orangeslice 1.7.2 → 1.7.3

package/docs/b2b-docs/B2B_TABLE_INDICES.ts

@@ -0,0 +1,496 @@
+/**
+ * B2B Database Table Index Mapping
+ * Generated from docs/B2B_SCHEMA.md
+ *
+ * Maps each table to its available indices for query optimization.
+ */
+
+export interface TableIndex {
+   name: string;
+   type: "UNIQUE btree" | "btree" | "GIN" | "GiST";
+   columns: string[];
+   notes?: string;
+   isPartial?: boolean;
+}
+
+export interface TableIndexMapping {
+   table: string;
+   recordCount?: string;
+   indices: TableIndex[];
+}
+
+export const B2B_TABLE_INDICES: TableIndexMapping[] = [
+   // ===================
+   // CORE TABLES
+   // ===================
+   {
+      table: "linkedin_company",
+      recordCount: "~millions",
+      indices: [
+         { name: "linkedin_company_pkey", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         {
+            name: "linkedin_company_universal_name_ix",
+            type: "btree",
+            columns: ["universal_name"],
+            notes: "Fast lookup"
+         },
+         { name: "ix_linkedin_company_domain", type: "btree", columns: ["domain"], notes: "Fast lookup" },
+         { name: "ix_linkedin_company_company_id", type: "btree", columns: ["company_id"], notes: "FK lookup" },
+         { name: "ix_linkedin_company_ticker", type: "btree", columns: ["ticker"], notes: "Stock lookup" },
+         {
+            name: "ix_linkedin_company_tsv",
+            type: "GIN",
+            columns: ["company_name", "universal_name"],
+            notes: "Full-text search"
+         },
+         {
+            name: "ix_linkedin_company_linkedin_org_id",
+            type: "btree",
+            columns: ["linkedin_org_id"],
+            notes: "Partial (NOT NULL)",
+            isPartial: true
+         }
+      ]
+   },
+   {
+      table: "linkedin_profile",
+      recordCount: "~1.15 billion",
+      indices: [
+         { name: "linkedin_profile_pkey", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         {
+            name: "ix_linkedin_profile_linkedin_user_id",
+            type: "btree",
+            columns: ["linkedin_user_id"],
+            notes: "Partial (NOT NULL)",
+            isPartial: true
+         },
+         { name: "linkedin_profile_updated_at_idx", type: "btree", columns: ["updated_at"], notes: "Recency queries" },
+         { name: "ix_linkedin_profile_org_tsv", type: "GIN", columns: ["org"], notes: "Full-text on org" },
+         {
+            name: "ix_linkedin_profile_random",
+            type: "btree",
+            columns: ["immutable_random(id)"],
+            notes: "Random sampling"
+         }
+      ]
+   },
+   {
+      table: "linkedin_profile_position3",
+      recordCount: "~2.6 billion",
+      indices: [
+         { name: "linkedin_profile_position3_pkey", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         {
+            name: "ix_linkedin_profile_position3_linkedin_profile_id",
+            type: "btree",
+            columns: ["linkedin_profile_id"],
+            notes: "Fast lookup by profile"
+         },
+         {
+            name: "ix_linkedin_profile_position3_linkedin_company_id",
+            type: "btree",
+            columns: ["linkedin_company_id"],
+            notes: "Fast lookup by company"
+         },
+         {
+            name: "ix_linkedin_profile_position3_key64",
+            type: "UNIQUE btree",
+            columns: ["key64(...)"],
+            notes: "Deduplication"
+         }
+      ]
+   },
+   {
+      table: "linkedin_job",
+      recordCount: "~1.48 billion",
+      indices: [
+         { name: "linkedin_job_pkey", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         { name: "ix_linkedin_job_job_id_uniq", type: "UNIQUE btree", columns: ["job_id"], notes: "LinkedIn job ID" },
+         {
+            name: "ix_linkedin_job_company_id_ix",
+            type: "btree",
+            columns: ["linkedin_company_id"],
+            notes: "Fast lookup by company"
+         },
+         { name: "ix_linkedin_job_title_id_ix", type: "btree", columns: ["title_id"], notes: "Title lookup" },
+         { name: "ix_linkedin_job_updated_at", type: "btree", columns: ["updated_at"], notes: "Recency queries" },
+         {
+            name: "ix_linkedin_job_recruiter_profile_id_ix",
+            type: "btree",
+            columns: ["recruiter_profile_id"],
+            notes: "Recruiter lookup"
+         }
+      ]
+   },
+   {
+      table: "linkedin_profile_education2",
+      recordCount: "~965 million",
+      indices: [
+         { name: "linkedin_profile_education2_pkey", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         {
+            name: "ix_linkedin_profile_education2_linkedin_profile_id",
+            type: "btree",
+            columns: ["linkedin_profile_id"],
+            notes: "Fast lookup by profile"
+         },
+         {
+            name: "ix_linkedin_profile_education2_linkedin_school_id",
+            type: "btree",
+            columns: ["linkedin_school_id"],
+            notes: "Partial (NOT NULL)",
+            isPartial: true
+         },
+         {
+            name: "linkedin_profile_education2_md5_bigint_key",
+            type: "UNIQUE btree",
+            columns: ["md5_bigint"],
+            notes: "Deduplication"
+         }
+      ]
+   },
+   {
+      table: "linkedin_crunchbase_funding",
+      indices: [
+         { name: "linkedin_crunchbase_funding_pkey", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         {
+            name: "ix_linkedin_crunchbase_funding_linkedin_company_id",
+            type: "btree",
+            columns: ["linkedin_company_id"],
+            notes: "Fast lookup by company"
+         }
+      ]
+   },
+   {
+      table: "linkedin_company_slug",
+      indices: [
+         { name: "linkedin_company_slug_pk", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         {
+            name: "linkedin_company_slug_slug_key64_uniq",
+            type: "UNIQUE btree",
+            columns: ["slug_key64"],
+            notes: "Fast lookup via key64()"
+         },
+         {
+            name: "linkedin_company_slug_linkedin_company_id_ix",
+            type: "btree",
+            columns: ["linkedin_company_id"],
+            notes: "Company lookup"
+         }
+      ]
+   },
+   {
+      table: "linkedin_profile_slug",
+      recordCount: "~1.14 billion",
+      indices: [
+         { name: "linkedin_profile_slug_pk", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         {
+            name: "linkedin_profile_slug_slug_key64_idx",
+            type: "btree",
+            columns: ["slug_key64"],
+            notes: "Fast lookup via key64()"
+         },
+         {
+            name: "linkedin_profile_slug_linkedin_profile_id_ix",
+            type: "btree",
+            columns: ["linkedin_profile_id"],
+            notes: "Profile lookup"
+         },
+         {
+            name: "linkedin_profile_slug_linkedin_user_id_ix",
+            type: "btree",
+            columns: ["linkedin_user_id"],
+            notes: "User ID lookup"
+         }
+      ]
+   },
+   {
+      table: "linkedin_industry",
+      indices: [
+         { name: "linkedin_industry_pkey", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         { name: "linkedin_industry_alt_names_uniq", type: "GiST", columns: ["alt_names"], notes: "Name matching" }
+      ]
+   },
+   {
+      table: "person",
+      recordCount: "~1.32 billion",
+      indices: [
+         { name: "person_pkey", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         {
+            name: "ix_person_linkedin_profile_id",
+            type: "btree",
+            columns: ["linkedin_profile_id"],
+            notes: "Fast lookup by profile"
+         },
+         {
+            name: "ix_person_first_last_name_tsv",
+            type: "GIN",
+            columns: ["first_name", "last_name", "formatted_name"],
+            notes: "Full-text name search"
+         }
+      ]
+   },
+   {
+      table: "company",
+      indices: [
+         { name: "company_pkey", type: "UNIQUE btree", columns: ["id"], notes: "Primary key" },
+         { name: "company_slug_key", type: "UNIQUE btree", columns: ["slug"], notes: "Slug lookup" },
+         { name: "ix_company_linkedin_id", type: "UNIQUE btree", columns: ["linkedin_id"], notes: "LinkedIn lookup" },
+         { name: "ix_company_employee_count", type: "btree", columns: ["employee_count"], notes: "Size queries" }
+      ]
+   },
+   {
+      table: "company_type",
+      indices: []
+   },
+   {
+      table: "linkedin_article",
+      indices: [{ name: "linkedin_article_linkedin_profile_id_ix", type: "btree", columns: ["linkedin_profile_id"] }]
+   },
+   {
+      table: "linkedin_patent",
+      indices: []
+   },
+   {
+      table: "linkedin_project",
+      indices: [{ name: "ix_linkedin_project_title", type: "btree", columns: ["title"] }]
+   },
+   {
+      table: "linkedin_publication2",
+      indices: []
+   },
+   {
+      table: "linkedin_proserve_service",
+      indices: []
+   },
+   {
+      table: "linkedin_industry_naics_code_mapping",
+      indices: []
+   },
+
+   // ===================
+   // REFERENCE TABLES
+   // ===================
+   {
+      table: "linkedin_school",
+      indices: []
+   },
+   {
+      table: "linkedin_language",
+      indices: []
+   },
+   {
+      table: "language_proficiency",
+      indices: []
+   },
+   {
+      table: "linkedin_specialty",
+      indices: []
+   },
+   {
+      table: "linkedin_work_modality",
+      indices: []
+   },
+   {
+      table: "country",
+      indices: []
+   },
+   {
+      table: "locality",
+      indices: []
+   },
+   {
+      table: "naics_code",
+      indices: []
+   },
+
+   // ===================
+   // JOB TAXONOMY TABLES
+   // ===================
+   {
+      table: "job_title",
+      indices: []
+   },
+   {
+      table: "job_function",
+      indices: []
+   },
+   {
+      table: "job_seniority",
+      indices: []
+   },
+   {
+      table: "job_employment_type",
+      indices: []
+   },
+   {
+      table: "job_academic_qualification",
+      indices: []
+   },
+
+   // ===================
+   // PROFILE DETAIL TABLES
+   // ===================
+   {
+      table: "linkedin_profile_certification",
+      indices: [
+         {
+            name: "ix_linkedin_profile_certification_linkedin_profile_id",
+            type: "btree",
+            columns: ["linkedin_profile_id"]
+         }
+      ]
+   },
+   {
+      table: "linkedin_profile_award",
+      indices: [
+         { name: "ix_linkedin_profile_award_linkedin_profile_id", type: "btree", columns: ["linkedin_profile_id"] }
+      ]
+   },
+   {
+      table: "linkedin_profile_project",
+      indices: [
+         { name: "ix_linkedin_profile_project_linkedin_profile_id", type: "btree", columns: ["linkedin_profile_id"] },
+         { name: "ix_linkedin_profile_project_linkedin_project_id", type: "btree", columns: ["linkedin_project_id"] }
+      ]
+   },
+   {
+      table: "linkedin_profile_volunteer_experience",
+      indices: [
+         {
+            name: "ix_linkedin_profile_volunteer_experience_linkedin_profile_id",
+            type: "btree",
+            columns: ["linkedin_profile_id"]
+         }
+      ]
+   },
+   {
+      table: "linkedin_profile_recommendation2",
+      indices: [
+         {
+            name: "ix_linkedin_profile_recommendation2_linkedin_profile_id",
+            type: "btree",
+            columns: ["linkedin_profile_id"]
+         },
+         {
+            name: "ix_linkedin_profile_recommendation2_recommender_linkedin_profile_id",
+            type: "btree",
+            columns: ["recommender_linkedin_profile_id"]
+         }
+      ]
+   },
+   {
+      table: "linkedin_profile_test_scores",
+      indices: []
+   },
+   {
+      table: "linkedin_profile_language_proficiency",
+      indices: []
+   },
+   {
+      table: "linkedin_profile_course",
+      indices: [
+         { name: "ix_linkedin_profile_course_linkedin_profile_id", type: "btree", columns: ["linkedin_profile_id"] }
+      ]
+   },
+   {
+      table: "linkedin_profile_patent",
+      indices: [
+         { name: "ix_linkedin_profile_patent_linkedin_profile_id", type: "btree", columns: ["linkedin_profile_id"] }
+      ]
+   },
+   {
+      table: "linkedin_profile_publication2",
+      indices: [
+         {
+            name: "ix_linkedin_profile_publication2_linkedin_profile_id",
+            type: "btree",
+            columns: ["linkedin_profile_id"]
+         }
+      ]
+   },
+   {
+      table: "linkedin_profile_url_resource",
+      indices: []
+   },
+   {
+      table: "linkedin_profile_url_resource_name",
+      indices: []
+   },
+   {
+      table: "linkedin_profile_priority_queue",
+      indices: []
+   },
+
+   // ===================
+   // COMPANY DETAIL TABLES
+   // ===================
+   {
+      table: "linkedin_company_address2",
+      indices: [
+         { name: "linkedin_company_address2_linkedin_company_idx", type: "btree", columns: ["linkedin_company_id"] }
+      ]
+   },
+   {
+      table: "linkedin_company_post",
+      indices: [
+         {
+            name: "ix_linkedin_company_post_linkedin_company_id_id",
+            type: "btree",
+            columns: ["linkedin_company_id", "id DESC"],
+            notes: "Composite index"
+         }
+      ]
+   },
+
+   // ===================
+   // DENORMALIZED VIEWS
+   // ===================
+   {
+      table: "lkd_company",
+      indices: []
+   },
+   {
+      table: "lkd_profile",
+      indices: []
+   }
+];
+
+/**
+ * Helper to get indices for a specific table
+ */
+export function getTableIndices(tableName: string): TableIndex[] {
+   const mapping = B2B_TABLE_INDICES.find((t) => t.table === tableName);
+   return mapping?.indices ?? [];
+}
+
+/**
+ * Helper to find tables with indices on a specific column
+ */
+export function findIndicesByColumn(columnName: string): { table: string; index: TableIndex }[] {
+   const results: { table: string; index: TableIndex }[] = [];
+   for (const mapping of B2B_TABLE_INDICES) {
+      for (const index of mapping.indices) {
+         if (index.columns.some((col) => col.includes(columnName))) {
+            results.push({ table: mapping.table, index });
+         }
+      }
+   }
+   return results;
+}
+
+/**
+ * Get all tables with their index count
+ */
+export function getTableIndexSummary(): { table: string; indexCount: number; recordCount?: string }[] {
+   return B2B_TABLE_INDICES.map((t) => ({
+      table: t.table,
+      indexCount: t.indices.length,
+      recordCount: t.recordCount
+   }));
+}
+
+/**
+ * Quick lookup map for table -> indexed columns
+ */
+export const TABLE_INDEXED_COLUMNS: Record<string, string[]> = Object.fromEntries(
+   B2B_TABLE_INDICES.map((t) => [t.table, [...new Set(t.indices.flatMap((i) => i.columns))]])
+);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "orangeslice",
-  "version": "1.7.2",
+  "version": "1.7.3",
   "description": "AI agent toolkit: B2B database, SERP, web scraping, browser automation, structured AI output, Apify actors, geocoding",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",