From dbb52e5c345aeafd3b7a2f142ca6bf2039616574 Mon Sep 17 00:00:00 2001
From: Siraaj Khandkar <siraaj@khandkar.net>
Date: Wed, 28 Nov 2018 17:12:13 -0500
Subject: [PATCH] Update the shell-equivalent implementation and motivation

---
 README.md | 37 +++++++++++++++++++++++++++++++------
 1 file changed, 31 insertions(+), 6 deletions(-)

diff --git a/README.md b/README.md
index 979eddc..04b84ec 100644
--- a/README.md
+++ b/README.md
@@ -6,16 +6,18 @@ having the same (and non-0) file size and MD5 hash digest.
 
 It is roughly equivalent to the following one-liner:
 ```sh
-find . -type f -exec md5sum '{}' \; | awk '{digest = $1; path = $2; paths[digest, ++count[digest]] = path} END {for (digest in count) {n = count[digest]; if (n > 1) {print(digest, n); for (i=1; i<=n; i++) {print "    ", paths[digest, i]} } } }'
+find . -type f -print0 | xargs -0 -P 6 -I % md5sum % | awk '{digest = $1;  sub("^" $1 " +", ""); path = $0; paths[digest, ++cnt[digest]] = path} END {for (digest in cnt) {n = cnt[digest]; if (n > 1) {print(digest, n); for (i=1; i<=n; i++) {printf "    %s\n", paths[digest, i]} } } }'
 ```
 
 which, when indented, looks like:
 ```sh
-find . -type f -exec md5sum '{}' \; \
+find . -type f -print0 \
+| xargs -0 -P $(nproc) md5sum \
 | awk '
     {
         digest = $1
-        path = $2
+        sub("^" $1 " +", "")
+        path = $0
         paths[digest, ++count[digest]] = path
     }
 
@@ -25,15 +27,38 @@ find . -type f -exec md5sum '{}' \; \
             if (n > 1) {
                 print(digest, n)
                 for (i=1; i<=n; i++) {
-                    print "    ", paths[digest, i]
+                    printf "    %s\n", paths[digest, i]
                 }
             }
         }
     }'
 ```
 
-and works well-enough, until you start getting weird file paths that are more
-of a pain to handle quoting for than re-writing this thing in OCaml :)
+and works well-enough, but is painfully slow (for instance, it takes around 8
+minutes to process my home directory, whereas `dups` takes around 8 seconds).
+
+Originally, my main motivation for rewriting the above script in OCaml was
+simply to avoid dealing with file paths containing newlines and spaces (the
+original rewrite was substantially simpler than it currently is).
+
+I since realized that, on the _input_, the problem is avoided by delimiting the
+found paths with the null byte, rather than a newline and in AWK doing an
+ostensible `shift` of the `$0` field (`sub("^" $1 " +", "")`).
+
+However, on the _output_, I still don't know of a _simple_ way to escape the
+newline in AWK (in OCaml, there's the `%S` in `printf` and in GNU `printf`
+there's `%q`).
+
+In any case, I now have 2 other reasons to continue with this project:
+1. The speed-up is a boon to my UX (thanks in large part to optimizations
+   suggested by @Aeronotix);
+2. I plan to extend the feature set, which is just too-unpleasant to manage in
+   a string-oriented PL:
+    1. byte-by-byte comparison of files that hash to the same digest, to make
+       super-duper sure they are indeed the same and do not just happen to
+       collide;
+    2. extend the metrics reporting;
+    3. output sorting options.
 
 Example
 -------
-- 
2.20.1