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/data/rmp_jurisdictions.csv

@@ -0,0 +1,167 @@
+State,County,Subdivision,Jurisdiction Type,FIPS,Website
+AK,,Bay_RMP,RMP_Resource_Management_Plan,58891654,
+AK,,Bering_Sea-Western_Interior_RMP,RMP_Resource_Management_Plan,50065852,
+AK,,Central_Yukon_RMP_Revision_proposed,RMP_Resource_Management_Plan,88244868,
+AK,,Draanjik_RMP,RMP_Resource_Management_Plan,76528034,
+AK,,East_Alaska_RMP,RMP_Resource_Management_Plan,12789119,
+AK,,Fort_Greely_RMP,RMP_Resource_Management_Plan,145246,
+AK,,Fort_Wainwright_RMP,RMP_Resource_Management_Plan,28335718,
+AK,,Kobuk_Seward_Peninsula_RMP,RMP_Resource_Management_Plan,73124680,
+AK,,Ring_of_Fire_RMP,RMP_Resource_Management_Plan,40701468,
+AZ,,Agua_Fria_NM_RMP,RMP_Resource_Management_Plan,91052984,
+AZ,,Arizona_Strip_Field_Office_RMP,RMP_Resource_Management_Plan,80496010,
+AZ,,Bradshaw_Harquahala_RMP,RMP_Resource_Management_Plan,65114804,
+AZ,,Grand_Canyon-Parashant_NM_RMP,RMP_Resource_Management_Plan,53431245,
+AZ,,Ironwood_Forest_NM_RMP,RMP_Resource_Management_Plan,41571498,
+AZ,,Kingman_RMP,RMP_Resource_Management_Plan,94532898,
+AZ,,Lake_Havasu_Field_Office_RMP,RMP_Resource_Management_Plan,737252,
+AZ,,Las_Cienegas_RMP,RMP_Resource_Management_Plan,71912775,
+AZ,,Phoenix_RMP,RMP_Resource_Management_Plan,61984530,
+AZ,,Safford_District_RMP,RMP_Resource_Management_Plan,99160383,
+AZ,,San_Pedro_Riparian_National_Conservation_Area_RMP,RMP_Resource_Management_Plan,98788395,
+AZ,,Sonoran_Desert_NM_RMP,RMP_Resource_Management_Plan,76471287,
+AZ,,Vermilion_Cliffs_NM_RMP,RMP_Resource_Management_Plan,29865380,
+AZ,,Yuma_Field_Office_RMP,RMP_Resource_Management_Plan,86211833,
+CA,,Alturas_RMP,RMP_Resource_Management_Plan,85572075,
+CA,,Bakersfield_ROD_&_Approved_RMP,RMP_Resource_Management_Plan,99911428,
+CA,,Bishop_Field_Office_RMP,RMP_Resource_Management_Plan,41389155,
+CA,,California_Coastal_NM_RMP,RMP_Resource_Management_Plan,74326555,
+CA,,California_Desert_Conservation_Area_Plan,RMP_Resource_Management_Plan,77257162,
+CA,,Carrizo_Plain_NM_Approved_RMP,RMP_Resource_Management_Plan,87487361,
+CA,,Clear_Creek_Management_Area_Proposed_RMP,RMP_Resource_Management_Plan,20322133,
+CA,,Eagle_Lake_Field_Office_RMP_and_EIS,RMP_Resource_Management_Plan,81360063,
+CA,,Eastern_San_Diego_County_RMP_&_ROD,RMP_Resource_Management_Plan,56019402,
+CA,,Headwaters_Forest_Reserve_RMP_and_ROD,RMP_Resource_Management_Plan,1925220,
+CA,,King_Range_NCA_RMP_and_ROD,RMP_Resource_Management_Plan,50294440,
+CA,,Northwest_California_Integrated_RMP_Revision_proposed,RMP_Resource_Management_Plan,28373463,
+CA,,Sierra_RMP_and_ROD,RMP_Resource_Management_Plan,64978099,
+CA,,South_Coast_RMP_and_ROD,RMP_Resource_Management_Plan,70257488,
+CA,,Southern_Diablo_Mountain_Range_and_Central_Coast_of_Californ,RMP_Resource_Management_Plan,1311864,
+CA,,Surprise_RMP_ROD,RMP_Resource_Management_Plan,5765283,
+CA,,Ukiah_RMP,RMP_Resource_Management_Plan,3279363,
+CO,,Browns_Canyon_NM_RMP,RMP_Resource_Management_Plan,22907548,
+CO,,Canyons_of_the_Ancients_NM_RMP,RMP_Resource_Management_Plan,27732763,
+CO,,Colorado_River_Valley_Field_Office_RMP,RMP_Resource_Management_Plan,97921160,
+CO,,Dominguez-Escalante_NCA_RMP,RMP_Resource_Management_Plan,32592106,
+CO,,Eastern_Colorado_RMP_Revision_proposed,RMP_Resource_Management_Plan,21568440,
+CO,,Grand_Junction_Field_Office_RMP,RMP_Resource_Management_Plan,67272052,
+CO,,Gunnison_Gorge_NCA_RMP,RMP_Resource_Management_Plan,72599971,
+CO,,Gunnison_Resource_Area_RMP,RMP_Resource_Management_Plan,36211355,
+CO,,Kremmling_RMP,RMP_Resource_Management_Plan,94278387,
+CO,,Little_Snake_RMP,RMP_Resource_Management_Plan,65371173,
+CO,,McInnis_Canyons_NCA_RMP,RMP_Resource_Management_Plan,44871622,
+CO,,San_Luis_Resource_Area_RMP,RMP_Resource_Management_Plan,88268922,
+CO,,Tres_Rios_Field_Office_RMP,RMP_Resource_Management_Plan,83907713,
+CO,,Uncompahgre_RMP,RMP_Resource_Management_Plan,34150003,
+CO,,White_River_Field_Office_RMP,RMP_Resource_Management_Plan,11902779,
+ES,,Alabama_RMP,RMP_Resource_Management_Plan,14797696,
+ES,,Florida_RMP,RMP_Resource_Management_Plan,95600383,
+ES,,Illinois_RMP,RMP_Resource_Management_Plan,93453295,
+ES,,Michigan_RMP,RMP_Resource_Management_Plan,26182278,
+ES,,Minnesota_MFP,RMP_Resource_Management_Plan,13778191,
+ES,,Missouri_RMP,RMP_Resource_Management_Plan,67092820,
+ES,,Wisconsin_RMP,RMP_Resource_Management_Plan,75739247,
+ID,,Bennett_Hills_Timmerman_Hills_MFP,RMP_Resource_Management_Plan,81790701,
+ID,,Big_Desert_MFP,RMP_Resource_Management_Plan,84522543,
+ID,,Big_Lost_MFP,RMP_Resource_Management_Plan,60403369,
+ID,,Bruneau_MFP,RMP_Resource_Management_Plan,18575811,
+ID,,Cassia_RMP,RMP_Resource_Management_Plan,80586106,
+ID,,Challis_RMP,RMP_Resource_Management_Plan,84849736,
+ID,,Coeur_d'Alene_RMP,RMP_Resource_Management_Plan,77822719,
+ID,,Cottonwood_RMP,RMP_Resource_Management_Plan,61739039,
+ID,,Craters_of_the_Moon_Monument_RMP,RMP_Resource_Management_Plan,99809881,
+ID,,Four_Rivers_Field_Office_RMP_Revision_proposed,RMP_Resource_Management_Plan,3941354,
+ID,,Jarbidge_RMP,RMP_Resource_Management_Plan,67339645,
+ID,,Lemhi_RMP,RMP_Resource_Management_Plan,37020393,
+ID,,Little_Lost_and_Birch_Creek_MFP,RMP_Resource_Management_Plan,4531220,
+ID,,Magic_MFP,RMP_Resource_Management_Plan,79311181,
+ID,,Medicine_Lodge_RMP,RMP_Resource_Management_Plan,35884558,
+ID,,Monument_RMP,RMP_Resource_Management_Plan,80677084,
+ID,,Morley_Nelson_Snake_River_Birds_of_Prey_RMP,RMP_Resource_Management_Plan,16215809,
+ID,,Owyhee_RMP,RMP_Resource_Management_Plan,61818628,
+ID,,Pocatello_RMP,RMP_Resource_Management_Plan,29729378,
+ID,,Sun_Valley_MFP,RMP_Resource_Management_Plan,43040853,
+MT,,Billings_Field_Office_RMP,RMP_Resource_Management_Plan,46147967,
+MT,,Butte_Field_Office_Approved_RMP,RMP_Resource_Management_Plan,28147387,
+MT,,Dillon_RMP,RMP_Resource_Management_Plan,34399012,
+MT,,HiLine_RMP,RMP_Resource_Management_Plan,73749051,
+MT,,Lewistown_RMP_and_EIS,RMP_Resource_Management_Plan,13183292,
+MT,,Miles_City_Field_Office_RMP,RMP_Resource_Management_Plan,97076346,
+MT,,Missoula_Field_Office_RMP_and_EIS,RMP_Resource_Management_Plan,17045292,
+MT,,North_Dakota_RMP_Revision_and_EIS_proposed,RMP_Resource_Management_Plan,54876233,
+MT,,Pompeys_Pillar_RMP,RMP_Resource_Management_Plan,28758168,
+MT,,South_Dakota_RMP,RMP_Resource_Management_Plan,79949953,
+MT,,Upper_Missouri_River_Breaks_NM_ROD_and_Approved_RMP,RMP_Resource_Management_Plan,40143545,
+NM,,Carlsbad_RMP,RMP_Resource_Management_Plan,85819939,
+NM,,El_Malpais_NCA_RMP,RMP_Resource_Management_Plan,78985957,
+NM,,Farmington_RMP,RMP_Resource_Management_Plan,98791451,
+NM,,Kasha-Katuwe_Tent_Rocks_NM_RMP,RMP_Resource_Management_Plan,97759534,
+NM,,Mimbres_RMP,RMP_Resource_Management_Plan,62451792,
+NM,,"Oklahoma,_Kansas,_and_Texas_RMP",RMP_Resource_Management_Plan,89755973,
+NM,,Organ_Mountains-Desert_Peaks_National_Monument_RMP_Developme,RMP_Resource_Management_Plan,72844604,
+NM,,Prehistoric_Trackways_NM_RMP,RMP_Resource_Management_Plan,53295310,
+NM,,Rio_Puerco_RMP_Revision_proposed,RMP_Resource_Management_Plan,84022284,
+NM,,Roswell_RMP,RMP_Resource_Management_Plan,12328195,
+NM,,Taos_RMP,RMP_Resource_Management_Plan,27096808,
+NM,,The_Socorro_Field_Office_RMP,RMP_Resource_Management_Plan,62817657,
+NM,,TriCounty_RMP_Revision_proposed,RMP_Resource_Management_Plan,1833400,
+NM,,White_Sands_RMP,RMP_Resource_Management_Plan,42605400,
+NV,,Black_Rock_Desert-High_Rock_Canyon_NCA_RMP,RMP_Resource_Management_Plan,67949345,
+NV,,Carson_City_District_RMP_Revision_proposed,RMP_Resource_Management_Plan,64923924,
+NV,,Carson_City_Field_Office_Consolidated_RMP,RMP_Resource_Management_Plan,82348138,
+NV,,Ely_RMP,RMP_Resource_Management_Plan,65721226,
+NV,,Las_Vegas_RMP,RMP_Resource_Management_Plan,29181757,
+NV,,Nevada_Test_and_Training_Range_RMP,RMP_Resource_Management_Plan,75571370,
+NV,,Red_Rock_Canyon_NCA_RMP,RMP_Resource_Management_Plan,39640186,
+NV,,Shoshone-Eureka_RMP,RMP_Resource_Management_Plan,9556286,
+NV,,Sloan_Canyon_NCA_RMP,RMP_Resource_Management_Plan,51698554,
+NV,,Tonopah_RMP,RMP_Resource_Management_Plan,15400829,
+NV,,Wells_RMP,RMP_Resource_Management_Plan,80574229,
+NV,,Winnemucca_District_RMP,RMP_Resource_Management_Plan,86689024,
+OR,,Andrews_Management_Area_RMP,RMP_Resource_Management_Plan,35881933,
+OR,,Baker_Field_Office_RMP,RMP_Resource_Management_Plan,49949509,
+OR,,Brothers_La_Pine_RMP_and_ROD,RMP_Resource_Management_Plan,25377003,
+OR,,Cascade-Siskiyou_NM_RMP_Revision_proposed,RMP_Resource_Management_Plan,43436158,
+OR,,John_Day_Basin_RMP,RMP_Resource_Management_Plan,1307849,
+OR,,Lakeview_RMP,RMP_Resource_Management_Plan,88555014,
+OR,,Northwestern_and_Coastal_Oregon_RMP,RMP_Resource_Management_Plan,76729348,
+OR,,San_Juan_Islands_NM_RMP_Development_proposed,RMP_Resource_Management_Plan,72230698,
+OR,,Spokane_District_RMP,RMP_Resource_Management_Plan,90070508,
+OR,,Steens_Mountain_Cooperative_Management_and_Protection_Area_R,RMP_Resource_Management_Plan,25804071,
+OR,,Three_Rivers_RMP_and_ROD,RMP_Resource_Management_Plan,61295012,
+OR,,Two_Rivers_RMP_and_ROD,RMP_Resource_Management_Plan,69118165,
+OR,,Upper_Deschutes_RMP,RMP_Resource_Management_Plan,86158415,
+OR,,Upper_Klamath_Basin_and_Wood_River_Wetland_ROD_and_RMP,RMP_Resource_Management_Plan,44586561,
+OR,,West_Eugene_Wetlands_RMP,RMP_Resource_Management_Plan,90944763,
+UT,,Bears_Ears_NM_RMP_Revision_proposed,RMP_Resource_Management_Plan,9969233,
+UT,,Box_Elder_RMP,RMP_Resource_Management_Plan,48914597,
+UT,,Cedar_Beaver_Garfield_Antimony_ROD_and_RMP,RMP_Resource_Management_Plan,47369722,
+UT,,Cedar_City_RMP_Revision_proposed,RMP_Resource_Management_Plan,54185302,
+UT,,Grand_Staircase-Escalante_NM_RMP_Revision_proposed,RMP_Resource_Management_Plan,75713142,
+UT,,House_Range_Resource_Area_RMP_and_ROD,RMP_Resource_Management_Plan,6748568,
+UT,,Isolated_Tract_Planning_Analysis,RMP_Resource_Management_Plan,29430495,
+UT,,Kanab_Field_Office_ROD_and_Approved_RMP,RMP_Resource_Management_Plan,30070478,
+UT,,Moab_Field_Office_ROD_and_Approved_RMP,RMP_Resource_Management_Plan,35629161,
+UT,,Monticello_RMP,RMP_Resource_Management_Plan,5029659,
+UT,,Park_City_MFP,RMP_Resource_Management_Plan,21186216,
+UT,,Pinyon_MFP,RMP_Resource_Management_Plan,53103483,
+UT,,Pony_Express_RMP,RMP_Resource_Management_Plan,22296049,
+UT,,Price_RMP,RMP_Resource_Management_Plan,59016563,
+UT,,Randolph_MFP,RMP_Resource_Management_Plan,23712262,
+UT,,Red_Cliffs_NCA_RMP,RMP_Resource_Management_Plan,23101027,
+UT,,Richfield_Field_Office_ROD_and_Approved_RMP,RMP_Resource_Management_Plan,45041856,
+UT,,St._George_Field_Office_ROD_and_RMP,RMP_Resource_Management_Plan,43863639,
+UT,,Vernal_Field_Office_ROD_and_Approved_RMP,RMP_Resource_Management_Plan,16333472,
+UT,,Warm_Springs_Resource_Area_Approved_RMP_ROD,RMP_Resource_Management_Plan,97347060,
+WY,,Buffalo_RMP,RMP_Resource_Management_Plan,40451061,
+WY,,Casper_RMP,RMP_Resource_Management_Plan,36548831,
+WY,,Cody_RMP,RMP_Resource_Management_Plan,11419523,
+WY,,Kemmerer_RMP,RMP_Resource_Management_Plan,18907881,
+WY,,Lander_RMP,RMP_Resource_Management_Plan,44340490,
+WY,,Nebraska_RMP,RMP_Resource_Management_Plan,44565171,
+WY,,Newcastle_&_Nebraska_RMPs_Revision_proposed,RMP_Resource_Management_Plan,91663595,
+WY,,Newcastle_RMP,RMP_Resource_Management_Plan,24790466,
+WY,,Pinedale_RMP,RMP_Resource_Management_Plan,91339324,
+WY,,Rawlins_RMP,RMP_Resource_Management_Plan,65447033,
+WY,,Rock_Springs_RMP_Revision_proposed,RMP_Resource_Management_Plan,30630964,
+WY,,Snake_River_RMP,RMP_Resource_Management_Plan,37751410,

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 +1,11 @@
+"""COMPASS Geothermal Resource Management Plan plugin"""
+
+import importlib.resources
+
+from compass.plugin import create_schema_based_one_shot_extraction_plugin
+
+
+COMPASSGeoRMPExtractor = create_schema_based_one_shot_extraction_plugin(
+    importlib.resources.files("compass.extraction.rmp") / "plugin_config.yaml",
+    tech="geo_rmp",
+)

compass/extraction/rmp/plugin_config.yaml

@@ -0,0 +1,83 @@
+# 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"
+  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"
+  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.
+
+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