# -*- coding: utf-8 -*-
# Copyright 2018, CS GROUP - France, https://www.csgroup.eu/
#
# This file is part of EODAG project
# https://www.github.com/CS-SI/EODAG
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import json
import logging
import re
from urllib.error import HTTPError as urllib_HTTPError
from urllib.request import urlopen
import requests
from jsonpath_ng.ext import parse
from lxml import etree
from eodag.api.product import EOProduct
from eodag.api.product.metadata_mapping import (
NOT_MAPPED,
format_metadata,
get_metadata_path,
get_metadata_path_value,
get_search_param,
properties_from_json,
properties_from_xml,
)
from eodag.config import load_stac_provider_config
from eodag.plugins.search.base import Search
from eodag.utils import (
GENERIC_PRODUCT_TYPE,
dict_items_recursive_apply,
format_dict_items,
quote,
update_nested_dict,
urlencode,
)
from eodag.utils.exceptions import AuthenticationError, MisconfiguredError, RequestError
logger = logging.getLogger("eodag.plugins.search.qssearch")
[docs]class QueryStringSearch(Search):
"""A plugin that helps implementing any kind of search protocol that relies on
query strings (e.g: opensearch).
The available configuration parameters for this kind of plugin are:
- **result_type**: (optional) One of "json" or "xml", depending on the
representation of the provider's search results. The default is "json"
- **results_entry**: (mandatory) The name of the key in the provider search
result that gives access to the result entries
- **api_endpoint**: (mandatory) The endpoint of the provider's search interface
- **literal_search_params**: (optional) A mapping of (search_param =>
search_value) pairs giving search parameters to be passed as is in the search
url query string. This is useful for example in situations where the user wants
to pass-in a search query as it is done on the provider interface. In such a case,
the user can put in his configuration file the query he needs to pass to the provider.
- **pagination**: (mandatory) The configuration of how the pagination is done
on the provider. It is a tree with the following nodes:
- *next_page_url_tpl*: The template for pagination requests. This is a simple
Python format string which will be resolved using the following keywords:
``url`` (the base url of the search endpoint), ``search`` (the query string
corresponding to the search request), ``items_per_page`` (the number of
items to return per page), ``skip`` (the number of items to skip) or
``skip_base_1`` (the number of items to skip, starting from 1) and
``page`` (which page to return).
- *total_items_nb_key_path*: (optional) An XPath or JsonPath leading to the
total number of results satisfying a request. This is used for providers
which provides the total results metadata along with the result of the
query and don't have an endpoint for querying the number of items
satisfying a request, or for providers for which the count endpoint returns
a json or xml document
- *count_endpoint*: (optional) The endpoint for counting the number of items
satisfying a request
- *next_page_url_key_path*: (optional) A JSONPATH expression used to retrieve
the URL of the next page in the response of the current page.
- **free_text_search_operations**: (optional) A tree structure of the form::
<search-param>: # e.g: $search
union: # how to join the operations below (e.g: ' AND ' -->
# '(op1 AND op2) AND (op3 OR op4)')
wrapper: # a pattern for how each operation will be wrapped
# (e.g: '({})' --> '(op1 AND op2)')
operations: # The operations to build
<opname>: # e.g: AND
- <op1> # e.g:
# 'sensingStartDate:[{startTimeFromAscendingNode}Z TO *]'
- <op2> # e.g:
# 'sensingStopDate:[* TO {completionTimeFromAscendingNode}Z]'
...
...
...
With the structure above, each operation will become a string of the form:
'(<op1> <opname> <op2>)', then the operations will be joined together using
the union string and finally if the number of operations is greater than 1,
they will be wrapped as specified by the wrapper config key.
The search plugins of this kind can detect when a metadata mapping is "query-able",
and get the semantics of how to format the query string parameter that enables to
make a query on the corresponding metadata. To make a metadata query-able, just
configure it in the metadata mapping to be a list of 2 items, the first one being
the specification of the query string search formatting. The later is a string
following the specification of Python string formatting, with a special behaviour
added to it. For example, an entry in the metadata mapping of this kind::
completionTimeFromAscendingNode:
- 'f=acquisition.endViewingDate:lte:{completionTimeFromAscendingNode#timestamp}'
- '$.properties.acquisition.endViewingDate'
means that the search url will have a query string parameter named *"f"* with a
value of *"acquisition.endViewingDate:lte:1543922280.0"* if the search was done
with the value of ``completionTimeFromAscendingNode`` being
``2018-12-04T12:18:00``. What happened is that
``{completionTimeFromAscendingNode#timestamp}`` was replaced with the timestamp
of the value of ``completionTimeFromAscendingNode``. This example shows all there
is to know about the semantics of the query string formatting introduced by this
plugin: any eodag search parameter can be referenced in the query string
with an additional optional conversion function that is separated from it by a
``#`` (see :func:`~eodag.utils.format_metadata` for further details on the
available converters). Note that for the values in the
``free_text_search_operations`` configuration parameter follow the same rule.
:param provider: An eodag providers configuration dictionary
:type provider: dict
:param config: Path to the user configuration file
:type config: str
"""
COMPLEX_QS_REGEX = re.compile(r"^(.+=)?([^=]*)({.+})+([^=&]*)$")
DEFAULT_ITEMS_PER_PAGE = 10
extract_properties = {"xml": properties_from_xml, "json": properties_from_json}
[docs] def __init__(self, provider, config):
super(QueryStringSearch, self).__init__(provider, config)
self.config.__dict__.setdefault("result_type", "json")
self.config.__dict__.setdefault("results_entry", "features")
self.config.__dict__.setdefault("pagination", {})
self.config.__dict__.setdefault("free_text_search_operations", {})
self.search_urls = []
self.query_params = dict()
self.query_string = ""
self.next_page_url = None
self.next_page_query_obj = None
self.next_page_merge = None
def clear(self):
"""Clear search context"""
super().clear()
self.search_urls.clear()
self.query_params.clear()
self.query_string = ""
self.next_page_url = None
self.next_page_query_obj = None
self.next_page_merge = None
def query(self, items_per_page=None, page=None, count=True, **kwargs):
"""Perform a search on an OpenSearch-like interface
:param items_per_page: (optional) The number of results that must appear in one
single page
:type items_per_page: int
:param page: (optional) The page number to return
:type page: int
:param count: (optional) To trigger a count request
:type count: bool
"""
product_type = kwargs.get("productType", None)
if product_type == GENERIC_PRODUCT_TYPE:
logger.warning(
"GENERIC_PRODUCT_TYPE is not a real product_type and should only be used internally as a template"
)
return [], 0
provider_product_type = self.map_product_type(product_type)
keywords = {k: v for k, v in kwargs.items() if k != "auth" and v is not None}
keywords["productType"] = (
provider_product_type
if (provider_product_type and provider_product_type != GENERIC_PRODUCT_TYPE)
else product_type
)
# provider product type specific conf
self.product_type_def_params = self.get_product_type_def_params(
product_type, **kwargs
)
# update config using provider product type definition metadata_mapping
# from another product
other_product_for_mapping = self.product_type_def_params.get(
"metadata_mapping_from_product", ""
)
if other_product_for_mapping:
other_product_type_def_params = self.get_product_type_def_params(
other_product_for_mapping, **kwargs
)
self.update_metadata_mapping(
other_product_type_def_params.get("metadata_mapping", {})
)
# from current product
self.update_metadata_mapping(
self.product_type_def_params.get("metadata_mapping", {})
)
# Add to the query, the queryable parameters set in the provider product type definition
self.product_type_def_params = self.get_product_type_def_params(
product_type, **kwargs
)
# if product_type_def_params is set, remove product_type as it may conflict with this conf
if self.product_type_def_params:
keywords.pop("productType", None)
keywords.update(
{
k: v
for k, v in self.product_type_def_params.items()
if k not in keywords.keys()
and k in self.config.metadata_mapping.keys()
and isinstance(self.config.metadata_mapping[k], list)
}
)
qp, qs = self.build_query_string(product_type, **keywords)
self.query_params = qp
self.query_string = qs
self.search_urls, total_items = self.collect_search_urls(
page=page, items_per_page=items_per_page, count=count, **kwargs
)
provider_results = self.do_search(items_per_page=items_per_page, **kwargs)
eo_products = self.normalize_results(provider_results, **kwargs)
total_items = len(eo_products) if total_items == 0 else total_items
return eo_products, total_items
def update_metadata_mapping(self, metadata_mapping):
"""Update plugin metadata_mapping with input metadata_mapping configuration"""
self.config.metadata_mapping.update(metadata_mapping)
for metadata in metadata_mapping:
path = get_metadata_path_value(self.config.metadata_mapping[metadata])
# check if path has already been parsed
if isinstance(path, str):
conversion, path = get_metadata_path(
self.config.metadata_mapping[metadata]
)
try:
# If the metadata is queryable (i.e a list of 2 elements), replace the value of the last item
if len(self.config.metadata_mapping[metadata]) == 2:
self.config.metadata_mapping[metadata][1] = (
conversion,
parse(path),
)
else:
self.config.metadata_mapping[metadata] = (
conversion,
parse(path),
)
except Exception: # jsonpath_ng does not provide a proper exception
# Assume the mapping is to be passed as is.
# Ignore any transformation specified. If a value is to be passed as is, we don't want to transform
# it further
_, text = get_metadata_path(self.config.metadata_mapping[metadata])
if len(self.config.metadata_mapping[metadata]) == 2:
self.config.metadata_mapping[metadata][1] = (None, text)
else:
self.config.metadata_mapping[metadata] = (None, text)
# Put the updated mapping at the end
self.config.metadata_mapping[
metadata
] = self.config.metadata_mapping.pop(metadata)
def build_query_string(self, product_type, **kwargs):
"""Build The query string using the search parameters"""
logger.debug("Building the query string that will be used for search")
if "raise_errors" in kwargs.keys():
del kwargs["raise_errors"]
# . not allowed in eodag_search_key, replaced with %2E
kwargs = {k.replace(".", "%2E"): v for k, v in kwargs.items()}
queryables = self.get_queryables(kwargs)
query_params = {}
# Get all the search parameters that are recognised as queryables by the
# provider (they appear in the queryables dictionary)
for eodag_search_key, provider_search_key in queryables.items():
user_input = kwargs[eodag_search_key]
if self.COMPLEX_QS_REGEX.match(provider_search_key):
parts = provider_search_key.split("=")
if len(parts) == 1:
formatted_query_param = format_metadata(
provider_search_key, product_type, **kwargs
)
if "{{" in provider_search_key:
# json query string (for POST request)
update_nested_dict(
query_params, json.loads(formatted_query_param)
)
else:
query_params[eodag_search_key] = formatted_query_param
else:
provider_search_key, provider_value = parts
query_params.setdefault(provider_search_key, []).append(
format_metadata(provider_value, product_type, **kwargs)
)
else:
query_params[provider_search_key] = user_input
# Now get all the literal search params (i.e params to be passed "as is"
# in the search request)
# ignore additional_params if it isn't a dictionary
literal_search_params = getattr(self.config, "literal_search_params", {})
if not isinstance(literal_search_params, dict):
literal_search_params = {}
# Now add formatted free text search parameters (this is for cases where a
# complex query through a free text search parameter is available for the
# provider and needed for the consumer)
literal_search_params.update(self.format_free_text_search(**kwargs))
for provider_search_key, provider_value in literal_search_params.items():
if isinstance(provider_value, list):
query_params.setdefault(provider_search_key, []).extend(provider_value)
else:
query_params.setdefault(provider_search_key, []).append(provider_value)
# Build the final query string, in one go without quoting it
# (some providers do not operate well with urlencoded and quoted query strings)
return (
query_params,
urlencode(
query_params, doseq=True, quote_via=lambda x, *_args, **_kwargs: x
),
)
def format_free_text_search(self, **kwargs):
"""Build the free text search parameter using the search parameters"""
query_params = {}
for param, operations_config in self.config.free_text_search_operations.items():
union = operations_config["union"]
wrapper = operations_config.get("wrapper", "{}")
formatted_query = []
for operator, operands in operations_config["operations"].items():
# The Operator string is the operator wrapped with spaces
operator = " {} ".format(operator)
# Build the operation string by joining the formatted operands together
# using the operation string
operation_string = operator.join(
format_metadata(operand, **kwargs)
for operand in operands
if any(
kw in operand and val is not None for kw, val in kwargs.items()
)
)
# Finally wrap the operation string as specified by the wrapper and add
# it to the list of queries (only if the operation string is not empty)
if operation_string:
query = wrapper.format(operation_string)
formatted_query.append(query)
# Join the formatted query using the "union" config parameter, and then
# wrap it with the Python format string specified in the "wrapper" config
# parameter
final_query = union.join(formatted_query)
if len(operations_config["operations"]) > 1 and len(formatted_query) > 1:
final_query = wrapper.format(query_params[param])
if final_query:
query_params[param] = final_query
return query_params
def get_queryables(self, search_params):
"""Retrieve the metadata mappings that are query-able"""
logger.debug("Retrieving queryable metadata from metadata_mapping")
queryables = {}
for eodag_search_key, user_input in search_params.items():
if user_input is not None:
md_mapping = self.config.metadata_mapping.get(
eodag_search_key, (None, NOT_MAPPED)
)
_, md_value = md_mapping
# query param from defined metadata_mapping
if md_mapping is not None and isinstance(md_mapping, list):
search_param = get_search_param(md_mapping)
if search_param is not None:
queryables[eodag_search_key] = search_param
# query param from metadata auto discovery
elif md_value == NOT_MAPPED and getattr(
self.config, "discover_metadata", {}
).get("auto_discovery", False):
pattern = re.compile(
self.config.discover_metadata.get("metadata_pattern", "")
)
search_param_cfg = self.config.discover_metadata.get(
"search_param", ""
)
if pattern.match(eodag_search_key) and isinstance(
search_param_cfg, str
):
search_param = search_param_cfg.format(
metadata=eodag_search_key
)
queryables[eodag_search_key] = search_param
elif pattern.match(eodag_search_key) and isinstance(
search_param_cfg, dict
):
search_param_cfg_parsed = dict_items_recursive_apply(
search_param_cfg,
lambda k, v: v.format(metadata=eodag_search_key),
)
for k, v in search_param_cfg_parsed.items():
if getattr(self.config, k, None):
update_nested_dict(
getattr(self.config, k), v, extend_list_values=True
)
else:
logger.warning(
"Could not use discover_metadata[search_param]: no entry for %s in plugin config",
k,
)
return queryables
def collect_search_urls(self, page=None, items_per_page=None, count=True, **kwargs):
"""Build paginated urls"""
urls = []
total_results = 0 if count else None
for collection in self.get_collections(**kwargs):
# skip empty collection if one is required in api_endpoint
if "{collection}" in self.config.api_endpoint and not collection:
continue
search_endpoint = self.config.api_endpoint.rstrip("/").format(
collection=collection
)
if page is not None and items_per_page is not None:
if count:
count_endpoint = self.config.pagination.get(
"count_endpoint", ""
).format(collection=collection)
if count_endpoint:
count_url = "{}?{}".format(count_endpoint, self.query_string)
_total_results = self.count_hits(
count_url, result_type=self.config.result_type
)
else:
# First do one request querying only one element (lightweight
# request to schedule the pagination)
next_url_tpl = self.config.pagination["next_page_url_tpl"]
count_url = next_url_tpl.format(
url=search_endpoint,
search=self.query_string,
items_per_page=1,
page=1,
skip=0,
skip_base_1=1,
)
_total_results = self.count_hits(
count_url, result_type=self.config.result_type
)
total_results += _total_results or 0
next_url = self.config.pagination["next_page_url_tpl"].format(
url=search_endpoint,
search=self.query_string,
items_per_page=items_per_page,
page=page,
skip=(page - 1) * items_per_page,
skip_base_1=(page - 1) * items_per_page + 1,
)
else:
next_url = "{}?{}".format(search_endpoint, self.query_string)
urls.append(next_url)
return urls, total_results
def do_search(self, items_per_page=None, **kwargs):
"""Perform the actual search request.
If there is a specified number of items per page, return the results as soon
as this number is reached
:param items_per_page: (optional) The number of items to return for one page
:type items_per_page: int
"""
results = []
for search_url in self.search_urls:
try:
response = self._request(
search_url,
info_message="Sending search request: {}".format(search_url),
exception_message="Skipping error while searching for {} {} "
"instance:".format(self.provider, self.__class__.__name__),
)
except RequestError:
# Signal the end of iteration (see PEP-479)
return
else:
next_page_url_key_path = self.config.pagination.get(
"next_page_url_key_path", None
)
next_page_query_obj_key_path = self.config.pagination.get(
"next_page_query_obj_key_path", None
)
next_page_merge_key_path = self.config.pagination.get(
"next_page_merge_key_path", None
)
if self.config.result_type == "xml":
root_node = etree.fromstring(response.content)
namespaces = {k or "ns": v for k, v in root_node.nsmap.items()}
result = [
etree.tostring(entry)
for entry in root_node.xpath(
self.config.results_entry, namespaces=namespaces
)
]
if next_page_url_key_path or next_page_query_obj_key_path:
raise NotImplementedError(
"Setting the next page url from an XML response has not "
"been implemented yet"
)
else:
resp_as_json = response.json()
if next_page_url_key_path:
path_parsed = parse(next_page_url_key_path)
try:
self.next_page_url = path_parsed.find(resp_as_json)[0].value
logger.debug(
"Next page URL collected and set for the next search",
)
except IndexError:
logger.debug("Next page URL could not be collected")
if next_page_query_obj_key_path:
path_parsed = parse(next_page_query_obj_key_path)
try:
self.next_page_query_obj = path_parsed.find(resp_as_json)[
0
].value
logger.debug(
"Next page Query-object collected and set for the next search",
)
except IndexError:
logger.debug(
"Next page Query-object could not be collected"
)
if next_page_merge_key_path:
path_parsed = parse(next_page_merge_key_path)
try:
self.next_page_merge = path_parsed.find(resp_as_json)[
0
].value
logger.debug(
"Next page merge collected and set for the next search",
)
except IndexError:
logger.debug("Next page merge could not be collected")
result = resp_as_json.get(self.config.results_entry, [])
if getattr(self.config, "merge_responses", False):
results = (
[dict(r, **result[i]) for i, r in enumerate(results)]
if results
else result
)
else:
results.extend(result)
if items_per_page is not None and len(results) == items_per_page:
return results
return results
def normalize_results(self, results, **kwargs):
"""Build EOProducts from provider results"""
normalize_remaining_count = len(results)
logger.debug(
"Adapting %s plugin results to eodag product representation"
% normalize_remaining_count
)
products = []
for result in results:
product = EOProduct(
self.provider,
QueryStringSearch.extract_properties[self.config.result_type](
result,
self.config.metadata_mapping,
discovery_pattern=getattr(self.config, "discover_metadata", {}).get(
"metadata_pattern", None
),
discovery_path=getattr(self.config, "discover_metadata", {}).get(
"metadata_path", "null"
),
),
**kwargs
)
# use product_type_config as default properties
product.properties = dict(
getattr(self.config, "product_type_config", {}), **product.properties
)
products.append(product)
return products
def count_hits(self, count_url, result_type="json"):
"""Count the number of results satisfying some criteria"""
# Handle a very annoying special case :'(
url = count_url.replace("$format=json&", "")
response = self._request(
url,
info_message="Sending count request: {}".format(url),
exception_message="Skipping error while counting results for {} {} "
"instance:".format(self.provider, self.__class__.__name__),
)
if result_type == "xml":
root_node = etree.fromstring(response.content)
total_nb_results = root_node.xpath(
self.config.pagination["total_items_nb_key_path"],
namespaces={k or "ns": v for k, v in root_node.nsmap.items()},
)[0]
total_results = int(total_nb_results)
else:
count_results = response.json()
if isinstance(count_results, dict):
path_parsed = parse(self.config.pagination["total_items_nb_key_path"])
total_results = path_parsed.find(count_results)[0].value
else: # interpret the result as a raw int
total_results = int(count_results)
return total_results
def get_collections(self, **kwargs):
"""Get the collection to which the product belongs"""
# See https://earth.esa.int/web/sentinel/missions/sentinel-2/news/-
# /asset_publisher/Ac0d/content/change-of
# -format-for-new-sentinel-2-level-1c-products-starting-on-6-december
product_type = kwargs.get("productType")
if product_type is None and not self.product_type_def_params:
collections = set()
collection = getattr(self.config, "collection", None)
if collection is None:
try:
for product_type, product_config in self.config.products.items():
if product_type != GENERIC_PRODUCT_TYPE:
collections.add(product_config["collection"])
else:
collections.add(
format_dict_items(product_config, **kwargs).get(
"collection", ""
)
)
except KeyError:
collections.add("")
else:
collections.add(collection)
return tuple(collections)
if self.provider == "peps":
if product_type == "S2_MSI_L1C":
date = kwargs.get("startTimeFromAscendingNode")
# If there is no criteria on date, we want to query all the collections
# known for providing L1C products
if date is None:
collections = ("S2", "S2ST")
else:
match = re.match(
r"(?P<year>\d{4})-(?P<month>\d{2})-(?P<day>\d{2})", date
).groupdict()
year, month, day = (
int(match["year"]),
int(match["month"]),
int(match["day"]),
)
if year > 2016 or (year == 2016 and month == 12 and day > 5):
collections = ("S2ST",)
else:
collections = ("S2", "S2ST")
else:
collections = (self.product_type_def_params.get("collection", ""),)
else:
collection = getattr(self.config, "collection", None)
if collection is None:
collection = (
self.product_type_def_params.get("collection", None) or product_type
)
collections = (
(collection,) if not isinstance(collection, list) else tuple(collection)
)
return collections
def map_product_type(self, product_type, **kwargs):
"""Map the eodag product type to the provider product type"""
if product_type is None:
return
logger.debug("Mapping eodag product type to provider product type")
return self.config.products.get(product_type, {}).get(
"productType", GENERIC_PRODUCT_TYPE
)
def get_product_type_def_params(self, product_type, **kwargs):
"""Get the provider product type definition parameters"""
if product_type in self.config.products.keys():
logger.debug(
"Getting provider product type definition parameters for %s",
product_type,
)
return self.config.products[product_type]
elif GENERIC_PRODUCT_TYPE in self.config.products.keys():
logger.debug(
"Getting genric provider product type definition parameters for %s",
product_type,
)
return {
k: v
for k, v in format_dict_items(
self.config.products[GENERIC_PRODUCT_TYPE], **kwargs
).items()
if v
}
else:
return {}
def _request(self, url, info_message=None, exception_message=None):
try:
# auth if needed
kwargs = {}
if (
getattr(self.config, "need_auth", False)
and hasattr(self, "auth")
and callable(self.auth)
):
kwargs["auth"] = self.auth
# requests auto quote url params, without any option to prevent it
# use urllib instead of requests if req must be sent unquoted
if hasattr(self.config, "dont_quote"):
# keep unquoted desired params
base_url, params = url.split("?")
qry = quote(params)
for keep_unquoted in self.config.dont_quote:
qry = qry.replace(quote(keep_unquoted), keep_unquoted)
# prepare req for Response building
req = requests.Request(method="GET", url=base_url, **kwargs)
prep = req.prepare()
prep.url = base_url + "?" + qry
# send urllib req
if info_message:
logger.info(info_message.replace(url, prep.url))
urllib_response = urlopen(prep.url)
# py2 compatibility : prevent AttributeError: addinfourl instance has no attribute 'reason'
if not hasattr(urllib_response, "reason"):
urllib_response.reason = ""
if not hasattr(urllib_response, "status") and hasattr(
urllib_response, "code"
):
urllib_response.status = urllib_response.code
# build Response
adapter = requests.adapters.HTTPAdapter()
response = adapter.build_response(prep, urllib_response)
else:
if info_message:
logger.info(info_message)
response = requests.get(url)
response.raise_for_status()
except (requests.HTTPError, urllib_HTTPError) as err:
err_msg = err.readlines() if hasattr(err, "readlines") else ""
if exception_message:
logger.exception("%s %s" % (exception_message, err_msg))
else:
logger.exception(
"Skipping error while requesting: %s (provider:%s, plugin:%s): %s",
url,
self.provider,
self.__class__.__name__,
err_msg,
)
raise RequestError(str(err))
return response
[docs]class AwsSearch(QueryStringSearch):
"""A specialisation of RestoSearch that modifies the way the EOProducts are built
from the search results"""
def normalize_results(self, results, **kwargs):
"""Transform metadata from provider representation to eodag representation"""
normalized = []
logger.debug("Adapting plugin results to eodag product representation")
for result in results:
ref = result["properties"]["title"].split("_")[5]
year = result["properties"]["completionDate"][0:4]
month = str(int(result["properties"]["completionDate"][5:7]))
day = str(int(result["properties"]["completionDate"][8:10]))
properties = QueryStringSearch.extract_properties[self.config.result_type](
result, self.config.metadata_mapping
)
properties["downloadLink"] = (
"s3://tiles/{ref[1]}{ref[2]}/{ref[3]}/{ref[4]}{ref[5]}/{year}/"
"{month}/{day}/0/"
).format(**locals())
normalized.append(EOProduct(self.provider, properties, **kwargs))
return normalized
[docs]class ODataV4Search(QueryStringSearch):
"""A specialisation of a QueryStringSearch that does a two step search to retrieve
all products metadata"""
def do_search(self, *args, **kwargs):
"""Do a two step search, as the metadata are not given into the search result"""
# TODO: This plugin is still very specific to the ONDA provider.
# Be careful to generalize it if needed when the chance to do so arrives
final_result = []
# Query the products entity set for basic metadata about the product
for entity in super(ODataV4Search, self).do_search(*args, **kwargs):
metadata_url = self.get_metadata_search_url(entity)
try:
logger.debug("Sending metadata request: %s", metadata_url)
response = requests.get(metadata_url)
response.raise_for_status()
except requests.HTTPError:
logger.exception(
"Skipping error while searching for %s %s instance:",
self.provider,
self.__class__.__name__,
)
else:
entity.update(
{item["id"]: item["value"] for item in response.json()["value"]}
)
final_result.append(entity)
return final_result
def get_metadata_search_url(self, entity):
"""Build the metadata link for the given entity"""
return "{}({})/Metadata".format(
self.config.api_endpoint.rstrip("/"), entity["id"]
)
[docs]class PostJsonSearch(QueryStringSearch):
"""A specialisation of a QueryStringSearch that uses POST method"""
[docs] def __init__(self, provider, config):
super(PostJsonSearch, self).__init__(provider, config)
self.config.results_entry = "results"
def query(self, items_per_page=None, page=None, count=True, **kwargs):
"""Perform a search on an OpenSearch-like interface"""
product_type = kwargs.get("productType", None)
provider_product_type = self.map_product_type(product_type, **kwargs)
keywords = {k: v for k, v in kwargs.items() if k != "auth" and v is not None}
if provider_product_type and provider_product_type != GENERIC_PRODUCT_TYPE:
keywords["productType"] = provider_product_type
elif product_type:
keywords["productType"] = product_type
# provider product type specific conf
self.product_type_def_params = self.get_product_type_def_params(
product_type, **kwargs
)
# update config using provider product type definition metadata_mapping
# from another product
other_product_for_mapping = self.product_type_def_params.get(
"metadata_mapping_from_product", ""
)
if other_product_for_mapping:
other_product_type_def_params = self.get_product_type_def_params(
other_product_for_mapping, **kwargs
)
self.update_metadata_mapping(
other_product_type_def_params.get("metadata_mapping", {})
)
# from current product
self.update_metadata_mapping(
self.product_type_def_params.get("metadata_mapping", {})
)
# Add to the query, the queryable parameters set in the provider product type definition
keywords.update(
{
k: v
for k, v in self.product_type_def_params.items()
if k not in keywords.keys()
and k in self.config.metadata_mapping.keys()
and isinstance(self.config.metadata_mapping[k], list)
}
)
qp, _ = self.build_query_string(product_type, **keywords)
for query_param, query_value in qp.items():
if (
query_param
in self.config.products.get(product_type, {}).get(
"specific_qssearch", {"parameters": []}
)["parameters"]
):
self.config.api_endpoint = query_value
self.config.metadata_mapping = self.config.products[product_type][
"specific_qssearch"
]["metadata_mapping"]
self.config.results_entry = self.config.products[product_type][
"specific_qssearch"
]["results_entry"]
self.config.collection = self.config.products[product_type][
"specific_qssearch"
].get("collection", None)
self.config.merge_responses = self.config.products[product_type][
"specific_qssearch"
].get("merge_responses", None)
super(PostJsonSearch, self).__init__(
provider=self.provider, config=self.config
)
self.count_hits = lambda *x, **y: 1
self._request = super(PostJsonSearch, self)._request
return super(PostJsonSearch, self).query(
items_per_page=items_per_page, page=page, **kwargs
)
# If we were not able to build query params but have search criteria, this means
# the provider does not support the search criteria given. If so, stop searching
# right away
if not qp and keywords:
return [], 0
self.query_params = qp
self.search_urls, total_items = self.collect_search_urls(
page=page, items_per_page=items_per_page, count=count, **kwargs
)
provider_results = self.do_search(items_per_page=items_per_page, **kwargs)
eo_products = self.normalize_results(provider_results, **kwargs)
total_items = len(eo_products) if total_items == 0 else total_items
return eo_products, total_items
def collect_search_urls(self, page=None, items_per_page=None, count=True, **kwargs):
"""Adds pagination to query parameters, and auth to url"""
urls = []
total_results = 0 if count else None
if hasattr(kwargs["auth"], "config"):
auth_conf_dict = getattr(kwargs["auth"].config, "credentials", {})
else:
auth_conf_dict = {}
for collection in self.get_collections(**kwargs):
try:
search_endpoint = self.config.api_endpoint.rstrip("/").format(
**dict(collection=collection, **auth_conf_dict)
)
except KeyError as e:
raise MisconfiguredError(
"Missing %s in %s configuration"
% (",".join(e.args), kwargs["auth"].provider)
)
if page is not None and items_per_page is not None:
if count:
count_endpoint = self.config.pagination.get(
"count_endpoint", ""
).format(**dict(collection=collection, **auth_conf_dict))
if count_endpoint:
_total_results = self.count_hits(
count_endpoint, result_type=self.config.result_type
)
else:
# Update the query params with a pagination requesting 1 product only
# which is enough to obtain the count hits.
count_pagination_params = dict(
items_per_page=1, page=1, skip=0, skip_base_1=1
)
count_params = json.loads(
self.config.pagination["next_page_query_obj"].format(
**count_pagination_params
)
)
update_nested_dict(self.query_params, count_params)
_total_results = self.count_hits(
search_endpoint, result_type=self.config.result_type
)
if getattr(self.config, "merge_responses", False):
total_results = _total_results or 0
else:
total_results += _total_results or 0
if isinstance(self.config.pagination["next_page_query_obj"], str):
# next_page_query_obj needs to be parsed
next_page_query_obj = self.config.pagination[
"next_page_query_obj"
].format(
items_per_page=items_per_page,
page=page,
skip=(page - 1) * items_per_page,
skip_base_1=(page - 1) * items_per_page + 1,
)
update_nested_dict(
self.query_params, json.loads(next_page_query_obj)
)
urls.append(search_endpoint)
return urls, total_results
def _request(self, url, info_message=None, exception_message=None):
try:
# auth if needed
kwargs = {}
if (
getattr(self.config, "need_auth", False)
and hasattr(self, "auth")
and callable(self.auth)
):
kwargs["auth"] = self.auth
# perform the request using the next page arguments if they are defined
if getattr(self, "next_page_query_obj", None):
self.query_params = self.next_page_query_obj
if info_message:
logger.info(info_message)
logger.debug("Query parameters: %s" % self.query_params)
response = requests.post(url, json=self.query_params, **kwargs)
response.raise_for_status()
except (requests.HTTPError, urllib_HTTPError) as err:
# check if error is identified as auth_error in provider conf
auth_errors = getattr(self.config, "auth_error_code", [None])
if not isinstance(auth_errors, list):
auth_errors = [auth_errors]
if err.response.status_code in auth_errors:
raise AuthenticationError(
"HTTP Error {} returned:\n{}\nPlease check your credentials for {}".format(
err.response.status_code,
err.response.text.strip(),
self.provider,
)
)
if exception_message:
logger.exception(exception_message)
else:
logger.exception(
"Skipping error while requesting: %s (provider:%s, plugin:%s):",
url,
self.provider,
self.__class__.__name__,
)
raise RequestError(str(err))
return response
[docs]class StacSearch(PostJsonSearch):
"""A specialisation of a QueryStringSearch that uses generic STAC configuration"""
[docs] def __init__(self, provider, config):
stac_provider_config = load_stac_provider_config()
# search config set to stac defaults overriden with provider config
config.__dict__["pagination"] = dict(
stac_provider_config.get("search", {}).get("pagination", {}),
**config.__dict__.get("pagination", {})
)
config.__dict__["discover_metadata"] = dict(
stac_provider_config.get("search", {}).get("discover_metadata", {}),
**config.__dict__.get("discover_metadata", {})
)
config.__dict__["metadata_mapping"] = dict(
stac_provider_config.get("search", {}).get("metadata_mapping", {}),
**config.__dict__.get("metadata_mapping", {})
)
super(StacSearch, self).__init__(provider, config)
self.config.results_entry = "features"
# TODO: fetch /collections to add dynamically product types
# response = self._request(
# config.api_endpoint.replace("/search", "/collections"),
# )
# collections = [c["id"] for c in response.json().get("collections", {}) if "id" in c.keys()]
# for c in collections:
# if c not in str(self.config.products):
# self.config.products[c] = {"productType": c}
def normalize_results(self, results, **kwargs):
"""Build EOProducts from provider results"""
products = super(StacSearch, self).normalize_results(results, **kwargs)
# move assets from properties to product's attr
for product in products:
product.assets = product.properties.pop("assets", [])
return products
