Merge branch 'master' of https://github.com/ipeaGIT/geobr

python-package/geobr/list_geobr.py

@@ -2,7 +2,7 @@
 import pandas as pd
 from io import StringIO
 from urllib.error import HTTPError
-
+import re
 
 def list_geobr():
     """Prints available functions, according to latest README.md file
@@ -19,8 +19,12 @@ def list_geobr():
 
     try:
         html_data = get("https://github.com/ipeaGIT/geobr/blob/master/README.md").text
-
-        df = pd.read_html(StringIO(html_data))[1]
+        find_emoji = html_data.index("👉")
+        html_data =  html_data[find_emoji:]
+        escaped_data = html_data.replace("\\u003c", "<").replace("\\u003e", ">")
+        tables = re.findall("<table>(.+?)</table>", escaped_data)
+        available_datasets = "<table>" + tables[0].replace("\\n", "") + "</table>"
+        df = pd.DataFrame(pd.read_html(StringIO(available_datasets))[0])
 
     except HTTPError:
         print(

python-package/geobr/read_health_facilities.py

@@ -1,7 +1,7 @@
 from geobr.utils import select_metadata, download_gpkg
 
 
-def read_health_facilities(verbose=False):
+def read_health_facilities(date=202303, verbose=False):
     """ Download geolocated data of health facilities as an sf object.
     
      Data comes from the National Registry of Healthcare facilities (Cadastro Nacional de Estabelecimentos de Saude - CNES),
@@ -19,6 +19,8 @@ def read_health_facilities(verbose=False):
 
     Parameters
     ----------
+    date : Numeric.  Date of the data in YYYYMM format. Defaults to `202303`,
+        which was the latest data available by the time of this update.
     verbose : bool, optional
         by default False
     
@@ -40,7 +42,7 @@ def read_health_facilities(verbose=False):
     >>> df = read_health_facilities()
     """
 
-    metadata = select_metadata("health_facilities", year=2015, simplified=False)
+    metadata = select_metadata("health_facilities", year=date, simplified=False)
 
     gdf = download_gpkg(metadata)
 

python-package/geobr/utils.py

@@ -111,7 +111,7 @@ def select_year(metadata, year):
         years = ", ".join([str(i) for i in metadata["year"].unique()])
 
         raise Exception(
-            "Error: Invalid Value to argument year. "
+            "Error: Invalid Value to argument 'year/date'. "
             "It must be one of the following: "
             f'{_get_unique_values(metadata, "year")}'
         )

python-package/pyproject.toml

@@ -25,6 +25,7 @@ shapely = "^1.7.0"
 python = "^3.9"
 requests = "^2.25.1"
 lxml = "^5.1.0"
+html5lib = "1.1"
 
 [tool.poetry.dev-dependencies]
 pytest = "^6.0"

python-package/tests/test_list_geobr.py

@@ -2,9 +2,7 @@
 
 
 def test_list_geobr(capsys):
-
     list_geobr()
-
     # Tests whether the function prints output
     captured = capsys.readouterr()
     assert len(captured.out) > 200

python-package/tests/test_read_health_facilities.py

@@ -9,7 +9,10 @@ def test_read_health_facilities():
 
     assert isinstance(df, gpd.geodataframe.GeoDataFrame)
 
-    assert len(df) == 360177
+    assert len(df) == 517629
 
     with pytest.raises(Exception):
         read_health_facilities(year=9999999)
+
+    with pytest.raises(Exception):
+        read_health_facilities(year="banana")