Merge pull request #1089 from open-contracting/1088-guatemala

feat: add guatemala_bulk

docs/spiders.rst

@@ -505,6 +505,16 @@ Greece
 
    scrapy crawl greece_digiwhist
 
+Guatemala
+~~~~~~~~~
+
+.. autoclass:: kingfisher_scrapy.spiders.guatemala_bulk.GuatemalaBulk
+   :no-members:
+
+.. code-block:: bash
+
+   scrapy crawl guatemala_bulk
+
 Honduras
 ~~~~~~~~
 

kingfisher_scrapy/spiders/guatemala_bulk.py

@@ -0,0 +1,65 @@
+from datetime import datetime
+
+import scrapy
+
+from kingfisher_scrapy.base_spiders import SimpleSpider
+from kingfisher_scrapy.util import components, handle_http_error
+
+
+class GuatemalaBulk(SimpleSpider):
+    """
+    Domain
+      Ministerio de Finanzas Públicas - Dirección General de Adquisiciones del Estado
+    Spider arguments
+      from_date
+        Download only data from this month onward (YYYY-MM format).
+        If ``until_date`` is provided, defaults to '2020-01'.
+      until_date
+        Download only data until this month (YYYY-MM format).
+        If ``from_date`` is provided, defaults to the current month.
+    API documentation
+      https://ocds.guatecompras.gt/api-ocds
+    Bulk download documentation
+      https://ocds.guatecompras.gt/descarga-datos
+    """
+    name = 'guatemala_bulk'
+
+    # BaseSpider
+    date_format = 'year-month'
+    default_from_date = '2020-01'
+
+    # SimpleSpider
+    data_type = 'record_package'
+
+    def start_requests(self):
+        url = 'https://ocds.guatecompras.gt/files'
+        yield scrapy.Request(url, meta={'file_name': 'list.json'}, callback=self.parse_list)
+
+    @handle_http_error
+    def parse_list(self, response):
+        # An example of expected response is:
+        # {
+        #   "id": "gc-{year}-{month}"
+        #   "results": [
+        #     {
+        #      "files": {
+        #        "csv": "...",
+        #        "sha": "...",
+        #        "json": "...",
+        #        "xlsx": "..."
+        #      },
+        #      "year": "values between 2020 to the current year",
+        #      "month": "values between 1 and 12",
+        #      "monthName": "values between enero to diciembre",
+        #      "source": "Guatecompras",
+        #      "timestamp": "last updated date in timestamp with time zone format"
+        #     }, ...
+        #   ]
+        # }
+        for item in response.json()["result"]:
+            if self.from_date and self.until_date:
+                date = datetime(int(item['year']), int(item['month']), 1)
+                if not (self.from_date <= date <= self.until_date):
+                    continue
+
+            yield self.build_request(item['files']['json'], formatter=components(-2))

kingfisher_scrapy/spiders/italy_ministry_of_infrastructure_and_transport.py

@@ -10,10 +10,10 @@ class ItalyMinistryOfInfrastructureAndTransport(SimpleSpider):
       Public Contracts Service (SCP) of the Ministry of Infrastructure and Transport
     Spider arguments
       from_date
-        Download only data from this time onward (YYYY-MM-DD format).
+        Download only data from this date onward (YYYY-MM-DD format).
         If ``until_date`` is provided, defaults to '2022-01-01'.
       until_date
-        Download only data until this time (YYYY-MM-DD format).
+        Download only data until this date (YYYY-MM-DD format).
         If ``from_date`` is provided, defaults to today.
     Swagger API documentation
       https://www.serviziocontrattipubblici.it/ocds-ms/swagger-ui.html

kingfisher_scrapy/spiders/mexico_puebla_state_seseap_plataforma_digital_nacional.py

@@ -1,7 +1,7 @@
 from kingfisher_scrapy.spiders.mexico_plataforma_digital_nacional_base import MexicoPlataformaDigitalNacionalBase
 
 
-class MexicoPueblaStateSESEAPlataformaDigitalNacional(MexicoPlataformaDigitalNacionalBase):
+class MexicoPueblaStateSESEAPPlataformaDigitalNacional(MexicoPlataformaDigitalNacionalBase):
     """
     Domain
       Secretaría Ejecutiva del Sistema Estatal Anticorrupción del Estado de Puebla (SESEAP) (Mexico) -