seer-pas-sdk 1.1.1__py3-none-any.whl → 1.2.1__py3-none-any.whl

seer_pas_sdk/common/__init__.py

@@ -99,7 +99,7 @@ def dict_to_df(data):
 
 
 # Most cases appear to be a .tsv file.
-def download_df(url, is_tsv=True, dtype={}):
+def download_df(url, is_tsv=True, dtype={}, usecols=None):
     """
     Fetches a TSV/CSV file from a URL and returns as a Pandas DataFrame.
 
@@ -114,6 +114,9 @@ def download_df(url, is_tsv=True, dtype={}):
     dtype : dict
         Data type conversion when intaking columns. e.g. {'a': str, 'b': np.float64}
 
+    usecols : list
+        Subset of columns to download. If not specified, downloads all columns.
+
     Returns
     -------
     pandas.core.frame.DataFrame
@@ -139,11 +142,9 @@ def download_df(url, is_tsv=True, dtype={}):
 
     if not url:
         return pd.DataFrame()
-    url_content = io.StringIO(requests.get(url).content.decode("utf-8"))
-    if is_tsv:
-        csv = pd.read_csv(url_content, sep="\t", dtype=dtype)
-    else:
-        csv = pd.read_csv(url_content, dtype=dtype)
+    csv = pd.read_csv(
+        url, dtype=dtype, sep="\t" if is_tsv else ",", usecols=usecols
+    )
     return csv
 
 
@@ -679,6 +680,52 @@ def camel_case(s):
     return "".join([s[0].lower(), s[1:]])
 
 
+def validate_d_zip_file(file):
+    """
+    Return True if a .d.zip file aligns with Seer requirements for PAS upload.
+
+    Parameters
+    ----------
+    file : str
+        The name of the zip file.
+
+    Returns
+    -------
+    bool
+        True if the .d.zip file is valid, False otherwise.
+    """
+
+    if not file.lower().endswith(".d.zip"):
+        return False
+
+    basename = os.path.basename(file)
+
+    # Remove the .zip extension to get the .d folder name
+    d_name = basename[:-4]
+
+    try:
+        with zipfile.ZipFile(file, "r") as zf:
+            names = zf.namelist()
+
+    except:
+        return False
+
+    if not names:
+        return False
+
+    # check for files at the root level
+    root_entries = [n for n in names if "/" not in n.rstrip("/")]
+    if root_entries:
+        return False
+
+    # find folders
+    top_level = {n.split("/")[0] for n in names}
+    if len(top_level) != 1 or d_name not in top_level:
+        return False
+
+    return True
+
+
 def rename_d_zip_file(source, destination):
     """
     Renames a .d.zip file. The function extracts the contents of the source zip file, renames the inner .d folder, and rezips the contents into the destination zip file.