Feature: Find Urls by Extensions and Patterns

asf_search/ASFProduct.py

@@ -3,6 +3,7 @@
 import warnings
 from shapely.geometry import shape, Point, Polygon, mapping
 import json
+import re
 
 from urllib import parse
 
@@ -290,6 +291,28 @@ def _get_additional_urls(self) -> List[str]:
             and url != self.properties['url']
         ]
 
+    def find_urls(self, extension: str = None, pattern: str = r'.*', directAccess: bool = False) -> List[str]:
+        """
+        Searches for all urls matching a given extension and/or pattern
+        param extension: the file extension to search for. (Defaults to `None`)
+            - Example: '.tiff'
+        param pattern: A regex pattern to search each url for.(Defaults to `False`)
+            - Example: `r'(QA_)+'` to find urls with 'QA_' at least once
+        param directAccess: should search in s3 bucket urls (Defaults to `False`)
+        """
+        search_list = self._get_s3_urls() if directAccess else self._get_additional_urls()
+
+        def _get_extension(file_url: str):
+            path = parse.urlparse(file_url).path
+            return os.path.splitext(path)[-1]
+
+        if extension is not None:
+            search_list = [url for url in search_list if _get_extension(url) == extension] 
+
+        regexp = re.compile(pattern=pattern)
+
+        return [url for url in search_list if regexp.search(url) is not None]
+
     def _get_s3_urls(self) -> List[str]:
         s3_urls = self._get_access_urls(
             ['GET DATA', 'EXTENDED METADATA', 'GET DATA VIA DIRECT ACCESS']

asf_search/ASFSearchResults.py

@@ -1,6 +1,7 @@
 from collections import UserList
 from multiprocessing import Pool
 import json
+from typing import List
 from asf_search import ASFSession, ASFSearchOptions
 from asf_search.download.file_download_type import FileDownloadType
 from asf_search.exceptions import ASFSearchError
@@ -42,6 +43,21 @@ def jsonlite(self):
     def jsonlite2(self):
         return results_to_jsonlite2(self)
 
+    def find_urls(self, extension: str = None, pattern: str = r'.*', directAccess: bool = False) -> List[str]:
+        """Returns a flat list of all https or s3 urls from all results matching an extension and/or regex pattern
+        param extension: the file extension to search for. (Defaults to `None`)
+            - Example: '.tiff'
+        param pattern: A regex pattern to search each url for.(Defaults to `False`)
+            - Example: `r'(QA_)+'` to find urls with 'QA_' at least once
+        param directAccess: should search in s3 bucket urls (Defaults to `False`)
+        """
+        urls = []
+
+        for product in self:
+            urls.extend(product.find_urls(extension=extension, pattern=pattern, directAccess=directAccess))
+
+        return urls
+
     def __str__(self):
         return json.dumps(self.geojson(), indent=2, sort_keys=True)