Rmp plugin

compass/__init__.py

@@ -1,16 +1,21 @@
 """Ordinance document download and structured data extraction"""
 
+from dotenv import load_dotenv
+
 from ._version import __version__
 from .utilities.logs import setup_logging_levels, COMPASS_DEBUG_LEVEL
 
 # Temporarily import to register plugins
 # Can drop once plugins register themselves
 from .extraction import (
     COMPASSGeoHeatPumpExtractor,
+    COMPASSGeoElectricityExtractor,
+    COMPASSGeoRMPExtractor,
     COMPASSSmallWindExtractor,
     COMPASSSolarExtractor,
     COMPASSWindExtractor,
     TexasWaterRightsExtractor,
 )
 
+load_dotenv()
 setup_logging_levels()

compass/_cli/finalize.py

@@ -6,6 +6,7 @@
 from rich.theme import Theme
 from rich.console import Console
 
+from compass.plugin.registry import PLUGIN_REGISTRY
 from compass.utilities import Directories
 from compass.utilities.io import load_config
 from compass.utilities.jurisdictions import Jurisdiction
@@ -66,7 +67,7 @@ def finalize(config):
     console.print("Compiling databases...")
     jurisdictions = jurisdictions.get("jurisdictions", [])
 
-    _compile_db(jurisdictions, dirs)
+    _compile_db(jurisdictions, dirs, tech)
 
     console.print("Saving meta info...")
     num_jurisdictions_searched = len(jurisdictions)
@@ -88,7 +89,7 @@ def finalize(config):
     console.print(f"✅ Finalized COMPASS run in {dirs.out!s}!")
 
 
-def _compile_db(jurisdictions, dirs):
+def _compile_db(jurisdictions, dirs, tech):
     """Merge all jurisdiction dbs into one"""
     all_doc_infos = []
     for jur_info in jurisdictions:
@@ -117,5 +118,6 @@ def _compile_db(jurisdictions, dirs):
             {"ord_db_fp": ord_db_fp, "jurisdiction": jurisdiction}
         )
 
-    db, __ = doc_infos_to_db(all_doc_infos)
-    save_db(db, dirs.out)
+    out_cols = PLUGIN_REGISTRY[tech].OUTPUT_COLUMNS
+    db, __ = doc_infos_to_db(all_doc_infos, out_cols)
+    save_db(db, dirs.out, out_cols)

compass/extraction/__init__.py

@@ -11,6 +11,8 @@
 # Temporarily import to register plugins
 # Can drop once plugins register themselves
 from .ghp import COMPASSGeoHeatPumpExtractor
+from .geothermal_electricity import COMPASSGeoElectricityExtractor
+from .rmp import COMPASSGeoRMPExtractor
 from .small_wind import COMPASSSmallWindExtractor
 from .solar import COMPASSSolarExtractor
 from .water import TexasWaterRightsExtractor

compass/extraction/geothermal_electricity/__init__.py

@@ -0,0 +1,14 @@
+"""COMPASS Geothermal Electricity plugin"""
+
+import importlib.resources
+
+from compass.plugin import create_schema_based_one_shot_extraction_plugin
+
+
+COMPASSGeoElectricityExtractor = (
+    create_schema_based_one_shot_extraction_plugin(
+        importlib.resources.files("compass.extraction.geothermal_electricity")
+        / "plugin_config.yaml",
+        tech="geothermal",
+    )
+)

compass/extraction/geothermal_electricity/geothermal_plugin_config.yaml → compass/extraction/geothermal_electricity/plugin_config.yaml

@@ -141,4 +141,4 @@ heuristic_keywords:
     - "cannabis cultivation"
     - "commercial cannabis"
 
-collection_prompts: True
+collection_prompts: True

compass/extraction/rmp/__init__.py

@@ -0,0 +1,37 @@
+"""COMPASS Geothermal Resource Management Plan plugin"""
+
+import importlib.resources
+from pathlib import Path
+
+from compass.plugin import create_schema_based_one_shot_extraction_plugin, OutputColumn
+from compass.utilities import doc_infos_to_db, save_db
+
+
+COMPASSGeoRMPExtractor = create_schema_based_one_shot_extraction_plugin(
+    importlib.resources.files("compass.extraction.rmp") / "plugin_config.yaml",
+    tech="geo_rmp",
+)
+
+_out_cols = getattr(COMPASSGeoRMPExtractor, "OUTPUT_COLUMNS", [])
+if not any(col.name == "restriction_level" for col in _out_cols):
+    _out_cols.append(OutputColumn("restriction_level"))
+if not any(col.name == "document_name" for col in _out_cols):
+    _out_cols.append(OutputColumn("document_name"))
+
+
+@classmethod
+def _save_structured_data(cls, doc_infos, out_dir):
+    """Save RMP extraction results, adding document_name from source path"""
+    output_cols = getattr(cls, "OUTPUT_COLUMNS", [])
+    db, num_docs_found = doc_infos_to_db(doc_infos, output_cols)
+
+    if not db.empty:
+        db["document_name"] = db["source"].apply(
+            lambda src: Path(src).name if isinstance(src, str) and src else None
+        )
+
+    save_db(db, out_dir, output_cols)
+    return num_docs_found
+
+
+COMPASSGeoRMPExtractor.save_structured_data = _save_structured_data

compass/extraction/rmp/plugin_config.yaml

@@ -0,0 +1,101 @@
+# One-shot plugin configuration for geothermal restriction extraction from
+# BLM Resource Management Plan (RMP) documents.
+#
+# Usage:
+#   compass process -c config.json5 -p plugin_config.yaml
+#
+# Or with pixi:
+#   pixi run --manifest-path "<path-to-COMPASS>/pixi.toml" \
+#       compass process -c config.json5 -p plugin_config.yaml
+
+schema: ./rmp_schema.json
+
+data_type_short_desc: geothermal leasing restriction in BLM Resource Management Plan
+
+# Keyword-based heuristic for quickly filtering text chunks before LLM calls.
+heuristic_keywords:
+  good_tech_keywords:
+    - "geothermal"
+    - "leasing"
+    - "stipulation"
+    - "closure"
+    - "drilling"
+    - "mineral"
+    - "nso"
+    - "energy"
+    - "map"
+  good_tech_acronyms:
+    - "nso"
+    - "rmp"
+    - "blm"
+    - "tlm"
+    - "esa"
+  good_tech_phrases:
+    - "no surface occupancy"
+    - "surface disturbance"
+    - "geothermal leasing"
+    - "geothermal development"
+    - "geothermal exploration"
+    - "mineral leasing"
+    - "no surface occupancy"
+    - "surface-disturbing activities"
+    - "timing limitation"
+    - "lease stipulation"
+    - "closed to leasing"
+    - "open to leasing"
+    - "geothermal resources"
+    - "resource management plan"
+    - "land withdrawal"
+    - "mineral rights"
+    - "energy development"
+    - "minerals and energy"
+    - "fluid mineral"
+    - "fluid minerals"
+    - "fluid mineral leasing"
+    - "fluid mineral development"
+    - "fluid leasable minerals"
+  not_tech_words:
+    - "grazing allotment"
+    - "livestock"
+    - "recreation permit"
+    - "campground"
+    - "trail maintenance"
+    - "off-highway vehicle"
+    - "ohv route"
+    - "land disposal"
+    - "land sale"
+    - "residential"
+    - "wildfire"
+    - "prescribed burn"
+    - "hunting permit"
+    - "fishing"
+
+# Collection prompts filter individual text chunks extracted from the PDF.
+# Chunks that pass are passed to the text extraction step.
+# Setting to True auto-generates prompts from the schema.
+collection_prompts: True
+
+# System prompt used during the final structured extraction LLM call.
+extraction_system_prompt: |-
+  You are an expert analyst specializing in BLM Resource Management Plans (RMPs)
+  and federal mineral leasing regulations. Your task is to extract structured data
+  about geothermal leasing restrictions from RMP document text.
+  Follow all instructions in the schema field descriptions carefully.
+  Only extract restrictions that directly constrain geothermal leasing, exploration,
+  drilling, or facility construction on BLM lands. Each distinct restriction should
+  be its own row in the outputs array. Use direct text excerpts and quotes in the
+  summary field wherever possible.
+  IMPORTANT: "Fluid minerals" is the BLM/Mineral Leasing Act regulatory term that
+  collectively covers oil, gas, AND geothermal resources. RMPs very frequently use
+  "fluid minerals" or "fluid mineral leasing/development" instead of spelling out
+  "oil, gas, and geothermal" — treat any restriction, closure, or stipulation on
+  fluid minerals as encompassing geothermal (geothermal_applicability = 'broad')
+  exactly as you would for an "oil, gas, and geothermal" or "mineral leasing" group
+  closure, unless the document explicitly excludes geothermal from the fluid
+  minerals category.
+
+cache_llm_generated_content: true
+
+# Allows a single jurisdiction entry to span multiple PDF documents,
+# which is typical for RMPs that are split across many files.
+doc_selection_method: multi-doc-all