When making self links absolute, also add file:// as the scheme

Author: jsignell

pystac/asset.py

@@ -8,7 +8,12 @@
 
 from pystac import MediaType, STACError, common_metadata, utils
 from pystac.html.jinja_env import get_jinja_env
-from pystac.utils import is_absolute_href, make_absolute_href, make_relative_href
+from pystac.utils import (
+    is_absolute_href,
+    make_absolute_href,
+    make_relative_href,
+    safe_urlparse,
+)
 
 if TYPE_CHECKING:
     from pystac.common_metadata import CommonMetadata
@@ -380,7 +385,7 @@ def get_self_href(self) -> str | None:
 
 def _absolute_href(href: str, owner: Assets | None, action: str = "access") -> str:
     if utils.is_absolute_href(href):
-        return href
+        return safe_urlparse(href).path
     else:
         item_self = owner.get_self_href() if owner else None
         if item_self is None:
@@ -389,4 +394,4 @@ def _absolute_href(href: str, owner: Assets | None, action: str = "access") -> s
                 "and owner item is not set. Hint: try using "
                 ":func:`~pystac.Item.make_asset_hrefs_absolute`"
             )
-        return utils.make_absolute_href(href, item_self)
+        return safe_urlparse(utils.make_absolute_href(href, item_self)).path

pystac/link.py

@@ -102,7 +102,7 @@ def __init__(
         self.rel = rel
         if isinstance(target, str):
             if rel == pystac.RelType.SELF:
-                self._target_href = make_absolute_href(target)
+                self._target_href = make_absolute_href(target, must_include_scheme=True)
             else:
                 self._target_href = make_posix_style(target)
             self._target_object = None

pystac/stac_io.py

@@ -303,6 +303,7 @@ def read_text_from_href(self, href: str) -> str:
             except HTTPError as e:
                 raise Exception(f"Could not read uri {href}") from e
         else:
+            href = safe_urlparse(href).path
             with open(href, encoding="utf-8") as f:
                 href_contents = f.read()
         return href_contents
@@ -328,7 +329,7 @@ def write_text_to_href(self, href: str, txt: str) -> None:
         """
         if _is_url(href):
             raise NotImplementedError("DefaultStacIO cannot write to urls")
-        href = os.fspath(href)
+        href = safe_urlparse(href).path
         dirname = os.path.dirname(href)
         if dirname != "" and not os.path.isdir(dirname):
             os.makedirs(dirname)
@@ -391,7 +392,7 @@ def _report_duplicate_object_names(
 
 def _is_url(href: str) -> bool:
     parsed = safe_urlparse(href)
-    return parsed.scheme != ""
+    return parsed.scheme != "" and parsed.scheme != "file"
 
 
 if HAS_URLLIB3:

pystac/utils.py

@@ -288,34 +288,54 @@ def _make_absolute_href_path(
     parsed_source: URLParseResult,
     parsed_start: URLParseResult,
     start_is_dir: bool = False,
+    must_include_scheme: bool = False,
 ) -> str:
-    # If the source is already absolute, just return it
+    # If the source is already absolute and doesn't need a  scheme just return it
     if os.path.isabs(parsed_source.path):
-        return urlunparse(parsed_source)
-
-    # If the start path is not a directory, get the parent directory
-    start_dir = (
-        parsed_start.path if start_is_dir else os.path.dirname(parsed_start.path)
-    )
+        if must_include_scheme:
+            abs_path = parsed_source.path
+        else:
+            return urlunparse(parsed_source)
+    else:
+        # If the start path is not a directory, get the parent directory
+        start_dir = (
+            parsed_start.path if start_is_dir else os.path.dirname(parsed_start.path)
+        )
 
-    # Join the start directory to the relative path and find the absolute path
-    abs_path = make_posix_style(
-        os.path.abspath(os.path.join(start_dir, parsed_source.path))
-    )
+        # Join the start directory to the relative path and find the absolute path
+        abs_path = make_posix_style(
+            os.path.abspath(os.path.join(start_dir, parsed_source.path))
+        )
 
-    # Account for the normalization of abspath for
-    # things like /vsitar// prefixes by replacing the
-    # original start_dir text when abspath modifies the start_dir.
-    if not start_dir == make_posix_style(os.path.abspath(start_dir)):
-        abs_path = abs_path.replace(
-            make_posix_style(os.path.abspath(start_dir)), start_dir
+        # Account for the normalization of abspath for
+        # things like /vsitar// prefixes by replacing the
+        # original start_dir text when abspath modifies the start_dir.
+        if not start_dir == make_posix_style(os.path.abspath(start_dir)):
+            abs_path = abs_path.replace(
+                make_posix_style(os.path.abspath(start_dir)), start_dir
+            )
+
+    # add a scheme if there isn't one already
+    if must_include_scheme:
+        return urlunparse(
+            (
+                "file",
+                parsed_start.netloc,
+                abs_path,
+                parsed_source.params,
+                parsed_source.query,
+                parsed_source.fragment,
+            )
         )
 
     return abs_path
 
 
 def make_absolute_href(
-    source_href: str, start_href: str | None = None, start_is_dir: bool = False
+    source_href: str,
+    start_href: str | None = None,
+    start_is_dir: bool = False,
+    must_include_scheme: bool = False,
 ) -> str:
     """Returns a new string that represents ``source_href`` as an absolute path. If
     ``source_href`` is already absolute it is returned unchanged. If ``source_href``
@@ -332,6 +352,8 @@ def make_absolute_href(
         start_is_dir : If ``True``, ``start_href`` is treated as a directory.
             Otherwise, ``start_href`` is considered to be a path to a file. Defaults to
             ``False``.
+        must_include_scheme : If ``True``, every output will have a scheme. This means
+            that local filepaths will be prefixed with `file://`. Defaults to ``False``.
 
     Returns:
         str: The absolute HREF.
@@ -349,7 +371,9 @@ def make_absolute_href(
     if parsed_source.scheme != "" or parsed_start.scheme != "":
         return _make_absolute_href_url(parsed_source, parsed_start, start_is_dir)
     else:
-        return _make_absolute_href_path(parsed_source, parsed_start, start_is_dir)
+        return _make_absolute_href_path(
+            parsed_source, parsed_start, start_is_dir, must_include_scheme
+        )
 
 
 def is_absolute_href(href: str) -> bool: