thea2Shattering_i18n_tools/compatibility-tools/convert_memoq_to_po.sh

#!/usr/bin/env bash

# This script export a multi-lanbguage RTF export from memoQ to a gettext .po

if [[ "$#" -ne 1 ]]; then
    echo "Please pass exactly 1 parameter: the rtf to convert"
    exit 1
fi

ORIGINAL_RTF="$1"
WORKING_RTF="$1.work"
PO_DIR="$(pwd)/"
TEMP_DIR="$(mktemp -d)"
PO=$(basename "${ORIGINAL_RTF%%.txt_fre.rtf}.po")
TEMP_PO="${TEMP_DIR}/${PO}"
PO="${PO_DIR}/${PO}"

cat "${ORIGINAL_RTF}" > "${WORKING_RTF}"

dos2unix "${WORKING_RTF}" &> /dev/null

#Remove all unused lines
sed -i '/brdrcf17/d' "${WORKING_RTF}"
sed -i '/ltrpar/d' "${WORKING_RTF}"
sed -i '/fs16/d' "${WORKING_RTF}"
sed -i '/\\row/d' "${WORKING_RTF}"
sed -i  's/\\\\//g' "${WORKING_RTF}"
sed -i  's/{\\rtlch\\fcs1\ \\ltrch\\fcs0//g' "${WORKING_RTF}"
sed -i  's/\\noproof \\cell }//g' "${WORKING_RTF}"
sed -i  's/\\cell }//g' "${WORKING_RTF}"
sed -i 's/{\\plain\\noproof\ \\cs99\\f0\\fs20\\cf13\ \\{MQ\\}}//g' "${WORKING_RTF}"
#Remove useless translations of EVENT, NODE, STORY
sed -i '/\lang1036 -- \[EVENT\]/d' "${WORKING_RTF}"
sed -i '/\lang1036 +\[NODE\]/d' "${WORKING_RTF}"
sed -i '/\lang1036 \[STORY\]/d' "${WORKING_RTF}"
sed -i '/\lang1036 \[\/EVENT\]/d' "${WORKING_RTF}"
sed -i '/\lang1036 \[\/NODE\]/d' "${WORKING_RTF}"
sed -i '/\lang1036 \[\/STORY\]/d' "${WORKING_RTF}"
# Remove headers
sed -i  '1,23d' "${WORKING_RTF}"

# Quick function to deal with \uc0\uXXX char in rtf format, as iconv and other tool cannot detect them...
# In addition to that, rtf adds a useless space after each encoded char that need to be corrected manually
function convert_char_to_utf8 {
    # Convert all unbreakable spaces in the line
    temp_line="${line//'\~'/' '}"
    i=$(echo "${temp_line}" | grep "uc0")
    while [[ "$i" != "" ]]
    do
        # Extract the utf16 code eg: \uc0\u171 -> 171
        temp=$(echo "${temp_line#*'\uc0\u'}")
        code=$(echo "${temp%%' '*}")
        # Convert the code to hexa eg: 171 -> 0x00ab
        char=$(printf "0x%04x" ${code})
        # Convert to unicode signal (first 0x should be \u) eg: 0x00ab -> \u00ab (substitution) -> « (echo -ne)
        # |tr -d '\0' remove a warning introduced in bash 4.4, see https://lists.gnu.org/archive/html/bug-bash/2016-09/msg00015.html
        char=$(echo -ne "${char/0x/'\u'}" |tr -d '\0')
        # Insert the new char in the line
        temp_line=$(echo "${temp_line}" | sed 's/\\uc0\\u'"$code"' /'"$char"'/g')
        # In case the char is end of line, insert with no space
        temp_line=$(echo "${temp_line}" | sed 's/\\uc0\\u'"$code"'/'"$char"'/g')
        i=$(echo "${temp_line}" | grep "uc0")
    done

    echo ${temp_line}
}

comment_event=""
comment_node=""
comment_story=""
comment_out=""
msgstr=""
msgid=""
out_incr=1
story_first_line=0
while read -r line; do
    if  [[ ${line} == *"-- [EVENT]"* ]]; then
        comment_event="${line}"
    elif [[ ${line} == *"+[NODE]"* ]]; then
        comment_node="${line}"
    elif [[ ${line} == *"[STORY]"* ]]; then
        comment_story="${comment_event}@@${comment_node}@@${line}"
        # Clean strings
        comment_story="${comment_story//'\noproof '/}"
        comment_story="${comment_story//'\cell }'/}"
        # Time for new story, write the comment to pot
        echo "#. ${comment_story}" | tr -d '\r' | tr -d '\n' >> "${TEMP_PO}"
        echo >> "${TEMP_PO}"
        story_first_line=1
    elif [[ ${line} == *"[/STORY]"* ]]; then
        # we reach end of story, insert the translated msgstr and reinit
        msgstr="${msgstr//'\lang1036 '/}"
        msgstr="${msgstr//'\cell }'/}"
        # If there is still lang1036 then it means the msgstr is untranslated, insert an empty translation
        if [[ ${msgstr} == *"lang1036"* ]]; then
            echo "msgstr \"\"" >> "${TEMP_PO}"
            echo >> "${TEMP_PO}"
        else
            echo "msgstr \"\"${msgstr}" >> "${TEMP_PO}"
            echo >> "${TEMP_PO}"
        fi
        comment_story=""
        msgstr=""
    elif [[ ${line} == *"\noproof [OUT]"* ]]; then
        # It's an out for english part (with noproof) so write the comment_out
        comment_out="${comment_event}@@${comment_node}@@[OUT]${out_incr}"
        comment_out="${comment_out//'\noproof '/}"
        comment_out="${comment_out//'\cell }'/}"
        # OUT are one line, let's write it to the po
        echo "#. ${comment_out}" | tr -d '\r' | tr -d '\n' >> "${TEMP_PO}"
        echo >> "${TEMP_PO}"
        line="${line//'\cell }'/}"
        line="${line//'\noproof '/}"
        # Get rid of special character
        line="${line//'"'/'\"'}"
        echo "msgid \"${line#\[OUT\]}\"" | tr -d '\r' | tr -d '\n' >> "${TEMP_PO}"
        echo >> "${TEMP_PO}"
        # Insert an empty msgstr in case of untranslated [OUT]
        # (in this case it never matches the \lang1036 [OUT] below so we need to insert it outside)
        echo "msgstr \"\"" >> "${TEMP_PO}"
        echo >> "${TEMP_PO}"
        ((out_incr++))
    elif [[ ${line} == *"\lang1036 [OUT]"* ]]; then
        # It's an out for translated part (with lang1036) so write the msgstr.
        line="${line//'\cell }'/}"
        line="${line//'\lang1036 '/}"
        # Get rid of special character
        line="${line//'"'/'\"'}"
        encoded_line=$(convert_char_to_utf8 "${line}")
        # Insert without the msgstr as it has been inserted with the msgid before.
        echo "\"${encoded_line#\[OUT\]}\"" | tr -d '\r' | tr -d '\n' >> "${TEMP_PO}"
        echo >> "${TEMP_PO}"
    elif [[ ${line} == *"[/NODE]"* ]]; then
        # We reach end of node, reinit
        comment_node=""
        comment_story=""
        comment_out=""
        out_incr=1
    elif [[ ${line} == *"[/EVENT]"* ]]; then
        # We reach end of event, reinit
        comment_event=""
        comment_node=""
        comment_story=""
        comment_out=""
        out_incr=1
    elif [[ "${comment_story}" != "" && ${story_first_line} == 1 ]]; then
        # As developer sometimes forgot empty lines at the end of STORY, skip empty lines
        if [[ "${line}" = "" ]]; then
            continue
        fi
        # If the line has nothing particular, and the comment_story is not empty, it's a story string
        # As the story_fist_line is set, it's the first line of the story, insert the msgid
        line="${line//'\cell }'/}"
        # Get rid of special character
        line="${line//'"'/'\"'}"
        echo "${line//'\noproof '/'msgid "'}\"" | tr -d '\r' | tr -d '\n' >> "${TEMP_PO}"
        echo >> "${TEMP_PO}"
        story_first_line=0
    elif [[ ${comment_story} != "" ]]; then
        # As developer sometimes forgot empty lines at the end of STORY, skip empty lines
        if [[ "${line}" = "" ]]; then
            continue
        fi
        # If it contains lang1036 it's a translated line, keep it for later
        if [[ $(echo "${line}" | grep "lang1036") ]]; then
            # Get rid of special character
            line="${line//'"'/'\"'}"
            encoded_line=$(convert_char_to_utf8 "${line}")
            # We need a newline between each translated lines
            msgstr="${msgstr}
\"${encoded_line}\n\""
            continue
        fi
        # Also as this is multiline, we need to add \n to the previous line (no way to detect it sooner...)
        sed -e '$s/\(.*\)"$/\1\\n"/' -i "${TEMP_PO}"
        line="${line//'\cell }'/}"
        line="${line//'\noproof '/}"
        # Get rid of special character
        line="${line//'"'/'\"'}"
        echo "\"${line}\"" | tr -d '\r' | tr -d '\n' >> "${TEMP_PO}"
        echo >> "${TEMP_PO}"
    else
        # It should be a newline, write a newline
        # If new case are added later, we need to deal with it manually, do not break the file by writing them
        echo "" >> "${TEMP_PO}"
    fi

done < "${WORKING_RTF}"

# Insert default po header
echo 'msgid ""
msgstr ""
"MIME-Version: 1.0\n"
"Content-Transfer-Encoding: 8bit\n"
"Content-Type: text/plain; charset=UTF-8\n"
"Project-Id-Version: \n"
"POT-Creation-Date: \n"
"PO-Revision-Date: \n"
"Last-Translator: \n"
"Language-Team: \n"' | cat - "${TEMP_PO}" > "${TEMP_PO}.tmp" && mv "${TEMP_PO}.tmp" "${TEMP_PO}"

# Unify duplicate due to pot syntax
msguniq --no-wrap "${TEMP_PO}" > "${PO}"