Add hyperlink references

This requires two passes, the first done by aux/emcollect,
the second by aux/emparse (the main awk script).

1 files changed, 14 insertions, 143 deletions

diff --git a/em b/em
index 010fa63..dc8684e 100755
--- a/em
+++ b/em
@@ -1,150 +1,21 @@
-#!/bin/awk -f
+#!/bin/rc
 
 # em -- limited hypertext markup language
 
-function newblock(name) {
-	openblock = name
-	openitem = 0
-	expectblock = 0
-	printf "<%s>", name
+if(~ $#* 0){
+	ramfs -p
+	file=/tmp/file
+	cat > $file
 }
+if not
+	file=$1
 
-function closeblock(name) {
-	if (name == "nl") printf "</ol>\n"
-	else printf "</%s>\n", name
+linktext=link
+hrefs=`'
+'{aux/emcollect $file}
+for(href in $hrefs){
+	parts=`'='{echo $href}
+	ref$parts(1)=$parts(2)
 }
 
-function breakblock() {
-	if (openitem) {
-		printf "</li>\n"
-		while (itemlevel-- > 1)
-			closeblock(openblock)
-		itemlevel = 1
-	}
-	if (opendef) printf "</dd>\n"
-	if (openblock) closeblock(openblock)
-	openitem = 0
-	opendef = 0
-	openblock = 0
-	expectblock = 1
-}
-
-function heading(level, line) {
-	sub("^=* ", "", line)
-	sub(" =*$", "", line)
-	printf "<h%d>%s</h%d>\n", level, line, level
-	# should inline formatting be supported in headings?
-}
-
-function item(level, type, line) {
-	if (openitem) printf "</li>"
-	openitem = 1
-	if (level > itemlevel) printf "<ul>\n"
-	if (level < itemlevel) printf "</ul>\n"
-	itemlevel = level
-	if (type == "ul") {
-		sub("^ +- ", "")
-		printf "<li>"
-	}
-	if (type == "ol") {
-		match($0, "[0-9]+\.")
-		n = substr($0, RSTART, RLENGTH-1)
-		sub("^ +[0-9]+\. ", "")
-		printf "<li value=\"%s\">", n
-	}
-	if (type == "nl") {
-		match($0, "\[[0-9]+\]")
-		n = substr($0, RSTART+1, RLENGTH-2)
-		sub("^ +\[[0-9]+\] ", "")
-		printf "<li value=\"%d\" id=\"ref%d\">", n, n
-	}
-}
-
-function term(line, t) { # t is a local variable
-	if (opendef) printf "</dd>"
-	opendef = 1
-	t = $0; d  = $0
-	sub("^ ", "", t)
-	sub(": .*$", "", t)
-	sub("^ [^:]+: ", "")
-	printf "<dt>%s</dt><dd>", t
-}
-
-function ref(n) {
-	return "[<a href=\"#ref" n "\">" n "</a>]"
-}
-
-function format(line) {
-	n = split(line, w, "[ 	]")
-	for (i = 0; i <= n; i++) {
-		if (w[i] == "") continue;
-
-		if (match(w[i], "^\[[0-9]+\]$"))
-			w[i] = ref(substr(w[i], 2, RLENGTH-2))
-		else if (match(w[i], "^\[[0-9]+\][.,:;?!]$"))
-			w[i] = ref(substr(w[i], 2, RLENGTH-3)) substr(w[i], RLENGTH)
-
-		if (match(w[i], "^`"))
-			w[i] = "<tt>" substr(w[i], 2)
-		else if (match(w[i], "^\\*"))
-			w[i] = "<i>" substr(w[i], 2)
-		else if (match(w[i], "^_"))
-			w[i] = "<b>" substr(w[i], 2)
-
-		if (match(w[i], "`$"))
-			w[i] = substr(w[i], 1, RSTART-1) "</tt>"
-		else if (match(w[i], "\\*$"))
-			w[i] = substr(w[i], 1, RSTART-1) "</i>"
-		else if (match(w[i], "_$"))
-			w[i] = substr(w[i], 1, RSTART-1) "</b>"
-		else if (match(w[i], "`[.,:;?!]$"))
-			w[i] = substr(w[i], 1, RSTART-1) "</tt>" substr(w[i], RSTART+1)
-		else if (match(w[i], "\\*\.$"))
-			w[i] = substr(w[i], 1, RSTART-1) "</i>" substr(w[i], RSTART+1)
-		else if (match(w[i], "_[.,:;?!]$"))
-			w[i] = substr(w[i], 1, RSTART-1) "</b>" substr(w[i], RSTART+1)
-
-		printf "%s", w[i]
-		if (i < n) printf " "
-	}
-	printf "\n"
-}
-
-BEGIN { expectblock = 1; itemlevel = 1 }
-END { breakblock() }
-
-/^$/							{ breakblock(); getline }
-expectblock && /^  /			{ newblock("table") }
-expectblock && /^ - /			{ newblock("ul") }
-expectblock && /^ [0-9]+\. /		{ newblock("ol") }
-expectblock && /^ \[[0-9]\]+ /	{ newblock("nl") }
-expectblock && /^ .*: /			{ newblock("dl") }
-expectblock && /^	/			{ newblock("pre"); sub("^	", "") }
-expectblock && /^---$/			{ expectblock = 0; printf "<hr/>\n"; next }
-expectblock && /^= .* =$/			{ heading(1, $0); next }
-expectblock && /^== .* ==$/			{ heading(3, $0); next }
-expectblock && /^=== .* ===$/		{ heading(3, $0); next }
-expectblock && /^==== .* ====$/		{ heading(4, $0); next }
-expectblock && /^===== .* =====$/	{ heading(5, $0); next }
-expectblock && /^====== .* ======$/	{ heading(6, $0); next }
-expectblock { newblock("p") }
-
-block = "ul" && /^ - /			{ item(1, "ul", line) }
-block = "ol" && /^ [0-9]+\. /		{ item(1, "ol", line) }
-block = "nl" && /^ \[[0-9]\]+ /		{ item(1, "nl", line) }
-block = "dl" && /^ .*: /			{ term(line) }
-
-block = "ul" && /^  -/			{ item(2, "ul", line) }
-block = "ol" && /^  [0-9]+\./		{ item(2, "ol", line) }
-block = "ul" && /^   -/			{ item(3, "ul", line) }
-block = "ol" && /^   [0-9]+\./		{ item(3, "ol", line) }
-block = "ul" && /^    -/			{ item(4, "ul", line) }
-block = "ol" && /^    [0-9]+\./		{ item(4, "ol", line) }
-block = "ul" && /^     -/			{ item(5, "ul", line) }
-block = "ol" && /^     [0-9]+\./		{ item(5, "ol", line) }
-block = "ul" && /^      -/			{ item(6, "ul", line) }
-block = "ol" && /^      [0-9]+\./		{ item(6, "ol", line) }
-
-block = "pre"  && /^	/	{ sub("^	", ""); printf "%s\n", $0; next }
-
-{ format($0) }
+aux/emparse $file