libsolv/0002-Add-support-for-storing-user-data-in-a-solv-file.patch

From 9b89a186e3769631b6cee859be9d69063cfdfb94 Mon Sep 17 00:00:00 2001
From: Michael Schroeder <mls@suse.de>
Date: Fri, 25 Feb 2022 16:47:21 +0100
Subject: [PATCH 2/3] Add support for storing user data in a solv file

Userdata can be arbritrary (binary)data with a maximum size of
65535 bytes. It can be read without reading the complete
solv file, but do not forget to rewind the fp after reading
the user data.

New functions:
void
    void repowriter_set_userdata(Repowriter *writer, const void *data, int len)
    int solv_read_userdata(FILE *fp, unsigned char **datap, int *lenp)
---
 src/libsolv.ver  |  2 ++
 src/pooltypes.h  |  6 ++++--
 src/repo_solv.c  | 54 ++++++++++++++++++++++++++++++++++++++++++++++++
 src/repo_solv.h  |  1 +
 src/repo_write.c | 24 ++++++++++++++++++++-
 src/repo_write.h |  3 +++
 tools/dumpsolv.c | 50 ++++++++++++++++++++++++++++++++++++++++----
 7 files changed, 133 insertions(+), 7 deletions(-)

diff --git a/src/libsolv.ver b/src/libsolv.ver
index ee40d0ad..4c6fbf4f 100644
--- a/src/libsolv.ver
+++ b/src/libsolv.ver
@@ -255,6 +255,7 @@ SOLV_1.0 {
 		repowriter_set_keyqueue;
 		repowriter_set_repodatarange;
 		repowriter_set_solvablerange;
+		repowriter_set_userdata;
 		repowriter_write;
 		selection_add;
 		selection_filter;
@@ -288,6 +289,7 @@ SOLV_1.0 {
 		solv_malloc;
 		solv_malloc2;
 		solv_oom;
+		solv_read_userdata;
 		solv_realloc;
 		solv_realloc2;
 		solv_replacebadutf8;
diff --git a/src/pooltypes.h b/src/pooltypes.h
index e1f77b0e..3bde155a 100644
--- a/src/pooltypes.h
+++ b/src/pooltypes.h
@@ -23,9 +23,11 @@
 #define SOLV_VERSION_6 6
 #define SOLV_VERSION_7 7
 #define SOLV_VERSION_8 8
+#define SOLV_VERSION_9 9
 
-#define SOLV_FLAG_PREFIX_POOL 4
-#define SOLV_FLAG_SIZE_BYTES  8
+#define SOLV_FLAG_PREFIX_POOL	4
+#define SOLV_FLAG_SIZE_BYTES	8
+#define SOLV_FLAG_USERDATA	16
 
 struct s_Stringpool;
 typedef struct s_Stringpool Stringpool;
diff --git a/src/repo_solv.c b/src/repo_solv.c
index 761d06e6..2ba602b2 100644
--- a/src/repo_solv.c
+++ b/src/repo_solv.c
@@ -514,6 +514,7 @@ repo_add_solv(Repo *repo, FILE *fp, int flags)
   switch (solvversion)
     {
       case SOLV_VERSION_8:
+      case SOLV_VERSION_9:
 	break;
       default:
         return pool_error(pool, SOLV_ERROR_UNSUPPORTED, "unsupported SOLV version");
@@ -552,6 +553,18 @@ repo_add_solv(Repo *repo, FILE *fp, int flags)
 	  return pool_error(pool, SOLV_ERROR_CORRUPT, "main repository contains holes, cannot extend");
     }
 
+  /*******  Part 0: skip optional userdata ******************************/
+
+  if (solvflags & SOLV_FLAG_USERDATA)
+    {
+      unsigned int userdatalen = read_u32(&data);
+      if (userdatalen >= 65536)
+        return pool_error(pool, SOLV_ERROR_CORRUPT, "illegal userdata length");
+      while (userdatalen--)
+	if (getc(data.fp) == EOF)
+	  return pool_error(pool, SOLV_ERROR_EOF, "unexpected EOF");
+    }
+
   /*******  Part 1: string IDs  *****************************************/
 
   sizeid = read_u32(&data);	       /* size of string space */
@@ -1353,3 +1366,44 @@ printf("=> %s %s %p\n", pool_id2str(pool, keys[key].name), pool_id2str(pool, key
   return 0;
 }
 
+int
+solv_read_userdata(FILE *fp, unsigned char **datap, int *lenp)
+{
+  unsigned char d[4 * 10], *ud = 0;
+  unsigned int n;
+  if (fread(d, sizeof(d), 1, fp) != 1)
+    return SOLV_ERROR_EOF;
+  n = d[0] << 24 | d[1] << 16 | d[2] << 8 | d[3];
+  if (n != ('S' << 24 | 'O' << 16 | 'L' << 8 | 'V'))
+    return SOLV_ERROR_NOT_SOLV;
+  n = d[4] << 24 | d[5] << 16 | d[6] << 8 | d[7];
+  switch(n)
+    {
+    case SOLV_VERSION_8:
+    case SOLV_VERSION_9:
+      break;
+    default:
+      return SOLV_ERROR_UNSUPPORTED;
+    }
+  n = d[32] << 24 | d[33] << 16 | d[34] << 8 | d[35];
+  if (!(n & SOLV_FLAG_USERDATA))
+    n = 0;
+  else
+    n = d[36] << 24 | d[37] << 16 | d[38] << 8 | d[39];
+  if (n >= 65536)
+    return SOLV_ERROR_CORRUPT;
+  if (n)
+    {
+      ud = solv_malloc(n + 1);
+      if (fread(ud, n, 1, fp) != 1)
+	{
+	  solv_free(ud);
+	  return SOLV_ERROR_EOF;
+	}
+      ud[n] = 0;
+    }
+  *datap = ud;
+  if (lenp)
+    *lenp = (int)n;
+  return 0;
+}
diff --git a/src/repo_solv.h b/src/repo_solv.h
index 0c663949..57bf1772 100644
--- a/src/repo_solv.h
+++ b/src/repo_solv.h
@@ -23,6 +23,7 @@ extern "C" {
 #endif
 
 extern int repo_add_solv(Repo *repo, FILE *fp, int flags);
+extern int solv_read_userdata(FILE *fp, unsigned char **datap, int *lenp);
 
 #define SOLV_ADD_NO_STUBS	(1 << 8)
 
diff --git a/src/repo_write.c b/src/repo_write.c
index af4e7599..a11de002 100644
--- a/src/repo_write.c
+++ b/src/repo_write.c
@@ -1071,6 +1071,7 @@ repowriter_create(Repo *repo)
 Repowriter *
 repowriter_free(Repowriter *writer)
 {
+  solv_free(writer->userdata);
   return solv_free(writer);
 }
 
@@ -1107,6 +1108,17 @@ repowriter_set_solvablerange(Repowriter *writer, int solvablestart, int solvable
   writer->solvableend = solvableend;
 }
 
+void
+repowriter_set_userdata(Repowriter *writer, const void *data, int len)
+{
+  writer->userdata = solv_free(writer->userdata);
+  writer->userdatalen = 0;
+  if (len < 0 || len >= 65536)
+    return;
+  writer->userdata = len ? solv_memdup(data, len) : 0;
+  writer->userdatalen = len;
+}
+
 /*
  * the code works the following way:
  *
@@ -1898,7 +1910,10 @@ for (i = 1; i < target.nkeys; i++)
 
   /* write file header */
   write_u32(&target, 'S' << 24 | 'O' << 16 | 'L' << 8 | 'V');
-  write_u32(&target, SOLV_VERSION_8);
+  if (writer->userdatalen)
+    write_u32(&target, SOLV_VERSION_9);
+  else
+    write_u32(&target, SOLV_VERSION_8);
 
 
   /* write counts */
@@ -1911,7 +1926,14 @@ for (i = 1; i < target.nkeys; i++)
   solv_flags = 0;
   solv_flags |= SOLV_FLAG_PREFIX_POOL;
   solv_flags |= SOLV_FLAG_SIZE_BYTES;
+  if (writer->userdatalen)
+    solv_flags |= SOLV_FLAG_USERDATA;
   write_u32(&target, solv_flags);
+  if (writer->userdatalen)
+    {
+      write_u32(&target, writer->userdatalen);
+      write_blob(&target, writer->userdata, writer->userdatalen);
+    }
 
   if (nstrings)
     {
diff --git a/src/repo_write.h b/src/repo_write.h
index 34716705..7734b013 100644
--- a/src/repo_write.h
+++ b/src/repo_write.h
@@ -32,6 +32,8 @@ typedef struct s_Repowriter {
   int (*keyfilter)(Repo *repo, Repokey *key, void *kfdata);
   void *kfdata;
   Queue *keyq;
+  void *userdata;
+  int userdatalen;
 } Repowriter;
 
 /* repowriter flags */
@@ -46,6 +48,7 @@ void repowriter_set_keyfilter(Repowriter *writer, int (*keyfilter)(Repo *repo, R
 void repowriter_set_keyqueue(Repowriter *writer, Queue *keyq);
 void repowriter_set_repodatarange(Repowriter *writer, int repodatastart, int repodataend);
 void repowriter_set_solvablerange(Repowriter *writer, int solvablestart, int solvableend);
+void repowriter_set_userdata(Repowriter *writer, const void *data, int len);
 int repowriter_write(Repowriter *writer, FILE *fp);
 
 /* convenience functions */
diff --git a/tools/dumpsolv.c b/tools/dumpsolv.c
index 13076574..49651fbe 100644
--- a/tools/dumpsolv.c
+++ b/tools/dumpsolv.c
@@ -13,6 +13,7 @@
 
 static int with_attr;
 static int dump_json;
+static int dump_userdata;
 
 #include "pool.h"
 #include "chksum.h"
@@ -394,10 +395,7 @@ int main(int argc, char **argv)
   int c, i, j, n;
   Solvable *s;
   
-  pool = pool_create();
-  pool_setloadcallback(pool, loadcallback, 0);
-
-  while ((c = getopt(argc, argv, "haj")) >= 0)
+  while ((c = getopt(argc, argv, "uhaj")) >= 0)
     {
       switch(c)
 	{
@@ -410,11 +408,55 @@ int main(int argc, char **argv)
 	case 'j':
 	  dump_json = 1;
 	  break;
+	case 'u':
+	  dump_userdata++;
+	  break;
 	default:
           usage(1);
           break;
 	}
     }
+  if (dump_userdata)
+    {
+      if (optind == argc)
+	argc++;
+      for (; optind < argc; optind++)
+	{
+	  unsigned char *userdata = 0;
+	  int r, userdatalen = 0;
+	  if (argv[optind] && freopen(argv[optind], "r", stdin) == 0)
+	    {
+	      perror(argv[optind]);
+	      exit(1);
+	    }
+	  r = solv_read_userdata(stdin, &userdata, &userdatalen);
+	  if (r)
+	    {
+	      fprintf(stderr, "could not read userdata: error %d\n", r);
+	      exit(1);
+	    }
+	  if (dump_userdata > 1)
+	    {
+	      /* dump raw */
+	      if (userdatalen && fwrite(userdata, userdatalen, 1, stdout) != 1)
+		{
+		  perror("fwrite");
+		  exit(1);
+		}
+	    }
+	  else
+	    {
+	      for (r = 0; r < userdatalen; r++)
+		printf("%02x", userdata[r]);
+	      printf("\n");
+	    }
+	  solv_free(userdata);
+	}
+      exit(0);
+    }
+
+  pool = pool_create();
+  pool_setloadcallback(pool, loadcallback, 0);
   if (!dump_json)
     pool_setdebuglevel(pool, 1);
   if (dump_json)
-- 
2.31.1