tnm de-duplication

src/fetchez/modules/base.py

@@ -386,6 +386,24 @@ def add_entry_to_results(self, url: str, dst_fn: str, data_type: Any, **kwargs):
         if hasattr(self, "stream_kwargs"):
             entry.update(self.stream_kwargs)
 
+        # Intercept standard spatial metadata fields
+        standard_metadata = {
+            "name": self.name,  # Module name (e.g., 'tnm', 'copernicus')
+            "title": kwargs.pop("title", "Unknown"),
+            "source": getattr(self, "meta_agency", "Unknown"),
+            "date": kwargs.pop("date", "Unknown"),
+            "data_type": data_type,
+            "resolution": kwargs.pop(
+                "resolution", getattr(self, "meta_resolution", "Unknown")
+            ),
+            "hdatum": kwargs.pop("hdatum", "Unknown"),
+            "vdatum": kwargs.pop("vdatum", "Unknown"),
+            "url": str(url),
+        }
+        entry_metadata = kwargs.pop("metadata", {})
+        standard_metadata.update(entry_metadata)
+        entry["metadata"] = standard_metadata
+
         entry.update(kwargs)
         self.results.append(entry)
 

src/fetchez/modules/tnm.py

@@ -222,8 +222,26 @@ def run(self):
                             item_bbox.get("maxY"),
                         )
 
-                    # Extract the tile footprint (e.g., 'n35w120')
-                    fn_bn = filename.split("_")[-2]
+                    # Extract the tile footprint or project ID based on dataset type
+                    if (
+                        "ned19" in filename.lower()
+                        or "opr" in filename.lower()
+                        or "lpc" in filename.lower()
+                    ):
+                        fn_bn = "_".join(filename.split("_")[:-1])
+                    elif bounds:
+                        fn_bn = f"{round(bounds[0], 4)}_{round(bounds[1], 4)}_{round(bounds[2], 4)}_{round(bounds[3], 4)}"
+                    else:
+                        fn_bn = item.get("title", filename)
+
+                    # date = item.get("publicationDate", "")
+                    # if bounds:
+                    #     bounds_str = f"{round(bounds[0], 4)}_{round(bounds[1], 4)}_{round(bounds[2], 4)}_{round(bounds[3], 4)}"
+                    #     project_id = "/".join(item.get("title", ""))
+                    #     fn_bn = f"{bounds_str}_{project_id}"
+                    # else:
+                    #     fn_bn = item.get("title", filename)
+
                     date = item.get("publicationDate", "")
 
                     item_data = {

src/fetchez/recipes/modifiers/exclude_module.py

@@ -26,7 +26,7 @@ class ExcludeModuleModifier(BaseModifier):
     """
 
     name = "exclude_module"
-    meta_desc = "Exclude specific modules from a recipe by name"
+    meta_desc = "Exclude specific modules from a recipe by name."
 
     def __init__(self, modules=None, **kwargs):
         super().__init__(**kwargs)

src/fetchez/recipes/modifiers/inject_args.py

@@ -0,0 +1,94 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+"""
+fetchez.recipes.modifiers.inject_args
+~~~~~~~~~~~~~~~~
+
+Recipe mutator to remove specifically named modules
+from a recipe.
+
+:copyright: (c) 2012 - 2026 CIRES Coastal DEM Team
+:license: MIT, see LICENSE for more details.
+"""
+
+from fetchez.recipes.modifiers.base import BaseModifier
+import logging
+
+logger = logging.getLogger(__name__)
+
+
+class InjectArgsModifier(BaseModifier):
+    """Injects arbitrary key:value arguments into matching modules or hooks.
+    Example: --modifier inject_args:match=stream_reproject,cache_dir=socal_data
+    """
+
+    name = "inject_args"
+    meta_desc = "Injects arbitrary key:value arguments into matching modules or hooks."
+
+    def __init__(self, match=None, **kwargs):
+        super().__init__(**kwargs)
+        self.match = match
+        self.inject_kwargs = kwargs
+
+    def _inject_into_item(self, item, match_name):
+        """Helper to safely mutate string or dict-based definitions."""
+
+        if isinstance(item, str) and item == match_name:
+            if self.inject_kwargs:
+                logger.info(
+                    f"Modifier '{self.name}': Upgrading '{item}' to inject args."
+                )
+                return {item: self.inject_kwargs.copy()}
+            return item
+
+        if isinstance(item, dict):
+            key = list(item.keys())[0]
+            if key == match_name:
+                val = item[key]
+                if val is None:
+                    val = {}
+                elif isinstance(val, str):
+                    val = {"_value": val}
+
+                if isinstance(val, dict):
+                    val.update(self.inject_kwargs)
+                    logger.info(
+                        f"Modifier '{self.name}': Injected {list(self.inject_kwargs.keys())} into '{key}'."
+                    )
+                item[key] = val
+
+        return item
+
+    def apply(self, config):
+        """Mutates the recipe config by injecting arguments into matches."""
+
+        if not self.match or not self.inject_kwargs:
+            logger.warning(
+                "InjectArgsModifier requires a 'match' target and arguments to inject. Skipping."
+            )
+            return config
+
+        if "hooks" in config:
+            config["hooks"] = [
+                self._inject_into_item(h, self.match) for h in config["hooks"]
+            ]
+
+        if "modules" in config:
+            updated_modules = []
+            for mod in config["modules"]:
+                mod = self._inject_into_item(mod, self.match)
+
+                if isinstance(mod, dict):
+                    mod_name = list(mod.keys())[0]
+                    mod_args = mod[mod_name]
+                    if isinstance(mod_args, dict) and "hooks" in mod_args:
+                        mod_args["hooks"] = [
+                            self._inject_into_item(h, self.match)
+                            for h in mod_args["hooks"]
+                        ]
+
+                updated_modules.append(mod)
+            config["modules"] = updated_modules
+
+        return config