tools/server-side/svn-populate-node-origins-index.c

   1 /*
   2  * svn-populate-node-origins-index.c : Populate the repository's node
   3  *                                     origins index.
   4  *
   5  * ====================================================================
   6  * Copyright (c) 2007 CollabNet.  All rights reserved.
   7  *
   8  * This software is licensed as described in the file COPYING, which
   9  * you should have received as part of this distribution.  The terms
  10  * are also available at http://subversion.tigris.org/license-1.html.
  11  * If newer versions of this license are posted there, you may use a
  12  * newer version instead, at your option.
  13  *
  14  * This software consists of voluntary contributions made by many
  15  * individuals.  For exact contribution history, see the revision
  16  * history and logs, available at http://subversion.tigris.org/.
  17  * ====================================================================
  18  */
  19
  20 #include "svn_pools.h"
  21 #include "svn_error.h"
  22 #include "svn_cmdline.h"
  23 #include "svn_path.h"
  24 #include "svn_repos.h"
  25 #include "svn_fs.h"
  26
  27 /* Used to terminate lines in large multi-line string literals. */
  28 #define NL APR_EOL_STR
  29
  30 static const char *usage_summary =
  31   "Crawl the Subversion repository located at REPOS-PATH in an attempt to"  NL
  32   "populate that repository's index of node origins.  "                     NL
  33   ""                                                                        NL
  34   "The node origins index is new as of Subversion 1.5, and behaves as a"    NL
  35   "cache to vastly speed up certain history-querying operations.  For"      NL
  36   "compatibility with repositories created with pre-1.5 versions of"        NL
  37   "Subversion, Subversion will gracefully handle cache misses by doing a"   NL
  38   "brute-force calculation of the query answer and lazily populating the"   NL
  39   "index with answers it calculates.  Unfortunately, calculating that"      NL
  40   "information using the brute-force method (instead of having the"         NL
  41   "information appear in the index organically) can be very costly."        NL
  42   ""                                                                        NL
  43   "This tool triggers the lazy index population logic built into"           NL
  44   "Subversion in a fashion far more efficient than is likely to happen"     NL
  45   "during typical repository usage.  It can be run while the repository"    NL
  46   "is online, too, without interrupting normal Subversion activities."      NL;
  47
  48 /* Print a usage message for this program (PROGNAME), possibly with an
  49    error message ERR_MSG, if not NULL.  */
  50 static void
  51 usage_maybe_with_err(const char *progname, const char *err_msg)
  52 {
  53   FILE *out;
  54
  55   out = err_msg ? stderr : stdout;
  56   fprintf(out, "Usage: %s REPOS-PATH\n\n%s", progname, usage_summary);
  57   if (err_msg)
  58     fprintf(out, "\nERROR: %s\n", err_msg);
  59 }
  60
  61 /* Build the node-origins index any newly added items introduced in
  62    REVISION in FS.  Set *COUNT to the number of new items found.  */
  63 static svn_error_t *
  64 index_revision_adds(int *count, svn_fs_t *fs,
  65                     svn_revnum_t revision, apr_pool_t *pool)
  66 {
  67   svn_fs_root_t *root;
  68   apr_hash_t *changes;
  69   apr_hash_index_t *hi;
  70   apr_pool_t *subpool;
  71
  72   *count = 0;
  73   SVN_ERR(svn_fs_revision_root(&root, fs, revision, pool));
  74   SVN_ERR(svn_fs_paths_changed(&changes, root, pool));
  75
  76   /* No paths changed in this revision?  Nothing to do.  */
  77   if (apr_hash_count(changes) == 0)
  78     return SVN_NO_ERROR;
  79
  80   subpool = svn_pool_create(pool);
  81   for (hi = apr_hash_first(pool, changes); hi; hi = apr_hash_next(hi))
  82     {
  83       const void *path;
  84       void *val;
  85       svn_fs_path_change_t *change;
  86
  87       svn_pool_clear(subpool);
  88       apr_hash_this(hi, &path, NULL, &val);
  89       change = val;
  90       if ((change->change_kind == svn_fs_path_change_add)
  91           || (change->change_kind == svn_fs_path_change_replace))
  92         {
  93           const char *copyfrom_path;
  94           svn_revnum_t copyfrom_rev;
  95
  96           SVN_ERR(svn_fs_copied_from(&copyfrom_rev, &copyfrom_path,
  97                                      root, path, subpool));
  98           if (! (copyfrom_path && SVN_IS_VALID_REVNUM(copyfrom_rev)))
  99             {
 100               svn_revnum_t origin;
 101               SVN_ERR(svn_fs_node_origin_rev(&origin, root, path, subpool));
 102               (*count)++;
 103             }
 104         }
 105     }
 106   svn_pool_destroy(subpool);
 107
 108   return SVN_NO_ERROR;
 109 }
 110
 111 /* Build the node-origins index for the repository located at REPOS_PATH.  */
 112 static svn_error_t *
 113 build_index(const char *repos_path, apr_pool_t *pool)
 114 {
 115   svn_repos_t *repos;
 116   svn_fs_t *fs;
 117   svn_revnum_t youngest_rev;
 118   int i, slotsize;
 119   const char *progress_fmt;
 120   apr_pool_t *subpool;
 121
 122   /* Open the repository. */
 123   SVN_ERR(svn_repos_open(&repos, repos_path, pool));
 124
 125   /* Get a filesystem object. */
 126   fs = svn_repos_fs(repos);
 127
 128   /* Fetch the youngest revision of the repository. */
 129   SVN_ERR(svn_fs_youngest_rev(&youngest_rev, fs, pool));
 130   slotsize = strlen(apr_ltoa(pool, youngest_rev));
 131   progress_fmt = apr_psprintf(pool,
 132                               "[%%%dd/%%%dd]  Found %%d new lines of history."
 133                               "\n", slotsize, slotsize);
 134
 135   /* Now, iterate over all the revisions, calling index_revision_adds(). */
 136   subpool = svn_pool_create(pool);
 137   for (i = 0; i < youngest_rev; i++)
 138     {
 139       int count;
 140       svn_pool_clear(subpool);
 141       SVN_ERR(index_revision_adds(&count, fs, i + 1, subpool));
 142       printf(progress_fmt, i + 1, youngest_rev, count);
 143     }
 144   svn_pool_destroy(subpool);
 145
 146   return SVN_NO_ERROR;
 147 }
 148
 149
 150 int
 151 main(int argc, const char **argv)
 152 {
 153   apr_pool_t *pool;
 154   svn_error_t *err = SVN_NO_ERROR;
 155   const char *repos_path;
 156
 157   /* Initialize the app.  Send all error messages to 'stderr'.  */
 158   if (svn_cmdline_init(argv[0], stderr) == EXIT_FAILURE)
 159     return EXIT_FAILURE;
 160
 161   pool = svn_pool_create(NULL);
 162
 163   if (argc <= 1)
 164     {
 165       usage_maybe_with_err(argv[0], "Not enough arguments.");
 166       goto cleanup;
 167     }
 168
 169   /* Convert argv[1] into a UTF8, internal-format, canonicalized path. */
 170   if ((err = svn_utf_cstring_to_utf8(&repos_path, argv[1], pool)))
 171     goto cleanup;
 172   repos_path = svn_path_internal_style(repos_path, pool);
 173   repos_path = svn_path_canonicalize(repos_path, pool);
 174
 175   if ((err = build_index(repos_path, pool)))
 176     goto cleanup;
 177
 178  cleanup:
 179   svn_pool_destroy(pool);
 180
 181   if (err)
 182     {
 183       svn_handle_error2(err, stderr, FALSE,
 184                         "svn-populate-node-origins-index: ");
 185       return EXIT_FAILURE;
 186     }
 187   return EXIT_SUCCESS;
 188 }