eodash_catalog 0.3.2__tar.gz → 0.3.4__tar.gz

{eodash_catalog-0.3.2 → eodash_catalog-0.3.4}/.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.3.2
+current_version = 0.3.4
 commit = True
 tag = True
 parse = (?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)(\-(?P<release>[a-z]+)\.(?P<build>\d+))?

{eodash_catalog-0.3.2 → eodash_catalog-0.3.4}/Dockerfile

@@ -29,4 +29,4 @@ RUN eodash_catalog --help
 
 CMD ["eodash_catalog"]
 
-LABEL version="0.3.2"
+LABEL version="0.3.4"

{eodash_catalog-0.3.2 → eodash_catalog-0.3.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: eodash_catalog
-Version: 0.3.2
+Version: 0.3.4
 Summary: This package is intended to help create a compatible STAC catalog for the eodash dashboard client. It supports configuration of multiple endpoint types for information extraction.
 Project-URL: Documentation, https://github.com/eodash/eodash_catalog#readme
 Project-URL: Issues, https://github.com/eodash/eodash_catalog/issues

{eodash_catalog-0.3.2 → eodash_catalog-0.3.4}/src/eodash_catalog/__about__.py

@@ -1,4 +1,4 @@
 # SPDX-FileCopyrightText: 2024-present Daniel Santillan <daniel.santillan@eox.at>
 #
 # SPDX-License-Identifier: MIT
-__version__ = "0.3.2"
+__version__ = "0.3.4"

{eodash_catalog-0.3.2 → eodash_catalog-0.3.4}/src/eodash_catalog/endpoints.py

@@ -538,6 +538,120 @@ def handle_rasdaman_endpoint(
     # add_example_info(collection, collection_config, endpoint_config, catalog_config)
     return collection
 
+def handle_GeoDB_Features_endpoint(
+    catalog_config: dict,
+    endpoint_config: dict,
+    collection_config: dict,
+    coll_path_rel_to_root_catalog: str,
+    catalog: Catalog,
+    options: Options,
+) -> Collection:
+
+    # ID of collection is data["Name"] instead of CollectionId to be able to
+    # create more STAC collections from one geoDB table
+    collection = get_or_create_collection(
+        catalog, collection_config["Name"], collection_config, catalog_config, endpoint_config
+    )
+    coll_path_rel_to_root_catalog = f'{coll_path_rel_to_root_catalog}/{collection_config["Name"]}'
+    select = f'?select={endpoint_config["TimeParameter"]}'
+    url = (
+        endpoint_config["EndPoint"]
+        + endpoint_config["Database"]
+        + "_{}".format(endpoint_config["CollectionId"])
+        + select
+    )
+    response = json.loads(requests.get(url).text)
+    # Use aggregation value to group datetime results
+    aggregation = endpoint_config.get("Aggregation", "day")
+    unique_datetimes = set()
+    for value in response:
+        time_object = datetime.fromisoformat(value[endpoint_config["TimeParameter"]])
+        match aggregation:
+            case "hour":
+                unique_datetimes.add(
+                    datetime(
+                        time_object.year,
+                        time_object.month,
+                        time_object.day,
+                        time_object.hour,
+                    )
+                )
+            case "day":
+                unique_datetimes.add(
+                    datetime(time_object.year, time_object.month, time_object.day).date()
+                )
+            case "month":
+                unique_datetimes.add(
+                    datetime(time_object.year, time_object.month, 1).date()
+                )
+            case "year":
+                unique_datetimes.add(
+                    datetime(time_object.year, 1, 1).date()
+                )
+            case _:
+                # default to day
+                unique_datetimes.add(
+                    datetime(time_object.year, time_object.month, time_object.day).date()
+                )
+    # go over unique datetimes and create items
+    items = []
+    for dt in sorted(unique_datetimes):
+        item_datetime = dt if isinstance(dt, datetime) else datetime(dt.year, dt.month, dt.day)
+        matching_string = ""
+        match aggregation:
+            case "hour":
+                matching_string = item_datetime.strftime("%Y-%m-%dT%H:00:00Z")
+            case "day":
+                matching_string = item_datetime.strftime("%Y-%m-%d")
+            case "month":
+                matching_string = item_datetime.strftime("%Y-%m")
+            case "year":
+                matching_string = item_datetime.strftime("%Y")
+        updated_query = endpoint_config["Query"].replace("{{date_time}}", matching_string)
+        assets = {
+            "geodbfeatures": Asset(
+            href=f"{endpoint_config['EndPoint']}{endpoint_config['Database']}_{endpoint_config['CollectionId']}?{updated_query}",
+            media_type="application/geodb+json",
+            roles=["data"],
+        )}
+        item = Item(
+            id=format_datetime_to_isostring_zulu(item_datetime),
+            bbox=endpoint_config.get("OverwriteBBox", [-180, -90, 180, 90]),
+            properties={},
+            geometry=create_geometry_from_bbox(
+                endpoint_config.get("OverwriteBBox", [-180, -90, 180, 90])
+            ),
+            datetime=item_datetime,
+            stac_extensions=[],
+            assets=assets,
+        )
+        # add eodash style visualization info if Style has been provided
+        if endpoint_config.get("Style"):
+            ep_st = endpoint_config.get("Style")
+            style_link = Link(
+                rel="style",
+                target=ep_st
+                if ep_st.startswith("http")
+                else f"{catalog_config['assets_endpoint']}/{ep_st}",
+                media_type="text/vector-styles",
+                extra_fields={
+                    "asset:keys": list(assets),
+                },
+            )
+            item.add_link(style_link)
+        add_projection_info(endpoint_config, item)
+        items.append(item)
+
+    save_items(
+        collection,
+        items,
+        options.outputpath,
+        catalog_config["id"],
+        f"{coll_path_rel_to_root_catalog}/{collection.id}",
+        options.gp,
+    )
+    return collection
+
 
 def handle_GeoDB_endpoint(
     catalog_config: dict,
@@ -605,33 +719,70 @@ def handle_GeoDB_endpoint(
         locations_collection = get_or_create_collection(
             collection, key, sc_config, catalog_config, endpoint_config
         )
-        if input_data:
+        # check if input data is none
+        if input_data is None:
+            input_data = []
+        if len(input_data) > 0 or endpoint_config.get("FeatureCollection"):
             items = []
             for v in values:
                 # add items based on inputData fields for each time step available in values
-                first_match = next(
+                first_match: dict = next(
                     (item for item in input_data if item.get("Identifier") == v["input_data"]), None
                 )
                 time_object = datetime.fromisoformat(v["time"])
+                if endpoint_config.get("MapReplaceDates"):
+                    # get mapping of AOI_ID to list of dates
+                    available_dates_for_aoi_id = endpoint_config.get("MapReplaceDates").get(
+                        v["aoi_id"]
+                    )
+                    if available_dates_for_aoi_id:
+                        formatted_datetime = time_object.strftime("%Y-%m-%d")
+                        if formatted_datetime not in available_dates_for_aoi_id:
+                            # discard this date because not in available map dates
+                            continue
                 # extract wkt geometry from sub_aoi
                 if "sub_aoi" in v and v["sub_aoi"] != "/":
                     # create geometry from wkt
-                    geometry = mapping(wkt.loads(v["sub_aoi"]))
+                    shapely_geometry = wkt.loads(v["sub_aoi"])
+                    geometry = mapping(shapely_geometry)
                     # converting multipolygon to polygon to avoid shapely throwing an exception
                     # in collection extent from geoparquet table generation
                     # while trying to create a multipolygon extent of all multipolygons
                     if geometry["type"] == "MultiPolygon":
                         geometry = {"type": "Polygon", "coordinates": geometry["coordinates"][0]}
+                    bbox = shapely_geometry.bounds
                 else:
                     geometry = create_geometry_from_bbox(bbox)
+
+                assets = {"dummy_asset": Asset(href="")}
+                if endpoint_config.get("FeatureCollection"):
+                    assets["geodbfeatures"] = Asset(
+                        href=f"{endpoint_config['EndPoint']}{endpoint_config['Database']}_{endpoint_config['FeatureCollection']}?aoi_id=eq.{v['aoi_id']}&time=eq.{v['time']}",
+                        media_type="application/geodb+json",
+                        roles=["data"],
+                    )
                 item = Item(
                     id=v["time"],
                     bbox=bbox,
                     properties={},
                     geometry=geometry,
                     datetime=time_object,
-                    assets={"dummy_asset": Asset(href="")},
+                    assets=assets,
                 )
+                # make sure to also add Style link if FeatureCollection and Style has been provided
+                if endpoint_config.get("FeatureCollection") and endpoint_config.get("Style"):
+                    ep_st = endpoint_config.get("Style")
+                    style_link = Link(
+                        rel="style",
+                        target=ep_st
+                        if ep_st.startswith("http")
+                        else f"{catalog_config['assets_endpoint']}/{ep_st}",
+                        media_type="text/vector-styles",
+                        extra_fields={
+                            "asset:keys": list(assets),
+                        },
+                    )
+                    item.add_link(style_link)
                 if first_match:
                     match first_match["Type"]:
                         case "WMS":
@@ -640,7 +791,7 @@ def handle_GeoDB_endpoint(
                                 "wms:layers": [first_match["Layers"]],
                                 "role": ["data"],
                             }
-                            if url.startswith("https://services.sentinel-hub.com/ogc/wms/"):
+                            if "sentinel-hub.com" in url:
                                 instanceId = os.getenv("SH_INSTANCE_ID")
                                 if "InstanceId" in endpoint_config:
                                     instanceId = endpoint_config["InstanceId"]
@@ -655,7 +806,7 @@ def handle_GeoDB_endpoint(
                                     {"wms:dimensions": {"TIME": f"{start_date}/{end_date}"}}
                                 )
                                 # we add the instance id to the url
-                                url = f"https://services.sentinel-hub.com/ogc/wms/{instanceId}"
+                                url = f"{url}{instanceId}"
                             else:
                                 extra_fields.update({"wms:dimensions": {"TIME": v["time"]}})
                             link = Link(
@@ -667,6 +818,50 @@ def handle_GeoDB_endpoint(
                             )
                             item.add_link(link)
                             items.append(item)
+                        case "XYZ":
+                            # handler for NASA apis
+                            url = first_match["Url"]
+                            extra_fields = {}
+                            # replace time to a formatted version
+                            date_formatted = time_object.strftime(
+                                first_match.get("DateFormat", "%Y_%m_%d")
+                            )
+                            target_url = url.replace("{time}", date_formatted)
+                            if SiteMapping := first_match.get("SiteMapping"):
+                                # match with aoi_id
+                                site = SiteMapping.get(v["aoi_id"])
+                                # replace in URL
+                                if site:
+                                    target_url = target_url.replace("{site}", site)
+                                else:
+                                    LOGGER.info(
+                                        f"Warning: no match for SiteMapping in config for {site}"
+                                    )
+                            link = Link(
+                                rel="xyz",
+                                target=target_url,
+                                media_type="image/png",
+                                title=collection_config["Name"],
+                                extra_fields=extra_fields,
+                            )
+                            item.add_link(link)
+                            items.append(item)
+                elif endpoint_config.get("FeatureCollection"):
+                    # no input data match found, just add the item with asset only
+                    assets["geodbfeatures"] = Asset(
+                        href=f"{endpoint_config['EndPoint']}{endpoint_config['Database']}_{endpoint_config['FeatureCollection']}?aoi_id=eq.{v['aoi_id']}&time=eq.{v['time']}",
+                        media_type="application/geodb+json",
+                        roles=["data"],
+                    )
+                    item = Item(
+                        id=v["time"],
+                        bbox=bbox,
+                        properties={},
+                        geometry=geometry,
+                        datetime=time_object,
+                        assets=assets,
+                    )
+                    items.append(item)
             save_items(
                 locations_collection,
                 items,
@@ -690,6 +885,7 @@ def handle_GeoDB_endpoint(
         link.extra_fields["latlng"] = latlon
         link.extra_fields["country"] = country
         link.extra_fields["name"] = city
+        add_collection_information(catalog_config, locations_collection, collection_config)
 
     if "yAxis" not in collection_config:
         # fetch yAxis and store it to data, preventing need to save it per dataset in yml
@@ -781,19 +977,24 @@ def handle_WMS_endpoint(
     # Create an item per time to allow visualization in stac clients
     if len(datetimes) > 0:
         for dt in datetimes:
+            # case of wms interval coming from config
+            dt_item = dt[0] if isinstance(dt, list) else dt
             item = Item(
-                id=format_datetime_to_isostring_zulu(dt),
+                id=format_datetime_to_isostring_zulu(dt_item),
                 bbox=spatial_extent,
                 properties={},
                 geometry=create_geometry_from_bbox(spatial_extent),
-                datetime=dt,
+                datetime=dt_item,
                 stac_extensions=[
                     "https://stac-extensions.github.io/web-map-links/v1.1.0/schema.json",
                 ],
                 assets={"dummy_asset": Asset(href="")},
             )
             add_projection_info(endpoint_config, item)
-            add_visualization_info(item, collection_config, endpoint_config, datetimes=[dt])
+            dt_visualization = dt if isinstance(dt, list) else [dt]
+            add_visualization_info(
+                item, collection_config, endpoint_config, datetimes=dt_visualization
+            )
             items.append(item)
     else:
         LOGGER.warn(f"NO datetimes returned for collection: {collection_config['Name']}!")
@@ -878,7 +1079,6 @@ def add_visualization_info(
             start_isostring = format_datetime_to_isostring_zulu(dt)
             # SH WMS for public collections needs time interval, we use full day here
             end = dt + timedelta(days=1) - timedelta(milliseconds=1)
-            # we have start_datetime and end_datetime
             if len(datetimes) == 2:
                 end = datetimes[1]
             end_isostring = format_datetime_to_isostring_zulu(end)
@@ -916,7 +1116,13 @@ def add_visualization_info(
                     )
                 dimensions[key] = value
         if datetimes is not None:
-            dimensions["TIME"] = format_datetime_to_isostring_zulu(datetimes[0])
+            if len(datetimes) > 1:
+                start = format_datetime_to_isostring_zulu(datetimes[0])
+                end = format_datetime_to_isostring_zulu(datetimes[1])
+                interval = f"{start}/{end}"
+                dimensions["TIME"] = interval
+            else:
+                dimensions["TIME"] = format_datetime_to_isostring_zulu(datetimes[0])
         if dimensions != {}:
             extra_fields["wms:dimensions"] = dimensions
         if endpoint_config.get("Styles"):

{eodash_catalog-0.3.2 → eodash_catalog-0.3.4}/src/eodash_catalog/generate_indicators.py

@@ -19,6 +19,7 @@ from eodash_catalog.endpoints import (
     handle_collection_only,
     handle_custom_endpoint,
     handle_GeoDB_endpoint,
+    handle_GeoDB_Features_endpoint,
     handle_rasdaman_endpoint,
     handle_raw_source,
     handle_SH_endpoint,
@@ -241,6 +242,15 @@ def process_collection_file(
                         catalog,
                         options,
                     )
+                elif endpoint_config["Name"] == "GeoDB Features":
+                    collection = handle_GeoDB_Features_endpoint(
+                        catalog_config,
+                        endpoint_config,
+                        collection_config,
+                        coll_path_rel_to_root_catalog,
+                        catalog,
+                        options,
+                    )
                 elif endpoint_config["Name"] == "VEDA":
                     collection = handle_VEDA_endpoint(
                         catalog_config,

{eodash_catalog-0.3.2 → eodash_catalog-0.3.4}/src/eodash_catalog/stac_handling.py

@@ -18,6 +18,7 @@ from structlog import get_logger
 from eodash_catalog.utils import (
     generateDatetimesFromInterval,
     get_full_url,
+    make_intervals,
     parse_datestring_to_tz_aware_datetime,
     read_config_file,
 )
@@ -535,16 +536,22 @@ def add_extra_fields(
 def get_collection_datetimes_from_config(endpoint_config: dict) -> list[datetime]:
     times_datetimes: list[datetime] = []
     if endpoint_config:
+        interval_between_dates = endpoint_config.get("WMSIntervalsBetweenDates")
         if endpoint_config.get("Times"):
             times = list(endpoint_config.get("Times", []))
             times_datetimes = sorted(
                 [parse_datestring_to_tz_aware_datetime(time) for time in times]
             )
+            if interval_between_dates:
+                # convert to list of datetime_start and datetime_end
+                times_datetimes = make_intervals(times_datetimes)
         elif endpoint_config.get("DateTimeInterval"):
             start = endpoint_config["DateTimeInterval"].get("Start", "2020-09-01T00:00:00Z")
             end = endpoint_config["DateTimeInterval"].get("End", "2020-10-01T00:00:00Z")
             timedelta_config = endpoint_config["DateTimeInterval"].get("Timedelta", {"days": 1})
-            times_datetimes = generateDatetimesFromInterval(start, end, timedelta_config)
+            times_datetimes = generateDatetimesFromInterval(
+                start, end, timedelta_config, interval_between_dates
+            )
     return times_datetimes
 
 

{eodash_catalog-0.3.2 → eodash_catalog-0.3.4}/src/eodash_catalog/utils.py

@@ -214,7 +214,7 @@ def parse_duration(datestring):
 
 
 def generateDatetimesFromInterval(
-    start: str, end: str, timedelta_config: dict | None = None
+    start: str, end: str, timedelta_config: dict | None = None, interval_between_dates: bool = False
 ) -> list[datetime]:
     if timedelta_config is None:
         timedelta_config = {}
@@ -226,7 +226,10 @@ def generateDatetimesFromInterval(
     delta = timedelta(**timedelta_config)
     dates = []
     while start_dt <= end_dt:
-        dates.append(start_dt)
+        if interval_between_dates:
+            dates.append([start_dt, start_dt + delta - timedelta(seconds=1)])
+        else:
+            dates.append(start_dt)
         start_dt += delta
     return dates
 
@@ -444,8 +447,9 @@ def update_extents_from_collection_children(collection: Collection):
         ):
             individual_datetimes.extend(c_child.extent.temporal.intervals[0])  # type: ignore
     individual_datetimes = list(filter(lambda x: x is not None, individual_datetimes))
-    time_extent = [min(individual_datetimes), max(individual_datetimes)]
-    collection.extent.temporal = TemporalExtent([time_extent])
+    if individual_datetimes:
+        time_extent = [min(individual_datetimes), max(individual_datetimes)]
+        collection.extent.temporal = TemporalExtent([time_extent])
 
 
 def extract_extent_from_geoparquet(table) -> tuple[TemporalExtent, SpatialExtent]:
@@ -615,3 +619,25 @@ def merge_bboxes(bboxes: list[list[float]]) -> list[float]:
     max_lat = max(b[3] for b in bboxes)
 
     return [min_lon, min_lat, max_lon, max_lat]
+
+
+def make_intervals(datetimes: list[datetime]) -> list[list[datetime]]:
+    """
+    Converts a list of datetimes into list of lists of datetimes in format of [start,end]
+    where end is next element in original list minus 1 second
+    """
+    intervals = []
+    n = len(datetimes)
+    for i in range(n):
+        start = datetimes[i]
+        if i < n - 1:
+            # end is next datetime minus one second
+            end = datetimes[i + 1] - timedelta(seconds=1)
+        else:
+            prev_interval = timedelta(seconds=0)
+            # last item: use previous interval length added to last start
+            if n > 1:
+                prev_interval = datetimes[-1] - datetimes[-2]
+            end = start + prev_interval
+        intervals.append([start, end])
+    return intervals