Removing duplicates from md5sum.txt and computing new MD5 for changed files

test/merge_debian_isos

@@ -434,6 +434,7 @@ do
 
   MOUNT_LIST="$MOUNT_LIST $mount_point"
 done
+echo "Done." >&2
 
 echo >&2
 echo "Copying dists directory and md5sum.txt from first ISO ..." >&2
@@ -473,6 +474,7 @@ else
   echo "--- Could not copy /md5sum.txt from first ISO." >&2
   cleanup_and_end 3
 fi
+echo "Done." >&2
 
 
 ## Helper functions
@@ -516,6 +518,84 @@ extract_release_head() {
   done
 }
 
+# Decide whether to re-compute MD5 and whether to put out the md5sum line
+# Parameters: prev_path prev_md5
+check_and_put_out_md5_grep() {
+
+  if echo "$prev_path" | grep '^\.\/pool\/' >/dev/null 2>&1
+  then
+    echo "$prev_md5  $prev_path"
+    return 0
+  fi
+  if echo "$prev_path" | grep '^\.\/firmware\/' >/dev/null 2>&1
+  then
+    if test "$was_multiple" = 1
+    then
+      # There is the risk that the surviving file does not match prev_md5.
+      # Better omit it.
+      dummy=dummy
+    else
+      echo "$prev_md5  $prev_path"
+    fi
+    return 0
+  fi
+
+  if echo "$prev_path" | grep '^\.\/dists\/' >/dev/null 2>&1
+  then
+    md5_path="$(echo -n "$prev_path" | sed -e 's/^\.\//merged_/')"
+    prev_md5="$(md5sum <"$md5_path" | awk '{print $1}')"
+  elif test "$prev_path" = ./README.txt
+  then
+    prev_md5="$(md5sum <merged_README.txt | awk '{print $1}')"
+  elif test "$was_multiple" = 1
+  then
+    # Unchanged file was taken from iso1
+    # md5sum.txt lines might be permuted though. So compute freshly
+    prev_md5="$(md5sum <"$mount_point_1"/"$prev_path" | awk '{print $1}')"
+  fi
+  if test -z "$prev_md5"
+  then
+    prev_md5="-"
+    return 1
+  else
+    echo "$prev_md5  $prev_path"
+  fi
+
+  return 0
+}
+
+# Put out merged_md5sum.txt while removing duplicates
+# and recomputing MD5 if needed
+polish_md5sum_txt() {
+  prev_path=
+  prev_md5=
+  was_multiple=0
+
+  cat merged_md5sum.txt | \
+  while read md5 path
+  do
+    if test "$path" = "$prev_path"
+    then
+      was_multiple=1
+  continue
+    fi
+    if test -n "$prev_path"
+    then
+      check_and_put_out_md5_grep
+    fi
+
+    prev_path="$path"
+    prev_md5="$md5"
+    was_multiple=0
+  done
+
+  if test -n "$prev_path"
+  then
+    check_and_put_out_md5_grep
+  fi
+
+  return 0
+}
 
 ## Determine which Debian release is on iso1
 echo >&2
@@ -619,12 +699,6 @@ echo "Done." >&2
 echo >&2
 echo "Merging package description files ..." >&2
 
-# /md5sum.txt seems to be the only overall package list
-for i in $MOUNT_LIST
-do
-  cat "$i"/md5sum.txt
-done | sort -k 2 >merged_md5sum.txt
-
 # Determine the files which are mentioned with checksum in main Release files
 path_list=$(for i in $MOUNT_LIST
             do
@@ -677,6 +751,7 @@ continue
   fi
   
 done
+echo "Done." >&2
 
 
 ## Update dists/"$dist"/Release
@@ -720,6 +795,20 @@ mv temp_file merged_dists/"$dist"/Release
 echo "Done." >&2
 
 
+## Merge md5sum.txt and compute the MD5s which might have changed
+echo >&2
+echo "Merging md5sum.txt files ..." >&2
+
+for i in $MOUNT_LIST
+do
+  cat "$i"/md5sum.txt
+done | sort -k 2 >merged_md5sum.txt
+
+polish_md5sum_txt >temp_file
+mv temp_file merged_md5sum.txt
+echo "Done." >&2
+
+
 ## Produce the new ISO image
 echo >&2
 echo "Producing result ISO image ..." >&2