b-capture_com.json

{
  "url": "https://b-capture.com",
  "refnavi_url": "https://osaka.refle.info/shop/b-capture/",
  "list_url": "https://b-capture.com/model.html",
  "next_page_url": {
    "xpath": null,
    "attr": null,
    "regex": null,
    "ng_words": null
  },
  "detail": {
    "xpath": "//*[contains(@class, 'gals_list')]//section//a[contains(@href, 'model_profile')]",
    "attr": "href",
    "regex": null,
    "ng_words": null
  },
  "therapist": {
    "name": {
      "xpath": "//h2[@class='today_flag']",
      "regex": "(.*)\\(",
      "ng_words": [
        "体験入店",
        "非公開",
        "店長"
      ]
    },
    "age": {
      "xpath": "//h2[@class='today_flag']",
      "regex": "\\((\\d{1,2})\\)",
      "ng_words": null
    },
    "height": {
      "xpath": "//h2[@class='today_flag']/span",
      "regex": "(\\d{1,3})cm",
      "ng_words": null
    },
    "comment": {
      "xpath": "//div[contains(@class, 'comment')]/div[contains(@class, 'text_free')]",
      "regex": null,
      "ng_words": null
    },
    "x_account": {
      "xpath": null,
      "attr": null,
      "regex": null,
      "ng_words": null
    },
    "image": {
      "xpath": "//*[@id='thumb_image']//img",
      "attr": "src",
      "regex": null,
      "ng_words": null
    },
    "schedule": {
      "date": {
        "month": {
          "xpath": "//*[contains(@class, 'schedule')]//li/span",
          "regex": "(\\d{1,2})/"
        },
        "day": {
          "xpath": "//*[contains(@class, 'schedule')]//li/span",
          "regex": "/(\\d{1,2})"
        }
      },
      "time": {
        "start": {
          "xpath": "//*[contains(@class, 'schedule')]//li",
          "regex": "^(\\d{1,2}):(\\d{1,2})"
        },
        "end": {
          "xpath": "//*[contains(@class, 'schedule')]//li",
          "regex": "(\\d{1,2}):(\\d{1,2})$"
        }
      }
    }
  }
}