Add dataset names for code completion
Browse files
src/get_results_for_task.py
CHANGED
@@ -60,6 +60,13 @@ def _process_urls(raw_urls: str) -> str:
|
|
60 |
html_urls = [model_hyperlink(*re.search(MD_LINK_PATTERN, url.strip()).groups()) for url in raw_urls.split(",")]
|
61 |
return ", ".join(html_urls)
|
62 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
63 |
|
64 |
def _get_results_dataset(task_id: str) -> pd.DataFrame:
|
65 |
results_df = load_dataset(
|
@@ -81,6 +88,7 @@ def _get_results_dataset(task_id: str) -> pd.DataFrame:
|
|
81 |
for link, model_name in zip(results_df["model_url"], results_df["Model Name"])
|
82 |
]
|
83 |
if task_id == 'project_code_completion':
|
|
|
84 |
results_df["Dataset"] = [_process_urls(urls) for urls in results_df["Dataset"]]
|
85 |
results_df["Resources"] = [_process_urls(urls) for urls in results_df["Resources"]]
|
86 |
results_df = results_df[get_columns_per_task(task_id)]
|
|
|
60 |
html_urls = [model_hyperlink(*re.search(MD_LINK_PATTERN, url.strip()).groups()) for url in raw_urls.split(",")]
|
61 |
return ", ".join(html_urls)
|
62 |
|
63 |
+
def _extract_dataset_name(raw_urls: str) -> str:
|
64 |
+
if not raw_urls:
|
65 |
+
return raw_urls
|
66 |
+
names = [re.search(MD_LINK_PATTERN, url.strip()).group(1) for url in raw_urls.split(",")]
|
67 |
+
return ", ".join(names)
|
68 |
+
|
69 |
+
|
70 |
|
71 |
def _get_results_dataset(task_id: str) -> pd.DataFrame:
|
72 |
results_df = load_dataset(
|
|
|
88 |
for link, model_name in zip(results_df["model_url"], results_df["Model Name"])
|
89 |
]
|
90 |
if task_id == 'project_code_completion':
|
91 |
+
results_df["Dataset_Names"] = [_extract_dataset_name(urls) for urls in results_df["Dataset"]]
|
92 |
results_df["Dataset"] = [_process_urls(urls) for urls in results_df["Dataset"]]
|
93 |
results_df["Resources"] = [_process_urls(urls) for urls in results_df["Resources"]]
|
94 |
results_df = results_df[get_columns_per_task(task_id)]
|