Fix parsing on github readme dataset table

ipeaGIT · Apr 23, 2024 · f6fd5af · f6fd5af
1 parent bda5a6a
commit f6fd5af
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 5 deletions.
diff --git a/python-package/geobr/list_geobr.py b/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(

diff --git a/python-package/tests/test_list_geobr.py b/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