day 04: add scripts to find incoming links

i said no automation, but this is somewhat innocent, i guess?

1 files changed, 43 insertions, 0 deletions

diff --git a/lib/inc.awk b/lib/inc.awk
new file mode 100644
index 0000000..b6978bc
--- /dev/null
+++ b/lib/inc.awk
@@ -0,0 +1,43 @@
+# usage: awk -f inc.awk *.html
+
+function nodot(name) {
+		if(name ~ /^\.\//){
+			name=substr(name,3);
+		}
+		return name;
+}
+
+BEGIN {
+	for(i=1;i<ARGC;i++){
+		files[nodot(ARGV[i])] = 1;
+	}
+}
+
+/<title>/ {
+	match($0, /<title>.*<\/title>/);
+	t=substr($0,RSTART+7,RLENGTH-7-8);
+	match(t, / — /);
+	title[nodot(FILENAME)]=substr(t,1,RSTART-1);
+}
+
+/<a href=/ {
+	if(match($0, /<a href="[a-zA-Z0-9].*\.html">/)){
+		l=substr($0,RSTART+9,RLENGTH-9-2);
+		incoming[l][nodot(FILENAME)]=1;
+		if(!(l in files)) {
+			printf("warning: redlink '%s' in '%s'\n", l, nodot(FILENAME));
+		}
+	}
+}
+
+END {
+	for(f in incoming){
+		split(f,fbase,".");
+		fname=fbase[1] "-inc.htm";
+		print "<ul>" > fname;
+		for(l in incoming[f]) {
+			printf("\t<li><a href=\"%s\">%s</a></li>\n",l,title[l]) >> fname;
+		}
+		printf "</ul>\n" >> fname;
+	}
+}