@mason-data/cli 0.8.0 → 0.9.0

package/dist/src/duck/resolved.js

@@ -14,6 +14,10 @@
  * defined is never in the catalog Postgres, so it never reaches here.
  */
 import { exprType } from "./pgtree.js";
+import { HELD_TYPE_INPUT } from "../api/postgres/schema.js";
+import { heldType, isHeldArray } from "./held-types.js";
+/** the key of a type the source made, held as its text (src/duck/held-types.ts) */
+const HELD_KEY = "made";
 const isNode = (v) => v !== null && v !== undefined && typeof v === "object" && !Array.isArray(v);
 /** an OID a customer's database made (an extension's type or operator, a table's row type) differs between databases:
  *  marked `u<oid>` here, and named by the catalog before it is a key (`nameUserKeys`: `ext:=(citext,citext)`) */
@@ -42,7 +46,9 @@ export async function nameUserKeys(rows, keys) {
     const read = async (set, sql) => set.size === 0 ? new Map() : new Map((await rows(sql([...set].join(", ")))).map((r) => [r[0], r[1]]));
     const proc = await read(oids.proc, (l) => `SELECT oid::text, 'ext:' || oid::regprocedure::text FROM pg_proc WHERE oid IN (${l})`);
     const op = await read(oids.op, (l) => `SELECT oid::text, 'ext:' || oid::regoperator::text FROM pg_operator WHERE oid IN (${l})`);
-    const type = await read(oids.type, (l) => `SELECT oid::text, CASE WHEN typtype = 'c' THEN 'rowtype' ELSE 'ext:' || typname END FROM pg_type WHERE oid IN (${l})`);
+    // a type the source made that the copy holds as its text is one key, whichever: what is verified over one is the text's
+    const type = await read(oids.type, (l) => `SELECT t.oid::text, CASE WHEN t.typtype = 'c' THEN 'rowtype' WHEN t.typinput::text LIKE '%${HELD_TYPE_INPUT}%' THEN '${HELD_KEY}'
+    WHEN e.typinput::text LIKE '%${HELD_TYPE_INPUT}%' THEN '${HELD_KEY}[]' ELSE 'ext:' || t.typname END FROM pg_type t LEFT JOIN pg_type e ON e.oid = t.typelem WHERE t.oid IN (${l})`);
     const name = (kind, p) => { const m = USER.exec(p); return m === null ? p : (kind === "operator" ? op : kind === "type" || kind === "cast" ? type : proc).get(m[1]) ?? p; };
     return new Set(list.map((k) => {
         const kind = k.slice(0, k.indexOf(":")), value = k.slice(k.indexOf(":") + 1);
@@ -84,6 +90,9 @@ export function resolvedKeys(tree) {
         const t = exprType(v);
         if (t !== null && t.type !== 0)
             keys.add(`type:${oidKey(t.type)}`);
+        // a system column (ctid, xmin, tableoid): a copy has none, whatever its type's being verified says. Never verified
+        if (name === "VAR" && Number(v.varattno) < 0)
+            keys.add("field:VAR.system");
         const resolved = RESOLVED[name];
         if (resolved) {
             const oid = v[resolved[0]];
@@ -120,6 +129,111 @@ export function resolvedKeys(tree) {
     visit(tree);
     return keys;
 }
+/** Each function, aggregate and window function an analysed tree calls, with the type of each argument it is given, as
+ *  `<function oid>:<type oid>` (the catalog's OIDs): what a function marked stable only for its arguments' text reads
+ *  (src/duck/generated-plan.ts `sessionReads`). */
+export function argumentTypes(tree) {
+    const out = new Set();
+    const visit = (v) => {
+        if (Array.isArray(v)) {
+            for (const x of v)
+                visit(x);
+            return;
+        }
+        if (!isNode(v))
+            return;
+        const r = RESOLVED[v._node];
+        const oid = r === undefined || r[1] === "operator" ? undefined : v[r[0]];
+        if (typeof oid === "string" && Array.isArray(v.args)) {
+            for (const a of v.args) {
+                const e = isNode(a) && a._node === "TARGETENTRY" ? a.expr : a;
+                const t = isNode(e) ? exprType(e) : null;
+                if (t !== null && t.type !== 0)
+                    out.add(`${oid}:${String(t.type)}`);
+            }
+        }
+        for (const [k, x] of Object.entries(v))
+            if (k !== "_node")
+                visit(x);
+    };
+    visit(tree);
+    return [...out];
+}
+/** What a statement does with a value the copy holds as its text (src/duck/held-types.ts): what the text cannot answer
+ *  as the type would. Each is `operator:<oid>`, `function:<oid>` … as a key; the planner reads an operator's name. */
+export function heldUses(tree) {
+    const out = [];
+    const typeOf = (x) => {
+        if (!isNode(x))
+            return null;
+        if (x._node === "TARGETENTRY")
+            return typeOf(x.expr);
+        // a literal is the statement's, not the copy's (`to_tsvector('simple', x)` reads a regconfig)
+        if (x._node === "CONST")
+            return null;
+        // a binary-compatible cast is its value (a cidr read by inet's operators)
+        if (x._node === "RELABELTYPE")
+            return typeOf(x.arg);
+        return exprType(x)?.type ?? null;
+    };
+    const visit = (v) => {
+        if (Array.isArray(v)) {
+            for (const x of v)
+                visit(x);
+            return;
+        }
+        if (!isNode(v))
+            return;
+        const resolved = RESOLVED[v._node];
+        if (resolved !== undefined && Array.isArray(v.args)) {
+            const oid = v[resolved[0]];
+            const type = v.args.map(typeOf).find((t) => heldType(t) !== undefined);
+            if (type != null && typeof oid === "string")
+                out.push({ kind: resolved[1] === "operator" ? "operator" : "function", oid, type });
+        }
+        if (v._node === "COERCEVIAIO" || v._node === "ARRAYCOERCEEXPR") {
+            const from = typeOf(v.arg), to = exprType(v)?.type ?? null;
+            // to another type, or a value read as one (a text cast to macaddr), which only Postgres's input function reads
+            if (heldType(from) !== undefined && to !== 25 && to !== 1043 && from !== to)
+                out.push({ kind: "cast", oid: String(to), type: from });
+            else if (heldType(to) !== undefined && from !== to)
+                out.push({ kind: "cast", oid: String(to), type: to });
+        }
+        // a regclass read as its oid: the copy holds its name
+        if (v._node === "RELABELTYPE") {
+            const from = typeOf(v.arg), to = exprType(v)?.type ?? null;
+            if (heldType(from) !== undefined && to !== 869 && heldType(to) === undefined && from !== to)
+                out.push({ kind: "cast", oid: String(to), type: from });
+        }
+        // a "char" read as a text by min and max, under the database's collation: the copy's is a byte's order
+        if (v._node === "AGGREF" && Array.isArray(v.args) && v.args.some((a) => isNode(a) && isNode(a.expr) && a.expr.funcid === "946"))
+            out.push({ kind: "cast", oid: "25", type: 18 });
+        // a point's x, a vector's element: a subscript of a type that is not an array
+        if (v._node === "SUBSCRIPTINGREF") {
+            const of = typeOf(v.refexpr);
+            if (heldType(of) !== undefined && !isHeldArray(of))
+                out.push({ kind: "cast", oid: String(of), type: of });
+        }
+        for (const [k, x] of Object.entries(v))
+            if (k !== "_node" && k !== "joinaliasvars")
+                visit(x);
+    };
+    visit(tree);
+    return out;
+}
+/** whether a statement tests a value of one of `types` with IS [NOT] NULL */
+export function nullTested(tree, types) {
+    const visit = (v) => {
+        if (Array.isArray(v))
+            return v.some(visit);
+        if (!isNode(v))
+            return false;
+        if (v._node === "NULLTEST" && isNode(v.arg) && types.has(exprType(v.arg)?.type ?? 0))
+            return true;
+        return Object.entries(v).some(([k, x]) => k !== "_node" && k !== "joinaliasvars" && visit(x));
+    };
+    return visit(tree);
+}
 // --------------------------------------------------------------------------------------------- what a key names
 /** each kind field's values in SQL's words (the enums of Postgres 18's primnodes.h / parsenodes.h, as outfuncs.c writes them) */
 const VALUES = {
@@ -148,6 +262,7 @@ const VALUES = {
     "COMMONTABLEEXPR.ctematerialized": { 0: "WITH", 1: "WITH … AS MATERIALIZED", 2: "WITH … AS NOT MATERIALIZED" },
 };
 const FLAGS = {
+    "VAR.system": "a system column (ctid, xmin, tableoid …), which a synced table does not have",
     "AGGREF.aggdistinct": "an aggregate of DISTINCT values", "AGGREF.aggorder": "an aggregate's own ORDER BY", "AGGREF.aggfilter": "an aggregate's FILTER (WHERE …)",
     "AGGREF.aggdirectargs": "an ordered-set aggregate's direct argument", "WINDOWFUNC.aggfilter": "a window aggregate's FILTER (WHERE …)",
     "QUERY.distinctClause": "SELECT DISTINCT", "QUERY.groupingSets": "GROUPING SETS / ROLLUP / CUBE", "QUERY.havingQual": "HAVING",
@@ -207,7 +322,7 @@ export async function describeKeys(rows, keys) {
     const proc = await read(procs, (l) => `SELECT oid::text, oid::regprocedure::text FROM pg_proc WHERE oid IN (${l})`);
     const op = await read(ops, (l) => `SELECT oid::text, oid::regoperator::text FROM pg_operator WHERE oid IN (${l})`);
     const type = await read(types, (l) => `SELECT oid::text, format_type(oid, NULL) FROM pg_type WHERE oid IN (${l})`);
-    const typeName = (t) => t === "rowtype" ? "a table's row type" : t.startsWith("ext:") ? `${t.slice(4)} (an extension's)` : type.get(t) ?? `type OID ${t}`;
+    const typeName = (t) => t === "rowtype" ? "a table's row type" : t === HELD_KEY ? "a type the source made" : t === `${HELD_KEY}[]` ? "an array of a type the source made" : t.startsWith("ext:") ? `${t.slice(4)} (an extension's)` : type.get(t) ?? `type OID ${t}`;
     const extName = (v) => v.startsWith("ext:") ? `${v.slice(4)} (an extension's)` : `OID ${v}`;
     for (const k of list) {
         const kind = k.slice(0, k.indexOf(":")), value = k.slice(k.indexOf(":") + 1);