{
  "version": "1.0.0",
  "exported_at": "2026-05-31T06:00:00.000Z",
  "project": {
    "name": "Yellow Pages Scraper Cloud",
    "description": "Scrapes YellowPages.com search results for battery businesses in Detroit, MI, exporting the Octoparse-equivalent fields: search name, location, business title, address, phone, website, and full YellowPages listing URL. Pagination is handled by clicking the enabled Next link until no further results page is available; data is appended to one CSV. Uses an XPath row selector to skip ad/placeholder rows without a non-empty business title.",
    "color": "bg-[#4589ff]",
    "template_id": "ai-generated"
  },
  "blocks": [
    {
      "block_id": "navigate-1",
      "block_type": "process",
      "title": "Navigate",
      "description": "Go to a URL",
      "position_x": 120,
      "position_y": 220,
      "config": {
        "url": "https://www.yellowpages.com/search?search_terms=battery&geo_location_terms=Detroit%2C%20MI",
        "color": "bg-[#4589ff]"
      }
    },
    {
      "block_id": "wait-for-page-load-1",
      "block_type": "process",
      "title": "Wait for Page Load",
      "description": "Wait for page to finish loading",
      "position_x": 480,
      "position_y": 220,
      "config": {
        "timeout": 30,
        "color": "bg-[#08bdba]"
      }
    },
    {
      "block_id": "wait-for-element-1",
      "block_type": "process",
      "title": "Wait for Element",
      "description": "Wait until element appears",
      "position_x": 840,
      "position_y": 220,
      "config": {
        "selector": ".result a.business-name",
        "timeout": 30,
        "visible": true,
        "color": "bg-[#08bdba]"
      }
    },
    {
      "block_id": "structured-export-1",
      "block_type": "process",
      "title": "Structured Export",
      "description": "Export data with custom columns",
      "position_x": 1200,
      "position_y": 220,
      "config": {
        "rowSelector": "//div[contains(concat(' ', normalize-space(@class), ' '), ' result ')][.//a[contains(concat(' ', normalize-space(@class), ' '), ' business-name ')][normalize-space(string()) != '']]",
        "fileName": "yellow-page-scraper-cloud.csv",
        "saveLocation": "C:\\Users\\theskd\\Documents\\UScraper\\templates",
        "includeHeaders": true,
        "fileMode": "append",
        "color": "bg-[#42be65]",
        "columns": [
          {
            "name": "name",
            "selector": "'battery'",
            "attribute": "text",
            "isJs": true
          },
          {
            "name": "location",
            "selector": "'Detroit'",
            "attribute": "text",
            "isJs": true
          },
          {
            "name": "title",
            "selector": "ROW.querySelector('a.business-name span, a.business-name')?.textContent?.replace(/\\s+/g, ' ').trim() || ''",
            "attribute": "text",
            "isJs": true
          },
          {
            "name": "address",
            "selector": "(() => { const clean = s => (s || '').replace(/\\s+/g, ' ').replace(/Serving the(?=\\S)/, 'Serving the ').trim(); const street = clean(ROW.querySelector('.street-address')?.textContent); const locality = clean(ROW.querySelector('.locality')?.textContent); const adr = clean(ROW.querySelector('.adr')?.textContent); return (street || locality) ? `${street} ${locality}`.trim() : adr; })()",
            "attribute": "text",
            "isJs": true
          },
          {
            "name": "phone",
            "selector": "ROW.querySelector('.phones')?.textContent?.replace(/\\s+/g, ' ').trim() || ''",
            "attribute": "text",
            "isJs": true
          },
          {
            "name": "website",
            "selector": "(() => { const a = ROW.querySelector('a.track-visit-website'); const href = a?.getAttribute('href') || ''; return href ? new URL(href, location.origin).href : ''; })()",
            "attribute": "text",
            "isJs": true
          },
          {
            "name": "url",
            "selector": "(() => { const a = ROW.querySelector('a.business-name'); const href = a?.getAttribute('href') || ''; return href ? new URL(href, location.origin).href : ''; })()",
            "attribute": "text",
            "isJs": true
          }
        ]
      }
    },
    {
      "block_id": "element-exists-1",
      "block_type": "process",
      "title": "Element Exists",
      "description": "Check if element exists",
      "position_x": 1560,
      "position_y": 220,
      "config": {
        "selector": ".pagination a.next:not(.disabled), a.next.ajax-page:not(.disabled), a[rel=\"next\"]",
        "color": "bg-[#ff832b]"
      }
    },
    {
      "block_id": "click-1",
      "block_type": "process",
      "title": "Click",
      "description": "Click on element",
      "position_x": 1560,
      "position_y": 560,
      "config": {
        "selector": ".pagination a.next:not(.disabled), a.next.ajax-page:not(.disabled), a[rel=\"next\"]",
        "timeout": 15,
        "color": "bg-[#ff832b]"
      }
    },
    {
      "block_id": "wait-for-page-load-2",
      "block_type": "process",
      "title": "Wait for Page Load",
      "description": "Wait for page to finish loading",
      "position_x": 1920,
      "position_y": 560,
      "config": {
        "timeout": 30,
        "color": "bg-[#08bdba]"
      }
    },
    {
      "block_id": "sleep-1",
      "block_type": "process",
      "title": "Sleep",
      "description": "Wait for specified time",
      "position_x": 2280,
      "position_y": 560,
      "config": {
        "duration": 2,
        "color": "bg-[#ff832b]"
      }
    },
    {
      "block_id": "wait-for-element-2",
      "block_type": "process",
      "title": "Wait for Element",
      "description": "Wait until element appears",
      "position_x": 2280,
      "position_y": 760,
      "config": {
        "selector": ".result a.business-name",
        "timeout": 30,
        "visible": true,
        "color": "bg-[#08bdba]"
      }
    },
    {
      "block_id": "end-1",
      "block_type": "output",
      "title": "End",
      "description": "Terminate execution flow",
      "position_x": 1560,
      "position_y": 900,
      "config": {
        "color": "bg-[#8d8d8d]"
      }
    }
  ],
  "connections": [
    {
      "from_block_id": "navigate-1",
      "from_connector_id": "right",
      "to_block_id": "wait-for-page-load-1",
      "to_connector_id": "left"
    },
    {
      "from_block_id": "wait-for-page-load-1",
      "from_connector_id": "right",
      "to_block_id": "wait-for-element-1",
      "to_connector_id": "left"
    },
    {
      "from_block_id": "wait-for-element-1",
      "from_connector_id": "right",
      "to_block_id": "structured-export-1",
      "to_connector_id": "left"
    },
    {
      "from_block_id": "structured-export-1",
      "from_connector_id": "right",
      "to_block_id": "element-exists-1",
      "to_connector_id": "left"
    },
    {
      "from_block_id": "element-exists-1",
      "from_connector_id": "true",
      "to_block_id": "click-1",
      "to_connector_id": "left"
    },
    {
      "from_block_id": "element-exists-1",
      "from_connector_id": "false",
      "to_block_id": "end-1",
      "to_connector_id": "left"
    },
    {
      "from_block_id": "click-1",
      "from_connector_id": "right",
      "to_block_id": "wait-for-page-load-2",
      "to_connector_id": "left"
    },
    {
      "from_block_id": "wait-for-page-load-2",
      "from_connector_id": "right",
      "to_block_id": "sleep-1",
      "to_connector_id": "left"
    },
    {
      "from_block_id": "sleep-1",
      "from_connector_id": "right",
      "to_block_id": "wait-for-element-2",
      "to_connector_id": "left"
    },
    {
      "from_block_id": "wait-for-element-2",
      "from_connector_id": "right",
      "to_block_id": "structured-export-1",
      "to_connector_id": "left"
    }
  ],
  "canvas_elements": [
    {
      "id": "group-load",
      "element_type": "group",
      "title": "Page Load",
      "color": "#08bdba",
      "position_x": 48,
      "position_y": 116,
      "width": 2480,
      "height": 836,
      "z_index": 20,
      "data": {
        "memberBlockIds": [
          "navigate-1",
          "wait-for-page-load-1",
          "wait-for-element-1",
          "wait-for-page-load-2",
          "sleep-1",
          "wait-for-element-2"
        ]
      }
    },
    {
      "id": "group-extract",
      "element_type": "group",
      "title": "Data Extraction",
      "color": "#42be65",
      "position_x": 1128,
      "position_y": 116,
      "width": 380,
      "height": 296,
      "z_index": 20,
      "data": {
        "memberBlockIds": [
          "structured-export-1"
        ]
      }
    },
    {
      "id": "group-pagination",
      "element_type": "group",
      "title": "Pagination Loop",
      "color": "#ff832b",
      "position_x": 1488,
      "position_y": 116,
      "width": 380,
      "height": 636,
      "z_index": 20,
      "data": {
        "memberBlockIds": [
          "element-exists-1",
          "click-1"
        ]
      }
    },
    {
      "id": "group-control",
      "element_type": "group",
      "title": "Control Flow",
      "color": "#8d8d8d",
      "position_x": 1488,
      "position_y": 796,
      "width": 380,
      "height": 296,
      "z_index": 20,
      "data": {
        "memberBlockIds": [
          "end-1"
        ]
      }
    },
    {
      "id": "note-overview",
      "element_type": "note",
      "title": "Overview",
      "content": "Scrapes YellowPages.com search results for battery businesses in Detroit, MI, exporting the Octoparse-equivalent fields: search name, location, business title, address, phone, website, and full YellowPages listing URL. Pagination is handled by clicking the enabled Next link until no further results page is available; data is appended to one CSV. Uses an XPath row selector to skip ad/placeholder rows without a non-empty business title.",
      "color": "#f1c21b",
      "position_x": 80,
      "position_y": 20,
      "width": 480,
      "height": 160,
      "z_index": 22,
      "data": {}
    },
    {
      "id": "note-block-structured-export-1",
      "element_type": "note",
      "title": "Note: Structured Export",
      "content": "Structured export with JS columns (name, location, title, address, phone). These selectors are fragile — update if the site layout changes.",
      "color": "#ee5396",
      "position_x": 1400,
      "position_y": 200,
      "width": 340,
      "height": 126,
      "z_index": 22,
      "data": {
        "block_id": "structured-export-1"
      }
    },
    {
      "id": "note-block-element-exists-1",
      "element_type": "note",
      "title": "Note: Element Exists",
      "content": "Condition block: checks `.pagination a.next:not(.disabled), a.next.ajax-page:not(.disabled), a[rel=\"next\"]`. True / False branches control which path runs next. Keep enough space between branches so both connector lines are visible.",
      "color": "#ee5396",
      "position_x": 1760,
      "position_y": 200,
      "width": 340,
      "height": 157,
      "z_index": 22,
      "data": {
        "block_id": "element-exists-1"
      }
    },
    {
      "id": "note-block-click-1",
      "element_type": "note",
      "title": "Note: Click",
      "content": "Pagination click — add waits after this block; the page reloads asynchronously.",
      "color": "#ee5396",
      "position_x": 1760,
      "position_y": 540,
      "width": 316,
      "height": 106,
      "z_index": 22,
      "data": {
        "block_id": "click-1"
      }
    }
  ]
}