Saka-NLP Web Extractor & Processor

An Apify Actor that uses every feature of Saka-NLP v0.2.6 to extract text from web pages and analyse it with Indonesian and regional-language NLP — plus a plug-and-play LLM integration that works with OpenAI, Gemini, Claude, LM Studio, Ollama, SumoPod, Groq, and any OpenAI-compatible endpoint.

README_ID.md

---

Saka-NLP Features Used

Saka-NLP Module	Function	Description
tokenize / async_tokenize	Tokenization	Split text into tokens; handles URLs, @mentions, #hashtags, emojis
get_token_count	Token count	Count total tokens in a text string
normalize / async_normalize	Slang normalization	Convert informal/slang words to standard Indonesian
parse_transaction_units	Numeric normalization	10k → 10000, 1jt → 1000000
extract_transaction_entities	Transaction NER	Tag numbers as PRICE or QUANTITY from context
demojize	Emoji → text	❤️ → cinta (Indonesian alias)
replace_emoticons	Emoticon → text	:) → Happy face
analyze / async_analyze	Morphological analysis	Root word, prefixes, suffixes, regional lexicon match, compound word splitting
get_stopwords	Multi-language stopwords	id, sunda, jawa, bali, minang, batak, en, jaksel, all
query_kbbi	KBBI dictionary lookup	Fetch word definitions from the official Indonesian dictionary
latin_to_aksara_sunda / aksara_sunda_to_latin	Sundanese script	Bidirectional Latin ↔ Sundanese script transliteration
latin_to_aksara_jawa / aksara_jawa_to_latin	Javanese script	Bidirectional Latin ↔ Hanacaraka (Javanese) transliteration
latin_to_aksara_bali / aksara_bali_to_latin	Balinese script	Bidirectional Latin ↔ Balinese script transliteration
OutputFormatter	Multi-format reports	Render results as markdown, html, csv, table, json, or text

---

LLM Integration (Plug-and-Play)

Connect any LLM provider by simply supplying a provider name and API key. The actor sends the extracted text and NLP results to the LLM for four optional tasks.

Token Optimization

This actor uses four strategies from Saka-NLP to minimize LLM token usage:

Strategy	Saka-NLP API	Token saving
Slang normalization before sending	build_prompt(optimize_text=True)	Removes redundant informal tokens from input
Hard input token cap	build_prompt(max_tokens=N)	Truncates text at exactly N tokens — no wasted budget
Output contract schema	build_prompt(output_contract={...})	LLM receives exact JSON schema → shorter, structured reply, no prose
Batch mode	single call, 4 tasks	All tasks in ONE API call → system prompt + article text sent once
CSV keyword context	OutputFormatter.format(top_words, 'csv')	Keywords as compact CSV string, not verbose JSON
Saka output parser	parse_llm_output(text, 'json')	Robust JSON extraction — no need for LLM to add markdown fences

Batch mode savings (default ON):

4 separate calls:  ~1,400 input tokens  (4× system + 4× article text)
1 batch call:      ~  500 input tokens  (1× system + 1× article text + compact context)
Saving:            ~  65% fewer input tokens

Batch mode can be disabled via llmBatchMode: false if your provider rejects multi-task prompts.

Supported Providers

llmProvider	Endpoint	Default Model
openai	api.openai.com/v1	gpt-4o-mini
gemini	generativelanguage.googleapis.com/v1beta/openai	gemini-2.0-flash
claude	api.anthropic.com/v1	claude-3-5-haiku-20241022
groq	api.groq.com/openai/v1	llama-3.3-70b-versatile
mistral	api.mistral.ai/v1	mistral-small-latest
together	api.together.xyz/v1	meta-llama/Llama-3-8b-chat-hf
openrouter	openrouter.ai/api/v1	openai/gpt-4o-mini
deepseek	api.deepseek.com/v1	deepseek-chat
sumopod	ai.sumopod.com/v1	gpt-4o-mini
lmstudio	localhost:1234/v1	local-model
ollama	localhost:11434/v1	llama3
custom	set via llmBaseUrl	set via llmModel

LLM Tasks

Task	Output key	Description
summary	llm_analysis.llm_summary	Fluent AI-generated article summary
sentiment	llm_analysis.llm_sentiment	Positive / negative / neutral + confidence score
keywords	llm_analysis.llm_keywords	Refined top-10 keywords chosen by the LLM
category	llm_analysis.llm_category	Topic classification (Technology, Politics, Health, etc.)

LLM analysis is fully optional — leave llmApiKey empty to skip it.

Example — SumoPod

curl https://ai.sumopod.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxx" \
  -d '{"model": "gpt-4o-mini", "messages": [{"role":"user","content":"Hello"}]}'

Equivalent actor input:

{
    "url": "https://kompas.com/artikel",
    "llmProvider": "sumopod",
    "llmApiKey": "sk-xxxx",
    "llmModel": "gpt-4o-mini",
    "llmTasks": "summary,sentiment,keywords,category"
}

Example — Custom / Self-hosted

{
    "url": "https://kompas.com/artikel",
    "llmProvider": "custom",
    "llmBaseUrl": "https://my-llm-server.com/v1",
    "llmApiKey": "my-token",
    "llmModel": "my-model",
    "llmTasks": "summary,sentiment"
}

Example — LM Studio (local)

{
    "url": "https://kompas.com/artikel",
    "llmProvider": "lmstudio",
    "llmApiKey": "lm-studio",
    "llmTasks": "summary,keywords"
}

---

Input

Core Parameters

Parameter	Type	Default	Description
url	String	—	Required. URL of the web page to process
lang	String	id	Stopword language: id, sunda, jawa, bali, minang, batak, en, jaksel, all
topN	Integer	20	Number of top words to return (5–100)
summaryLines	Integer	5	Number of sentences in the extractive summary (1–20)
kbbiLookup	Integer	5	Number of top words to look up in KBBI (0 = disabled)
aksaraWords	Integer	5	Number of top words to transliterate into regional scripts (0 = disabled)
outputFormat	String	json	Format for formatted_reports: json, markdown, html, csv, table, text
includeFullText	Boolean	false	Include the full extracted text in the output

LLM Parameters

Parameter	Type	Default	Description
llmProvider	String	openai	LLM provider name (see table above)
llmApiKey	String	—	Bearer token / API key. Leave empty to skip LLM
llmBaseUrl	String	—	Override base URL (required for custom provider)
llmModel	String	—	Model name. Uses provider default when empty
llmTasks	String	all	Comma-separated tasks: summary, sentiment, keywords, category
llmTimeout	Integer	60	Max seconds to wait for LLM response (10–300)
llmBatchMode	Boolean	true	Send all tasks in 1 API call (saves ~65% input tokens)

Full Example Input

{
    "url": "https://www.kompas.com/sains/read/2024/01/01/example-article",
    "lang": "id",
    "topN": 25,
    "summaryLines": 5,
    "kbbiLookup": 5,
    "aksaraWords": 5,
    "outputFormat": "markdown",
    "includeFullText": false,

    "llmProvider": "openai",
    "llmApiKey": "sk-xxxx",
    "llmModel": "gpt-4o-mini",
    "llmTasks": "summary,sentiment,keywords,category",
    "llmTimeout": 60
}

---

Example A — NLP only (no LLM)

{
    "url": "https://www.kompas.com/tekno/read/2024/01/15/kecerdasan-buatan-indonesia",
    "lang": "id",
    "topN": 20,
    "summaryLines": 5,
    "kbbiLookup": 5,
    "aksaraWords": 5,
    "outputFormat": "markdown",
    "includeFullText": false
}

Example B — NLP + OpenAI

{
    "url": "https://www.kompas.com/tekno/read/2024/01/15/kecerdasan-buatan-indonesia",
    "lang": "id",
    "topN": 20,
    "summaryLines": 5,
    "kbbiLookup": 5,
    "aksaraWords": 5,
    "outputFormat": "markdown",
    "includeFullText": false,
    "llmProvider": "openai",
    "llmApiKey": "sk-proj-xxxx",
    "llmModel": "gpt-4o-mini",
    "llmTasks": "summary,sentiment,keywords,category",
    "llmBatchMode": true,
    "llmTimeout": 60
}

Example C — LM Studio (local, no cloud key)

{
    "url": "https://www.kompas.com/tekno/read/2024/01/15/kecerdasan-buatan-indonesia",
    "lang": "id",
    "llmProvider": "lmstudio",
    "llmApiKey": "local",
    "llmTasks": "summary,sentiment"
}

Example D — SumoPod custom endpoint

{
    "url": "https://www.detik.com/finance/berita-ekonomi-bisnis/artikel",
    "lang": "id",
    "topN": 30,
    "llmProvider": "custom",
    "llmBaseUrl": "https://ai.sumopod.com/v1",
    "llmApiKey": "sk-xxxx",
    "llmModel": "gpt-4o-mini",
    "llmTasks": "summary,keywords,category",
    "llmBatchMode": true
}

---

Output

Example A — NLP only (llm_analysis is null)

{
    "url": "https://www.kompas.com/tekno/read/2024/01/15/kecerdasan-buatan-indonesia",
    "domain": "kompas.com",
    "title": "Kecerdasan Buatan Dorong Transformasi Digital Indonesia - Kompas.com",
    "language": "id",
    "text_preview": "Kecerdasan buatan (AI) kini menjadi pendorong utama transformasi digital di Indonesia. Pemerintah menargetkan Indonesia masuk 10 besar negara ekonomi digital dunia pada 2045...",
    "text_normalized": "Kecerdasan buatan sekarang menjadi pendorong utama transformasi digital di Indonesia. Pemerintah menargetkan Indonesia masuk 10 besar negara ekonomi digital dunia pada 2045...",
    "extractive_summary": "Kecerdasan buatan (AI) kini menjadi pendorong utama transformasi digital di Indonesia. Investasi teknologi AI meningkat 340% dalam dua tahun terakhir. Pemerintah menargetkan Indonesia masuk 10 besar ekonomi digital dunia pada 2045.",
    "stats": {
        "char_count": 8420,
        "token_count_raw": 1384,
        "token_count_clean": 847,
        "unique_token_count": 318,
        "sentence_count": 52,
        "avg_sentence_length": 26.6,
        "lexical_density_pct": 37.54,
        "regional_word_count": 14
    },
    "stopword_coverage": {
        "id": 460, "sunda": 58, "jawa": 45, "bali": 47,
        "minang": 40, "batak": 90, "en": 180, "jaksel": 37, "all": 1057
    },
    "tokens_raw_sample": ["Kecerdasan", "buatan", "AI", "kini", "menjadi", "pendorong", "utama", "transformasi"],
    "tokens_clean_sample": ["kecerdasan", "buatan", "pendorong", "transformasi", "digital", "indonesia", "teknologi"],
    "top_words": [
        { "word": "teknologi",    "count": 28 },
        { "word": "digital",      "count": 24 },
        { "word": "kecerdasan",   "count": 19 },
        { "word": "indonesia",    "count": 17 },
        { "word": "transformasi", "count": 14 },
        { "word": "pemerintah",   "count": 11 },
        { "word": "investasi",    "count":  9 },
        { "word": "inovasi",      "count":  7 }
    ],
    "top_bigrams": [
        { "bigram": "kecerdasan buatan",    "count": 17 },
        { "bigram": "transformasi digital", "count": 12 },
        { "bigram": "ekonomi digital",      "count":  7 }
    ],
    "top_trigrams": [
        { "trigram": "era kecerdasan buatan",          "count": 6 },
        { "trigram": "transformasi digital indonesia", "count": 4 }
    ],
    "morphology_sample": [
        { "word": "mempertanggungjawabkan", "root": "tanggung",    "prefixes": ["memper"], "suffixes": ["kan"], "type": "regional", "regional_matches": ["jawa"] },
        { "word": "mentransformasikan",     "root": "transformasi","prefixes": ["men"],    "suffixes": ["kan"], "type": "unknown",  "regional_matches": [] },
        { "word": "menyebarluaskan",        "root": "sebar luas",  "prefixes": ["meny"],   "suffixes": ["kan"], "type": "unknown",  "regional_matches": [] }
    ],
    "top_roots": [
        { "root": "teknologi",    "count": 8 },
        { "root": "digital",      "count": 6 },
        { "root": "transformasi", "count": 4 }
    ],
    "regional_words": [
        { "word": "mangga", "root": "mangga", "regional_matches": ["sunda", "jawa"] },
        { "word": "horas",  "root": "horas",  "regional_matches": ["batak"] }
    ],
    "transaction_analysis": {
        "transaction_normalized_preview": "Investasi senilai 10000000 rupiah dialokasikan untuk 3 proyek startup.",
        "transaction_entities": [
            { "value": 10000000, "type": "PRICE",    "original_token": "10000000" },
            { "value": 3,        "type": "QUANTITY", "original_token": "3" }
        ],
        "price_count": 1,
        "quantity_count": 1
    },
    "kbbi_lookup": [
        { "word": "teknologi", "status": "found",     "definitions": ["metode ilmiah untuk mencapai tujuan praktis"] },
        { "word": "inovasi",   "status": "found",     "definitions": ["pemasukan hal-hal baru", "pembaruan"] },
        { "word": "startup",   "status": "not_found", "definitions": [] }
    ],
    "aksara_transliteration": [
        {
            "word": "teknologi",
            "aksara_sunda": "ᮒᮨᮊ᮪ᮔᮧᮜᮧᮌᮤ", "aksara_sunda_back": "teknologi",
            "aksara_jawa":  "ꦠꦺꦏ꧀ꦤꦺꦴꦭꦺꦴꦒꦶ", "aksara_jawa_back":  "teknologi",
            "aksara_bali":  "ᬢᭂᬓ᭄ᬦᭀᬮᭀᬕᬶ",    "aksara_bali_back":  "teknologi"
        },
        {
            "word": "digital",
            "aksara_sunda": "ᮓᮤᮌᮤᮒᮜ᮪", "aksara_sunda_back": "digital",
            "aksara_jawa":  "ꦢꦶꦒꦶꦠꦭ꧀", "aksara_jawa_back":  "digital",
            "aksara_bali":  "ᬤᬶᬕᬶᬢᬮ᭄",   "aksara_bali_back":  "digital"
        }
    ],
    "formatted_reports": {
        "top_words_markdown": "| word | count |\n|---|---|\n| teknologi | 28 |\n| digital | 24 |\n| kecerdasan | 19 |",
        "top_words_csv":      "word,count\nteknologi,28\ndigital,24\nkecerdasan,19\n",
        "top_words_table":    "word | count\n-----------\nteknologi | 28\ndigital   | 24",
        "morphology_html":    "<table border=\"1\"><tr><th>word</th><th>root</th></tr><tr><td>mempertanggungjawabkan</td><td>tanggung</td></tr></table>",
        "morphology_json":    "[{\"word\":\"mempertanggungjawabkan\",\"root\":\"tanggung\",\"prefixes\":[\"memper\"],\"suffixes\":[\"kan\"]}]"
    },
    "llm_analysis": null
}

Example B — NLP + LLM (llm_analysis populated)

{
    "url": "https://www.kompas.com/tekno/read/2024/01/15/kecerdasan-buatan-indonesia",
    "domain": "kompas.com",
    "title": "Kecerdasan Buatan Dorong Transformasi Digital Indonesia - Kompas.com",
    "language": "id",
    "text_preview": "Kecerdasan buatan (AI) kini menjadi pendorong utama transformasi digital di Indonesia...",
    "text_normalized": "Kecerdasan buatan sekarang menjadi pendorong utama transformasi digital di Indonesia...",
    "extractive_summary": "Kecerdasan buatan (AI) kini menjadi pendorong utama transformasi digital di Indonesia. Investasi teknologi AI meningkat 340% dalam dua tahun terakhir.",
    "stats": {
        "char_count": 8420, "token_count_raw": 1384, "token_count_clean": 847,
        "unique_token_count": 318, "sentence_count": 52,
        "avg_sentence_length": 26.6, "lexical_density_pct": 37.54, "regional_word_count": 14
    },
    "stopword_coverage": { "id": 460, "sunda": 58, "jawa": 45, "all": 1057 },
    "tokens_raw_sample": ["Kecerdasan", "buatan", "AI", "kini", "menjadi"],
    "tokens_clean_sample": ["kecerdasan", "buatan", "transformasi", "digital", "indonesia"],
    "top_words": [
        { "word": "teknologi",  "count": 28 }, { "word": "digital",    "count": 24 },
        { "word": "kecerdasan", "count": 19 }, { "word": "indonesia",  "count": 17 }
    ],
    "top_bigrams":  [{ "bigram": "kecerdasan buatan",    "count": 17 }],
    "top_trigrams": [{ "trigram": "era kecerdasan buatan", "count": 6 }],
    "morphology_sample": [
        { "word": "mempertanggungjawabkan", "root": "tanggung", "prefixes": ["memper"], "suffixes": ["kan"], "type": "regional", "regional_matches": ["jawa"] }
    ],
    "top_roots":      [{ "root": "teknologi", "count": 8 }],
    "regional_words": [{ "word": "mangga", "root": "mangga", "regional_matches": ["sunda", "jawa"] }],
    "transaction_analysis": {
        "transaction_normalized_preview": "Investasi senilai 10000000 rupiah dialokasikan untuk 3 proyek startup.",
        "transaction_entities": [
            { "value": 10000000, "type": "PRICE",    "original_token": "10000000" },
            { "value": 3,        "type": "QUANTITY", "original_token": "3" }
        ],
        "price_count": 1, "quantity_count": 1
    },
    "kbbi_lookup": [
        { "word": "teknologi", "status": "found", "definitions": ["metode ilmiah untuk mencapai tujuan praktis"] },
        { "word": "startup",   "status": "not_found", "definitions": [] }
    ],
    "aksara_transliteration": [
        {
            "word": "teknologi",
            "aksara_sunda": "ᮒᮨᮊ᮪ᮔᮧᮜᮧᮌᮤ", "aksara_sunda_back": "teknologi",
            "aksara_jawa":  "ꦠꦺꦏ꧀ꦤꦺꦴꦭꦺꦴꦒꦶ", "aksara_jawa_back":  "teknologi",
            "aksara_bali":  "ᬢᭂᬓ᭄ᬦᭀᬮᭀᬕᬶ",    "aksara_bali_back":  "teknologi"
        }
    ],
    "formatted_reports": {
        "top_words_markdown": "| word | count |\n|---|---|\n| teknologi | 28 |\n| digital | 24 |",
        "top_words_csv":      "word,count\nteknologi,28\ndigital,24\n",
        "morphology_html":    "<table border=\"1\"><tr><th>word</th><th>root</th></tr><tr><td>mempertanggungjawabkan</td><td>tanggung</td></tr></table>",
        "morphology_json":    "[{\"word\":\"mempertanggungjawabkan\",\"root\":\"tanggung\",\"prefixes\":[\"memper\"],\"suffixes\":[\"kan\"]}]"
    },
    "llm_analysis": {
        "provider_model": "gpt-4o-mini",
        "batch_mode": true,
        "token_usage": {
            "prompt_tokens": 487,
            "completion_tokens": 198,
            "total_tokens": 685,
            "prompt_tokens_estimated": 312
        },
        "llm_summary": {
            "text": "Artikel ini membahas bagaimana kecerdasan buatan menjadi pendorong utama transformasi digital Indonesia. Pemerintah menargetkan Indonesia masuk 10 besar ekonomi digital dunia pada 2045, didukung investasi teknologi yang meningkat 340% dalam dua tahun terakhir. Berbagai startup lokal memanfaatkan AI untuk mengotomatisasi layanan dan meningkatkan efisiensi.",
            "model": "gpt-4o-mini"
        },
        "llm_sentiment": {
            "parsed": { "label": "positive", "score": 0.88 },
            "model": "gpt-4o-mini"
        },
        "llm_keywords": {
            "parsed": {
                "keywords": ["kecerdasan buatan", "transformasi digital", "startup",
                             "investasi teknologi", "ekonomi digital", "inovasi",
                             "otomatisasi", "infrastruktur", "data", "pemerintah"]
            },
            "model": "gpt-4o-mini"
        },
        "llm_category": {
            "parsed": { "label": "Technology", "score": 0.96 },
            "model": "gpt-4o-mini"
        }
    }
}

---

Processing Pipeline

URL Input
  │
  ▼
[1] Fetch + BeautifulSoup         remove nav / footer / script / style
  │  page_title, full_text
  ▼
[2] saka.normalize                slang normalization + 10k → 10000
    saka.demojize                 emoji → Indonesian text alias
    saka.replace_emoticons        :) → Happy face
  │  clean_text
  ▼
[3] saka.async_tokenize           tokenization
    saka.get_stopwords            stopword filter (9 language variants)
  │  tokens_clean
  ▼
[4] Counter + n-gram              word frequency, bigrams, trigrams
[5] saka.async_analyze × N        concurrent morphology: root, affixes, regional dict, compounds
[6] saka.extract_transaction_entities   PRICE / QUANTITY NER
[7] saka.query_kbbi               KBBI dictionary definitions
[8] saka.latin_to_aksara_*        transliterate to 3 regional scripts
    OutputFormatter               render reports in 6 formats
  │
  ▼
[9] LLMClient (plug-and-play)     summary · sentiment · keywords · category
    └─ any OpenAI-compatible API  OpenAI / Gemini / Claude / Groq / SumoPod / ...
  │
  ▼
Apify Dataset (JSON)

---

Display Information

Field	Value
Tagline	Full-stack Indonesian NLP on any web page — powered by Saka-NLP
Categories	Content Moderation, Text Processing, Language Processing
SEO Title	Saka-NLP Web Extractor & Processor — Indonesian NLP Actor
SEO Description	Apify Actor for full Indonesian NLP: tokenization, morphology, slang normalization, KBBI lookup, regional script transliteration, transaction NER, n-grams, multi-format reports, and plug-and-play LLM integration — powered by Saka-NLP v0.2.6.

---

Monetization

This Actor uses a Pay Per Result pricing model.

Field	Value
Pricing model	Pay per result
Price per unit	$0.01 USD
Unit	Per page successfully analysed
Billed when	A run produces at least one dataset item

---

Output Schema

The full schema is defined in .actor/output_schema.json (output). Dataset views are configured inline in .actor/actor.json under storages.dataset. Four dataset views are available in the Apify Console:

View	Fields shown
Overview	url, domain, title, language, extractive_summary, stats
Word Frequency	url, top_words, top_bigrams, top_trigrams
Morphological Analysis	url, morphology_sample, top_roots, regional_words
LLM Analysis	url, llm_analysis

Top-level output fields:

Field	Type	Description
url	String	Processed URL
domain	String	Hostname extracted from the URL
title	String	HTML <title> of the page
language	String	Stopword language code used
text_preview	String	First 2,000 characters of extracted text
text_normalized	String	First 500 characters of slang-normalized text
extractive_summary	String	Auto-generated extractive summary
stats	Object	char_count, token_count_raw/clean, unique_token_count, sentence_count, avg_sentence_length, lexical_density_pct, regional_word_count
stopword_coverage	Object	Stopword list size per language
tokens_raw_sample	Array	First 50 raw tokens
tokens_clean_sample	Array	First 50 clean tokens
top_words	Array	{word, count} — top-N most frequent words
top_bigrams	Array	{bigram, count} — top 10 two-word sequences
top_trigrams	Array	{trigram, count} — top 10 three-word sequences
morphology_sample	Array	{word, root, prefixes, suffixes, type, regional_matches} for up to 15 tokens
top_roots	Array	{root, count} — most common stem forms
regional_words	Array	Tokens matched in a regional dictionary
transaction_analysis	Object	transaction_entities, price_count, quantity_count
kbbi_lookup	Array	{word, status, definitions} from KBBI
aksara_transliteration	Array	{word, aksara_sunda, aksara_jawa, aksara_bali, …_back}
formatted_reports	Object	Reports in markdown, csv, table, html, json formats
llm_analysis	Object	LLM results: llm_summary, llm_sentiment, llm_keywords, llm_category (optional)
full_text	String	Full page text (only when includeFullText: true)

---

Dependencies

Package	Version	Purpose
apify	1.7.2	Apify Python SDK
saka-nlp	0.2.6	Indonesian NLP framework
beautifulsoup4	4.12.3	HTML parsing
requests	2.32.3	HTTP client + LLM API calls
lxml	5.2.2	Fast HTML/XML parser backend for BeautifulSoup
emoji	2.12.1	Emoji conversion (required by saka.demojize)
emot	3.1.0	Emoticon conversion (required by saka.replace_emoticons)

---

Deploy to Apify

Option 1 — Apify Console (GitHub Integration)

1. Create a new Actor at Apify Console
2. Choose Git repository as the source
3. Enter your GitHub repository URL
4. Enable GitHub Integration for automatic deploys on push

Option 2 — Apify CLI

npm install -g apify-cli
apify login
apify push

---

Project Structure

saka-nlp-actor/
├── .actor/
│   ├── actor.json            # Apify actor metadata + inline dataset views
│   └── output_schema.json    # Output schema (actorOutputSchemaVersion)
├── main.py                   # Actor entry point
├── llm_client.py             # Plug-and-play LLM integration
├── INPUT_SCHEMA.json         # Input definition
├── requirements.txt          # Python dependencies
├── Dockerfile                # Docker configuration
├── LICENSE                   # MIT License
├── README.md                 # This file (English)
└── README_ID.md              # Indonesian documentation

---

License

This project is licensed under the LICENSE.

MIT License

Copyright (c) 2026 Muhammad Ikhwan Fathulloh

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.