[fix] webscraper merge function now starts looking for numbers starti…

…ng from the end
calblueprint · Jan 16, 2025 · 6fa6239 · 6fa6239
1 parent aa917bf
commit 6fa6239
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 8 deletions.
diff --git a/api/webscraper/__pycache__/nyserda_scraper.cpython-312.pyc b/api/webscraper/__pycache__/nyserda_scraper.cpython-312.pyc
diff --git a/api/webscraper/database.py b/api/webscraper/database.py
@@ -980,7 +980,7 @@ def merge_projects():
                 # otherwise, combine fields of current project with duplicate's data
                 update = combine_projects(update, matching_project)
 
-                # add sizes of duplicate proejcts together
+                # add sizes of duplicate projects together
                 if (
                     update.get("size", None) is not None
                     and matching_project.get("size", None) is not None

diff --git a/api/webscraper/utils/scraper_utils.py b/api/webscraper/utils/scraper_utils.py
@@ -161,13 +161,14 @@ def find_keyword(project_name):
         i = project_name.lower().find("wind")
         return project_name[:i].strip()
     else:
-        j = 0
-        while j < len(project_name):
-            if project_name[j].isdigit():
-                break
-            else:
-                j += 1
-        return project_name[:j].strip()
+        last_non_digit_index = len(project_name) - 1
+        while last_non_digit_index > 0 and project_name[last_non_digit_index].isdigit():
+            last_non_digit_index -= 1
+        return (
+            project_name[: last_non_digit_index + 1].strip()
+            if last_non_digit_index < len(project_name) - 1
+            else project_name.strip()
+        )
 
 
 def combine_projects(existing_project: dict, new_project: dict) -> dict: