include gri_meta_json in convenience tables

Author: mlin

bindings/python/genomicsqlite.py

@@ -123,6 +123,7 @@ class ReferenceSequence(NamedTuple):
     length: int
     assembly: Optional[str]
     refget_id: Optional[str]
+    meta: Dict[str, Any]
 
 
 def get_reference_sequences_by_rid(
@@ -132,7 +133,7 @@ def get_reference_sequences_by_rid(
     if schema:
         table = f"{schema}.{table}"
     sql = (
-        "SELECT _gri_rid, gri_refseq_name, gri_refseq_length, gri_assembly, gri_refget_id FROM "
+        "SELECT _gri_rid, gri_refseq_name, gri_refseq_length, gri_assembly, gri_refget_id, gri_refseq_meta_json FROM "
         + table
     )
     params = []
@@ -145,7 +146,12 @@ def get_reference_sequences_by_rid(
             isinstance(row[0], int) and row[0] not in ans
         ), "genomicsqlite: invalid or duplicate reference sequence rid"
         ans[row[0]] = ReferenceSequence(
-            rid=row[0], name=row[1], length=row[2], assembly=row[3], refget_id=row[4]
+            rid=row[0],
+            name=row[1],
+            length=row[2],
+            assembly=row[3],
+            refget_id=row[4],
+            meta=(json.loads(row[5]) if row[5] else {}),
         )
     return ans
 

include/genomicsqlite.h

@@ -124,10 +124,10 @@ std::string PutGenomicReferenceSequenceSQL(const std::string &name, sqlite3_int6
 
 /* Lookup helpers for stored reference sequence metadata (assumes it's finalized) */
 struct gri_refseq_t {
-    unsigned long long rid, length;
-    std::string name, assembly, refget_id;
+    long long rid, length;
+    std::string name, assembly, refget_id, meta_json = "{}";
 };
-std::map<unsigned long long, gri_refseq_t>
+std::map<long long, gri_refseq_t>
 GetGenomicReferenceSequencesByRid(sqlite3 *dbconn, const std::string &assembly = "",
                                   const std::string &attached_schema = "");
 std::map<std::string, gri_refseq_t>

src/genomicsqlite.cc

@@ -384,7 +384,7 @@ string GenomicSQLiteVacuumIntoSQL(const string &destfile, const string &config_j
         throw std::runtime_error("error processing config JSON $.inner_page_size");
     int inner_page_size = extract->getColumn(0).getInt();
 
-    string desturi = GenomicSQLiteURI(destfile, config_json);
+    string desturi = GenomicSQLiteURI(destfile, config_json) + "&outer_unsafe=true";
 
     ostringstream ans;
     ans << "PRAGMA page_size = " << inner_page_size << ";\nPRAGMA auto_vacuum = FULL"
@@ -990,13 +990,13 @@ static void sqlfn_put_genomic_reference_assembly_sql(sqlite3_context *ctx, int a
     SQL_WRAPPER(PutGenomicReferenceAssemblySQL(assembly, schema))
 }
 
-map<unsigned long long, gri_refseq_t>
+map<long long, gri_refseq_t>
 GetGenomicReferenceSequencesByRid(sqlite3 *dbconn, const string &assembly, const string &schema) {
-    map<unsigned long long, gri_refseq_t> ans;
+    map<long long, gri_refseq_t> ans;
     string schema_prefix = schema.empty() ? "" : (schema + ".");
 
     string query =
-        "SELECT _gri_rid, gri_refseq_name, gri_refseq_length, gri_assembly, gri_refget_id FROM " +
+        "SELECT _gri_rid, gri_refseq_name, gri_refseq_length, gri_assembly, gri_refget_id, gri_refseq_meta_json FROM " +
         schema_prefix + "__gri_refseq";
     if (!assembly.empty()) {
         query += " WHERE gri_assembly = ?";
@@ -1026,6 +1026,9 @@ GetGenomicReferenceSequencesByRid(sqlite3 *dbconn, const string &assembly, const
         if (sqlite3_column_type(stmt.get(), 4) == SQLITE_TEXT) {
             item.refget_id = (const char *)sqlite3_column_text(stmt.get(), 4);
         }
+        if (sqlite3_column_type(stmt.get(), 4) == SQLITE_TEXT) {
+            item.meta_json = (const char *)sqlite3_column_text(stmt.get(), 5);
+        }
         ans[item.rid] = item;
     }
     if (rc != SQLITE_DONE) {