Merge 361d6ba into 39cca97

coursera-dl · Apr 5, 2018 · 2a371cb · 2a371cb
2 parents 39cca97 + 361d6ba
commit 2a371cb
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 5 deletions.
diff --git a/edx_dl/edx_dl.py b/edx_dl/edx_dl.py
@@ -1019,12 +1019,11 @@ def main():
     _display_selections(selections)
 
     # Extract the unit information (downloadable resources)
-    # This parses the HTML of all the subsection.url and extracts
+    # This parses the HTML of one page in each section and extracts
     # the URLs of the resources as Units.
-    all_urls = [subsection.url
+    all_urls = [selected_section.subsections[0].url
                 for selected_sections in selections.values()
-                for selected_section in selected_sections
-                for subsection in selected_section.subsections]
+                for selected_section in selected_sections]
 
     extractor = extract_all_units_in_parallel
     if args.sequential:

diff --git a/edx_dl/parsing.py b/edx_dl/parsing.py
@@ -376,7 +376,7 @@ def _make_url(section_soup):  # FIXME: Extract from here and test
 
         def _get_section_name(section_soup):  # FIXME: Extract from here and test
             try:
-                return section_soup.button.h3.string.strip()
+                return section_soup.find('span',class_='subsection-title').string.strip()
             except AttributeError:
                 return None