Convert Planet product bundles description into JSON

bundles.json

{
  "bundles": {
    "analytic": {
      "name": "Analytic Radiance",
      "description": "Orthorectified product, calibrated to at-sensor radiance",
      "assets": {
        "Landsat8L1G": [
          "analytic_b1",
          "analytic_b2",
          "analytic_b3",
          "analytic_b4",
          "analytic_b5",
          "analytic_b6",
          "analytic_b7",
          "analytic_b8",
          "analytic_b9",
          "analytic_b10",
          "analytic_b11",
          "analytic_bqa",
          "metadata_txt"
        ],
        "PSOrthoTile": [
          "analytic",
          "udm",
          "analytic_xml"
        ],
        "PSScene3Band": [
          "analytic",
          "analytic_xml",
          "udm"
        ],
        "PSScene4Band": [
          "analytic",
          "analytic_xml",
          "udm"
        ],
        "REOrthoTile": [
          "analytic",
          "analytic_xml",
          "udm"
        ],
        "Sentinel1": [
          "ortho_analytic_vv",
          "ortho_analytic_vh"
        ],
        "Sentinel2L1C": [
          "analytic_b1",
          "analytic_b2",
          "analytic_b3",
          "analytic_b4",
          "analytic_b5",
          "analytic_b6",
          "analytic_b7",
          "analytic_b8",
          "analytic_b8a",
          "analytic_b9",
          "analytic_b10",
          "analytic_b11",
          "analytic_b12",
          "metadata_aux"
        ],
        "SkySatScene": [
          "ortho_analytic",
          "ortho_analytic_udm"
        ],
        "SkySatCollect": [
          "ortho_analytic",
          "ortho_analytic_udm"
        ]
      }
    },
    "analytic_udm2": {
      "name": "Analytic Radiance with UDM2",
      "description": "Orthorectified product, calibrated to at-sensor radiance. Includes udm2.",
      "assets": {
        "PSOrthoTile": [
          "analytic",
          "udm",
          "udm2",
          "analytic_xml"
        ],
        "PSScene4Band": [
          "analytic",
          "analytic_xml",
          "udm",
          "udm2"
        ],
        "SkySatScene": [
          "ortho_analytic",
          "ortho_analytic_udm",
          "ortho_analytic_udm2"
        ],
        "SkySatCollect": [
          "ortho_analytic",
          "ortho_analytic_udm",
          "ortho_analytic_udm2"
        ]
      }
    },
    "visual": {
      "name": "Visual",
      "description": "Orthorectified product, visually enhanced for optimal appearance",
      "assets": {
        "Landsat8L1G": [
          "visual"
        ],
        "PSOrthoTile": [
          "visual",
          "visual_xml"
        ],
        "PSScene3Band": [
          "visual",
          "visual_xml"
        ],
        "REOrthoTile": [
          "visual",
          "visual_xml"
        ],
        "Sentinel2L1C": [
          "visual"
        ],
        "SkySatCollect": [
          "ortho_visual"
        ],
        "SkySatScene": [
          "ortho_visual"
        ]
      }
    },
    "uncalibrated_dn": {
      "name": "Uncalibrated DN",
      "description": "Orthorectified product, non-radiometrically-calibrated, raw digital number",
      "assets": {
        "PSOrthoTile": [
          "analytic_dn",
          "udm",
          "analytic_dn_xml"
        ],
        "PSScene3Band": [
          "analytic_dn",
          "analytic_dn_xml",
          "udm"
        ],
        "PSScene4Band": [
          "analytic_dn",
          "analytic_dn_xml",
          "udm"
        ],
        "SkySatCollect": [
          "ortho_analytic_dn",
          "ortho_analytic_udm"
        ],
        "SkySatScene": [
          "ortho_analytic_dn",
          "ortho_analytic_udm"
        ]
      }
    },
    "uncalibrated_dn_udm2": {
      "name": "Uncalibrated DN with UDM2",
      "description": "Orthorectified product, non-radiometrically-calibrated, raw digital number. Includes udm2.",
      "assets": {
        "PSOrthoTile": [
          "analytic_dn",
          "analytic_dn_xml",
          "udm",
          "udm2"
        ],
        "PSScene4Band": [
          "analytic_dn",
          "analytic_dn_xml",
          "udm",
          "udm2"
        ],
        "SkySatCollect": [
          "ortho_analytic_dn",
          "ortho_analytic_udm",
          "ortho_analytic_udm2"
        ],
        "SkySatScene": [
          "ortho_analytic_dn",
          "ortho_analytic_udm",
          "ortho_analytic_udm2"
        ]
      }
    },
    "basic_analytic": {
      "name": "Non-Orthorectified Analytic Radiance",
      "description": "Non-orthorectified, calibrated to at-sensor radiance",
      "assets": {
        "PSScene3Band": [
          "basic_analytic",
          "basic_udm",
          "basic_analytic_rpc",
          "basic_analytic_xml"
        ],
        "PSScene4Band": [
          "basic_analytic",
          "basic_udm",
          "basic_analytic_rpc",
          "basic_analytic_xml"
        ],
        "REScene": [
          "basic_analytic_b1",
          "basic_analytic_b2",
          "basic_analytic_b3",
          "basic_analytic_b4",
          "basic_analytic_b5",
          "basic_analytic_xml",
          "basic_analytic_rpc",
          "basic_udm",
          "basic_analytic_sci",
          "browse"
        ],
        "SkySatScene": [
          "basic_analytic",
          "basic_analytic_rpc"
        ]
      }
    },
    "basic_analytic_udm2": {
      "name": "Non-Orthorectified Analytic Radiance with UDM2",
      "description": "Non-orthorectified, calibrated to at-sensor radiance. Includes udm2.",
      "assets": {
        "PSScene4Band": [
          "basic_analytic",
          "basic_udm",
          "basic_udm2",
          "basic_analytic_rpc",
          "basic_analytic_xml"
        ],
        "SkySatScene": [
          "basic_analytic",
          "basic_analytic_rpc"
        ]
      }
    },
    "basic_uncalibrated_dn": {
      "name": "Non-Orthorectified, Uncalibrated DN",
      "description": "Non-orthorectified, non-radiometrically-calibrated, raw digital number",
      "assets": {
        "PSScene3Band": [
          "basic_analytic_dn",
          "basic_analytic_dn_xml",
          "basic_udm",
          "basic_analytic_dn_rpc"
        ],
        "PSScene4Band": [
          "basic_analytic_dn",
          "basic_analytic_dn_xml",
          "basic_udm",
          "basic_analytic_rpc"
        ],
        "SkySatScene": [
          "basic_analytic_dn",
          "basic_analytic_dn_rpc"
        ]
      }
    },
    "basic_uncalibrated_dn_udm2": {
      "name": "Non-Orthorectified, Uncalibrated DN with UDM2",
      "description": "Non-orthorectified, non-radiometrically-calibrated, raw digital number. Includes udm2.",
      "assets": {
        "PSScene4Band": [
          "basic_analytic_dn",
          "basic_analytic_dn_xml",
          "basic_udm",
          "basic_udm2",
          "basic_analytic_rpc"
        ],
        "SkySatScene": [
          "basic_analytic_dn",
          "basic_analytic_dn_rpc"
        ]
      }
    },
    "analytic_sr": {
      "name": "Analytic Surface Reflectance",
      "description": "Orthorectified product, radiometrically calibrated and atmospherically corrected to surface reflectance",
      "assets": {
        "PSScene4Band": [
          "analytic_sr",
          "udm",
          "analytic_xml"
        ],
        "MOD09GQ": [
          "analytic_num_observations",
          "analytic_orbit_pnt",
          "analytic_granule_pnt",
          "analytic_sur_refl_b01",
          "analytic_sur_refl_b02",
          "analytic_qc_250m",
          "analytic_obscov",
          "analytic_iobs_res"
        ],
        "MYD09GQ": [
          "analytic_num_observations",
          "analytic_orbit_pnt",
          "analytic_granule_pnt",
          "analytic_sur_refl_b01",
          "analytic_sur_refl_b02",
          "analytic_qc_250m",
          "analytic_obscov",
          "analytic_iobs_res"
        ],
        "MOD09GA": [
          "analytic_num_observations_500m",
          "analytic_num_observations_1km",
          "analytic_state_1km",
          "analytic_sensor_zenith",
          "analytic_sensor_azimuth",
          "analytic_range",
          "analytic_solar_zenith",
          "analytic_solar_azimuth",
          "analytic_gflags",
          "analytic_orbit_pnt",
          "analytic_granule_pnt",
          "analytic_sur_refl_b01",
          "analytic_sur_refl_b02",
          "analytic_sur_refl_b03",
          "analytic_sur_refl_b04",
          "analytic_sur_refl_b05",
          "analytic_sur_refl_b06",
          "analytic_sur_refl_b07",
          "analytic_qc_500m",
          "analytic_obscov_500m",
          "analytic_iobs_res",
          "analytic_q_scan"
        ],
        "MYD09GA": [
          "analytic_num_observations_500m",
          "analytic_num_observations_1km",
          "analytic_state_1km",
          "analytic_sensor_zenith",
          "analytic_sensor_azimuth",
          "analytic_range",
          "analytic_solar_zenith",
          "analytic_solar_azimuth",
          "analytic_gflags",
          "analytic_orbit_pnt",
          "analytic_granule_pnt",
          "analytic_sur_refl_b01",
          "analytic_sur_refl_b02",
          "analytic_sur_refl_b03",
          "analytic_sur_refl_b04",
          "analytic_sur_refl_b05",
          "analytic_sur_refl_b06",
          "analytic_sur_refl_b07",
          "analytic_qc_500m",
          "analytic_obscov_500m",
          "analytic_iobs_res",
          "analytic_q_scan"
        ]
      }
    },
    "analytic_sr_udm2": {
      "name": "Analytic Surface Reflectance with UDM2",
      "description": "Orthorectified product, radiometrically calibrated and atmospherically corrected to surface reflectance. Includes udm2.",
      "assets": {
        "PSScene4Band": [
          "analytic_sr",
          "analytic_xml",
          "udm",
          "udm2"
        ]
      }
    },
    "basic_uncalibrated_dn_nitf": {
      "name": "Non-Orthorectified, Uncalibrated DN NITF",
      "description": "Non-orthorectified, non-radiometrically-calibrated, raw digital number in NITF format",
      "assets": {
        "PSScene4Band": [
          "basic_analytic_dn_nitf",
          "basic_analytic_dn_xml_nitf",
          "basic_analytic_dn_rpc_nitf",
          "basic_udm"
        ]
      }
    },
    "basic_uncalibrated_dn_nitf_udm2": {
      "name": "Non-Orthorectified, Uncalibrated DN NITF with UDM2",
      "description": "Non-orthorectified, non-radiometrically-calibrated, raw digital number in NITF format. Includes udm2.",
      "assets": {
        "PSScene4Band": [
          "basic_analytic_dn_nitf",
          "basic_analytic_dn_xml_nitf",
          "basic_analytic_dn_rpc_nitf",
          "basic_udm",
          "basic_udm2"
        ]
      }
    },
    "basic_analytic_nitf": {
      "name": "Non-Orthorectified Analytic Radiance NITF",
      "description": "Non-orthorectified, calibrated to at-sensor radiance, in NITF format",
      "assets": {
        "PSScene4Band": [
          "basic_analytic_nitf",
          "basic_analytic_rpc_nitf",
          "basic_analytic_xml_nitf",
          "basic_udm"
        ],
        "REScene": [
          "basic_analytic_b1_nitf",
          "basic_analytic_b2_nitf",
          "basic_analytic_b3_nitf",
          "basic_analytic_b4_nitf",
          "basic_analytic_b5_nitf",
          "basic_analytic_xml_nitf",
          "basic_analytic_rpc",
          "basic_udm",
          "basic_analytic_sci",
          "browse"
        ]
      }
    },
    "basic_analytic_nitf_udm2": {
      "name": "Non-Orthorectified Analytic Radiance NITF with UDM2",
      "description": "Non-orthorectified, calibrated to at-sensor radiance, in NITF format. Includes udm2.",
      "assets": {
        "PSScene4Band": [
          "basic_analytic_nitf",
          "basic_analytic_rpc_nitf",
          "basic_analytic_xml_nitf",
          "basic_udm",
          "basic_udm2"
        ]
      }
    },
    "basic_panchromatic": {
      "name": "Non-orthorectified, scaled radiance image",
      "description": "Non-orthorectified radiometrically-calibrated panchromatic imagery",
      "assets": {
        "SkySatScene": [
          "basic_panchromatic",
          "basic_panchromatic_rpc"
        ]
      }
    },
    "basic_panchromatic_dn": {
      "name": "Non-orthorectified panchromatic imagery, non-radiometrically calibrated, raw digital number",
      "description": "Unorthorectified panchromatic imagery, non-radiometrically calibrated, raw digital number",
      "assets": {
        "SkySatScene": [
          "basic_panchromatic_dn",
          "basic_panchromatic_dn_rpc"
        ]
      }
    },
    "panchromatic": {
      "name": "Panchromatic",
      "description": "Orthorectified radiometrically-calibrated panchromatic image stored as 16-bit scaled radiance",
      "assets": {
        "SkySatCollect": [
          "ortho_panchromatic",
          "ortho_panchromatic_udm"
        ],
        "SkySatScene": [
          "ortho_panchromatic",
          "ortho_panchromatic_udm"
        ]
      }
    },
    "panchromatic_dn": {
      "name": "Panchromatic DN",
      "description": "Orthorectified panchromatic imagery, non-radiometrically calibrated, raw digital number",
      "assets": {
        "SkySatCollect": [
          "ortho_panchromatic_dn",
          "ortho_panchromatic_udm"
        ],
        "SkySatScene": [
          "ortho_panchromatic_dn",
          "ortho_panchromatic_udm"
        ]
      }
    },
    "panchromatic_dn_udm2": {
      "name": "Panchromatic DN with UDM2",
      "description": "Orthorectified panchromatic imagery, non-radiometrically calibrated, raw digital number",
      "assets": {
        "SkySatCollect": [
          "ortho_panchromatic_dn",
          "ortho_panchromatic_udm",
          "ortho_panchromatic_udm2"
        ],
        "SkySatScene": [
          "ortho_panchromatic_dn",
          "ortho_panchromatic_udm",
          "ortho_panchromatic_udm2"
        ]
      }
    },
    "pansharpened": {
      "name": "Pansharpened",
      "description": "Orthorectifed, pan-sharpened imagery, non-radiometrically calibrated, raw digital number",
      "assets": {
        "SkySatCollect": [
          "ortho_pansharpened",
          "ortho_pansharpened_udm"
        ],
        "SkySatScene": [
          "ortho_pansharpened",
          "ortho_pansharpened_udm"
        ]
      }
    },
    "pansharpened_udm2": {
      "name": "Pansharpened with UDM2",
      "description": "Orthorectifed, pan-sharpened imagery, non-radiometrically calibrated, raw digital number",
      "assets": {
        "SkySatCollect": [
          "ortho_pansharpened",
          "ortho_pansharpened_udm",
          "ortho_pansharpened_udm2"
        ],
        "SkySatScene": [
          "ortho_pansharpened",
          "ortho_pansharpened_udm",
          "ortho_pansharpened_udm2"
        ]
      }
    },
    "basic_l1a_dn": {
      "name": "Non-Orthorectified, Level 1A Uncalibrated DN",
      "description": "Non-orthorectified, non-radiometrically-calibrated, raw digital number, Level 1A",
      "assets": {
        "SkySatScene": [
          "basic_l1a_panchromatic_dn",
          "basic_l1a_panchromatic_dn_rpc"
        ]
      }
    },
    "all": {
      "name": "All assets",
      "description": "All assets that can be produced for an item",
      "assets": {
        "Landsat8L1G": [
          "analytic_b1",
          "analytic_b2",
          "analytic_b3",
          "analytic_b4",
          "analytic_b5",
          "analytic_b6",
          "analytic_b7",
          "analytic_b8",
          "analytic_b9",
          "analytic_b10",
          "analytic_b11",
          "analytic_bqa",
          "metadata_txt",
          "visual"
        ],
        "MOD09GA": [
          "analytic_num_observations_500m",
          "analytic_num_observations_1km",
          "analytic_state_1km",
          "analytic_sensor_zenith",
          "analytic_sensor_azimuth",
          "analytic_range",
          "analytic_solar_zenith",
          "analytic_solar_azimuth",
          "analytic_gflags",
          "analytic_orbit_pnt",
          "analytic_granule_pnt",
          "analytic_sur_refl_b01",
          "analytic_sur_refl_b02",
          "analytic_sur_refl_b03",
          "analytic_sur_refl_b04",
          "analytic_sur_refl_b05",
          "analytic_sur_refl_b06",
          "analytic_sur_refl_b07",
          "analytic_qc_500m",
          "analytic_obscov_500m",
          "analytic_iobs_res",
          "analytic_q_scan"
        ],
        "MOD09GQ": [
          "analytic_num_observations",
          "analytic_orbit_pnt",
          "analytic_granule_pnt",
          "analytic_sur_refl_b01",
          "analytic_sur_refl_b02",
          "analytic_qc_250m",
          "analytic_obscov",
          "analytic_iobs_res"
        ],
        "MYD09GA": [
          "analytic_num_observations_500m",
          "analytic_num_observations_1km",
          "analytic_state_1km",
          "analytic_sensor_zenith",
          "analytic_sensor_azimuth",
          "analytic_range",
          "analytic_solar_zenith",
          "analytic_solar_azimuth",
          "analytic_gflags",
          "analytic_orbit_pnt",
          "analytic_granule_pnt",
          "analytic_sur_refl_b01",
          "analytic_sur_refl_b02",
          "analytic_sur_refl_b03",
          "analytic_sur_refl_b04",
          "analytic_sur_refl_b05",
          "analytic_sur_refl_b06",
          "analytic_sur_refl_b07",
          "analytic_qc_500m",
          "analytic_obscov_500m",
          "analytic_iobs_res",
          "analytic_q_scan"
        ],
        "MYD09GQ": [
          "analytic_num_observations",
          "analytic_orbit_pnt",
          "analytic_granule_pnt",
          "analytic_sur_refl_b01",
          "analytic_sur_refl_b02",
          "analytic_qc_250m",
          "analytic_obscov",
          "analytic_iobs_res"
        ],
        "PSOrthoTile": [
          "analytic_dn",
          "analytic_dn_xml",
          "udm",
          "analytic",
          "analytic_xml",
          "visual",
          "visual_xml"
        ],
        "PSScene3Band": [
          "analytic_dn",
          "analytic_dn_xml",
          "udm",
          "analytic",
          "analytic_xml",
          "basic_analytic",
          "basic_udm",
          "basic_analytic_rpc",
          "basic_analytic_xml",
          "visual",
          "visual_xml",
          "basic_analytic_dn",
          "basic_analytic_dn_rpc",
          "basic_analytic_dn_xml"
        ],
        "PSScene4Band": [
          "basic_udm",
          "analytic_dn",
          "analytic_dn_xml",
          "udm",
          "analytic",
          "analytic_xml",
          "basic_analytic",
          "basic_analytic_rpc",
          "basic_analytic_xml",
          "basic_analytic_dn",
          "basic_analytic_dn_xml",
          "analytic_sr"
        ],
        "REOrthoTile": [
          "analytic",
          "analytic_xml",
          "udm",
          "visual",
          "visual_xml"
        ],
        "REScene": [
          "basic_analytic_rpc",
          "basic_udm",
          "basic_analytic_sci",
          "browse",
          "basic_analytic_b1",
          "basic_analytic_b2",
          "basic_analytic_b3",
          "basic_analytic_b4",
          "basic_analytic_b5",
          "basic_analytic_xml"
        ],
        "Sentinel1": [
          "ortho_analytic_vv",
          "ortho_analytic_vh"
        ],
        "Sentinel2L1C": [
          "analytic_b1",
          "analytic_b2",
          "analytic_b3",
          "analytic_b4",
          "analytic_b5",
          "analytic_b6",
          "analytic_b7",
          "analytic_b8",
          "analytic_b8a",
          "analytic_b9",
          "analytic_b10",
          "analytic_b11",
          "analytic_b12",
          "metadata_aux",
          "visual"
        ],
        "SkySatCollect": [
          "ortho_analytic_dn",
          "ortho_analytic_udm",
          "ortho_panchromatic",
          "ortho_panchromatic_dn",
          "ortho_panchromatic_udm",
          "ortho_visual",
          "ortho_pansharpened",
          "ortho_pansharpened_udm",
          "ortho_analytic"
        ],
        "SkySatScene": [
          "ortho_analytic",
          "ortho_analytic_udm",
          "ortho_visual",
          "ortho_analytic_dn",
          "ortho_panchromatic",
          "ortho_panchromatic_dn",
          "ortho_panchromatic_udm",
          "basic_panchromatic",
          "basic_panchromatic_dn",
          "basic_panchromatic_dn_rpc",
          "basic_analytic",
          "basic_analytic_dn",
          "basic_analytic_dn_rpc",
          "ortho_pansharpened",
          "ortho_pansharpened_udm"
        ]
      }
    },
    "all_udm2": {
      "name": "All assets with UDM2",
      "description": "All assets that can be produced for an item. Includes udm2.",
      "assets": {
        "PSOrthoTile": [
          "analytic_dn",
          "analytic_dn_xml",
          "udm",
          "udm2",
          "analytic",
          "analytic_xml",
          "visual",
          "visual_xml"
        ],
        "PSScene4Band": [
          "basic_udm",
          "basic_udm2",
          "analytic_dn",
          "analytic_dn_xml",
          "udm",
          "udm2",
          "analytic",
          "analytic_xml",
          "basic_analytic",
          "basic_analytic_rpc",
          "basic_analytic_xml",
          "basic_analytic_dn",
          "basic_analytic_dn_xml",
          "analytic_sr"
        ],
        "SkySatCollect": [
          "ortho_panchromatic",
          "ortho_panchromatic_dn",
          "ortho_panchromatic_udm",
          "ortho_panchromatic_udm2",
          "ortho_pansharpened",
          "ortho_pansharpened_udm",
          "ortho_pansharpened_udm2",
          "ortho_visual",
          "ortho_analytic",
          "ortho_analytic_dn",
          "ortho_analytic_udm",
          "ortho_analytic_udm2"
        ],
        "SkySatScene": [
          "basic_analytic",
          "basic_analytic_dn",
          "basic_analytic_dn_rpc",
          "basic_panchromatic",
          "basic_panchromatic_dn",
          "basic_panchromatic_dn_rpc",
          "basic_panchromatic_udm2",
          "ortho_panchromatic",
          "ortho_panchromatic_dn",
          "ortho_panchromatic_udm",
          "ortho_panchromatic_udm2",
          "ortho_pansharpened",
          "ortho_pansharpened_udm",
          "ortho_pansharpened_udm2",
          "ortho_visual",
          "ortho_analytic",
          "ortho_analytic_dn",
          "ortho_analytic_udm",
          "ortho_analytic_udm2"
        ]
      }
    }
  },
  "version": "2019-09-25"
}

parse.py

import json
import re

import requests
from bs4 import BeautifulSoup

if __name__ == "__main__":
    url = "https://developers.planet.com/docs/orders/product-bundles-reference/"
    html = requests.get(url).text
    soup = BeautifulSoup(html, "html.parser")

    bundles = {}
    for detail in soup.find_all("details"):
        bundle_id = detail.find("summary").text
        bundle_name = detail.find_next("h4").text
        bundle_desc = (
            detail.find_next("p").find_next("p").text
        )  # Ignore first (empty) p-tag

        for item_type_node in detail.select("tr td:not([align=right])"):
            item_type = item_type_node.text

            for asset_type_node in item_type_node.find_next_sibling():

                if not isinstance(asset_type_node, str):
                    continue

                bundles.setdefault(
                    bundle_id,
                    {"name": bundle_name, "description": bundle_desc, "assets": {}},
                )["assets"].setdefault(item_type, []).append(asset_type_node)

    print(
        json.dumps(
            {
                "bundles": bundles,
                "version": soup.find("em", text=re.compile("last updated: .*"))
                .text.split(":")
                .pop()
                .strip(),
            }
        )
    )