卷曲从 Smartling 下载所有文件

Question

我想编写一个简单的解决方案来备份我们目前在 smarting 上拥有的所有字符串文件。这包括任何语言环境文件。

目前我还有许多其他问题，但主要目标是下载所有文件。

单个文件的当前curl为：

curl -d "apiKey=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxx
&fileUri=%2Fpath%2Fto%2Ffolder%2Fstrings.xml
&projectId=1234567890" "https://api.smartling.com/v1/file/get/"

这将为我提供一个 strings.xml 文件。我可以使用&locale=DE 进一步定义语言环境，以获得与翻译相同的文件。

我可以使用通配符下载所有 *.xml 文件吗？

Answer 1

Smartling 有一个与 Smrtling API 交互的 shell 脚本，可以执行此类操作（以及更多）。假设您要下载所有已翻译文件以及已发布的翻译文件，其命令如下所示：

./download-smartling-files.sh -t published -a xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxx -p 1234567890 -l "nl-NL ru-RU"

确保使用您为 Smartling 项目配置的语言环境更新 locales -l 参数。

如果您想下载所有文件，包括任何和所有待处理的翻译以及已发布的翻译，那么该命令将类似于

./download-smartling-files.sh -t pending -a xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxx -p 1234567890 -l "nl-NL ru-RU"

这两个命令都将与 Smartling API 交互以获取当前上传到 Smartling 的所有文件的列表，然后遍历它们，将翻译后的文件下载到 translate/[locale] 目录

您可以通过脚本执行许多其他操作，包括下载所有翻译的 TMX 文件。要了解如何执行此操作以及其他操作，只需运行

./download-smartling-files.sh

另一种选择是使用Smartling Maven Plugin。这通常用于将本地化整合到您的构建过程中，并且您希望在构建和部署事件期间上传和/或下载文件。

这里是下载-smartling-files.sh 脚本

#!/bin/bash
# author: Eric Negron (enegron@smartling.com)
# last updated February 20, 2014

shopt -s nullglob


function delete_files () {

    # print the files first to make damn sure they know what they are deleting
    COUNT=1
    for CURRENT_FILE in ${uris[@]}
    do
        echo $'\n'$COUNT" "$CURRENT_FILE
        ((COUNT++))
        done

    echo "Are you SURE you want to delete all "${#uris[@]}" files?"

    # confirm by getting the user to enter 'YES' to input, and if YES then delete these files
    COUNT=1
    read delete_confirmation
    if [ $delete_confirmation != "YES" ]; then
        echo "sorry, must answer YES to delete. exiting"
    exit

    else
        for CURRENT_FILE in ${uris[@]}
        do
            echo $'\n'$COUNT" "$CURRENT_FILE
            curl -sS -d "apiKey=$SL_APIKEY&projectId=$SL_PROJECT&fileUri=$CURRENT_FILE" "https://$SERVER_URL/v1/file/delete"
            ((COUNT++))
        done
    fi
}

function usage {
cat << EOF
    This script uses the Smartling API
    OPTIONS:
    -h      Show this message
    -t (REQUIRED) download type. Acceptable types: original | published | p100 | pseudo | pending | TMX | DELETE
    -a (REQUIRED) API key
    -p (REQUIRED) project ID
    -u (OPTIONAL) URI mask
    -l (OPTIONAL) locales list - a qouted array of Smartling Locales. E.G. "es-ES ja-JP"
    -x (OPTIONAL) TMX download type: full (default) | published - only valid with -t TMX
    -d (OPTIONAL) when downloading translations or TMX set this option to false to append filename with locale instead of creating locale folders. Not valid with -t original option.
    -E (OPTIONAL) DELETE files -E must be true and type = DELETE and must use -u MASK option
EOF
}

# hardcode to regular API, not sandbox
SERVER_URL=api.smartling.com

#opt t
DL_TYPE=

#opt a
SL_APIKEY=

#opt p
SL_PROJECT=

#opt u
URI_MASK=

#opt u
LOCALES=

#opt x
TMX_FLAG="full"

#opt d
USE_LOCALE_DIR="true"

#opt E
DELETE_CONFIRM="false"

while getopts "ht:a:p:u:l:x:d:E:" OPTION
do
    case $OPTION in
        h)
            usage
            exit 1
            ;;
        t)
            DL_TYPE=$OPTARG
            ;;
        a)
            SL_APIKEY=$OPTARG
            ;;
        p)
            SL_PROJECT=$OPTARG
            ;;
        u)
            URI_MASK=$OPTARG
            ;;
        l)
            LOCALES=($OPTARG)
            ;;
        x)
            TMX_FLAG=$OPTARG
            ;;
        d)
            USE_LOCALE_DIR=$OPTARG
            ;;
        E)
            DELETE_CONFIRM=$OPTARG
            ;;

    esac
done

# make sure the required paramaters are set to something
# TODO check if getops handles this natively in some way
if [ "$DL_TYPE" == "" ] || [ "$SL_APIKEY" == "" ] || [ "$SL_PROJECT" == "" ]; then
    usage
    exit
fi

# make sure download type is valid
if [ "$DL_TYPE" != "original" ]\
 && [ "$DL_TYPE" != "published" ]\
 && [ "$DL_TYPE" != "p100" ]\
 && [ "$DL_TYPE" != "pseudo" ]\
 && [ "$DL_TYPE" != "pending" ]\
 && [ "$DL_TYPE" != "TMX" ]\
 && [ "$DL_TYPE" != "DELETE" ]; then
echo "INVALID TYPE. Acceptable types: original | published | p100 | pseudo | pending | TMX.   Exiting."
exit
fi

# if user has specified mask confirm it - note mask means nothing to TMX
if [ "$URI_MASK" != "" ]; then
    if [ $DL_TYPE == "TMX" ]; then
        echo "Mask option has no effect when downloading TMX. Ignoring."
        URI_MASK=""
    else
        echo "mask was set!"$'\n'
    fi
fi

IFS=$'\n'

#make sure there are no API configuration errors by doing a file/list call just to check for error conditions
errors=($(curl -sS -d "apiKey=$SL_APIKEY&projectId=$SL_PROJECT&uriMask=$URI_MASK" "https://$SERVER_URL/v1/file/list" |  grep -Eo "ERROR"))
error_count=${#errors[@]}$'\n'
if [ $error_count -gt 0 ]; then
    echo "error with either the API key or Project ID - verify your values."
    exit
fi

# figure out how many total files there - because if more than standard 500 returned in list, then need to paginate to build the list of uris
# need to move this check since if downloading TMX it's possible there are no files.
total_files=($(curl -sS -d "apiKey=$SL_APIKEY&projectId=$SL_PROJECT&uriMask=$URI_MASK" "https://$SERVER_URL/v1/file/list" |  grep -Eo "fileCount\":[0-9]+" | sed -e 's/fileCount\"://g'))


# The /file/list API has a limit of 500 items, if the total needed is more, need to make multiple calls to build the full URI list

listLimit=500

# figure out how many passes in batches of 500 needed

let "passes=$total_files/$listLimit"

# if more than 1 batch needed, then make as many passes as needed to build the list

pass=0

while [ $pass -le $passes ]; do

let "passOffset=$pass*$listLimit"

    uris+=($(curl -sS -d "apiKey=$SL_APIKEY&projectId=$SL_PROJECT&uriMask=$URI_MASK&offset=$passOffset" "https://$SERVER_URL/v1/file/list" |  grep -Eo "fileUri\":\"[^\"]*" | sed -e 's/fileUri":"//g'))
    ((pass++))

done

# exit if requesting to download files, but nothing matched at all, otherwise show the count of files that matched and the names
# TODO: refactor since TMX handling is different with getopts

file_count=${#uris[@]}
if [ $file_count = "0" ] && [ $DL_TYPE != "TMX" ]; then echo "nothing to operate on, exiting!"; exit; fi

# if deleting check all the parameters set and if so pass of URI mask list to delete function
if [ "$DL_TYPE" == "DELETE" ]; then
    if [ "$DELETE_CONFIRM" == "true" ] && [ "$URI_MASK" != "" ]; then
        delete_files $uris
        exit
    else
        echo "Must set -u to a uri mask value and must set -E true to confirm deletion. Exiting"
        exit
    fi
fi

# if not downloading TMX - then list the files we found
echo "total files to download "$total_files$'\n'
if [ $DL_TYPE != "TMX" ]; then echo "URIs: "${uris[@]}$'\n'; fi


#BEGIN ORIGINALS
if [ "$DL_TYPE" == "original" ] ; then
    echo "downloading originals"
    echo "files to download: "$file_count$'\n'

# create the originals folder if it doesn't exist
    if [ ! -d "originals" ]; then
        echo "making originals folder"$'\n'
        mkdir originals
    fi

    COUNT=0
    # go through the list of files to download (in each language)
    for CURRENT_FILE in ${uris[@]}
    do
        echo $CURRENT_FILE
        ((COUNT++))
        echo $COUNT

        # since the URI includes full URI including default smartling prefix '/files/' or any other prefix specified - strip that and just use the last part after the last /
        # e.g. if the URI is /files/filename.ext then BF_NAME will be filename.ext - this what we use for the local file name (in the folder)
        BF_NAME=${CURRENT_FILE##*/}
        # This is curl call that downloads the originals. Since no LOCALE is set that is the behavior of /file/get

        # becuase we are flattening the URIs to basename it's possible filenames could be duplicated - so to avoid this check if the file already exist and if it does use count to append name to create unique
        # TODO USE COUNT
        if [ -e originals/$BF_NAME ]; then
            BF_NAME=$COUNT.$BF_NAME
            echo "Updated BF_NAME"
        fi

        curl -sS -d "apiKey=$SL_APIKEY&projectId=$SL_PROJECT&fileUri=$CURRENT_FILE" "https://$SERVER_URL/v1/file/get" > originals/$BF_NAME
        echo "downloaded URI: "$CURRENT_FILE" to: originals/"$BF_NAME$'\n'

    done
exit
fi
#end of originals


# if not getting originals - then getting translations OR TMX - so get locales and then use that to download all
# get the list of locales for this project.  Similar to above grep but the locale key value then sed for just the value

# if user has not set the locales array in the call, then get the full list via API
# TODO - validate the locales are good - otherwise user will just get empty files
if [ "$LOCALES" = "" ]; then
    LOCALES=($(curl -sS -d "apiKey=$SL_APIKEY&projectId=$SL_PROJECT" "https://$SERVER_URL/v1/project/locale/list" |  grep -Eo "locale\":\"[^\"]*" | sed -e 's/locale":"//g'))
fi

echo "locales: "${#LOCALES[@]}$'\n'
# count of the locales
echo ${LOCALES[@]}$'\n'

#script puts the translated' versions in 'translated folder' with sub-folders for locales
if [ ! -d "translated" ]; then
    echo "making translated folder"$'\n'
    mkdir translated
fi

if [ $USE_LOCALE_DIR == "true" ]; then
#if [ "$USE_LOCALE_DIR" == "true" ]; then
    # make the subfolders in the translated folder if they don't exist
    for CURRENT_LOCALE in ${LOCALES[@]}
        do
        if [ ! -d translated/$CURRENT_LOCALE ]; then
            echo "making "$CURRENT_LOCALE" folder" $'\n'
            mkdir translated/$CURRENT_LOCALE
        fi
    done
fi

#TODO need to refactor TMX / MASK / PUBLISHED / FULL
# Download TMX
# Downloading TMX uses a different API than downloading tranlated files
if [ "$DL_TYPE" == "TMX" ] ; then
    echo "downloading TMX "$TMX_FLAG
    for CURRENT_LOCALE in ${LOCALES[@]}
    do
        # if using folders - set the folder locale
        # else using file-names so set the filename locale
        if [ "$USE_LOCALE_DIR" == "true" ]; then
            LOCALE_DIR=$CURRENT_LOCALE"/"
            FILE_LOCALE="TMX-"
        else
            FILE_LOCALE=$CURRENT_LOCALE"-"
            LOCALE_DIR=""
        fi

        curl -sS -d "apiKey=$SL_APIKEY&projectId=$SL_PROJECT&locale=$CURRENT_LOCALE&format=TMX&dataSet=$TMX_FLAG" "https://$SERVER_URL/v1/translations/download" > translated/$LOCALE_DIR$FILE_LOCALE"TMX.xml"
        echo "downloaded " $CURRENT_LOCALE "TMX" $'\n'
    done

    exit
fi


# downloading translations of files
# by this point the file list has already been filtered and the and the locale list has been created
# go through the list of files to download (in each language)

COUNT=0
P100=0

# since I use DL_TYPE as actual parameter to API call need to set this back to actual "published" before entering loop to download
if [ $DL_TYPE == "p100" ]; then
    DL_TYPE="published"
    P100=1
fi


for CURRENT_FILE in ${uris[@]}
do
    echo $CURRENT_FILE
    ((COUNT++))
    echo $COUNT
    # loop through all the locales for each file

    for CURRENT_LOCALE in ${LOCALES[@]}
    do

    # if using folders - set the folder locale
    # else using file-names so set the filename locale
    if [ $USE_LOCALE_DIR == "true" ]; then
        LOCALE_DIR=$CURRENT_LOCALE"/"
        FILE_LOCALE=""
    else
        FILE_LOCALE=$CURRENT_LOCALE"-"
        LOCALE_DIR=""
    fi

        # since the URI includes full URI including default smartling prefix '/files/' or any other prefix specified - strip that and just use the last part after the last /
        # e.g. if the URI is /files/filename.ext then BF_NAME will be filename.ext - this what we use for the local file name (in the locale folder)
        BF_NAME=${CURRENT_FILE##*/}

        EXT=${BF_NAME##*.}
        # rename .pot files to .po - we really should be checking the header from the /file/get API call but this is just simpler
        if [ "$EXT" == "pot" ]; then
            filename=${BF_NAME%.*}
            BF_NAME=$filename".po"
        echo "renamed pot to po"
        fi

        # because we are flattening the URIs to basename it's possible filenames could be duplicated - so to avoid this check if the file already exist and if it does use COUNT to append name to create unique
        if [ -e translated/$LOCALE_DIR$FILE_LOCALE$BF_NAME ]; then
            BF_NAME=$COUNT.$BF_NAME
            echo "Updated BF_NAME"
        fi

# Here if user is doing 100% published need to test and if it's not then skip it
# probably should do this earlier in this loops to avoid other tests but thare not needed if we are just going to skip the file

if [ $P100 == 1 ]; then

#    echo "requested 100 percent only"

    # use api status call and pull out the part of the response that has the string count and completed string count
    completeStatus=($(curl -sS -d "apiKey=$SL_APIKEY&projectId=$SL_PROJECT&fileUri=$CURRENT_FILE&locale=$CURRENT_LOCALE" "https://$SERVER_URL/v1/file/status" | grep -Eo "stringCount\":[0-9]+.*\"completedStringCount\":[0-9]+" ))

    # strip those out as separate variables to test
    stringCount=($(echo $completeStatus | sed -E 's/stringCount\":([0-9]+).*/\1/g'))
    completedStringCount=($(echo $completeStatus | sed -E 's/.*completedStringCount\":([0-9]+).*/\1/g'))
#echo $stringCount","$completedStringCount

    # set download to 1 only if 100%
    if [ $stringCount == $completedStringCount ]; then
        download=1
        echo $CURRENT_FILE" in "$CURRENT_LOCALE" is 100% complete!"$'\n'

    else
        download=0
        echo "100% complete requested. Skipping URI: "$CURRENT_FILE" in "$CURRENT_LOCALE" total:"$stringCount", completed:"$completedStringCount$'\n'
    fi

else
    # didn't ask for 100% so always download
    download=1

fi


if [ $download == 1 ]; then
        # This is the actual curl command that downloads the given file for given locale, and state as specified when called
        curl -sS -d "apiKey=$SL_APIKEY&projectId=$SL_PROJECT&fileUri=$CURRENT_FILE&locale=$CURRENT_LOCALE&retrievalType=$DL_TYPE" "https://$SERVER_URL/v1/file/get" > translated/$LOCALE_DIR$FILE_LOCALE$BF_NAME
        echo "downloaded URI: "$CURRENT_FILE" to: translated/"$LOCALE_DIR$FILE_LOCALE$BF_NAME$'\n'
fi


    done

done

exit