metadata: full vgi-lint conformance at latest linter; bump deps; add browsable table

Bring vgi-pii to a clean vgi-lint gate (score 64 -> 95, 0 findings at
fail-on: info) against the LATEST vgi-lint-check, and drop the pinned
`version: "0.37.0"` from the CI action so the fleet tracks latest.

Metadata fixes:
- Fully qualify every example query and agent-test reference_sql as
  `pii.main.<object>` (was `pii.<object>`); under-qualified refs silently
  failed coverage attribution and --execute. Fixes VGI511/VGI520 across all
  six objects (example_coverage 0.0 -> 1.0, test_coverage 0.0 -> 1.0).
- Migrate the retired `vgi.result_columns_md` tag (VGI414) to the structured
  `vgi.result_columns_schema` JSON on both table functions.
- Remove complete runnable SELECT ... FROM queries from vgi.doc_md
  (VGI179) on has_pii / detect_pii / supported_entities; replace with
  signature + column prose. Examples live in vgi.example_queries.
- Rewrite the two bare `SELECT *` examples into projected/filtered analytical
  queries (VGI514).
- Add a browsable `entity_types` discovery table (VALUES-style, backed by the
  existing supported_entities generator) so the worker is not table-functions-
  only (VGI146), with its own docs, category, example queries, and a
  recognizes-credit-card agent task.

Dependencies:
- Bump vgi-python 0.9 -> 0.14 (pinned <0.15) and vgi-rpc 0.22 -> 0.24; commit
  the refreshed uv.lock. vgi-python 0.15 changes the table RPC schema
  (required_filters: list<list<string>>) which the currently released signed
  vgi extension does not yet accept, so a worker exposing a table cannot attach
  at 0.15; 0.14 matches the conformant sibling vgi-conform and is green.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: rustyconover

.github/workflows/ci.yml

@@ -102,7 +102,6 @@ jobs:
       - name: vgi-lint
         uses: Query-farm/vgi-lint-check@v1
         with:
-          version: "0.37.0"
           location: "uv run --python 3.13 ${{ github.workspace }}/pii_worker.py"
           fail-on: info
 

pii_worker.py

@@ -1,7 +1,7 @@
 # /// script
 # requires-python = ">=3.13"
 # dependencies = [
-#     "vgi-python[http]>=0.9.0",
+#     "vgi-python[http]>=0.14.0,<0.15.0",
 #     "presidio-analyzer>=2.2",
 #     "presidio-anonymizer>=2.2",
 #     "spacy>=3.7",
@@ -25,12 +25,12 @@
     INSTALL vgi FROM community; LOAD vgi;
     ATTACH 'pii' (TYPE vgi, LOCATION 'uv run pii_worker.py');
 
-    SELECT pii.has_pii('Call John Smith at john@example.com');        -- true
-    SELECT pii.redact('Call John Smith at john@example.com');         -- 'Call <PERSON> at <EMAIL_ADDRESS>'
-    SELECT pii.anonymize('Call John Smith at john@example.com');      -- 'Call **** at ****************'
-    SELECT pii.pii_types('Call John Smith at john@example.com');      -- ['EMAIL_ADDRESS', 'PERSON']
-    SELECT * FROM pii.detect_pii('Call John Smith at john@example.com');
-    SELECT * FROM pii.supported_entities() ORDER BY entity_type;
+    SELECT pii.main.has_pii('Call John Smith at john@example.com');        -- true
+    SELECT pii.main.redact('Call John Smith at john@example.com');         -- 'Call <PERSON> at <EMAIL_ADDRESS>'
+    SELECT pii.main.anonymize('Call John Smith at john@example.com');      -- 'Call **** at ****************'
+    SELECT pii.main.pii_types('Call John Smith at john@example.com');      -- ['EMAIL_ADDRESS', 'PERSON']
+    SELECT * FROM pii.main.detect_pii('Call John Smith at john@example.com');
+    SELECT * FROM pii.main.supported_entities() ORDER BY entity_type;
 """
 
 from __future__ import annotations
@@ -43,7 +43,7 @@
 
 from vgi_pii import engine
 from vgi_pii.scalars import SCALAR_FUNCTIONS
-from vgi_pii.tables import TABLE_FUNCTIONS
+from vgi_pii.tables import DISCOVERY_TABLES, TABLE_FUNCTIONS
 
 _CATALOG_DESCRIPTION_LLM = (
     "Detect and redact personally-identifiable information (PII) in free text directly in SQL, "
@@ -86,7 +86,7 @@
     "redaction powers safe sharing, export, and archival. Every operation accepts an optional "
     "ISO language argument (defaulting to English) and treats NULL or blank input gracefully.\n\n"
     "```sql\n"
-    "SELECT pii.redact('Call John Smith at john@example.com');\n"
+    "SELECT pii.main.redact('Call John Smith at john@example.com');\n"
     "-- 'Call <PERSON> at <EMAIL_ADDRESS>'\n"
     "```\n\n"
     "Learn more from the [Presidio documentation](https://microsoft.github.io/presidio/), the "
@@ -218,7 +218,7 @@
             "Does the text 'Contact Jane Doe at jane@example.com' contain any "
             "personally-identifiable information? Return a single boolean value."
         ),
-        "reference_sql": "SELECT pii.has_pii('Contact Jane Doe at jane@example.com')",
+        "reference_sql": "SELECT pii.main.has_pii('Contact Jane Doe at jane@example.com')",
         "success_criteria": (
             "Returns a single boolean value that is true, obtained by calling the worker's "
             "PII-presence predicate on the given text."
@@ -228,7 +228,7 @@
     {
         "name": "list-pii-types",
         "prompt": ("List the distinct PII entity types present in the text 'Contact Jane Doe at jane@example.com'."),
-        "reference_sql": "SELECT pii.pii_types('Contact Jane Doe at jane@example.com')",
+        "reference_sql": "SELECT pii.main.pii_types('Contact Jane Doe at jane@example.com')",
         "success_criteria": (
             "Returns the sorted array of distinct PII entity types (such as EMAIL_ADDRESS and "
             "PERSON) present in the text."
@@ -241,7 +241,7 @@
             "Produce a copy of the text 'Email jane@example.com now' in which every PII value "
             "is replaced by a tag naming its entity type (for example <EMAIL_ADDRESS>)."
         ),
-        "reference_sql": "SELECT pii.redact('Email jane@example.com now')",
+        "reference_sql": "SELECT pii.main.redact('Email jane@example.com now')",
         "success_criteria": (
             "Returns the input text with each detected PII entity replaced by a <TYPE> tag naming its entity type."
         ),
@@ -253,7 +253,7 @@
             "Produce a copy of the text 'Email jane@example.com now' in which every PII value's "
             "characters are masked with asterisks."
         ),
-        "reference_sql": "SELECT pii.anonymize('Email jane@example.com now')",
+        "reference_sql": "SELECT pii.main.anonymize('Email jane@example.com now')",
         "success_criteria": (
             "Returns the input text with each detected PII entity's characters overwritten by '*' asterisks."
         ),
@@ -267,7 +267,7 @@
         ),
         "reference_sql": (
             "SELECT entity_type, text FROM "
-            "pii.detect_pii('Contact Jane Doe at jane@example.com') ORDER BY entity_type, text"
+            "pii.main.detect_pii('Contact Jane Doe at jane@example.com') ORDER BY entity_type, text"
         ),
         "success_criteria": (
             "Returns one row per detected PII entity, each showing the entity type and the "
@@ -279,13 +279,26 @@
     {
         "name": "count-supported-entity-types",
         "prompt": "How many distinct PII entity types can this worker's analyzer detect?",
-        "reference_sql": "SELECT count(*) FROM pii.supported_entities()",
+        "reference_sql": "SELECT count(*) FROM pii.main.supported_entities()",
         "success_criteria": (
             "Returns the number of distinct PII entity types the analyzer supports, obtained "
             "from the worker's supported-entities discovery function."
         ),
         "ignore_column_names": True,
     },
+    {
+        "name": "recognizes-credit-card",
+        "prompt": (
+            "Does this worker's list of recognizable PII entity types include credit-card "
+            "numbers (the CREDIT_CARD type)? Return a single boolean value."
+        ),
+        "reference_sql": ("SELECT count(*) > 0 FROM pii.main.entity_types WHERE entity_type = 'CREDIT_CARD'"),
+        "success_criteria": (
+            "Returns a single boolean that is true, determined by looking up CREDIT_CARD in the "
+            "worker's browsable table of recognizable entity types."
+        ),
+        "ignore_column_names": True,
+    },
 ]
 
 _CATALOG_TAGS = {
@@ -324,6 +337,7 @@
                 "topic": "pii-detection-and-redaction",
             },
             functions=[*SCALAR_FUNCTIONS, *TABLE_FUNCTIONS],
+            tables=list(DISCOVERY_TABLES),
         ),
     ],
 )

pyproject.toml

@@ -18,7 +18,7 @@ classifiers = [
     "Typing :: Typed",
 ]
 dependencies = [
-    "vgi-python>=0.9.0",
+    "vgi-python>=0.14.0,<0.15.0",
     "presidio-analyzer>=2.2",
     "presidio-anonymizer>=2.2",
     "spacy>=3.7",
@@ -30,7 +30,7 @@ dependencies = [
 # addition to stdio. That path needs vgi-python's `http` extra (waitress).
 # CI's http transport leg installs this via `uv sync --extra http`.
 http = [
-    "vgi-python[http]>=0.9.0",
+    "vgi-python[http]>=0.14.0,<0.15.0",
 ]
 dev = [
     "pytest>=8",

uv.lock

@@ -61,15 +61,16 @@
     ),
     doc_md=(
         "# has_pii\n\n"
-        "A scalar predicate that answers a single question: *does this text contain PII?*\n\n"
-        "## Usage\n\n"
-        "```sql\n"
-        "SELECT pii.has_pii('Call John Smith at john@example.com');  -- true\n"
-        "SELECT id FROM messages WHERE pii.has_pii(body);            -- rows needing scrubbing\n"
-        "```\n\n"
-        "## Notes\n\n"
-        "Pass an explicit language as the second argument (`has_pii(text, 'en')`) to scan "
-        "non-English text. NULL or blank input yields NULL rather than `false`."
+        "A scalar predicate that answers a single question: *does this text contain PII?* "
+        "Call it inline in a projection to tag rows, or in a `WHERE` clause to keep only "
+        "the rows that need privacy handling before they are shared, exported, or archived.\n\n"
+        "## Behaviour\n\n"
+        "- Returns `true` as soon as any entity (a person name, email, phone number, credit "
+        "card, US SSN, location, URL, IP address, ...) is detected, otherwise `false`.\n"
+        "- Pass an explicit ISO language as the optional second argument "
+        "(`has_pii(text, 'en')`) to scan non-English text.\n"
+        "- `NULL`, empty, or whitespace-only input yields `NULL` rather than `false`.\n\n"
+        "See this function's example queries for a runnable demonstration."
     ),
     keywords=[
         "pii",
@@ -109,7 +110,7 @@
         "surrounding text intact.\n\n"
         "## Usage\n\n"
         "```sql\n"
-        "SELECT pii.redact('Call John Smith at john@example.com');\n"
+        "SELECT pii.main.redact('Call John Smith at john@example.com');\n"
         "-- 'Call <PERSON> at <EMAIL_ADDRESS>'\n"
         "```\n\n"
         "## Notes\n\n"
@@ -156,7 +157,7 @@
         "the rest of the text readable.\n\n"
         "## Usage\n\n"
         "```sql\n"
-        "SELECT pii.anonymize('Call John Smith at john@example.com');\n"
+        "SELECT pii.main.anonymize('Call John Smith at john@example.com');\n"
         "-- 'Call **** at ****************'\n"
         "```\n\n"
         "## Notes\n\n"
@@ -202,7 +203,7 @@
         "`VARCHAR[]`.\n\n"
         "## Usage\n\n"
         "```sql\n"
-        "SELECT pii.pii_types('Call John Smith at john@example.com');\n"
+        "SELECT pii.main.pii_types('Call John Smith at john@example.com');\n"
         "-- ['EMAIL_ADDRESS', 'PERSON']\n"
         "```\n\n"
         "## Notes\n\n"
@@ -263,7 +264,7 @@ class Meta:
         tags = _HAS_PII_TAGS
         examples = [
             FunctionExample(
-                sql="SELECT pii.has_pii('Call John Smith at john@example.com')",
+                sql="SELECT pii.main.has_pii('Call John Smith at john@example.com')",
                 description="Detect whether text contains PII",
             ),
         ]
@@ -288,7 +289,7 @@ class Meta:
         tags = _HAS_PII_TAGS
         examples = [
             FunctionExample(
-                sql="SELECT pii.has_pii('Call John Smith at john@example.com', 'en')",
+                sql="SELECT pii.main.has_pii('Call John Smith at john@example.com', 'en')",
                 description="Detect PII with an explicit language",
             ),
         ]
@@ -320,7 +321,7 @@ class Meta:
         tags = _REDACT_TAGS
         examples = [
             FunctionExample(
-                sql="SELECT pii.redact('Call John Smith at john@example.com')",
+                sql="SELECT pii.main.redact('Call John Smith at john@example.com')",
                 description="Tag-redact PII (-> 'Call <PERSON> at <EMAIL_ADDRESS>')",
             ),
         ]
@@ -345,7 +346,7 @@ class Meta:
         tags = _REDACT_TAGS
         examples = [
             FunctionExample(
-                sql="SELECT pii.redact('Call John Smith at john@example.com', 'en')",
+                sql="SELECT pii.main.redact('Call John Smith at john@example.com', 'en')",
                 description="Tag-redact PII with an explicit language",
             ),
         ]
@@ -377,7 +378,7 @@ class Meta:
         tags = _ANONYMIZE_TAGS
         examples = [
             FunctionExample(
-                sql="SELECT pii.anonymize('Call John Smith at john@example.com')",
+                sql="SELECT pii.main.anonymize('Call John Smith at john@example.com')",
                 description="Mask PII (-> 'Call **** at ****************')",
             ),
         ]
@@ -402,7 +403,7 @@ class Meta:
         tags = _ANONYMIZE_TAGS
         examples = [
             FunctionExample(
-                sql="SELECT pii.anonymize('Call John Smith at john@example.com', 'en')",
+                sql="SELECT pii.main.anonymize('Call John Smith at john@example.com', 'en')",
                 description="Mask PII with an explicit language",
             ),
         ]
@@ -434,7 +435,7 @@ class Meta:
         tags = _PII_TYPES_TAGS
         examples = [
             FunctionExample(
-                sql="SELECT pii.pii_types('Call John Smith at john@example.com')",
+                sql="SELECT pii.main.pii_types('Call John Smith at john@example.com')",
                 description="The distinct PII types in text (-> ['EMAIL_ADDRESS', 'PERSON'])",
             ),
         ]
@@ -459,7 +460,7 @@ class Meta:
         tags = _PII_TYPES_TAGS
         examples = [
             FunctionExample(
-                sql="SELECT pii.pii_types('Call John Smith at john@example.com', 'en')",
+                sql="SELECT pii.main.pii_types('Call John Smith at john@example.com', 'en')",
                 description="The distinct PII types in text, explicit language",
             ),
         ]