Retire the legacy SQLite targets table and everything that reads or writes it

.github/workflows/ci.yml

@@ -112,11 +112,10 @@ jobs:
       - name: Test OpenTimestamps anchoring
         run: uv run pytest -q --durations=30 tests/test_ots_anchor.py
 
-      - name: Build source input database
+      - name: Smoke-test the source-artifact database CLI
         run: |
-          uv run chronicle --db /tmp/chronicle-targets-ci.db init
-          uv run chronicle --db /tmp/chronicle-targets-ci.db load all
-          uv run chronicle --db /tmp/chronicle-targets-ci.db stats
+          uv run chronicle --db /tmp/chronicle-source-ci.db init
+          uv run chronicle --db /tmp/chronicle-source-ci.db source-stats
 
       - name: Build wheel
         run: |
@@ -134,13 +133,17 @@ jobs:
           import policyengine_chronicle.normalization
           import policyengine_chronicle.sources
           import policyengine_chronicle.targets
+          from policyengine_chronicle.targets import hard_target_package_aliases
           from policyengine_chronicle.schema import (
               CONSUMER_FACT_SCHEMA_SHA256,
               CONSUMER_FACT_SCHEMA_SHA256_BY_VERSION,
               consumer_fact_schema,
           )
 
           assert policyengine_chronicle.__name__ == "policyengine_chronicle"
+          # Target source coverage ships; the legacy target tables do not.
+          assert "soi-table-1-1" in hard_target_package_aliases()
+          assert not hasattr(policyengine_chronicle.targets, "query_targets")
           # Every consumer-fact contract must ship inside the wheel: frozen v1,
           # the v2 labels contract, the v3 geography-name contract, and the v4
           # contract rows are emitted under.

README.md

@@ -108,12 +108,12 @@ chronicle/
 │   ├── sources/             # Source lineage helpers
 │   ├── facts/               # Source-backed facts
 │   ├── normalization/       # Low-assumption representation helpers
-│   ├── targets/             # Target input schema, client, loaders
+│   ├── targets/             # Target source coverage (US poverty/nonfiler)
 │   ├── jurisdictions/       # Temporary in-repo jurisdiction source prototypes
-├── db/                      # SQLModel persistence and source loaders
-│   ├── schema.py            # SQLModel: Target, Stratum, StratumConstraint
+├── db/                      # Source-artifact store and packaged source files
+│   ├── schema.py            # SQLModel: SourceArtifact, SourceTable, SourceRow
 │   ├── supabase_client.py   # Supabase client helpers
-│   └── etl_*.py             # Source-specific ETL pipelines
+│   └── data/                # Publisher files that source packages read
 ├── data/                    # Cached data files
 └── docs/                    # Architecture and source documentation
 ```
@@ -127,23 +127,16 @@ Solver execution and calibrated dataset construction belong in Microcosm.
 
 ### 1. Install
 
+Chronicle is not published on PyPI. Clone this repository into a
+Chronicle-named directory and install it with its development extras:
+
 ```bash
-pip install policyengine-chronicle
-# Or for development, clone this repository into a Chronicle-named directory:
 git clone <current repository URL> chronicle
 cd chronicle
-pip install -e ".[dev]"
-```
-
-### 2. Initialize and Load Legacy Target Inputs
-
-```bash
-chronicle init
-chronicle load soi --years 2021
-chronicle stats
+uv sync --all-extras
 ```
 
-### 3. Validate Fixture Facts
+### 2. Validate Fixture Facts
 
 The standalone Chronicle fact harness validates JSONL aggregate facts and emits a
 JSON report with fact counts, QA counts, warnings, and validation errors:
@@ -516,7 +509,7 @@ same `--axiom-cli` and `--axiom-root` flags, plus
 `--require-axiom-validation` when skipped concept checks should fail agent
 acceptance.
 
-### 4. Run Chronicle Explorer
+### 3. Run Chronicle Explorer
 
 Chronicle Explorer is a Next/Tailwind app that reads the fixture fact JSONL and
 source-cell JSONL, then shows aggregate facts, source-cell lineage, and
@@ -538,25 +531,19 @@ By default, the workbench reads the current local suite outputs at
 LEDGER_EXPLORER_DATA_DIRS=/tmp/chronicle-build-a,/tmp/chronicle-build-b npm run dev -- --port 3090
 ```
 
-### 5. Query Target Inputs in Python
-
-```python
-from policyengine_chronicle.targets import DataSource, Target, TargetType, query_targets
-
-target_rows = query_targets(jurisdiction="us", year=2024)
-```
-
-## Target Input Schema
-
-Target inputs use a three-table schema:
-
-- **strata**: Population subgroups, such as California filers with AGI between
-  $50k and $75k.
-- **stratum_constraints**: Rules defining each stratum.
-- **targets**: Source-published aggregate values linked to strata.
+## Retired legacy target tables
 
-These are source-backed inputs. Microcosm owns the contracts that select them,
-the active support-aware subset, and calibrated solver execution.
+Chronicle used to load hand-written ETLs into a local SQLite `targets` /
+`strata` / `stratum_constraints` schema (`chronicle load`, `chronicle stats`,
+`chronicle query`) and to read a Supabase copy through
+`policyengine_chronicle.targets.query_targets`. No live consumer read those
+tables, and most of their values did not trace to publisher bytes, so they were
+retired.
+Source-backed facts reach consumers through source packages, the bundle and
+the consumer-fact export. Microcosm owns the contracts that select them, the
+active support-aware subset, and calibrated solver execution. See
+[`docs/us-legacy-etl-cleanup.md`](docs/us-legacy-etl-cleanup.md) for what each
+retired loader held and which package publishes it now.
 
 ## Identifier Epochs
 
@@ -623,7 +610,7 @@ normalized_fact = convert_units(fact, 1000, "count")
 
 ## Current Coverage
 
-### Aggregate Facts And Target Inputs
+### Aggregate facts
 
 | Source | Coverage | Description |
 |--------|----------|-------------|

changelog.d/legacy-targets-table-retired.removed.md

@@ -0,0 +1,22 @@
+Retire the legacy SQLite target tables (`targets`, `strata`,
+`stratum_constraints`) and everything that wrote or read them:
+- the `chronicle load`, `chronicle stats` and `chronicle query` commands
+- the last two loaders, `db/etl_soi.py` and `db/etl_snap.py`
+- the Supabase `query_targets` / `query_strata` / `insert_targets_batch` helpers
+- `scripts/export_to_json.py`
+- the `as_target` / `target_kwargs` / `TargetBlueprint` normalization helpers
+- the legacy names in `chronicle.targets` and `policyengine_chronicle.targets`
+
+Nothing reads these tables today. The values the last two loaders wrote (IRS
+SOI Tables 1.1 and 1.4, and SNAP) are published as source-package facts through
+the bundle and consumer-fact export, which are unchanged.
+
+`chronicle init` now creates only the source-artifact tables. The AGI-bracket
+tables the SOI harness used move into `chronicle/jurisdictions/us/soi.py`.
+`policyengine_chronicle.targets` keeps the US poverty/nonfiler target source
+coverage.
+
+Migration: Chronicle was never published on PyPI, and no live repository
+imports these names. Archived Cosilico-era repositories once did. Code that called `query_targets` should read consumer facts from a
+pinned bundle (`chronicle build-bundle`, `chronicle build-consumer-artifact`)
+instead.

chronicle/facts/models.py

@@ -20,7 +20,7 @@ class DerivationStep:
 
 @dataclass(frozen=True)
 class SourceFact:
-    """A source-backed fact before target-input materialization."""
+    """A source-backed fact with its structural normalization lineage."""
 
     name: str
     value: float

chronicle/jurisdictions/us/__init__.py

@@ -1,11 +1,7 @@
-"""United States Chronicle target loaders."""
+"""United States Chronicle source builders.
 
-from chronicle.targets.loaders import (
-    load_snap_targets,
-    load_soi_targets,
-)
+US measures are source packages. The legacy DB target loaders were retired
+(see docs/us-legacy-etl-cleanup.md), so this module exports no loaders.
+"""
 
-__all__ = [
-    "load_snap_targets",
-    "load_soi_targets",
-]
+__all__: list[str] = []

chronicle/jurisdictions/us/soi.py

@@ -24,7 +24,6 @@
     SourceRecordSpec,
 )
 from chronicle.yaml_io import safe_load as safe_load_yaml
-from db.etl_soi import AGI_BRACKETS, TABLE_1_1_AGI_LABEL_TO_BRACKET
 
 SOI_TABLE_1_1_SOURCE_NAME = "irs_soi"
 SOI_TABLE_1_1_SOURCE_TABLE = "Publication 1304 Table 1.1"
@@ -48,6 +47,51 @@
     "tax-year source record."
 )
 
+# AGI bracket definitions (lower, upper) in dollars.
+AGI_BRACKETS = {
+    "under_1": (float("-inf"), 1),
+    "1_to_5k": (1, 5_000),
+    "5k_to_10k": (5_000, 10_000),
+    "10k_to_15k": (10_000, 15_000),
+    "15k_to_20k": (15_000, 20_000),
+    "20k_to_25k": (20_000, 25_000),
+    "25k_to_30k": (25_000, 30_000),
+    "30k_to_40k": (30_000, 40_000),
+    "40k_to_50k": (40_000, 50_000),
+    "50k_to_75k": (50_000, 75_000),
+    "75k_to_100k": (75_000, 100_000),
+    "100k_to_200k": (100_000, 200_000),
+    "200k_to_500k": (200_000, 500_000),
+    "500k_to_1m": (500_000, 1_000_000),
+    "1m_to_1_5m": (1_000_000, 1_500_000),
+    "1_5m_to_2m": (1_500_000, 2_000_000),
+    "2m_to_5m": (2_000_000, 5_000_000),
+    "5m_to_10m": (5_000_000, 10_000_000),
+    "10m_plus": (10_000_000, float("inf")),
+}
+
+TABLE_1_1_AGI_LABEL_TO_BRACKET = {
+    "No adjusted gross income": "under_1",
+    "$1 under $5,000": "1_to_5k",
+    "$5,000 under $10,000": "5k_to_10k",
+    "$10,000 under $15,000": "10k_to_15k",
+    "$15,000 under $20,000": "15k_to_20k",
+    "$20,000 under $25,000": "20k_to_25k",
+    "$25,000 under $30,000": "25k_to_30k",
+    "$30,000 under $40,000": "30k_to_40k",
+    "$40,000 under $50,000": "40k_to_50k",
+    "$50,000 under $75,000": "50k_to_75k",
+    "$75,000 under $100,000": "75k_to_100k",
+    "$100,000 under $200,000": "100k_to_200k",
+    "$200,000 under $500,000": "200k_to_500k",
+    "$500,000 under $1,000,000": "500k_to_1m",
+    "$1,000,000 under $1,500,000": "1m_to_1_5m",
+    "$1,500,000 under $2,000,000": "1_5m_to_2m",
+    "$2,000,000 under $5,000,000": "2m_to_5m",
+    "$5,000,000 under $10,000,000": "5m_to_10m",
+    "$10,000,000 or more": "10m_plus",
+}
+
 
 def build_soi_table_1_1_facts(year: int = 2023) -> list[AggregateFact]:
     """Build source-lineaged Chronicle facts from SOI Table 1.1."""

chronicle/normalization/__init__.py

@@ -7,24 +7,18 @@
 """
 
 from chronicle.facts import DerivationStep, SourceFact
-from .targets import (
-    TargetBlueprint,
+from .transforms import (
     apply_share,
-    as_target,
     convert_units,
     format_derivation,
     scale_value,
-    target_kwargs,
 )
 
 __all__ = [
     "DerivationStep",
     "SourceFact",
-    "TargetBlueprint",
     "apply_share",
-    "as_target",
     "convert_units",
     "format_derivation",
     "scale_value",
-    "target_kwargs",
 ]