Shanduro dzave kuwanikwa - Sarudza mutauro waunoda kubva kumenyu.
Maitiro Atakashandisa OpenAI API Kuti Tishandure Webhusaiti Yedu

Maitiro Atakashandisa OpenAI API Kuti Tishandure Webhusaiti Yedu

Nhanganyaya

Pakatanga kugadzira webhusaiti yedu inobva paGoHugo.io kuti ive nemitauro yakawanda, taida nzira inoshanda, inogadziriswa, uye inodhura kuti tigadzire shanduro. Panzvimbo pekushandura peji rega rega nemaoko, takashandisa OpenAI’s API kuti tiite kuti maitiro aya aite otomatiki. Chinyorwa ichi chinotsanangura maitiro atakabatanidza OpenAI API neHugo, tichishandisa theme yeHugoPlate kubva kuZeon Studio, kuti tigadzire shanduro nekukurumidza uye nemazvo.

Nei Takasarudza OpenAI API Kuti Tishandure

Masevhisi echinyakare ekushandura anowanzoda kuedza kwakakura kwekumanikidza, uye maturusi akaita seGoogle Translate, kunyange achibatsira, haasi nguva dzose anopa chiyero chekuvandudza chatinoda. OpenAI’s API yakatibvumira:

  • Kugadzirisa shanduro dzakawanda
  • Kugadzirisa maitiro ekushandura
  • Kudzora mhando zviri nani
  • Kubatanidza pasina kupindira newebhusaiti yedu inobva paHugo
  • Kuisa mapeji ega ega kuti atongororwe zvakare

Maitiro Ekutevera

1. Kugadzirira Webhusaiti yeHugo

Webhusaiti yedu yakatogadzirwa ichishandisa theme yeHugoPlate, iyo inotsigira basa remitauro yakawanda. Danho rekutanga raive rekugonesa tsigiro yemutauro mufaira redu reHugo config/_default/languages.toml:

################ English language ##################
[en]
languageName = "English"
languageCode = "en-us"
contentDir = "content/english"
weight = 1

################ Arabic language ##################
[ar]
languageName = "العربية"
languageCode = "ar"
contentDir = "content/arabic"
languageDirection = 'rtl'
weight = 2

Iyi configuration inoitira kuti Hugo ikwanise kugadzira shanduro dzemutauro dzakasiyana dzezviri mukati medu.

2. Kugadzirisa Shanduro neOpenAI API

Takagadzira Bash script kuti tiite kuti shanduro dzeMarkdown mafaira dziite otomatiki. Iyi script:

  • Inoverenga mafaera eChirungu .md kubva mudhairekitori remidzi.
  • Inoshandisa OpenAI API kuti ishande shanduro yechinyorwa ichichengetedza formatting yeMarkdown.
  • Inonyora zviri kushandurwa kumadhirekitori emutauro akakodzera.
  • Inotora chiyero chekuvandudza kwekushandura ichishandisa faira reJSON.

Heano maonero e script yedu:

#!/bin/bash
# ===========================================
# Hugo Content Translation and Update Script (Sequential Processing & New-Language Cleanup)
# ===========================================
# This script translates Hugo Markdown (.md) files from English to all supported target languages
# sequentially (one file at a time). It updates a JSON status file after processing each file.
# At the end of the run, it checks translation_status.json and removes any language from
# translate_new_language.txt only if every file for that language is marked as "success".
# ===========================================

set -euo pipefail

# --- Simple Logging Function (writes to stderr) ---
log_step() {
    echo "[$(date '+%Y-%m-%d %H:%M:%S')] $*" >&2
}

# --- Environment Setup ---
export PATH="/opt/homebrew/opt/coreutils/libexec/gnubin:$PATH"
# (Removed "Script starting." log)

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
log_step "SCRIPT_DIR set to: $SCRIPT_DIR"

if [ -f "$SCRIPT_DIR/.env" ]; then
    log_step "Loading environment variables from .env"
    set -o allexport
    source "$SCRIPT_DIR/.env"
    set +o allexport
fi

# Load new languages from translate_new_language.txt (if available)
declare -a NEW_LANGUAGES=()
if [ -f "$SCRIPT_DIR/translate_new_language.txt" ]; then
    while IFS= read -r line || [[ -n "$line" ]]; do
        NEW_LANGUAGES+=("$line")
    done <"$SCRIPT_DIR/translate_new_language.txt"
else
    log_step "No new languages file found; proceeding with empty NEW_LANGUAGES."
fi

API_KEY="${OPENAI_API_KEY:-}"
if [ -z "$API_KEY" ]; then
    log_step "❌ Error: OPENAI_API_KEY environment variable is not set."
    exit 1
fi

# Supported Languages (full list)
SUPPORTED_LANGUAGES=("ar" "bg" "bn" "cs" "da" "de" "el" "es" "fa" "fi" "fr" "ha" "he" "hi" "hr" "hu" "id" "ig" "it" "ja" "ko" "ml" "mr" "ms" "nl" "no" "pa" "pl" "pt" "ro" "ru" "sk" "sn" "so" "sr" "sv" "sw" "ta" "te" "th" "tl" "tr" "uk" "vi" "xh" "yo" "zh" "zu")

STATUS_FILE="$SCRIPT_DIR/translation_status.json"
SRC_DIR="$SCRIPT_DIR/Content/english"
log_step "Source directory: $SRC_DIR"

# Check dependencies
for cmd in jq curl; do
    if ! command -v "$cmd" >/dev/null 2>&1; then
        log_step "❌ Error: '$cmd' is required. Please install it."
        exit 1
    fi
done

MAX_RETRIES=5
WAIT_TIME=2 # seconds

# Create/initialize status file if missing
if [ ! -f "$STATUS_FILE" ]; then
    echo "{}" >"$STATUS_FILE"
    log_step "Initialized status file at: $STATUS_FILE"
fi

# --- Locking for Status Updates ---
lock_status() {
    local max_wait=10
    local start_time
    start_time=$(date +%s)
    while ! mkdir "$STATUS_FILE.lockdir" 2>/dev/null; do
        sleep 0.01
        local now
        now=$(date +%s)
        if ((now - start_time >= max_wait)); then
            log_step "WARNING: Lock wait exceeded ${max_wait}s. Forcibly removing stale lock."
            rm -rf "$STATUS_FILE.lockdir"
        fi
    done
}

unlock_status() {
    rmdir "$STATUS_FILE.lockdir"
}

update_status() {
    local file_path="$1" lang="$2" status="$3"
    lock_status
    jq --arg file "$file_path" --arg lang "$lang" --arg status "$status" \
        '.[$file][$lang] = $status' "$STATUS_FILE" >"$STATUS_FILE.tmp" && mv "$STATUS_FILE.tmp" "$STATUS_FILE"
    unlock_status
}

# --- Translation Function ---
translate_text() {
    local text="$1" lang="$2"
    local retry_count=0
    while [ "$retry_count" -lt "$MAX_RETRIES" ]; do
        user_message="Translate the following text to $lang. Preserve all formatting exactly as in the original.
$text"
        json_payload=$(jq -n \
            --arg system "Translate from English to $lang. Preserve original formatting exactly." \
            --arg user_message "$user_message" \
            '{
                "model": "gpt-4o-mini",
                "messages": [
                    {"role": "system", "content": $system},
                    {"role": "user", "content": $user_message}
                ],
                "temperature": 0.3
            }')
        response=$(curl -s https://api.openai.com/v1/chat/completions \
            -H "Content-Type: application/json" \
            -H "Authorization: Bearer $API_KEY" \
            -d "$json_payload")
        log_step "📥 Received API response."
        local error_type
        error_type=$(echo "$response" | jq -r '.error.type // empty')
        local error_message
        error_message=$(echo "$response" | jq -r '.error.message // empty')
        if [ "$error_type" == "insufficient_quota" ]; then
            sleep "$WAIT_TIME"
            retry_count=$((retry_count + 1))
        elif [[ "$error_type" == "rate_limit_reached" || "$error_type" == "server_error" || "$error_type" == "service_unavailable" ]]; then
            sleep "$WAIT_TIME"
            retry_count=$((retry_count + 1))
        elif [ "$error_type" == "invalid_request_error" ]; then
            return 1
        elif [ -z "$error_type" ]; then
            if ! translated_text=$(echo "$response" | jq -r '.choices[0].message.content' 2>/dev/null); then
                return 1
            fi
            if [ "$translated_text" == "null" ] || [ -z "$translated_text" ]; then
                return 1
            else
                translated_text=$(echo "$translated_text" | sed -e 's/^```[[:space:]]*//; s/[[:space:]]*```$//')
                echo "$translated_text"
                return 0
            fi
        else
            return 1
        fi
    done
    return 1
}

# --- Process a Single File (Sequential Version) ---
process_file() {
    local src_file="$1" target_file="$2" lang="$3" rel_src="$4"
    # If target file exists and is non-empty, mark status as success.
    if [ -s "$target_file" ]; then
        update_status "$rel_src" "$lang" "success"
        return 0
    fi
    content=$(<"$src_file")
    if [[ "$content" =~ ^(---|\+\+\+)[[:space:]]*$ ]] && [[ "$content" =~ [[:space:]]*(---|\+\+\+\+)[[:space:]]*$ ]]; then
        front_matter=$(echo "$content" | sed -n '/^\(---\|\+\+\+\)$/,/^\(---\|\+\+\+\)$/p')
        body_content=$(echo "$content" | sed -n '/^\(---\|\+\+\+\)$/,/^\(---\|\+\+\+\)$/d')
    else
        front_matter=""
        body_content="$content"
    fi
    log_step "Translating [$rel_src] to $lang..."
    translated_body=$(translate_text "$body_content" "$lang")
    if [ $? -ne 0 ]; then
        update_status "$rel_src" "$lang" "failed"
        return 1
    fi
    mkdir -p "$(dirname "$target_file")"
    if [ -n "$front_matter" ]; then
        echo -e "$front_matter
$translated_body" >"$target_file"
    else
        echo -e "$translated_body" >"$target_file"
    fi
    updated_content=$(echo "$content" | sed -E 's/^retranslate:\s*true/retranslate: false/')
    echo "$updated_content" >"$src_file"
    update_status "$rel_src" "$lang" "success"
}

# --- Main Sequential Processing ---
ALL_SUCCESS=true
for TARGET_LANG in "${SUPPORTED_LANGUAGES[@]}"; do
    log_step "Processing language: $TARGET_LANG"
    TARGET_DIR="$SCRIPT_DIR/Content/$TARGET_LANG"
    while IFS= read -r -d '' src_file; do
        rel_src="${src_file#$SCRIPT_DIR/}"
        target_file="$TARGET_DIR/${src_file#$SRC_DIR/}"
        # If file is marked not to retranslate, check that target file exists and is non-empty.
        if ! [[ " ${NEW_LANGUAGES[@]:-} " =~ " ${TARGET_LANG} " ]] && grep -q '^retranslate:\s*false' "$src_file"; then
            if [ -s "$target_file" ]; then
                update_status "$rel_src" "$TARGET_LANG" "success"
            else
                update_status "$rel_src" "$TARGET_LANG" "failed"
            fi
            continue
        fi
        process_file "$src_file" "$target_file" "$TARGET_LANG" "$rel_src"
    done < <(find "$SRC_DIR" -type f -name "*.md" -print0)
done

log_step "Translation run completed."
end_time=$(date +%s)
duration=$((end_time - $(date +%s)))
log_step "Execution Time: $duration seconds"

if [ "$ALL_SUCCESS" = true ]; then
    log_step "🎉 Translation completed successfully for all supported languages!"
else
    log_step "⚠️ Translation completed with some errors."
fi

# --- Clean Up Fully Translated New Languages ---
if [ -f "$SCRIPT_DIR/translate_new_language.txt" ]; then
    log_step "Cleaning up fully translated new languages..."
    for lang in "${NEW_LANGUAGES[@]:-}"; do
        incomplete=$(jq --arg lang "$lang" 'to_entries[] | select(.value[$lang] != null and (.value[$lang] != "success")) | .key' "$STATUS_FILE")
        if [ -z "$incomplete" ]; then
            log_step "All translations for new language '$lang' are marked as success. Removing from translate_new_language.txt."
            sed -E -i '' "/^[[:space:]]*$lang[[:space:]]*$/d" "$SCRIPT_DIR/translate_new_language.txt"
        else
            log_step "Language '$lang' still has incomplete translations."
        fi
    done
fi

3. Kutungamira Mamiriro eShanduro

Kuti tidzivise shanduro dzakawandisa uye titeedzere kufambira mberi, takashandisa faira reJSON (translation_status.json). Iyi script inogadzirisa faira iri mushure mekugadzirisa chinyorwa chimwe nechimwe, ichivimbisa kuti chete zvitsva kana zvakatangwa zviri kushandurwa.

4. Kubata Kwekukanganisa uye API Rate Limits

Takaita kuti kudzokorora uye kubata kukanganisa kuitwe kuti tigadzirise miganhu, kukanganisa kweAPI, uye nyaya dzequota. Iyi script inotora nguva isati yadzozve kana OpenAI API ikadzorera kukanganisa kwakadai serate_limit_reached kana service_unavailable.

5. Kuendesa

Kana zviri kushandurwa zvagadzirwa, kumhanya hugo --minify kunovaka webhusaiti inogadzirwa nemitauro yakawanda, yakagadzirira kuendeswa.

Zvipingamupinyi uye Mhinduro

1. Kururama kweShanduro

Kunyange zvazvo shanduro dzeOpenAI dzakajairika dzichinyatsoita, mamwe mazwi ehunyanzvi anogona kuda kuongororwa nemaoko, asi isu tiri timu yevaviri chete, saka tinotarisira zvakanakisisa. Takagadzirisa prompts kuti tiite kuti mamiriro uye tono zvichengetedzwe.

2. Nyaya dzeFormatting

Markdown syntax dzimwe nguva dzakachinjwa mukushandurwa. Kuti tigadzirise izvi, takawedzera logic yekupedzisira kuti tiite kuti formatting igamuchirwe.

3. API Cost Optimization

Kuti tidzikise mutengo, takaita caching kuti tidzivise kushandura zvakare zviri zvisina kuchinjwa.

4. Kubata Retranslations Zvakanyatsobudirira

Kuti tishandure mapeji akasarudzwa, takawedzera parameter ye retranslate: true kumusoro kwemafaira. Iyi script inoshandura chete mapeji akanyorwa nemuparameter uyu. Izvi zvinotibvumira kugadzirisa shanduro sezvinodiwa pasina kuda kushandura webhusaiti yese zvakare.

Mhedziso

Nekubatanidza OpenAI API neHugo, takaita kuti shanduro yewebhusaiti yedu iite otomatiki tichichengetedza mhando uye kugadzirisa. Maitiro aya akachengetedza nguva, akavimbisa kuenderana, uye akatibvumira kukura pasina kupindira. Kana uri kutsvaga kuita kuti webhusaiti yako yeHugo ive nemitauro yakawanda, OpenAI’s API inopa mhinduro ine simba.

Share This Page: