wiki/src/contribute/l10n_tricks/language_statistics.sh

   1 #!/bin/bash
   2 # -*- mode: sh; sh-basic-offset: 4; indent-tabs-mode: nil; -*-
   3 # vim: set filetype=sh sw=4 sts=4 expandtab autoindent:
   4
   5 set -eu
   6 set -o pipefail
   7
   8 LANGUAGES=${@:-de fr pt}
   9
  10 count_msgids () {
  11     cat | grep -E '^msgid\s+' | wc -l
  12 }
  13
  14 count_original_words () {
  15     cat | grep ^msgid | sed 's/^msgid "//g;s/"$//g' | wc -w
  16 }
  17
  18 count_translated_words () {
  19     cat | grep ^msgstr | sed 's/^msgstr "//g;s/"$//g' | wc -w
  20 }
  21
  22 statistics () {
  23     PO_MESSAGES="$(mktemp -t XXXXXX.$lang.po)"
  24     msgcat --files-from=$PO_FILES --output=$PO_MESSAGES
  25     TOTAL=$(msgattrib --no-obsolete $PO_MESSAGES | count_msgids)
  26     TOTAL_WC=$(
  27         msgattrib --no-obsolete --no-wrap $PO_MESSAGES | count_original_words
  28     )
  29     FUZZY=$(msgattrib --only-fuzzy --no-obsolete $PO_MESSAGES | count_msgids)
  30     TRANSLATED=$(
  31         msgattrib --translated --no-fuzzy --no-obsolete $PO_MESSAGES \
  32             | count_msgids
  33     )
  34     TRANSLATED_WC=$(
  35         msgattrib --translated --no-fuzzy --no-obsolete --no-wrap $PO_MESSAGES \
  36             | count_translated_words
  37     )
  38     echo "  - $lang: $(($TRANSLATED*100/$TOTAL))% ($TRANSLATED) strings translated, $(($FUZZY*100/$TOTAL))% strings fuzzy, $(($TRANSLATED_WC*100/$TOTAL_WC))% words translated"
  39     rm -f $PO_FILES $PO_MESSAGES
  40 }
  41
  42 # sanity checks
  43
  44 if pwd | grep -qs 'wiki/src$' ; then
  45     WEBSITE_ROOT_DIR='.'
  46 elif [ -d '.git' ] ; then
  47     WEBSITE_ROOT_DIR='wiki/src'
  48 else
  49     echo >&2 "Error: $(basename $0) is meant to be run either from the wiki/src directory,"
  50     echo >&2 "       or from the root of the source tree"
  51     exit 1
  52 fi
  53
  54 # all PO files
  55 echo "All website PO files"
  56 echo "===================="
  57 echo ""
  58
  59 for lang in $LANGUAGES ; do
  60     PO_FILES="$(mktemp -t XXXXXX.$lang)"
  61     find "$WEBSITE_ROOT_DIR" -iname "*.$lang.po" > $PO_FILES
  62     find "$WEBSITE_ROOT_DIR" -path "*/locale/$lang/LC_MESSAGES/*.po" >> $PO_FILES
  63     statistics $PO_FILES
  64 done
  65
  66 # core PO files
  67 echo ""
  68 echo "[[Core PO files|contribute/l10n_tricks/core_po_files.txt]]"
  69 echo "=========================================================="
  70 echo ""
  71
  72 for lang in $LANGUAGES ; do
  73     PO_FILES="$(mktemp -t XXXXXX.$lang)"
  74     cat "$WEBSITE_ROOT_DIR"/contribute/l10n_tricks/core_po_files.txt \
  75         | sed "s/$/.$lang.po/g" \
  76         | sed "s,^,$WEBSITE_ROOT_DIR/," \
  77         > $PO_FILES
  78     statistics $PO_FILES
  79 done