feat: basic collection summaries for #45

Author: rushgeo

src/stactools/cli/__init__.py

@@ -19,6 +19,7 @@ def register_plugin(registry: "Registry") -> None:
         lint,
         merge,
         migrate,
+        summary,
         update_geometry,
         validate,
         version,
@@ -35,6 +36,7 @@ def register_plugin(registry: "Registry") -> None:
     registry.register_subcommand(layout.create_layout_command)
     registry.register_subcommand(lint.create_lint_command)
     registry.register_subcommand(merge.create_merge_command)
+    registry.register_subcommand(summary.create_summary_command)
     registry.register_subcommand(validate.create_validate_command)
     registry.register_subcommand(version.create_version_command)
     registry.register_subcommand(update_geometry.create_update_geometry_command)

src/stactools/cli/commands/summary.py

@@ -0,0 +1,57 @@
+from typing import Any
+
+import click
+import pystac
+from pystac import Catalog, Collection
+
+
+def format_summary(summary: dict[str, Any], indent: int = 4) -> str:
+    out = ""
+    for var in summary:
+        if type(summary[var]) == dict:
+            out += var + ": \n" + " " * indent + str(summary[var]) + "\n"
+        else:
+            out += var + ": " + str(summary[var]) + "\n"
+    return out
+
+
+def print_summaries(collection_path: str, force_resummary: bool = False) -> None:
+    col = pystac.read_file(collection_path)
+    if isinstance(col, Collection):
+        orig_summaries = col.summaries.to_dict()
+        if force_resummary or len(orig_summaries) == 0:
+            new_summaries = pystac.summaries.Summarizer().summarize(col).to_dict()
+            if len(orig_summaries) == 0:
+                print(
+                    "Collection's summaries were empty. Printing recalculated summaries:"
+                )
+            print(format_summary(new_summaries))
+        else:
+            print(format_summary(orig_summaries))
+    elif isinstance(col, Catalog):
+        print("Input is a catalog, not a collection.")
+        print(
+            "Run 'stac describe -h {}' to discover collections in this catalog".format(
+                collection_path
+            )
+        )
+        return
+    else:
+        print("{} is not a collection".format(collection_path))
+        return
+
+
+def create_summary_command(cli: click.Group) -> click.Command:
+    @cli.command("summary", short_help="Summarize a STAC collection's contents.")
+    @click.option(
+        "-f",
+        "--force_resummary",
+        is_flag=True,
+        default=False,
+        help="Ignore existing summaries in the collection and print recalculated summaries",
+    )
+    @click.argument("collection_path")
+    def summary_command(collection_path: str, force_resummary: bool) -> None:
+        print_summaries(collection_path, force_resummary)
+
+    return summary_command