44ccfacbe3a3d4b300f80d48651c77837a4b571e
galt
  Tue Apr 26 11:12:02 2022 -0700
SQL INJECTION Prevention Version 2 - this improves our methods by making subclauses of SQL that get passed around be both easy and correct to use. The way that was achieved was by getting rid of the obscure and not well used functions sqlSafefFrag and sqlDyStringPrintfFrag and replacing them with the plain versions of those functions, since these are not needed anymore. The new version checks for NOSQLINJ in unquoted %-s which is used to include SQL clauses, and will give an error the NOSQLINJ clause is not present, and this will automatically require the correct behavior by developers. sqlDyStringPrint is a very useful function, however because it was not enforced, users could use various other dyString functions and they operated without any awareness or checking for SQL correct use. Now those dyString functions are prohibited and it will produce an error if you try to use a dyString function on a SQL string, which is simply detected by the presence of the NOSQLINJ prefix.

diff --git src/hg/near/hgClusterGenes/hgClusterGenes.c src/hg/near/hgClusterGenes/hgClusterGenes.c
index 6204846..12ddc81 100644
--- src/hg/near/hgClusterGenes/hgClusterGenes.c
+++ src/hg/near/hgClusterGenes/hgClusterGenes.c
@@ -248,47 +248,47 @@
     cluster = node->val;
     slAddHead(&clusterList, cluster);
     }
 slSort(&clusterList, clusterCmp);
 
 /* Clean up and go home. */
 binKeeperFree(&bk);
 dlListFree(&clusters);
 return clusterList;
 }
 
 void createClusterTable(struct sqlConnection *conn, 
 	char *tableName, int longestName)
 /* Create cluster table. */
 {
-struct dyString *dy = newDyString(1024);
+struct dyString *dy = dyStringNew(1024);
 sqlDyStringPrintf(dy,
     "CREATE TABLE %s (\n"
     " clusterId int unsigned not null,\n"
     " transcript varchar(255) not null,\n"
     " INDEX(transcript(%d)),\n"
     " INDEX(clusterId))\n"
     , tableName, longestName);
 sqlRemakeTable(conn, tableName, dy->string);
 dyStringFree(&dy);
 }
 
 void createCannonicalTable(struct sqlConnection *conn, 
 	char *tableName, int longestName)
 /* Create cannonical representative of cluster table. */
 {
-struct dyString *dy = newDyString(1024);
+struct dyString *dy = dyStringNew(1024);
 sqlDyStringPrintf(dy,
     "CREATE TABLE %s (\n"
     " chrom varchar(255) not null,\n"
     " chromStart int not null,\n"
     " chromEnd int not null,\n"
     " clusterId int unsigned not null,\n"
     " transcript varchar(255) not null,\n"
     " protein varchar(255) not null,\n"
     " UNIQUE(clusterId),\n"
     " INDEX(transcript(%d)),\n"
     " INDEX(protein(8)))\n"
     , tableName, longestName);
 sqlRemakeTable(conn, tableName, dy->string);
 dyStringFree(&dy);
 }
@@ -320,31 +320,31 @@
     slAddHead(&gpList, gp);
     if (nameLen > longestName)
         longestName = nameLen;
     ++totalGeneCount;
     }
 sqlFreeResult(&sr);
 
 /* Build hash to map between transcript names and protein IDs. */
 if (!noProt)
     {
     protHash = newHash(16);
     if (sangerLinks)
 	sqlSafef(query, sizeof(query), "select orfName,protName from sangerLinks");
     else if (isNotEmpty(protAccQuery))
 	// accepting special user query without checking
-	sqlSafef(query, sizeof(query), "%-s", protAccQuery);  
+	sqlSafef(query, sizeof(query), protAccQuery, NULL);  
     else
 	sqlSafef(query, sizeof(query), "select name, proteinId from %s", geneTable);
     sr = sqlGetResult(conn, query);
     while ((row = sqlNextRow(sr)) != NULL)
 	hashAdd(protHash, row[0], cloneString(row[1]));
     sqlFreeResult(&sr);
     }
 
 slReverse(&gpList);
 clusterList = makeCluster(gpList, hChromSize(database, chrom));
 for (cluster = clusterList; cluster != NULL; cluster = cluster->next)
     {
     struct hashEl *helList = hashElListHash(cluster->geneHash);
     struct hashEl *hel;
     struct genePred *cannonical = NULL;