Updates post Solar Validation

compass/common/__init__.py

@@ -2,6 +2,7 @@
 
 from .base import (
     EXTRACT_ORIGINAL_TEXT_PROMPT,
+    SYSTEM_SIZE_REMINDER,
     BaseTextExtractor,
     empty_output,
     llm_response_starts_with_no,

compass/extraction/apply.py

@@ -22,6 +22,7 @@ async def check_for_ordinance_info(
     doc,
     model_config,
     heuristic,
+    tech,
     ordinance_text_collector_class,
     permitted_use_text_collector_class=None,
     usage_tracker=None,
@@ -36,6 +37,9 @@ async def check_for_ordinance_info(
         ``"contains_ord_info"`` key, it will not be processed. To force
         a document to be processed by this function, remove that key
         from the documents attrs.
+    tech : str
+        Technology of interest (e.g. "solar", "wind", etc). This is
+        used to set up some document validation decision trees.
     text_splitter : obj
         Instance of an object that implements a `split_text` method.
         The method should take text as input (str) and return a list
@@ -65,6 +69,7 @@ async def check_for_ordinance_info(
     chunks = model_config.text_splitter.split_text(doc.text)
     chunk_parser = ParseChunksWithMemory(chunks, num_to_recall=2)
     legal_text_validator = LegalTextValidator(
+        tech=tech,
         llm_service=model_config.llm_service,
         usage_tracker=usage_tracker,
         doc_is_from_ocr=doc.attrs.get("from_ocr", False),

compass/extraction/date.py

@@ -44,8 +44,10 @@ def __init__(self, structured_llm_caller, text_splitter=None):
         structured_llm_caller : compass.llm.StructuredLLMCaller
             StructuredLLMCaller instance. Used for structured validation
             queries.
-        text_splitter : langchain.text_splitter.TextSplitter, optional
-            Optional text splitter instance to attach to doc (used for
+        text_splitter : TextSplitter, optional
+            Optional
+            :class:`langchain_text_splitters.character.TextSplitter`
+            text splitter instance to attach to doc (used for
             splitting out pages in an HTML document).
             By default, ``None``.
         """

compass/extraction/solar/__init__.py

@@ -47,4 +47,5 @@
     "renewable": 3,
     "municipal": 1,
     "department": 1,
+    # TODO: add board???
 }

compass/extraction/solar/graphs.py

@@ -4,6 +4,7 @@
     setup_graph_no_nodes,
     llm_response_starts_with_yes,
     llm_response_starts_with_no,
+    SYSTEM_SIZE_REMINDER,
 )
 
 
@@ -45,22 +46,101 @@ def setup_graph_sef_types(**kwargs):
             "What are the different solar energy farm sizes **regulated by "
             "this ordinance**? List them in order of increasing size. "
             "Include any relevant numerical qualifiers in the name, if "
-            "appropriate. Only include solar energy farm types; do not "
-            "include generic types or other energy system types."
+            "appropriate. Only include systems that resemble ground-mounted "
+            "solar energy farms; do not include other solar energy system "
+            "types lice CSP or roof-mounted systems or other technologies "
+            "like wind energy systems, geothermal energy systems, etc."
         ),
     )
-    G.add_edge("get_text", "final")
+    G.add_edge("get_text", "get_regulated")
     G.add_node(
-        "final",
+        "get_regulated",
+        prompt=(
+            "Are any of these systems **not** regulated by this ordinance?"
+        ),
+    )
+
+    G.add_edge("get_regulated", "get_largest")
+    G.add_node(
+        "get_largest",
+        prompt=(
+            "What is the **largest** solar energy farm size that **is "
+            "regulated by this ordinance**?"
+        ),
+    )
+
+    G.add_edge("get_largest", "check_matches_definition")
+    G.add_node(
+        "check_matches_definition",
+        prompt=(
+            "Does the ordinance explicitly define this system as large, "
+            "commercial, utility-scale, or something akin to that? "
+            "Please start your response with either 'Yes' or 'No' and briefly "
+            "explain your answer."
+        ),
+    )
+
+    G.add_edge(
+        "check_matches_definition",
+        "final_large",
+        condition=llm_response_starts_with_yes,
+    )
+    G.add_edge(
+        "check_matches_definition",
+        "check_scale_reason",
+        condition=llm_response_starts_with_no,
+    )
+    G.add_node(
+        "check_scale_reason",
+        prompt=(
+            "Would a reasonable person classify this kind of system as a "
+            "**large**, commercial, or even utility-scale** solar energy farm "
+            "(e.g. with the primary purpose of generating electricity for "
+            "sale, as opposed to small, residential, roof-mounted, private, "
+            "or other kinds of 'small' systems)? "
+            "Please start your response with either 'Yes' or 'No' and briefly "
+            "explain your answer."
+        ),
+    )
+
+    G.add_edge(
+        "check_scale_reason",
+        "final_large",
+        condition=llm_response_starts_with_yes,
+    )
+    G.add_edge(
+        "check_scale_reason",
+        "final_small",
+        condition=llm_response_starts_with_no,
+    )
+    G.add_node(
+        "final_large",
         prompt=(
             "Respond based on our entire conversation so far. Return your "
             "answer as a dictionary in JSON format (not markdown). Your "
             "JSON file must include exactly two keys. The keys are "
-            "'largest_sef_type' and 'explanation'. The value of the "
-            "'largest_sef_type' key should be a string that labels the "
+            "'largest_sef_type' and 'explanation', and 'is_large'. The value "
+            "of the 'largest_sef_type' key should be a string that labels the "
             "largest solar energy system size **regulated by this "
             "ordinance**. The value of the 'explanation' key should be a "
-            "string containing a short explanation for your choice."
+            "string containing a short explanation for your choice. The value "
+            "of the 'is_large' key should be the boolean value `true`, since "
+            "we determined this is a large-scale system."
+        ),
+    )
+    G.add_node(
+        "final_small",
+        prompt=(
+            "Respond based on our entire conversation so far. Return your "
+            "answer as a dictionary in JSON format (not markdown). Your "
+            "JSON file must include exactly two keys. The keys are "
+            "'largest_sef_type' and 'explanation', and 'is_large'. The value "
+            "of the 'largest_sef_type' key should be a string that labels the "
+            "largest solar energy system size **regulated by this "
+            "ordinance**. The value of the 'explanation' key should be a "
+            "string containing a short explanation for your choice. The value "
+            "of the 'is_large' key should be the boolean value `false`, since "
+            "we determined this is not a large-scale system."
         ),
     )
     return G
@@ -90,14 +170,10 @@ def setup_multiplier(**kwargs):
             "Does the text mention a multiplier that should be applied to the "
             "structure height to compute the setback distance from {feature} "
             "for {tech}? "
-            "Focus only on {feature}; do not respond based on any text "
+            "Please consider only {feature}; do not respond based on any text "
             "related to {ignore_features}. "
-            "Please only consider setbacks specifically for systems that "
-            "would typically be defined as {tech} based on the text itself "
-            "— for example, systems intended for electricity generation or "
-            "sale, or those above thresholds such as height or rated "
-            "capacity. Ignore any requirements that apply only to smaller "
-            "or clearly non-commercial systems. "
+            "Please also only consider setbacks specifically for "
+            f"{SYSTEM_SIZE_REMINDER}"
             "Please start your response with either 'Yes' or 'No' and briefly "
             "explain your answer."
         ),
@@ -108,14 +184,10 @@ def setup_multiplier(**kwargs):
         prompt=(
             "Does the ordinance give the setback from {feature} as a fixed "
             "distance value? "
-            "Focus only on {feature}; do not respond based on any text "
-            "related to {ignore_features}. "
-            "Please only consider setbacks specifically for systems that "
-            "would typically be defined as {tech} based on the text itself "
-            "— for example, systems intended for electricity generation or "
-            "sale, or those above thresholds such as height or rated "
-            "capacity. Ignore any requirements that apply only to smaller "
-            "or clearly non-commercial systems. "
+            "Please consider only on {feature}; do not respond based on any "
+            "text related to {ignore_features}. "
+            "Please also only consider setbacks specifically for "
+            f"{SYSTEM_SIZE_REMINDER}"
             "Please start your response with either 'Yes' or "
             "'No' and briefly explain your answer."
         ),
@@ -156,8 +228,8 @@ def setup_multiplier(**kwargs):
             "of the 'units' key should be a string corresponding to the "
             "(standard) units of the setback distance value from {feature} "
             "or `null` if there was no such value. "
-            "As before, focus only on setbacks specifically for systems that "
-            "would typically be defined as {tech} based on the text itself. "
+            "As before, focus only on setbacks that would apply for"
+            f"{SYSTEM_SIZE_REMINDER}"
             "{SUMMARY_PROMPT} {SECTION_PROMPT}"
         ),
     )
@@ -169,12 +241,8 @@ def setup_multiplier(**kwargs):
             "Are multiple values given for the multiplier used to "
             "compute the setback distance value from {feature} for {tech}? "
             "Remember to ignore any text related to {ignore_features}. "
-            "Focus only on setbacks specifically for systems that would "
-            "typically be defined as {tech} based on the text itself — for "
-            "example, systems intended for electricity generation or sale, "
-            "or those above thresholds such as height or rated capacity. "
-            "Ignore any requirements that apply only to smaller or clearly "
-            "non-commercial systems. "
+            "Please only consider setbacks specifically for "
+            f"{SYSTEM_SIZE_REMINDER}"
             "If so, select and state the largest one. Otherwise, repeat the "
             "single multiplier value that was given in the text. "
         ),
@@ -187,12 +255,8 @@ def setup_multiplier(**kwargs):
             "static distance value that should be added to the result of "
             "the multiplication? "
             "Remember to ignore any text related to {ignore_features}. "
-            "Focus only on setbacks specifically for systems that would "
-            "typically be defined as {tech} based on the text itself — for "
-            "example, systems intended for electricity generation or sale, "
-            "or those above thresholds such as height or rated capacity. "
-            "Ignore any requirements that apply only to smaller or clearly "
-            "non-commercial systems. "
+            "Please only consider setbacks specifically for "
+            f"{SYSTEM_SIZE_REMINDER}"
             "Do not confuse this value with static setback requirements. "
             "Ignore text with clauses such as 'no lesser than', 'no greater "
             "than', 'the lesser of', or 'the greater of'. Please start your "

compass/extraction/solar/ordinance.py

@@ -21,15 +21,16 @@
     "facilities (SEF), solar energy farms (SEF), solar farms (SF), "
     "utility-scale solar energy systems (USES), commercial solar energy "
     "systems (CSES), ground-mounted solar energy systems (GSES), "
-    "alternate energy systems (AES), or similar"
+    "alternate energy systems (AES), commercial energy production "
+    "systems (CEPCS), or similar"
 )
 _SEARCH_TERMS_AND = (
     "zoning, siting, setback, system design, and operational "
     "requirements/restrictions"
 )
 _SEARCH_TERMS_OR = _SEARCH_TERMS_AND.replace("and", "or")
 _IGNORE_TYPES = (
-    "private, residential, roof-mounted, micro, small, or medium sized"
+    "CSP, private, residential, roof-mounted, micro, small, or medium sized"
 )
 
 
@@ -175,7 +176,8 @@ def contains_ord_info(self):
     def ordinance_text(self):
         """str: Combined ordinance text from the individual chunks"""
         logger.debug(
-            "Grabbing %d chunk(s) from original text at these indices: %s",
+            "Grabbing %d ordinance chunk(s) from original text at these "
+            "indices: %s",
             len(self._ordinance_chunks),
             list(self._ordinance_chunks),
         )
@@ -212,9 +214,12 @@ class SolarPermittedUseDistrictsTextCollector(StructuredLLMCaller):
 
     DISTRICT_PROMPT = (
         "You are a legal scholar that reads ordinance text and determines "
-        "whether the text explicitly details the districts where large "
-        "solar energy farms are a permitted use. Large solar energy systems "
-        f"(SES) may also be referred to as {_LARGE_SEF_SYNONYMS}. "
+        "whether it explicitly contains relevant information to determine the "
+        "districts (and especially the district names) where large solar "
+        "energy farms are a permitted use (primary, special, accessory, or "
+        "otherwise), as well as the districts where large solar energy farms "
+        "are prohibited entirely. Large solar energy systems (SES) may also "
+        f"be referred to as {_LARGE_SEF_SYNONYMS}. "
         "Do not make any inferences; only answer based on information that "
         "is explicitly stated in the text. "
         "Note that relevant information may sometimes be found in tables. "
@@ -287,7 +292,8 @@ def contains_district_info(self):
     def permitted_use_district_text(self):
         """str: Combined permitted use districts text from the chunks"""
         logger.debug(
-            "Grabbing %d chunk(s) from original text at these indices: %s",
+            "Grabbing %d permitted use chunk(s) from original text at these "
+            "indices: %s",
             len(self._district_chunks),
             list(self._district_chunks),
         )
@@ -333,6 +339,8 @@ class SolarOrdinanceTextExtractor(BaseTextExtractor):
         f"\t{_LARGE_SEF_SYNONYMS.capitalize()}.\n"
         "- Explicitly include any text related to **bans or prohibitions** "
         "on solar energy systems.\n"
+        "- Explicitly include any text related to the adoption or enactment "
+        "date of the ordinance (if any).\n"
         "\n2. ## Exclusions ##:\n"
         "- Do **not** include text that does not pertain to solar energy "
         "systems.\n"
@@ -426,7 +434,8 @@ class SolarPermittedUseDistrictsTextExtractor(BaseTextExtractor):
         "\n1. ## Scope of Extraction ##:\n"
         "- Retain all text defining permitted use(s) for a district, "
         "including:\n"
-        "\t- **Primary, Special, Accessory, and other permitted use types.**\n"
+        "\t- **Primary, Special, Conditional, Accessory, Prohibited, and "
+        "any other use types.**\n"
         "\t- **District names and zoning classifications.**\n"
         "- Pay extra attention to any references to **solar energy "
         "facilities** or related terms.\n"
@@ -471,7 +480,8 @@ class SolarPermittedUseDistrictsTextExtractor(BaseTextExtractor):
         "\n1. ## Scope of Extraction ##:\n"
         "- Retain all text defining permitted use(s) for a district, "
         "including:\n"
-        "\t- **Primary, Special, Accessory, and other permitted use types.**\n"
+        "\t- **Primary, Special, Conditional, Accessory, Prohibited, and "
+        "any other use types.**\n"
         "\t- **District names and zoning classifications.**\n"
         "- Ensure that **tables, lists, and structured elements** are "
         "preserved as they may contain relevant details.\n"