From a03e011345413a9b3749deb9da895bb4e8ec9974 Mon Sep 17 00:00:00 2001
From: Tom Ryder <tom@sanctum.geek.nz>
Date: Sat, 3 Dec 2016 02:02:16 +1300
Subject: Rewrite unf(1df) in awk

Also update manual page references to it, and improve it by making it
ignore body content
---
 ISSUES.markdown    |  2 --
 bin/unf.awk        | 34 ++++++++++++++++++++++++++++++++++
 bin/unf.sed        | 17 -----------------
 man/man1/unf.1df   |  8 +++++---
 man/man1/urlh.1df  |  2 +-
 man/man1/urlmt.1df |  2 +-
 6 files changed, 41 insertions(+), 24 deletions(-)
 create mode 100644 bin/unf.awk
 delete mode 100644 bin/unf.sed

diff --git a/ISSUES.markdown b/ISSUES.markdown
index 19177f71..6808fab3 100644
--- a/ISSUES.markdown
+++ b/ISSUES.markdown
@@ -24,5 +24,3 @@ Known issues
     13:33:32 <tejr> yep it does
 
 *   dr(1df) is probably more practical in awk
-*   unf(1df) doesn't work on NetBSD sed(1); might be time to give up and
-    rewrite it in awk
diff --git a/bin/unf.awk b/bin/unf.awk
new file mode 100644
index 00000000..22a10aa8
--- /dev/null
+++ b/bin/unf.awk
@@ -0,0 +1,34 @@
+# Unfold header lines in an internet message, don't touch the body
+
+# Function to write and empty the buffer
+function wrbuf() {
+    if (length(buf))
+        print buf
+    buf = ""
+}
+
+# Flag to stop processing once we hit the first blank line
+!length {
+    wrbuf() 
+    body = 1
+}
+body {
+    print
+    next
+}
+
+# Write any buffer contents once we hit a line not starting with a space
+/^[^ \t]/ {
+    wrbuf()
+}
+
+# Append the current line to the buffer
+{
+    sub(/^[ \t]+/, " ")
+    buf = buf $0
+}
+
+# Write the buffer out again when we hit the end
+END {
+    wrbuf()
+}
diff --git a/bin/unf.sed b/bin/unf.sed
deleted file mode 100644
index 23270fc6..00000000
--- a/bin/unf.sed
+++ /dev/null
@@ -1,17 +0,0 @@
-# Unfold lines with leading spaces (e.g. RFC 822 headers)
-/^[ \t]/!{
-  1!{
-    x
-    p
-    x
-  }
-  h
-}
-/^[ \t]/{
-  H
-  x
-  s/[\r\n]//g
-  x
-}
-$!d
-x
diff --git a/man/man1/unf.1df b/man/man1/unf.1df
index b74ef9ce..288ddf5f 100644
--- a/man/man1/unf.1df
+++ b/man/man1/unf.1df
@@ -1,7 +1,7 @@
 .TH UNF 1df "August 2016" "Manual page for unf"
 .SH NAME
 .B unf
-\- join lines with leading spaces to the previous line
+\- join header lines in an internet message
 .SH USAGE
 .B unf
 FILE
@@ -16,7 +16,9 @@ curl -I http://www.example.net/ |
 .B
 unf
 .SH DESCRIPTION
-Joins lines with leading spaces in the output to the previous line,
-such as folded header fields in RFC 822 format messages.
+Joins long lines denoted with leading spaces in the folded headers of an RFC
+822 style internet message, such as an email message or an HTTP response.
+.SH SEE ALSO
+urlmt(1df)
 .SH AUTHOR
 Tom Ryder <tom@sanctum.geek.nz>
diff --git a/man/man1/urlh.1df b/man/man1/urlh.1df
index d17f0a74..8eeb359a 100644
--- a/man/man1/urlh.1df
+++ b/man/man1/urlh.1df
@@ -12,6 +12,6 @@ makes a cURL HEAD request for the given URL, and searches the headers for a key
 matching the given name, case-insensitively. It prints any matching values to
 stdout.
 .SH SEE ALSO
-curl(1), urlmt(1df)
+curl(1), unf(1df), urlmt(1df)
 .SH AUTHOR
 Tom Ryder <tom@sanctum.geek.nz>
diff --git a/man/man1/urlmt.1df b/man/man1/urlmt.1df
index 892b10d4..cbc31bed 100644
--- a/man/man1/urlmt.1df
+++ b/man/man1/urlmt.1df
@@ -10,6 +10,6 @@ https://www.sanctum.geek.nz/
 uses urlh(1df) to search for a Content-Type header for the given URL, and prints
 it with any trailing data (e.g. charset) trimmed off.
 .SH SEE ALSO
-curl(1), urlh(1df)
+curl(1), unf(1df), urlh(1df)
 .SH AUTHOR
 Tom Ryder <tom@sanctum.geek.nz>
-- 
cgit v1.2.3