# Expanso Pipeline: keyword-extract (CLI mode)
# =============================================
#
# Usage:
#   cat article.txt | expanso-edge run pipeline-cli.yaml
#
# Requires OPENAI_API_KEY.

name: "keyword-extract-cli"
type: pipeline

config:
  input:
    stdin:
      codec: all
      max_buffer: 1048576

  pipeline:
    processors:
      - mapping: |
          meta trace_id = uuid_v4()
          meta input_hash = content().hash("sha256").encode("hex")
          meta input_length = content().length()
          let max_keywords = env("MAX_KEYWORDS").or("10").number().or(10)

          root.messages = [
            {
              "role": "system",
              "content": "You are a keyword extraction expert. Extract keywords from text and return JSON with: 'keywords' (array of single words with 'word' and 'relevance' 0-1), 'phrases' (array of 2-3 word key phrases), 'topics' (array of main themes). Limit to " + $max_keywords.string() + " keywords. Focus on nouns and important terms."
            },
            {
              "role": "user",
              "content": content()
            }
          ]

      - openai_chat_completion:
          api_key: "${OPENAI_API_KEY}"
          model: gpt-4o-mini

      - mapping: |
          let response = this.choices.0.message.content.parse_json().catch({
            "keywords": [],
            "phrases": [],
            "topics": []
          })

          root.keywords = $response.keywords.or([])
          root.phrases = $response.phrases.or([])
          root.topics = $response.topics.or([])
          root.keyword_count = $response.keywords.or([]).length()
          root.metadata = {
            "skill": "keyword-extract",
            "mode": "cli",
            "model": "gpt-4o-mini",
            "input_hash": meta("input_hash"),
            "input_length": meta("input_length"),
            "trace_id": meta("trace_id"),
            "timestamp": now()
          }

      - log:
          level: INFO
          message: |
            [keyword-extract] Extracted ${! root.keyword_count } keywords (trace: ${! meta("trace_id").slice(0, 8) })

  output:
    stdout:
      codec: json_object