kdeps-components-searchWeb

version: 1.0.1
requires:
    schemas/v1: '>=1.0.3'
files:
    component.yaml: |
        apiVersion: kdeps.io/v1
        kind: Component
        metadata:
          name: searchWeb
          description: Search the web - DuckDuckGo by default, Brave, Bing or Tavily with an API key - and return title/url/snippet results
          version: "1.0.1"
          targetActionId: searchWebResults
        interface:
          inputs:
            - name: query
              type: string
              required: true
              description: Search query
            - name: provider
              type: string
              default: ddg
              description: ddg (no key), brave, bing or tavily
            - name: connectionName
              type: string
              default: ""
              connection: search
              description: Name under search_connections in ~/.kdeps/config.yaml holding the provider's apiKey (brave, bing, tavily)
            - name: maxResults
              type: integer
              default: 5
              description: Results to return at most
            - name: timeout
              type: string
              default: 15s
              description: Request timeout as a duration
          output:
            results: "{{ get('searchWeb.results') }}"
            count: "{{ len(get('searchWeb.results')) }}"
            query: "{{ get('searchWeb.query') }}"
            provider: "{{ lower(trim(get('searchWeb.provider'))) }}"
        resources:
          - actionId: searchWebDDG
            name: DuckDuckGo
            validations:
              skip:
                - "lower(trim(get('searchWeb.provider'))) != 'ddg'"
            httpClient:
              method: GET
              url: "{{ env('KDEPS_DDG_URL', 'https://html.duckduckgo.com') }}/html/?q={{ urlencode(get('searchWeb.query')) }}"
              headers:
                User-Agent: "Mozilla/5.0 (compatible; kdeps/2.0)"
                Accept: text/html
              timeout: "{{ get('searchWeb.timeout') }}"
              failOnError: true
          - actionId: searchWebBrave
            name: Brave
            validations:
              skip:
                - "lower(trim(get('searchWeb.provider'))) != 'brave'"
              check:
                - "get('config.search_connections.' + get('searchWeb.connectionName') + '.apiKey', '') != ''"
              error:
                code: 400
                message: "searchWeb: connectionName required for brave - set search_connections.<name>.apiKey in ~/.kdeps/config.yaml"
            httpClient:
              method: GET
              url: "{{ env('KDEPS_BRAVE_URL', 'https://api.search.brave.com') }}/res/v1/web/search?q={{ urlencode(get('searchWeb.query')) }}&count={{ get('searchWeb.maxResults') }}"
              headers:
                Accept: application/json
                X-Subscription-Token: "{{ get('config.search_connections.' + get('searchWeb.connectionName') + '.apiKey') }}"
              timeout: "{{ get('searchWeb.timeout') }}"
              failOnError: true
          - actionId: searchWebBing
            name: Bing
            validations:
              skip:
                - "lower(trim(get('searchWeb.provider'))) != 'bing'"
              check:
                - "get('config.search_connections.' + get('searchWeb.connectionName') + '.apiKey', '') != ''"
              error:
                code: 400
                message: "searchWeb: connectionName required for bing - set search_connections.<name>.apiKey in ~/.kdeps/config.yaml"
            httpClient:
              method: GET
              url: "{{ env('KDEPS_BING_URL', 'https://api.bing.microsoft.com') }}/v7.0/search?q={{ urlencode(get('searchWeb.query')) }}&count={{ get('searchWeb.maxResults') }}"
              headers:
                Ocp-Apim-Subscription-Key: "{{ get('config.search_connections.' + get('searchWeb.connectionName') + '.apiKey') }}"
              timeout: "{{ get('searchWeb.timeout') }}"
              failOnError: true
          - actionId: searchWebTavily
            name: Tavily
            validations:
              skip:
                - "lower(trim(get('searchWeb.provider'))) != 'tavily'"
              check:
                - "get('config.search_connections.' + get('searchWeb.connectionName') + '.apiKey', '') != ''"
              error:
                code: 400
                message: "searchWeb: connectionName required for tavily - set search_connections.<name>.apiKey in ~/.kdeps/config.yaml"
            httpClient:
              method: POST
              url: "{{ env('KDEPS_TAVILY_URL', 'https://api.tavily.com') }}/search"
              headers:
                Content-Type: application/json
              data:
                api_key: "{{ get('config.search_connections.' + get('searchWeb.connectionName') + '.apiKey') }}"
                query: "{{ get('searchWeb.query') }}"
                max_results: "{{ int(get('searchWeb.maxResults')) }}"
              timeout: "{{ get('searchWeb.timeout') }}"
              failOnError: true
          # Normalizes whichever provider ran to [{title, url, snippet}], at most maxResults.
          - actionId: searchWebResults
            name: Results
            validations:
              check:
                - "lower(trim(get('searchWeb.provider'))) in ['ddg', 'brave', 'bing', 'tavily']"
              error:
                code: 400
                message: "searchWeb: unknown provider (use ddg, brave, bing or tavily)"
            before:
              - "set('searchWeb.results', lower(trim(get('searchWeb.provider'))) == 'brave' ? map((output('searchWebBrave').data.web.results ?? [])[:int(get('searchWeb.maxResults'))], ({title: #.title ?? '', url: #.url ?? '', snippet: #.description ?? ''})) : lower(trim(get('searchWeb.provider'))) == 'bing' ? map((output('searchWebBing').data.webPages.value ?? [])[:int(get('searchWeb.maxResults'))], ({title: #.name ?? '', url: #.url ?? '', snippet: #.snippet ?? ''})) : lower(trim(get('searchWeb.provider'))) == 'tavily' ? map((output('searchWebTavily').data.results ?? [])[:int(get('searchWeb.maxResults'))], ({title: #.title ?? '', url: #.url ?? '', snippet: #.content ?? ''})) : map(filter(htmlFind(output('searchWebDDG').body, 'a.result__a'), (#.text ?? '') != '' || (#.href ?? '') != '')[:int(get('searchWeb.maxResults'))], ({title: #.text ?? '', url: (#['data-href'] ?? '') != '' ? #['data-href'] : (#.href ?? ''), snippet: ''})))"