From 0ff7de22a75c8cf359bf090d11ce598dddf114f1 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Th=C3=A9r=C3=A8se=20Godefroy?= <godef.th@free.fr>
Date: Fri, 15 Aug 2014 01:20:20 +0200
Subject: [PATCH] fr/kitchen/color-wdiff: fix a few bugs, reword the
 explanations.

---
 fr/kitchen/color-wdiff | 200 ++++++++++++++++-------------------------
 1 file changed, 75 insertions(+), 125 deletions(-)

diff --git a/fr/kitchen/color-wdiff b/fr/kitchen/color-wdiff
index 7648a267..9d84eeec 100755
--- a/fr/kitchen/color-wdiff
+++ b/fr/kitchen/color-wdiff
@@ -1,132 +1,82 @@
 #!/bin/bash
 
-# color-wdiff visualizes the differences between 2 versions of the same page.
-# This script was extracted from GNUN's GNUmakefile (function "mark-outdated")
-# and adapted. GNUN (http://www.gnu.org/software/gnun/) is under GPLv3.
-
-# Howto:
-# - The script is in [...]/enc/fr/kitchen.
-# - cd to that directory and run:
-#              ./color-wdiff FILE1 FILE2
-#   (FILE1 and FILE2 are the paths to the files to be compared.)
-# - The diff file is created in the directory where FILE1 resides.
+# NAME
+#    color-wdiff - visualize differences between text files
+
+# SYNOPSIS
+#    color-wdiff FILE0 FILE1
+
+# DEPENDENCIES
+#    wdiff, reformat-html in $HOME/bin.
+
+# DESCRIPTION
+#    1. If the input files are HTML, they are reformatted, to remove
+#       indentation among other things.
+#    2. The markup is inactivated by replacing angle brackets with the
+#       corresponding entities.
+#    3. The files are compared with wdiff, using options which label
+#       insertions and deletions. The labels are HTML tags with specific
+#       classes.
+#    4. The diff is turned into a valid HTML page by adding the required
+#       markup, plus CSS style for the insertion and deletion classes.
 #
-# For example, compare old and new versions of index.t.html:
-#              ./color-diff index.t.html index-new.t.html
-
-# Note: the diff is much easier to use if the HTML is not indented.
-# Reformatting is done by the function "page_reformat", which leaves the
-# original page intact.
-
-
-file=$1
-file1=$2
-diff_file=${file%.html}-diff.html
-
-page_reformat () {
-  #reformats the original pages of emailselfdefense.fsf.org
-
-  cp $1 $1.tmp
-
-  # Remove javascript, which shouldn't be reformatted.
-  sed -i '/jquery-1.11.0.min.js/,$d' $1.tmp
-
-  # Remove leading and trailing spaces/tabs.
-  sed -i 's,\t, ,g' $1.tmp
-  sed -i 's,^ *,,'  $1.tmp
-  sed -i 's, *$,,'  $1.tmp
-
-  # Remove LF after </a>.
-  sed -i '/<\/a>$/ {N; s,<\/a>\n\([^<]\),<\/a>\1,}' $1.tmp
-
-  # One string per paragraph, header or list item.
-  for tag in li p strong a h3; do
-    sed -i "/<$tag[^>]*>$/ {N; s,\\n, ,}" $1.tmp
-  done
-  for tag in a strong; do
-    sed -i "/<\\/$tag>$/ {N; s,\\n, ,}" $1.tmp
-  done
-  # This command may need to be repeated. Adjust the number of repeats. This
-  # could be done by looping back to a sed marker, but a while loop seems
-  # quicker.
-  i=0
-  while (( i < 2 )); do
-    sed -i '/[^<>]$/ {N; s,\([^<>]\)\n,\1 ,}' $1.tmp
-    let i=i+1
-  done
-
-  sed -i '/ \/>$/ {N; s,\( \/>\)\n,\1 ,}' $1.tmp
-  sed -i '/ <a[^>]*>$/ {N; s,\(<a[^>]*>\)\n\([^<]\),\1 \2,}' $1.tmp
-
-  # Make sure there is only one paragraph per string. This command may need to
-  # be repeated. Adjust the number of repeats.
-  i=0
-  while (( i < 2 )); do
-    sed -i 's,</p>\(.\+\)$,</p>\n\1,' $1.tmp
-    let i=i+1
-  done
-
-  # Single out the tags which include p (will also work for pre).
-  sed -i 's,\(.\)<p,\1\n<p,' $1.tmp
-
-  # Single-out input meta and link.
-  for tag in input meta link link; do
-    sed -i "s,> <$tag,>\n<$tag," $1.tmp
-  done
-  # Remove leading and trailing spaces, double spaces and blank lines.
-  sed -i 's,^ *,,' $1.tmp
-  sed -i 's, *$,,' $1.tmp
-  sed -i 's,  , ,g' $1.tmp
-  sed -i '/^$/d' $1.tmp
-
-  # Fuse comment with </p>.
-  sed -i '/<\/p>$/ {N;s,\n\(<!-- [^~]\),\1,}' $1.tmp
-
-  # Separate truncated "~~~" comment from fused tag.
-  sed -i 's,~~~[ ]\?[-]\?[-]\?[ ]\?<,~~~\n<,' $1.tmp
-
-  # Fuse header, section and footer with the corresponding div.
-  for tag in header section footer; do
-    sed -i "/^<$tag/ {N; s,\\(<$tag[^>]*>\\)\\n<div>,\\1<div>,}" $1.tmp
-    sed -i "/^<\\/div>$/ {N; s,<\\/div>\\n\\(<\\/$tag>\\),</div>\\1,}" $1.tmp
-  done
-
-  # Add LF before main sections and commented-out parts.
-  sed -i 's,<!-- ~~,\n<!-- ~~,' $1.tmp
-  sed -i '/COMMENTED OUT/ s,^,\n,' $1.tmp
-
-  # Make the text more readable.
-  for tag in p h1 h2 h3 h4 dl title form; do
-    sed -i "s,<$tag,\\n&," $1.tmp
-  done
-  for tag in p dl ul h1 h2 h3 h4 title head footer form script; do
-    sed -i "/<\\/$tag>/s,$,\\n," $1.tmp
-  done
-  sed -i '/<\/dd>/ {N; s,</dd>\n<dt,</dd>\n\n<dt,}' $1.tmp
-  sed -i '/<\/dt>/ {N; s,</dt>\n<dd,</dt>\n\n<dd,}' $1.tmp
-  sed -i 's,</p></span>$,</p>\n</span>,' $1.tmp
-
-  sed -i 's, alt=,\nalt=,g' $1.tmp
-  sed -i 's, | , |\n,g' $1.tmp
-
-  # Remove extra LFs, if any.
-  sed -i ':a /^$/ {N; s,\n$,,; ba}' $1.tmp
-  sed -i ':a /^\n*$/ {$d; N; ba}' $1.tmp
-
-  # Wrap the text.
-  fmt -s -w 95 $1.tmp > $1.r
-}
+#    The diff file is created in the directory where FILE0 resides.
+
+# ORIGIN OF THE SCRIPT
+#    This script was extracted from GNUN's GNUmakefile (function
+#    "mark-outdated"), and adapted.
+#    GNUN (http://www.gnu.org/software/gnun/) is under GPLv3.
+
+# =============================================================================
 
+# Command-line arguments
+arg=($1 $2)
+
+set -e
+
+close_term () {
+  printf '\n%s' '*** Close the terminal window or press Return.'; read OK
+  test -z "$OK" && exit $1
+}
 
-page_reformat ${file}
-# Replace chevrons with HTML entities; the files are treated as simple text.
-sed "s/</\&lt;/g;s/>/\&gt;/g" ${file}.r  > ${file}.tmp
-page_reformat ${file1}
-sed "s/</\&lt;/g;s/>/\&gt;/g" ${file1}.r > ${file1}.tmp
+f[0]=$(mktemp -t cdif.XXXXXX) || close_or_exit 1
+f[1]=$(mktemp -t cdif.XXXXXX) || close_or_exit 1
+trap 'rm -f "${f[0]}" "${f[1]}"' EXIT
+
+## Prepare the pages to be compared.
+
+for n in 0 1; do
+  # Input a valid file.
+  input=${arg[$n]}
+  if test ! -f "$input" -o ! -s "$input"; then
+    echo "*** color-wdiff - Please enter file $n."; read input
+    input=${input%\'}; input=${input#\'}
+    test -f "$input" -a -s "$input" \
+    || (echo 1>&2 "!!! This file doesn't exist or is empty."; close_term 1)
+  fi
+
+  # Name the diff after file 0.
+  test "$n" == "0" && diff_file=${input%.html}-diff.html
+
+  # If the file is an HTML but not a diff, process it:
+  if test "${input%.html}" != "$input" -a "${input%-diff.html}" == "$input";
+  then
+    # - Standardize the format for easier reading of the diff.
+    if test -f "$HOME/bin/reformat-html"; then
+      $HOME/bin/reformat-html $input ${f[$n]}
+    else
+      cp $input ${f[$n]}
+    fi
+    # - Replace chevrons with HTML entities. The page becomes simple text.
+    sed -i "s/</\&lt;/g;s/>/\&gt;/g" ${f[$n]}
+  fi
+done
+
+## Build the diff page.
 
 # Add an HTML header to the wdiff output, with style for visualizing the
 # insertions and deletions, and write the title of the page.
-cat > ${diff_file} << EOF
+cat > $diff_file << EOF
 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
     "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
 <!-- Generated by GNUN -->
@@ -147,10 +97,10 @@ wdiff --start-delete '<span class="removed"><del><strong>' \
       --end-delete '</strong></del></span>' \
       --start-insert '<span class="inserted"><ins><em>' \
       --end-insert '</em></ins></span>' \
-      ${file}.tmp ${file1}.tmp >> $diff_file
+      ${f[0]} ${f[1]} >> $diff_file || true
 
 # Add the closing tags.
-echo '</pre></body></html>' >> $diff_file
+echo '</pre></body></html>' >> ${diff_file}
 
-# Clean up.
-rm -f ${file}.tmp ${file1}.tmp ${file}.r ${file1}.r
+echo -e "\n    The diff file is $diff_file."
+close_term 0
-- 
2.25.1