@opendatalabs/personal-server-ts-core 1.34.2 → 1.35.0

package/dist/legacy-projection/bindings.js

@@ -825,6 +825,15 @@ const instagramPosts = {
         post_likes: ["post_id", "user_id", "username"],
     },
     primaryKey: { posts: ["id"], post_likes: ["post_id", "liker_ordinal"] },
+    // Each post is built from its own post_likes.
+    window: {
+        parentStream: "posts",
+        childStream: "post_likes",
+        childKey: "post_id",
+        arrayKey: "posts",
+        sumKeys: [],
+        parentOrder: primaryKeyOrder("id"),
+    },
     lossy: [
         "nullable media_url, caption, and like_count use the retained collector's empty-string and zero sentinels; every source post is retained",
         "media_url is only available for images and the first image in a carousel; video posts and later carousel images are absent from the artifact's parser output",
@@ -1170,6 +1179,7 @@ const githubHistory = {
         ],
     },
     primaryKey: { user: ["id"], issues: ["id"], pull_requests: ["id"] },
+    memoryBound: "Issues and pull requests are two independent lists, each one array, so there is no parent and child to split.",
     lossy: [
         "legacy IDs used gh-issue-/gh-pr- prefixes; projection uses the signed PDPP record ID directly",
         "body is truncated to the legacy connector's 500-character limit",
@@ -1322,6 +1332,7 @@ const githubProfile = {
         pinned_repositories: ["id"],
         organizations: ["id"],
     },
+    memoryBound: "The output is one profile object; user_stats gains a row per day and is read in full to pick the latest, and the window has no parent and child to split.",
     lossy: [
         "profileUrl is derived from the authenticated user's login using GitHub's canonical profile URL pattern",
         "organization label and URL are omitted because the organizations stream carries neither field; description is not substituted for label",
@@ -1618,6 +1629,16 @@ const chatgptConversations = {
         ],
     },
     primaryKey: { conversations: ["id"], messages: ["id"] },
+    // Each conversation is built from its own messages; the list is ordered by
+    // the conversation's own update_time and id.
+    window: {
+        parentStream: "conversations",
+        childStream: "messages",
+        childKey: "conversation_id",
+        arrayKey: "conversations",
+        sumKeys: ["total"],
+        parentOrder: newestUpdateFirst,
+    },
     lossy: [
         "Conversation create/update times pass through as null when the source lacks them, matching legacy's listed?.create_time ?? fetched.create_time ?? null fallback.",
         "fetched_at is the projection time (options.now, else the wall clock), matching legacy's per-run toConversationRecord timestamp.",
@@ -1851,6 +1872,15 @@ function newestUpdateFirst(a, b) {
         return byTime;
     return String(a.id) < String(b.id) ? -1 : String(a.id) > String(b.id) ? 1 : 0;
 }
+/**
+ * The output order of a binding that lists its parents in primary-key order,
+ * which is what `orderByPrimaryKey` gives it: the same key `inPrimaryKeyOrder`
+ * sorts by.
+ */
+function primaryKeyOrder(field) {
+    const key = (row) => JSON.stringify([row[field] ?? null]);
+    return (a, b) => (key(a) < key(b) ? -1 : key(a) > key(b) ? 1 : 0);
+}
 // This is the merged 0.1.3 source declaration. Admission remains a separate
 // decision; this local manifest hash is not the signed OCI artifact digest.
 export const ANTHROPIC_CONNECTOR_ID = "https://registry.pdpp.dev/connectors/anthropic";
@@ -1904,22 +1934,58 @@ function claudeMetadata(scope, records) {
         organizationId: profile.organization_id,
     };
 }
-function claudeConversationsBinding() {
+function claudeLiveMetadata(scope, records) {
+    const plans = byStream(records, "account_plan");
+    const plan = plans[0];
+    if (plans.length !== 1 ||
+        !plan ||
+        typeof plan.id !== "string" ||
+        !plan.id.trim() ||
+        typeof plan.account_id !== "string" ||
+        !plan.account_id.trim() ||
+        !(typeof plan.plan_display_name === "string" ||
+            plan.plan_display_name === null)) {
+        return {
+            ok: false,
+            error: {
+                kind: "incomplete_scope",
+                scope,
+                reason: "A single verified organization account_plan is required",
+            },
+        };
+    }
+    return {
+        ok: true,
+        organizationId: plan.account_id,
+        profile: { name: null, plan: plan.plan_display_name },
+    };
+}
+function claudeConversationsBinding(live = false) {
     const scope = "claude.conversations";
     return {
         scope,
-        pdppSource: ANTHROPIC_CONNECTOR_ID,
-        pdppStreams: ["account_profile", "conversations", "messages"],
+        pdppSource: live
+            ? "https://registry.pdpp.dev/connectors/anthropic-live"
+            : ANTHROPIC_CONNECTOR_ID,
+        pdppStreams: [
+            live ? "account_plan" : "account_profile",
+            "conversations",
+            "messages",
+        ],
         legacySchemaPath: "claude.conversations.json",
         fieldsRead: {
-            account_profile: [
-                "id",
-                "organization_id",
-                "full_name",
-                "plan",
-                "name_source",
-                "metadata_status",
-            ],
+            ...(live
+                ? { account_plan: ["id", "account_id", "plan_display_name"] }
+                : {
+                    account_profile: [
+                        "id",
+                        "organization_id",
+                        "full_name",
+                        "plan",
+                        "name_source",
+                        "metadata_status",
+                    ],
+                }),
             conversations: [
                 "id",
                 "title",
@@ -1941,23 +2007,47 @@ function claudeConversationsBinding() {
             ],
         },
         primaryKey: {
-            account_profile: ["id"],
+            [live ? "account_plan" : "account_profile"]: ["id"],
             conversations: ["id"],
             messages: ["id"],
         },
+        // Each conversation is built from its own messages. The list is in
+        // primary-key order, the order `inPrimaryKeyOrder` gives it.
+        window: {
+            parentStream: "conversations",
+            childStream: "messages",
+            childKey: "conversation_id",
+            arrayKey: "conversations",
+            sumKeys: ["total", "messageTotal"],
+            parentOrder: primaryKeyOrder("id"),
+        },
         lossy: [
             "href is reconstructed from the retained Claude route convention; rawContent and unavailable fetchError are omitted.",
         ],
-        provenance: anthropicProjectionProvenance,
+        // The live declaration's conversation blob_ref is the raw envelope; its
+        // typed record and messages are complete without it.
+        ...(live ? { ancillaryBlobStreams: ["conversations"] } : {}),
+        provenance: live
+            ? [
+                {
+                    kind: "oci",
+                    ref: "ghcr.io/pdp-connect/connector/anthropic-live:0.3.6",
+                    digest: "sha256:817716e8c732c9c991be507049147a0cc887b154363428bc4509cc91ed5d17a6",
+                    path: "declarations/anthropic-live-0.3.6.collection-profile.json",
+                },
+            ]
+            : anthropicProjectionProvenance,
         project(records, options) {
             const guard = requireStreams(scope, options, [
-                "account_profile",
+                live ? "account_plan" : "account_profile",
                 "conversations",
                 "messages",
             ]);
             if (guard)
                 return guard;
-            const metadata = claudeMetadata(scope, records);
+            const metadata = live
+                ? claudeLiveMetadata(scope, records)
+                : claudeMetadata(scope, records);
             if (!metadata.ok)
                 return metadata;
             const conversations = byStream(records, "conversations");
@@ -1987,6 +2077,7 @@ function claudeConversationsBinding() {
                     },
                 };
             const projected = [];
+            let unknownMessageCounts = 0;
             for (const conversation of conversations) {
                 if (typeof conversation.id !== "string" ||
                     !conversation.id.trim() ||
@@ -2027,10 +2118,16 @@ function claudeConversationsBinding() {
                         attachments: message.attachments ?? [],
                     };
                 });
-                const messageCount = typeof conversation.message_count === "number"
-                    ? conversation.message_count
-                    : messages.length;
-                if (messageCount !== messages.length)
+                // Only the source's own count proves the fetched messages are the
+                // whole conversation. Without one (anthropic-live writes null when
+                // the run did not read the detail) the conversation is reported
+                // partial, never certified complete. messageCount stays the number
+                // of messages collected, as the legacy schema defines it.
+                const sourceCount = conversation.message_count ?? null;
+                const messageCount = messages.length;
+                if (sourceCount === null)
+                    unknownMessageCounts += 1;
+                else if (sourceCount !== messageCount)
                     return {
                         ok: false,
                         error: {
@@ -2051,20 +2148,23 @@ function claudeConversationsBinding() {
                     messages,
                 });
             }
+            const diagnostics = partialRows(scope, "conversations", "message_count_unknown", unknownMessageCounts);
             return {
                 ok: true,
+                ...(diagnostics.length > 0 ? { diagnostics } : {}),
                 payload: {
                     profile: metadata.profile,
                     organizationId: metadata.organizationId,
                     conversations: projected,
                     total: projected.length,
                     messageTotal: projected.reduce((total, conversation) => total + conversation.messageCount, 0),
-                    source: "official-export",
+                    source: live ? "live-session" : "official-export",
                 },
             };
         },
     };
 }
+export const ANTHROPIC_LIVE_BINDINGS = new Map([["claude.conversations", claudeConversationsBinding(true)]]);
 const claudeConversations = claudeConversationsBinding();
 const claudeProjects = {
     scope: "claude.projects",
@@ -2108,6 +2208,15 @@ const claudeProjects = {
         projects: ["id"],
         project_documents: ["id"],
     },
+    // Each project is built from its own project_documents.
+    window: {
+        parentStream: "projects",
+        childStream: "project_documents",
+        childKey: "project_id",
+        arrayKey: "projects",
+        sumKeys: ["total"],
+        parentOrder: primaryKeyOrder("id"),
+    },
     lossy: [
         "detail carries only declared project and raw document fields. Other raw-only keys and documents dropped by source parsing cannot be recovered; href and label follow the legacy connector's route and label rules.",
     ],
@@ -2750,6 +2859,15 @@ const spotifyPlaylists = {
         playlists: ["id"],
         playlist_items: ["id"],
     },
+    // Each playlist is built from its own playlist_items.
+    window: {
+        parentStream: "playlists",
+        childStream: "playlist_items",
+        childKey: "playlist_id",
+        arrayKey: "playlists",
+        sumKeys: ["total"],
+        parentOrder: primaryKeyOrder("id"),
+    },
     lossy: [
         "tracks[].uri: raw playlist item URIs are retained when available; older pinned records fall back to a canonical Spotify track ID, and a null ID uses the retained collector's empty-string sentinel.",
         "tracks_total counts joined fetched tracks, as in the retained collector; catalog track_count is not a fetched-row count.",
@@ -2970,9 +3088,10 @@ const ouraActivity = {
             "contributors",
         ],
     },
+    optionalFieldsRead: { activity: ["timestamp"] },
     primaryKey: { activity: ["id"] },
     lossy: [
-        "timestamp: no source field; omitted (legacy schema makes it optional)",
+        "timestamp: preserved when supplied; absent or null source timestamps are omitted",
         "target_calories: has no legacy field and is not projected",
         "Legacy integer fields: fractional source numbers are rounded with Math.round at projection time",
     ],
@@ -2996,6 +3115,9 @@ const ouraActivity = {
                 days: rows.map((r) => ({
                     id: r.id,
                     day: r.day,
+                    ...(typeof r.timestamp === "string"
+                        ? { timestamp: r.timestamp }
+                        : {}),
                     score: ouraInteger(r.score),
                     ...ifDefinedOuraInteger(r, "active_calories", "activeCalories"),
                     ...ifDefinedOuraInteger(r, "total_calories", "totalCalories"),
@@ -3028,9 +3150,10 @@ const ouraReadiness = {
             "contributors",
         ],
     },
+    optionalFieldsRead: { readiness: ["timestamp"] },
     primaryKey: { readiness: ["id"] },
     lossy: [
-        "timestamp: no source field; omitted (legacy schema makes it optional)",
+        "timestamp: preserved when supplied; absent or null source timestamps are omitted",
         "Legacy integer fields: fractional source numbers are rounded with Math.round at projection time",
     ],
     provenance: [
@@ -3053,6 +3176,9 @@ const ouraReadiness = {
                 days: rows.map((r) => ({
                     id: r.id,
                     day: r.day,
+                    ...(typeof r.timestamp === "string"
+                        ? { timestamp: r.timestamp }
+                        : {}),
                     score: ouraInteger(r.score),
                     ...ifDefined(r, "temperature_deviation", "temperatureDeviation"),
                     ...ifDefined(r, "temperature_trend_deviation", "temperatureTrendDeviation"),
@@ -3205,6 +3331,15 @@ const amazonOrders = {
         order_items: ["order_id", "name", "url", "unit_price"],
     },
     primaryKey: { orders: ["id"], order_items: ["id"] },
+    // Each order is built from its own order_items.
+    window: {
+        parentStream: "orders",
+        childStream: "order_items",
+        childKey: "order_id",
+        arrayKey: "orders",
+        sumKeys: ["total"],
+        parentOrder: primaryKeyOrder("id"),
+    },
     lossy: [
         "items[].quantity: no legacy field (legacy schema does not carry quantity at all; PDPP-side quantity is simply not projected)",
         "items[].price (N-S2): legacy's own connector ALWAYS emits the literal constant '' for this field (amazon-playwright.js:222, `price: ''` — not a fallback for a missing value, the connector never scrapes a real price here at all). This projection instead emits PDPP's real `unit_price` when present, so an app reading this field gets a populated value where legacy always gave it an empty string — a value change, not a gap-fill, and the comment at the call site below previously implied the two matched when they do not.",
@@ -3399,6 +3534,7 @@ const hebNutrition = {
         order_items: ["product_id", "name", "product_url"],
     },
     primaryKey: { nutrition: ["id"], orders: ["id"], order_items: ["id"] },
+    memoryBound: "The output is one map keyed by product with whole-scope coverage totals and checks, not one row per parent.",
     lossy: [
         "The legacy map key is the declared H-E-B product_id; no global UPC identity is inferred.",
         "fetched_at is omitted.",
@@ -3592,6 +3728,7 @@ const hebOrders = {
         ],
     },
     primaryKey: { orders: ["id"], order_items: ["id"] },
+    memoryBound: "Rejects the whole scope when every order is dropped, a decision a window of orders cannot make.",
     lossy: [
         "orderUrl and address: the signed profile has no equivalent fields; fulfillment_location is not treated as an address.",
         "status_code, store_name, timeslot_start/end, unfulfilled_count, fulfillment_method, and fetched_at are not represented by the legacy schema.",
@@ -3803,6 +3940,7 @@ const wholefoodsOrders = {
         ],
     },
     primaryKey: { orders: ["id"], order_items: ["id"] },
+    memoryBound: "Rejects the whole scope when every order is dropped, a decision a window of orders cannot make.",
     lossy: [
         "The retained collector emitted only product-linked named lines; a source line without either field is dropped and counted, and its order is delivered without it.",
         "An order with no usable line and an item_count that is not a known zero is left out and counted; an order whose line count differs from item_count, or whose item_count is unknown, is delivered and reported partial. A missing or duplicate order id still fails the scope.",
@@ -3955,6 +4093,7 @@ const wholefoodsNutrition = {
         order_items: ["product_id", "name", "product_url", "image_url"],
     },
     primaryKey: { nutrition: ["product_id"], order_items: ["id"] },
+    memoryBound: "The output is one map keyed by product with whole-scope coverage totals and checks, not one row per parent.",
     lossy: [
         "ingredients and allergens are not collected by the current nutrition stream.",
         "servingsPerContainer is converted from the declared number to the legacy string field.",
@@ -4090,9 +4229,6 @@ const wholefoodsNutrition = {
         };
     },
 };
-// amazon.profile is a documented GAP: no `profile` stream exists on the
-// amazon collection-profile.json at all (verified: exactly 2 streams,
-// orders and order_items) — legacy's `name`/`isPrime` have no PDPP source.
 // iCloud Notes is bound to the signed 0.1.0 collection profile. Desktop
 // admission and ingest projection are separate gates.
 const icloudNoteFolders = {
@@ -4155,6 +4291,7 @@ const icloudNotes = {
         profile: ["id", "user_name"],
     },
     primaryKey: { notes: ["id"], folders: ["id"], profile: ["id"] },
+    memoryBound: "One flat list of notes; folders is a name lookup, so there are no per-parent children to split.",
     lossy: [
         "userName is null when the profile stream has no readable user_name (legacy's fullName default); it is never inferred from an Apple ID or the account id.",
         "folder uses the legacy display-name shape; unresolved folder references fall back to folder_id.",
@@ -4372,7 +4509,4 @@ export const LEGACY_SCOPE_BINDINGS = new Map(Object.entries({
     "strava.activities": stravaActivities,
     "strava.coverage_diagnostics": stravaCoverageDiagnostics,
 }));
-/** Known legacy contracts checked against signed declarations but not
- * projectable without fabricating fields or changing the retained contract. */
-export const LEGACY_SCOPE_GAPS = new Map([]);
 //# sourceMappingURL=bindings.js.map