delay for out-of-date mark; GRACE computation bugs

[Ineiev]

It seems to be necessary to introduce another grace period,
a delay for out-of-date mark; it will default to 60 days.

When writing this patch, I've noticed that the GRACE period is measured
from the latest POT creation time rather than from the latest
full translation; the patch fixes it, too.

Another change is to make the detection of incomplete PO files more
precise: the earlier revisions relied on presence of fuzzy strings
alone, which produced wrong results when a new passage was added
without modification of any old ones.

The new revision maintains and depends on the presence of extracted
dates for the latest full translation in ${file}.po-latest. in order
to initially generate it, I'm going to use a script like the attached.

The script will also be used to revert the HTMLs of outdated translations
to the latest full translation, which was planned ages ago.

Index: GNUmakefile
===================================================================
RCS file: /sources/trans-coord/trans-coord/gnun/server/gnun/GNUmakefile,v
retrieving revision 1.85
diff -U 2 -r1.85 GNUmakefile
--- GNUmakefile 30 Aug 2011 14:59:44 -0000      1.85
+++ GNUmakefile 31 Aug 2011 13:36:08 -0000
@@ -39,4 +39,7 @@
                    -o "translated=W<pre>"
 
+# Period to delay notice about translation being out of date
+OUTDATED-GRACE := 60
+
 # Determine the VCS to use.  The conditional is for efficiency --
 # there's no need to spawn a shell to determine the value when all VCS
@@ -175,9 +178,12 @@
 endef
 
+define po-is-outdated
+($(MSGFMT) --statistics -o /dev/null $(1) 2>&1 \
+  | $(EGREP) '(fuzzy|untranslated)' > /dev/null)
+endef
+
 # Add notice about the translation being out of date
 define mark-outdated
-  $(MSGFMT) --statistics -o /dev/null $$$$PO 2>&1 \
-    | ($(EGREP) '(fuzzy|untranslated)' > /dev/null \
-       && $(SED) --in-place \
+  $(SED) --in-place \
             "/^<!--#set var=\"PO_FILE\"/,+3d;\
 /^<!--#include virtual=\"\/server\/banner/a\
@@ -185,6 +191,6 @@
     value='<a href=\"http://www.gnu.org$$$${PO#$(rootdir)}\">\n\
            http://www.gnu.org$$$${PO#$(rootdir)}</a>' -->\n\
-<!--#include virtual=\"/server/outdated.$(2).html\" -->" $(1); \
-      ) || true
+<!--#include virtual=\"/server/outdated.$(2).html\" -->" $(1) \
+   || true
 endef
 
@@ -526,5 +532,6 @@
          $(rootdir)/po/home.$(1).m4
        $(M4) $(rootdir)/po/home.$(1).m4 > $$@
-       PO=$(rootdir)/po/home.$(1).po; $(call mark-outdated,$$@,$(1))
+       PO=$(rootdir)/po/home.$(1).po; $(call po-is-outdated,$$$$PO) \
+         && $(call mark-outdated,$$@,$(1))
        $(VALIDATESKIP) $(VALIDATE-HTML-NOTIFY) $(NOTIFYSKIP) $(transl-addr) \
          $$@ || (PO=$(rootdir)/po/home.$(1).po ; $$(touch-hook))
@@ -562,4 +569,12 @@
 endef
 
+define extract-pot-date
+$(AWK) '/POT-Creation-Date/ { gsub(/-/, ""); print $$$$2; exit 0; }' $(1)
+endef
+
+define grace-is-over
+test "`date --date="$(2) days ago" +%Y%m%d`" -ge "`cat $(1)-latest`"
+endef
+
 define article-rules
 $(1).po: $(basename $(1)).pot
@@ -624,6 +639,13 @@
 # If GRACE is not defined, which is the usual case for local manual
 # builds, update the target and validate the result.
+       rm -f $(1).mark; $(call po-is-outdated,$(1).po) && touch $(1).mark
+# Save the date of the date of the latest POT file corresponding to full
+# translation (or the date for the current POT if we have never seen
+# a full translation)
+       if ! test -f $(1).po-latest || ! test -f $(1).mark; then \
+         $(call extract-pot-date,$(1).po) > $(1).po-latest; \
+       fi
 ifndef GRACE
-       $(output-article-html);
+       $(output-article-html)
 # Check if the article is not in `no-grace-articles'.
 else ifneq ($(basename $(1)), \
@@ -631,16 +653,16 @@
 # If there are no fuzzy strings, there might be untranslated or
 # obsolete, so proceed as usual.  If there are fuzzy strings, compare
-# the POT-Creation-Date in the PO with the current value of the grace
-# period and invoke regeneration only if the grace period is over.
+# the POT-Creation-Date in the $(1).po-latest with the current value
+# of the $(GRACE) period and invoke regeneration only if the grace period
+# is over. Likewise, check against $(OUTDATED-GRACE) and insert a notice
+# into the HTML file when the grace period is over.
 # If the target is of zero size, it means that it was just touched by
 # $(addfile) -- GRACE is ignored in this case to avoid committing
 # empty files to the repository.
-       if ! $(GREP) --quiet "^#, fuzzy" $(1).po || ! test -s $$@; \
+       if test -f $(1).mark || ! test -s $$@; \
          then \
            $(output-article-html); \
          else \
-         if [ "`date --date="$(GRACE) days ago" +%Y%m%d`" -ge \
-              "`$(AWK) '/POT-Creation-Date/ { gsub(/-/, ""); print $$$$2 }' \
-                $(1).po`" ]; \
+         if $(call grace-is-over,$(1).po,$(GRACE)) ; \
            then \
              $(output-article-html); \
@@ -649,9 +671,11 @@
          fi; \
        fi
+       $(call grace-is-over,$(1).po,$(OUTDATED-GRACE)) || rm -f $(1).mark
 else
        @echo 'Ignoring grace period for article "$(notdir $(basename $(1)))"'
-       $(output-article-html);
+       $(output-article-html)
 endif
-       PO=$(1).po; $(call mark-outdated,$$@,$(2))
+       PO=$(1).po; test -f $(1).mark && $(call mark-outdated,$$@,$(2))
+       rm -f $(1).mark
 endef
 

#! /bin/bash

log_file=${HOME}/logs/find-latest.log
export log_file

function extract-pot-date {
  awk '/POT-Creation-Date/ { gsub(/-/, ""); print $2; exit 0; }' \
    $1 > $1-latest
  awk '/PO-Revision-Date/ { gsub(/-/, ""); print $2; exit 0; }' \
    $1 > $1-latest-po
  echo ${maior}.${minor} >> $1-latest-po
  cvs log $1 2>/dev/null | sed '/^revision *'"${maior}.${minor}"'$/ 
{N;s/.*date:\([^;]*\);.*/\1/;q};d' >> $1-latest-po
}

function check-translation {
  if msgfmt --statistics -o /dev/null $1 2>&1 \
    | egrep '(fuzzy|untranslated)' > /dev/null; then
    return 1; 
  else
    if test "x$2" = xskip;then
      return 0;
    fi
    extract-pot-date $1
    echo PO rev: $2 >> ${log_file}
    cvs log $html 2>/dev/null | awk -v po_time="$(cat $1-latest-po | tail -n 
1)"  '
BEGIN {
  gsub(/:/, " ", po_time);
  sub(/-/, " ", po_time); sub(/-/, " ", po_time);
  po_seconds=mktime(po_time);
  revision="";
}
/^----------------------------$/ {
  getline;
  if ($0 !~ /^revision /)
    next;
  next_revision=$2;
  getline;
  if ($0 !~ /^date: /)
    next;
  next_html_time=$2 " " $3 " " substr($4,1,5);
  gsub(/:/, " ",next_html_time);
  sub(/-/, " ",next_html_time);
  sub(/-/, " ",next_html_time);
  next_html_seconds=mktime(next_html_time)
  print next_html_time " " next_html_seconds " " po_time " " po_seconds;
  if (po_seconds < next_html_seconds + 600 || length(revision) == 0)
    {
      if (length (revision) == 0 && po_seconds > next_html_seconds + 600)
        system("echo HTML seems to have never been regenerated >> ${log_file}");
      revision=next_revision;
      html_seconds=next_html_seconds;
      html_time=next_html_time;
      next;
    }
  exit 0;
}
END {
  print revision;
  system("echo -n PO: " po_time ", HTML: " html_time " >> ${log_file}");
  system("echo -n , DIFF: " html_seconds-po_seconds " >> ${log_file}");
  system("echo , HTML rev: " revision " >> ${log_file}");
}
' 
    return 0;
  fi
}

if ! test -f $1;then exit 1;fi

html=$(echo $1 | sed 's \<po/  ;s/\.po$/.html/')
echo ${html} >> ${log_file}

cvs update -AC $1
if check-translation $1 skip; then
  echo $1 up-to-date
  echo up-to-date >> ${log_file}
  exit 0;
fi

head=`cvs log $1  2>/dev/null | sed "/^head/{s/head://;q};d"`
minor=${head##*.}
maior=${head%.*}
echo head ${head}

while test ${minor} -gt 0; do
  echo checking ${maior}.${minor}...
  cvs update -C -r ${maior}.${minor} $1
  check-translation $1 ${maior}.${minor} && exit 0
  minor=$((${minor}-1))
done
echo no full translations found | tee ${log_file}
cvs update -C -r ${head} $1
exit 1;