#!/usr/local/bin/gawk -f
# plain text file to html converter.
# Copyright 1999 Shuji AONUMA (aonuma@lcc-toulouse.fr)
# There is absolutely no warranty for the program.
# You can redistribute it and/or modify it under the terms of
# the GNU General Public License version 2 dated June, 1991.
#
# $Id: plain2html.awk,v 1.7 1999/11/24 10:04:16 aonuma Exp aonuma $
#<![CDATA[
BEGIN{
## How to make input plain text file.
## Spaces and tabs at a beginning and end of a line are ignored.
## A command line is
## a line begun with a single period followed
## by COMMAND (that is a commnad for plain2html.awk) (eg .SAMPLE foo bar)
## PREAMBLE is a special part of the file.
## Some commands shuold be put in PREAMBLE (marked as [P]
## in this manual), and some show same ([PB]) or different behavior in PREAMBLE or not
## ([P]=blah blah blah, [B]=tac tac tac),
## and others will be ignored in PREAMBLE ([B]).
## PREAMBLE is from a top line of a input file
## to the line before non-COMMAND line. In addition,
## PREAMBLE commands can be specified from a gawk command line,
## for example, gawk -v SAMPLE="foo bar".
## BODY is a part after the first non-command line.
# header, footer
#TI title #[P]=generate a <HEAD> part, <TITLE>, <H1>; [B]=<H1> heading only.
#SC, SSC, SSSC; section, subsection, subsubsectionの<H2>, <H3>, <H4>
# LANG="ja"
# CHARSET is set by SetCharset(LANG)
# if (xxx) @xxx --> yyy substitutions
if (!BASE) BASE="." #e.g. @BASE/index.html, / should be followed.
if (!ROOT) ROOT=".." # @ROOT/
#FOOT (internal) END処理中にfooter part 出力
if (!ADDRESS) ADDRESS="default@address" # mail address (footer part 出力)
if (!NAME) NAME="default name" # name (NAME{}{}の方と紛らわしいけど)
if (!FOOTERFILE) FOOTERFILE="@ROOT/etc/footer.@LANG.txt" # FOOTERFILEがあれば<address>...</address>の代わりにこちらを使う。
#NOFOOTER=1 (FOOTERは出力しない)
#ONLYFOOTER=1 (FOOTERのみ出力)
if (ONLYFOOTER) {PrintFooter();exit}
#TABLE関係。
if (!TS) TS = "not@defined%yet" #
if (!TH) TH = "#" # TH行を示すのに使う
## If a field in a TH line begin with "=", this colum is aligned center.
## If a field in a TH line begin with ">", this colum is aligned right.
## NOTH, no title line (TH) in TABLE
if (!TABLEHEADER) TABLEHEADER="<TABLE BORDER=5 CELLPADDING=0 CELLSPACING=0>"
TABLE=0 #(internal) TABLE処理中はTABLE>0
TBLANK="---" # defaut item for blank data
#TTAG=0 # if 1, create NAME anchor
# CAPTION
# リスト
if (!LISTTYPE) LISTTYPE="UL"
if (!LI) LI="@" # LI行を示すのに使う。
LIST=0 #(internal) LIST処理中はLIST=1
# DL (internal) description list mode
# 以上の処理の対象にならなかった行の処理
if (!OUT) OUT=1 #出力しない 1出力する (古いヴァージョンとの互換性のため)
# else OUT=0
# TEXT (internal) <P>の挿入
#\\, HREF{}{}, NAME, MARK{}{}
# MENU=1 #menu mode
MENU=1
if (NOMENU) MENU=0
if (MENU+0) MENU="MENULIST" # MENU=1のように数字だった場合。
if (!BODY) BODY="<body BGCOLOR=\"#ffffC0\" TEXT=\"#102c5c\" LINK=\"#0000FF\" VLINK=\"#339930\" ALINK=\"#FF3300\"><script type="text/javascript" src="http://hb.lycos.com/hb.js"></script>
<script type="text/javascript" src="http://ratings.lycos.com/ratings/lycosRating.js.php"></script>
<script type="text/javascript"><!--//--><![CDATA[//><!--
var cm_role = "live";
var cm_host = "tripod.lycos.com";
var cm_taxid = "/memberembedded";
var tripod_member_name = "aonuma";
var tripod_member_page = "aonuma/scripts/plain2html.awk";
var tripod_ratings_hash = "1258351847:8a6f21dc2c6a7c10c77903a2187e589a";
var lycos_ad_category = {"dmoz":"recreation\/food","ontarget":"&CAT=family%20and%20lifestyles&L2CAT=food%20and%20wine","find_what":"cooking school"};
var lycos_ad_remote_addr = "38.107.191.105";
var lycos_ad_www_server = "www.tripod.lycos.com";
var lycos_ad_track_small = "http://members.tripod.com/adm/img/common/ot_smallframe.gif?rand=845427";
var lycos_ad_track_served = "http://members.tripod.com/adm/img/common/ot_adserved.gif?rand=845427";
//--><!]]]]><![CDATA[></script>
<script type="text/javascript" src="http://scripts.lycos.com/catman/init.js"></script>
<script type="text/javascript" src="http://members.tripod.com/adm/ad/code-start.js"></script>
<script type="text/javascript" src="http://members.tripod.com/adm/ad/code-middle.js"></script>
<script type="text/javascript" src="http://members.tripod.com/adm/ad/code-end.js"></script>
<noscript>
<img src="http://members.tripod.com/adm/img/common/ot_noscript.gif?rand=845427" alt="" width="1" height="1" />
<!-- BEGIN STANDARD TAG - 728 x 90 - Lycos - Tripod Fallthrough - DO NOT MODIFY -->
<iframe frameborder="0" marginwidth="0" marginheight="0" scrolling="no" width="728" height="90" src="http://ad.yieldmanager.com/st?ad_type=iframe&ad_size=728x90§ion=209094"></iframe>
<!-- END TAG -->
</noscript>
"
if (!MAINPAGE) {
MAINPAGE="@ROOT/index.@LANG.html" # MENUのBACKで指定するページ
}
# DESC ## [P] <META NAME="description" ...>
# KW ## [P] <META NAME="keywords" ...>
PREAMBLE=1 # (internal)
}
END{
if (TABLE) print "</TABLE>"
if (LIST) printf("</%s>\n\n", LISTTYPE)
if (MENU) {
printf("<CENTER><A NAME=\"%s\"><SMALL>auto menu</SMALL></A>\n", MENU)
printf("[<A HREF=\"#ToP\"><SMALL>TOP of this page</SMALL></A>]\n")
for (i=0;i<menu;i++) print Menu[i]
print"<BR>"
if (MAINPAGE) {
MAINPAGE=Gsub(MAINPAGE)
printf("<A HREF=\"%s\">HOME</A>\n", MAINPAGE)
}
print "</CENTER>"
}
if (FOOT && !NOFOOTER) PrintFooter()
}
(DEBUG) {
print "<"$0">"
}
{
Proc()
}
## functions
function PrintHeader(X) {
print"<!DOCTYPE HTML PUBLIC \"-//IETF//DTD HTML//EN\">"
if (LANG) printf("<HTML LANG=\"%s\">\n", LANG)
else print"<HTML>"
print"<HEAD>"
print"\t<!-- This HTML file has been created by plain2html.awk -->"
if (CHARSET) printf("\t<META HTTP-EQUIV=\"Content-Type\" CONTENT=\"text/html; charset=%s\">\n", CHARSET)
if (DESC) printf("\t<META NAME=\"description\" CONTENT=\"%s\">\n", DESC)
if (KW) printf("\t<META NAME=\"keywords\" CONTENT=\"%s\">\n", KW)
printf("\t<TITLE>%s</TITLE>\n", X)
print"</HEAD>\n"
if (BODY) print BODY; else "<BODY>"
printf("<CENTER><H1><A NAME=\"ToP\">%s</A></H1></CENTER>\n", X)
FOOT=1
TI=""
}
function PrintFooter() {
if (FOOTERFILE) {
FOOTERFILE=Gsub(FOOTERFILE)
while ((getline < FOOTERFILE) > 0) Proc()
}
else {
if (ADDRESS) {
if (!NAME) NAME=ADDRESS
printf("<ADDRESS><A HREF=\"mailto:%s\">%s</A></ADDRESS>\n", ADDRESS, NAME)
}
}
print"<HR>"
if (ROOT) printf("<FONT SIZE=-3>Converted by <A HREF=\"%s/scripts/plain2html.awk\">plain2html.awk</A> on %s</FONT>\n", ROOT, strftime("%Y-%m-%dT%T+0900"))
else printf("<FONT SIZE=-3>Converted by plain2html.awk on %s</FONT>\n", strftime("%Y-%m-%dT%T+0900"))
print"</body>"
print"</HTML>"
}
function PrintSection(A, n) {
if (A ~ "<A NAME=\"") {
NTAG=substr(A, index(A, "<A NAME=\"")+9)
NTAG=substr(NTAG, 1, index(NTAG, "\"")-1)
printf("\n<H%i>%s</H%i>\n", n, A, n)
}
else {
NTAG=A
gsub("[ \t]+", "_", NTAG)
if (NTAGS ~ ":"NTAG":") NTAG=sprintf("%s%03i", NTAG, NR)
if (NTAG ~ /[^.,:;\+-_0-9A-Za-z]/) NTAG=sprintf("%s%03i", "t", NR)
printf("\n<H%i><A NAME=\"%s\">%s</A></H%i>\n", n, NTAG, A, n)
}
NTAGS=NTAGS":"NTAG":"
if (n==1) Menu[menu++]=sprintf("[<A HREF=\"#%s\"><BIG>%s</bIG></A>]", NTAG, A)
if (n==2) Menu[menu++]=sprintf("[<A HREF=\"#%s\"><STRONG>%s</STRONG></A>]", NTAG, A)
if (n==3) Menu[menu++]=sprintf("[<A HREF=\"#%s\">%s</A>]", NTAG, A)
if (n==4) Menu[menu++]=sprintf("[<A HREF=\"#%s\"><SMALL>%s</SMALL></A>]", NTAG, A)
}
function PrintMenuTop(MENU) {
if (NTAGS) return
NTAGS=":"MENU":ToP"
printf("<CENTER><A HREF=\"#%s\"><SMALL>go to menu</SMALL></A></CENTER>\n", MENU)
}
function Expand(EXTAG, EXn, EXIN) {
# EXTAG{}..{} --> EX[i]
# EXIN is splitted to EXX + (EXX[i], n=1,..EXn) + EXY
if (substr(EXIN, 1, length(EXTAG))==EXTAG) EXX=""
else EXX=substr(EXIN, 1, index(EXIN, EXTAG"{")-1)
EXY=substr(EXIN, index(EXIN, EXTAG"{")+length(EXTAG))
EXi=1
# to split EX[i] + EXY
while (EXY ~ /^{/ && EXi<=EXn) {
EX[EXi]=EXY
EXY=substr(EX[EXi], index(EX[EXi], "}")+1)
EX[EXi]=substr(EX[EXi], 2, index(EX[EXi], "}")-2)
while (TestBraces("{", "}", EX[EXi])) { #check looped {}
EX[EXi]=sprintf("%s}%s", EX[EXi], substr(EXY, 1, index(EXY, "}")-1))
EXY=substr(EXY, index(EXY, "}")+1)
}
EXi++
}
while (EXi<=EXn) EX[EXi++]=""
}
function TestBraces(TA, TB, TC) {
return (gsub(TA, TA, TC) > gsub(TB, TB, TC))
}
function ProcDL (){
if ($0=="/DL") {
if (DD) print"\t</DD>\n"
print "</DL>\n"
DL=DD=0
$0=""
return
}
if (substr($0, 1, 1)==LI) {
if (DD) print"\t</DD>\n"
$0=substr($0, 2)
printf("\t<DT>%s</DT>\n\t<DD>", $0)
}
else {
if ($0) {
print
DD=1
}
else {
if (DD) print"\t</DD>\n"
DD=0
}
}
$0=""
return
}
function SetCharset(Lang) {
#CHARSET="us-ascii"
CHARSET="ISO-8859-1"
if (Lang=="ja") CHARSET="EUC-JP"
if (Lang=="fr") CHARSET="ISO-8859-1"
}
function Gsub(gs) {
if (BASE) gsub("@BASE/", BASE"/", gs)
else gsub("@BASE/", "", gs)
if (ROOT) gsub("@ROOT/", ROOT"/", gs)
else gsub("@ROOT/", "", gs)
if (LANG) gsub("@LANG", LANG, gs)
else gsub("\\.@LANG", "", gs)
return(gs)
}
function Proc() {
if ($0 ~ "@"){ # for @xxx --> yyy substitutions
$0=Gsub($0)
}
if ($0) {
gsub("^[ \t]+", "")
gsub("[ \t]+$", "")
sub("\\\\\\\\$", "<BR>")
while ($0 ~ "HREF{") {
Expand("HREF", 2, $0)
if (!EX[2]) EX[2]=EX[1]
$0=sprintf("%s<A HREF=\"%s\">%s</A>%s", EXX, EX[1], EX[2], EXY)
}
while ($0 ~ "NAME{") {
Expand("NAME", 2, $0)
if (!EX[2]) EX[2]=EX[1]
$0=sprintf("%s<A NAME=\"%s\">%s</A>%s", EXX, EX[1], EX[2], EXY)
}
while ($0 ~ "MARK{") {
Expand("MARK", 3, $0)
if (!EX[3]) {
if (EX[1] ~ " ") EX[3]=substr(EXX, 1, index(EXX, " ")-1)
else EX[3]=EX[1]
}
$0=sprintf("%s<%s>%s</%s>%s", EXX, EX[1], EX[2], EX[3], EXY)
}
}
if (DL) {ProcDL()}
if ($0=="" && TEXT && !LIST) {
print "<P>\n"
TEXT=""
}
if ($0 ~ /^\./) {
A=substr($0, 2)
X[1]=substr(A, 1, index(A" ", " ")-1)
X[2]=substr(A, index(A, " ")+1)
if (X[2]==0) X[2]=""
if (X[1]=="TS") {TS=X[2]; $0=""}
if (X[1]=="TH") {TH=X[2]; $0=""}
if (X[1]=="NOTH") {NOTH=X[2]; $0=""}
if (X[1]=="TABLEHEADER") {TABLEHEADER=X[2]; $0=""}
if (X[1]=="TI") {TI=X[2]; $0=""}
if (X[1]=="TBLANK") {TBLANK=X[2]; $0=""}
if (X[1]=="TTAG") {TTAG=1; $0=""}
if (X[1]=="CAPTION") {CAPTION=X[2]; $0=""}
if (X[1]=="LANG") {LANG=X[2]; $0=""; SetCharset(LANG)}
if (X[1]=="CHARSET") {CHARSET=X[2]; $0=""}
if (X[1]=="DESC") {DESC=X[2]; $0=""}
if (X[1]=="KW") {KW=X[2]; $0=""}
if (X[1]=="BODY") {BODY=X[2]; $0=""}
if (X[1]=="NOMENU") {MENU=0; $0=""}
if (X[1]=="MENU") {
MENU=X[2]
PrintMenuTop(MENU)
$0=""
}
if (X[1]=="SC") {SC[2]=X[2]; $0=""}
if (X[1]=="SSC") {SC[3]=X[2]; $0=""}
if (X[1]=="SSSC") {SC[4]=X[2]; $0=""}
if (X[1]=="ADDRESS") {ADDRESS=X[2]; $0=""}
if (X[1]=="NAME") {NAME=X[2]; $0=""}
if (X[1]=="FOOTERFILE") {FOOTERFILE=X[2]; $0=""}
if (X[1]=="OUT") {OUT=X[2]; $0=""}
if (X[1]=="LISTTYPE") {LISTTYPE=X[2]; $0=""}
if (X[1]=="LI") {LI=X[2]; $0=""}
if (X[1]=="DL") {print "<DL>"; DL=1; $0=""}
if (X[1]=="HREF") {HREF=X[2]; $0=""}
if (X[1]=="INCLUDE") {# kanji code should be euc-jis
INCLUDE=Gsub(X[2])
while ((getline < INCLUDE) > 0) Proc()
$0=""
}
# for @xxx --> yyy substitutions
if (X[1]=="BASE") {BASE=X[2]; $0=""}
if (X[1]=="ROOT") {ROOT=X[2]; $0=""}
# if (X[1]=="") =X[2]
}
if ($0 && PREAMBLE) {PREAMBLE=0}
if (TI) {
if (PREAMBLE || NR==1) {
PrintHeader(TI)
# PREAMBLE=0
if (MENU) PrintMenuTop(MENU)
}
else PrintSection(TI, 1)
TI=""
}
if (SC[2]) { PrintSection(SC[2], 2); delete SC}
if (SC[3]) { PrintSection(SC[3], 3); delete SC}
if (SC[4]) { PrintSection(SC[4], 4); delete SC}
if ($0=="" && LIST) {
LIST=0
printf("</%s>\n\n", LISTTYPE)
}
if ($0 !~ TS && TABLE) {
TABLE=0
print "</TABLE>"
}
if ($0 ~ TS) Table()
if (substr($0, 1, 1)==LI) {
$0=substr($0, 2)
if (LIST==0) printf("<%s>\n", LISTTYPE)
LIST=1
# printf("\t<LI>%s</LI>\n", $0)
printf("\t<LI>%s\n", $0)
$0=""
}
if (OUT && $0) {
print
if (!LIST) TEXT=1
}
}
function Table() {
n=split($0, X, TS)
if (substr($0, 1, 1)==TH || !TABLE) {
if (substr($0, 1, 1)==TH) X[1]=substr(X[1], 2)
if (TABLE) print "</TABLE>"
print TABLEHEADER
if (CAPTION) {printf("\t<CAPTION>%s<CAPTION>\n", CAPTION); CAPTION=""}
TABLE=n
SEP = "TH"
if (NOTH) SEP="TD"
for (i in X) {
TALIGN[i]=""
if (substr(X[i], 1, 1) == "=") {
TALIGN[i] = "center";
X[i] = substr(X[i], 2)
}
else {
if (substr(X[i], 1, 1) == ">") {
TALIGN[i] = "right"
X[i] = substr(X[i], 2)
}
}
}
}
else SEP = "TD"
print "\t<TR>"
for (i=1; i<=TABLE; i++) {
gsub("^[ \t]+", "", X[i])
gsub("[ \t]+$", "", X[i])
if (X[i]=="") X[i]=TBLANK
BSEP=SEP
if (i==1 && TTAG) {
NTAG=X[i]
gsub("[ \t]+", "_", NTAG)
gsub("/", "%2E", NTAG)
if (NTAGS ~ ":"NTAG":") NTAG=sprintf("%s%03i", NTAG, NR)
if (NTAG ~ /[^*+\-.,0-9A-Z_a-z]/) NTAG=sprintf("%s%03i", "t", NR)
NTAGS=NTAGS":"NTAG":"
printf("\t\t<A NAME=\"%s\"> </A>\n", NTAG)
}
if (TALIGN[i]) BSEP=sprintf("%s ALIGN=\"%s\"", SEP, TALIGN[i])
if (i<=n) printf("\t\t<%s>%s</%s>\n", BSEP, X[i], SEP)
else printf("\t\t<%s>%s</%s>\n", BSEP, TBLANK, SEP)
}
print "\t</TR>"
$0=""
}
]]></plaintext></span>