3]
}
(CLINE==READLINES && !SKIP) {# Analyze file type
TYPE=DESC=""
if (EXT=="Makefile") TypeMakefile()
if (EXT ~ /,v$/ ) TypeRcs()
if (EXT=="nn") TypeNn()
if (EXT=="tex") TypeTex()
if (EXT=="html" || EXT=="htm") TypeHtml()
if (LINE[1] ~ /^%!/ && LINE[1] ~ "PS-Adobe") TypeFile()
if (LINE[2] ~ /^#/) TypeFile()
if (!TYPE) TypeASCII()
if (!TYPE) TypeFile()
sub("^[ \t]+", "", DESC)
sub("[ \t]+$", "", DESC)
CLINE++
}
function PrintOutList(){
printf("@%s", CURRENTFILE, CURRENTFILE)
TYPE=TYPE" "DESC
if (FRCS) TYPE=sprintf("%s (RCS rev. %s)", TYPE, FRCS)
printf(" %50s%10s, %10s
\n", TYPE, FSIZE, FDATE)
DESC=TYPE=""
}
function PrintOutTable(){
printf("%s///", CURRENTFILE, CURRENTFILE)
printf("%s, %s", TYPE, DESC)
if (FRCS) printf(" (RCS rev. %s)", FRCS)
printf("///%s///%s", FSIZE, FDATE)
printf "\n"
DESC=TYPE=""
}
function Type(type){
TYPE=type
}
function TypeMakefile(){
TYPE="Makefile text"
if (LINE[1] ~ /^#/) DESC=substr(LINE[1], 2)
}
function TypeHtml(){
TYPE="html document"
i=1
while (LINE[i] !~ "" && i<= READLINES) i++
j=i
while (LINE[j] !~ "" && j<= READLINES) j++
if (i==j) DESC=LINE[i]
else for (n=i; n<=j; n++) DESC=sprintf("%s %s", DESC, LINE[n])
sub("]*>", "", DESC)
sub("", "", DESC)
if (CURRENTFILE ~ "guide." && DESC ~ /^\[SA\] /) DESC=DESC", index of this directory"
i=1
while (LINE[i] !~ "%s", i, i)
}
function TypePs(){
TYPE="postscript"
i=1
while (LINE[i] !~ /^%%Title: / && i<= READLINES) i++
DESC=substr(LINE[i], 10)
}
function TypeTex(){
if (EXT=="tex") {
TYPE="plain TeX"
i=1
while (LINE[i] !~ "\\document" && i<= READLINES) i++
if (LINE[i] ~ "\\documentstyle") TYPE="LaTeX209"
if (LINE[i] ~ "\\documentclass") TYPE="LaTeX"
while (LINE[i] !~ "\\title" && i<= READLINES) i++
DESC=LINE[i]
return
}
TYPE="TeX"
}
function TypeASCII() {
MUL=PLN=TAB=TXT=TEX=TITLE=0
if (EXT=="txt") {MUL+=2; PLN+=2; TAB+=2; TXT+=2}
if (LINE[1] ~ "[a-z][a-z]/[a-z][a-z]") {MUL+=2; TXT++}
if (LINE[1] ~ /^\.TI/) {PLN+=2; TXT++; TITLE=substr(LINE[1], 4)}
if (LINE[1] ~ /^#/) {TAB+=1; TXT++}
if (MUL && LINE[2] ~ /\.TI/) {MUL+=2; TXT++; TITLE=substr(LINE[2], 4)}
# print MUL, PLN, TAB, TXT
gsub("^[ \t]+", "", TITLE)
for (i in LINE) {
if (LINE[i] ~ /^[a-z][a-z][\>]/) {MUL++; TXT++}
if (LINE[i] ~ /^\.[A-Z][A-Z]/) {MUL++; PLN++; TXT++}
if (LINE[i] ~ /[;/]/) {TAB++; TXT++;}
if (LINE[i] ~ /[\{\}\\]/) {TEX++; TXT++;}
}
# print MUL, PLN, TAB, TXT
if (TAB>=CLINE && MUL+PLNPLN) {
TYPE="plain txt"
DESC="multilingual source file for html"
if (TITLE) DESC=sprintf("%s (%s)", DESC, TITLE)
return
}
if (PLN>CLINE/2) {
TYPE="plain txt"
DESC="source file for html"
if (TITLE) DESC=sprintf("%s (%s)", DESC, TITLE)
return
}
if (TEX>CLINE/2) {
TYPE="TeX"
DESC=""
return
}
}
function TypeFile(){
"file -L "FILENAME | getline TYPE
TYPE=substr(TYPE, index(TYPE, ":")+2)
if (LINE[2] ~ /^#/) DESC=substr(LINE[2], 2)
}
function PrintFooter() {
printf("Generated by magic.awk on %s\n", strftime("%Y-%m-%dT%T+0900"))
}