From 7e219a9e1c9eb18e840a41390bce0876ebc22792 Mon Sep 17 00:00:00 2001
From: Jason Lee <jasonlee@lanl.gov>
Date: Tue, 23 Jan 2024 11:11:06 -0700
Subject: [PATCH] gufi_query -Q <dbname> <attachname> <sql>

attach single db file to all threads
file is attached at the start of gufi_query instead of being reattached at every single directory
---
 include/bf.h                          |  6 ++++++
 scripts/gufi_common.py                |  1 +
 src/bf.c                              | 18 ++++++++++++++++++
 src/gufi_query/PoolArgs.c             | 19 +++++++++++++++++++
 test/unit/googletest/aggregate.cpp.in |  2 +-
 5 files changed, 45 insertions(+), 1 deletion(-)

diff --git a/include/bf.h b/include/bf.h
index 16ae5ace0..9e1b874e4 100644
--- a/include/bf.h
+++ b/include/bf.h
@@ -233,6 +233,12 @@ struct input {
 
    /* compress work items (if compression library was found) */
    int compress;
+
+   struct {
+       refstr_t dbname;       /* name of single db file */
+       refstr_t attachname;   /* attach name of db file */
+       refstr_t sql;          /* SQL used to create view(s) on attached data */
+   } attach_single;
 };
 
 void print_help(const char *prog_name,
diff --git a/scripts/gufi_common.py b/scripts/gufi_common.py
index 49d11410a..e4d9e1ca9 100644
--- a/scripts/gufi_common.py
+++ b/scripts/gufi_common.py
@@ -73,6 +73,7 @@
     from shlex import quote as sanitize # new in Python 3.3
 
 # table names
+ENTRIES     = 'entries'
 SUMMARY     = 'summary'
 PENTRIES    = 'pentries'
 XSUMMARY    = 'xsummary'
diff --git a/src/bf.c b/src/bf.c
index f3501c46f..af79ddad9 100644
--- a/src/bf.c
+++ b/src/bf.c
@@ -143,6 +143,9 @@ void print_help(const char* prog_name,
 #if HAVE_ZLIB
       case 'e': printf("  -e                     compress work items"); break;
 #endif
+      case 'Q': printf("  -Q <dbname>\n"
+                       "     <attachname>\n"
+                       "     <sql>               Attach a single file across every single db in the index and use the SQL to create views"); break;
       default: printf("print_help(): unrecognized option '%c'", (char)ch);
       }
       printf("\n");
@@ -197,6 +200,9 @@ void show_input(struct input* in, int retval) {
    printf("in.target_memory_footprint  = %" PRIu64 "\n",   in->target_memory_footprint);
    printf("in.subdir_limit             = %zu\n",           in->subdir_limit);
    printf("in.compress                 = %d\n",            in->compress);
+   printf("in.attach_single.dbname     = %s\n",            in->attach_single.dbname.data);
+   printf("in.attach_single.attachname = %s\n",            in->attach_single.attachname.data);
+   printf("in.attach_single.sql        = %s\n",            in->attach_single.sql.data);
    printf("\n");
    printf("retval                      = %d\n",    retval);
    printf("\n");
@@ -432,6 +438,18 @@ int parse_cmd_line(int         argc,
           break;
 #endif
 
+      case 'Q':
+          INSTALL_STR(&in->attach_single.dbname, optarg);
+          optarg = argv[optind];
+
+          INSTALL_STR(&in->attach_single.attachname, optarg);
+          optarg = argv[++optind];
+
+          INSTALL_STR(&in->attach_single.sql, optarg);
+          optarg = argv[++optind];
+
+          break;
+
       case '?':
          // getopt returns '?' when there is a problem.  In this case it
          // also prints, e.g. "getopt_test: illegal option -- z"
diff --git a/src/gufi_query/PoolArgs.c b/src/gufi_query/PoolArgs.c
index 7cbeb47e8..9c92568aa 100644
--- a/src/gufi_query/PoolArgs.c
+++ b/src/gufi_query/PoolArgs.c
@@ -109,6 +109,25 @@ int PoolArgs_init(PoolArgs_t *pa, struct input *in, pthread_mutex_t *global_mute
         }
         #endif
 
+        /* handle -Q */
+        if (in->attach_single.dbname.len) {
+            /* attach single db to thread instance */
+            if (!attachdb(in->attach_single.dbname.data, ta->outdb,
+                          in->attach_single.attachname.data,
+                          SQLITE_OPEN_READONLY, 1)) {
+                break;
+            }
+
+            /* set up views */
+            char *err = NULL;
+            if (sqlite3_exec(ta->outdb, in->attach_single.sql.data, NULL, NULL, &err) != SQLITE_OK) {
+                fprintf(stderr, "Error: Could not run SQL Init \"%s\" on %s: %s\n",
+                        in->sql.init.data, ta->dbname, err);
+                sqlite3_free(err);
+                break;
+            }
+        }
+
         /* run -I */
         if (in->sql.init.len) {
             char *err = NULL;
diff --git a/test/unit/googletest/aggregate.cpp.in b/test/unit/googletest/aggregate.cpp.in
index e8aec66d0..0282c4f63 100644
--- a/test/unit/googletest/aggregate.cpp.in
+++ b/test/unit/googletest/aggregate.cpp.in
@@ -95,7 +95,7 @@ TEST(gufi_query, aggregate) {
 
     const size_t row_count = std::uniform_int_distribution <uint32_t> (1, 8)(gen);
 
-    struct input in;
+    struct input in{};
     in.output                = STDOUT;
     in.maxthreads            = std::uniform_int_distribution <uint32_t> (1, 8)(gen);
     in.delim                 = ' ';