From 3eadc5e345fc9c6260da89e5b24d63bda5781774 Mon Sep 17 00:00:00 2001
From: Michael Schroeder <mls@suse.de>
Date: Mon, 9 Jul 2018 12:59:20 +0200
Subject: [PATCH] Refactor internal stringpool_resize_hash function

---
 src/repo_solv.c | 29 +++++-----------------------
 src/strpool.c   | 50 +++++++++++++++++++++++++++++++------------------
 src/strpool.h   |  1 +
 3 files changed, 38 insertions(+), 42 deletions(-)

diff --git a/src/repo_solv.c b/src/repo_solv.c
index 034d2fab..3509e866 100644
--- a/src/repo_solv.c
+++ b/src/repo_solv.c
@@ -682,35 +682,16 @@ repo_add_solv(Repo *repo, FILE *fp, int flags)
       /* alloc id map for name and rel Ids. this maps ids in the solv files
        * to the ids in our pool */
       idmap = solv_calloc(numid + numrel, sizeof(Id));
-
-      /* grow hash if needed, otherwise reuse */
-      hashmask = mkmask(spool->nstrings + numid);
+      stringpool_resize_hash(spool, numid);
+      hashtbl = spool->stringhashtbl;
+      hashmask = spool->stringhashmask;
 #if 0
       POOL_DEBUG(SOLV_DEBUG_STATS, "read %d strings\n", numid);
-      POOL_DEBUG(SOLV_DEBUG_STATS, "string hash buckets: %d, old %d\n", hashmask + 1, spool->stringhashmask + 1);
+      POOL_DEBUG(SOLV_DEBUG_STATS, "string hash buckets: %d\n", hashmask + 1);
 #endif
-      if (hashmask > spool->stringhashmask)
-	{
-	  spool->stringhashtbl = solv_free(spool->stringhashtbl);
-	  spool->stringhashmask = hashmask;
-          spool->stringhashtbl = hashtbl = solv_calloc(hashmask + 1, sizeof(Id));
-	  for (i = 1; i < spool->nstrings; i++)
-	    {
-	      h = strhash(spool->stringspace + spool->strings[i]) & hashmask;
-	      hh = HASHCHAIN_START;
-	      while (hashtbl[h])
-		h = HASHCHAIN_NEXT(h, hh, hashmask);
-	      hashtbl[h] = i;
-	    }
-	}
-      else
-	{
-	  hashtbl = spool->stringhashtbl;
-	  hashmask = spool->stringhashmask;
-	}
-
       /*
        * run over strings and merge with pool.
+       * we could use stringpool_str2id, but this is faster.
        * also populate id map (maps solv Id -> pool Id)
        */
       for (i = 1; i < numid; i++)
diff --git a/src/strpool.c b/src/strpool.c
index af43e01d..5e879183 100644
--- a/src/strpool.c
+++ b/src/strpool.c
@@ -76,11 +76,39 @@ stringpool_clone(Stringpool *ss, Stringpool *from)
   ss->sstrings = from->sstrings;
 }
 
+void
+stringpool_resize_hash(Stringpool *ss, int numnew)
+{
+  Hashval h, hh, hashmask;
+  Hashtable hashtbl;
+  int i;
+
+  if (numnew <= 0)
+    return;
+  hashmask = mkmask(ss->nstrings + numnew);
+  if (hashmask <= ss->stringhashmask)
+    return;	/* same as before */
+
+  /* realloc hash table */
+  ss->stringhashmask = hashmask;
+  solv_free(ss->stringhashtbl);
+  ss->stringhashtbl = hashtbl = (Hashtable)solv_calloc(hashmask + 1, sizeof(Id));
+  
+  /* rehash all strings into new hashtable */
+  for (i = 1; i < ss->nstrings; i++)
+    {
+      h = strhash(ss->stringspace + ss->strings[i]) & hashmask;
+      hh = HASHCHAIN_START;
+      while (hashtbl[h] != 0)
+	h = HASHCHAIN_NEXT(h, hh, hashmask);
+      hashtbl[h] = i;
+    }
+}
+
 Id
 stringpool_strn2id(Stringpool *ss, const char *str, unsigned int len, int create)
 {
   Hashval h, hh, hashmask, oldhashmask;
-  int i;
   Id id;
   Hashtable hashtbl;
 
@@ -90,27 +118,13 @@ stringpool_strn2id(Stringpool *ss, const char *str, unsigned int len, int create
     return STRID_EMPTY;
 
   hashmask = oldhashmask = ss->stringhashmask;
-  hashtbl = ss->stringhashtbl;
-
   /* expand hashtable if needed */
   if ((Hashval)ss->nstrings * 2 > hashmask)
     {
-      solv_free(hashtbl);
-
-      /* realloc hash table */
-      ss->stringhashmask = hashmask = mkmask(ss->nstrings + STRING_BLOCK);
-      ss->stringhashtbl = hashtbl = (Hashtable)solv_calloc(hashmask + 1, sizeof(Id));
-
-      /* rehash all strings into new hashtable */
-      for (i = 1; i < ss->nstrings; i++)
-	{
-	  h = strhash(ss->stringspace + ss->strings[i]) & hashmask;
-	  hh = HASHCHAIN_START;
-	  while (hashtbl[h] != 0)
-	    h = HASHCHAIN_NEXT(h, hh, hashmask);
-	  hashtbl[h] = i;
-	}
+      stringpool_resize_hash(ss, STRING_BLOCK);
+      hashmask = ss->stringhashmask;
     }
+  hashtbl = ss->stringhashtbl;
 
   /* compute hash and check for match */
   h = strnhash(str, len) & hashmask;
diff --git a/src/strpool.h b/src/strpool.h
index c97b8737..f96c5c18 100644
--- a/src/strpool.h
+++ b/src/strpool.h
@@ -33,6 +33,7 @@ void stringpool_init_empty(Stringpool *ss);
 void stringpool_clone(Stringpool *ss, Stringpool *from);
 void stringpool_free(Stringpool *ss);
 void stringpool_freehash(Stringpool *ss);
+void stringpool_resize_hash(Stringpool *ss, int numnew);
 
 Id stringpool_str2id(Stringpool *ss, const char *str, int create);
 Id stringpool_strn2id(Stringpool *ss, const char *str, unsigned int len, int create);
-- 
2.47.2