[GNUnet-SVN] r19157 - gnunet/src/fs

[gnunet]

Author: grothoff
Date: 2012-01-15 15:55:13 +0100 (Sun, 15 Jan 2012)
New Revision: 19157

Modified:
   gnunet/src/fs/fs_dirmetascan.c
   gnunet/src/fs/gnunet-publish.c
Log:
-LRN: use filescanner in gnunet-publish

Modified: gnunet/src/fs/fs_dirmetascan.c
===================================================================
--- gnunet/src/fs/fs_dirmetascan.c      2012-01-15 12:16:42 UTC (rev 19156)
+++ gnunet/src/fs/fs_dirmetascan.c      2012-01-15 14:55:13 UTC (rev 19157)
@@ -568,6 +568,13 @@
                                      EXTRACTOR_METATYPE_FILENAME,
                                      EXTRACTOR_METAFORMAT_UTF8, "text/plain",
                                      short_fn, strlen (short_fn) + 1);
+  if (ads->parent == NULL)
+  {
+    /* we're finished with the scan, make sure caller gets the top-level
+     * directory pointer
+     */
+    ads->adc->toplevel = item;
+  }
 }
 
 /**
@@ -869,7 +876,7 @@
  * The function from which the scanner thread starts
  */
 #if WINDOWS
-static DWORD
+DWORD
 #else
 static void *
 #endif
@@ -1043,12 +1050,7 @@
 
   if (0 != STAT (filename, &sbuf))
     return NULL;
-  /* TODO: consider generalizing this for files too! */
-  if (!S_ISDIR (sbuf.st_mode))
-  {
-    GNUNET_break (0);
-    return NULL;
-  }
+
   /* scan_directory() is guaranteed to be given expanded filenames,
    * so expand we will!
    */
@@ -1211,30 +1213,35 @@
     /* process a child entry (a file or a directory) and move to the next one*/
     if (stack->item->is_directory)
       stack->end_directory = GNUNET_NO;
-    stack->dir_entry_count++;
-    GNUNET_CONTAINER_meta_data_iterate (stack->item->meta, 
&add_to_meta_counter, stack->metacounter);
+    if (stack->ctx->toplevel->is_directory)
+    {
+      stack->dir_entry_count++;
+      GNUNET_CONTAINER_meta_data_iterate (stack->item->meta, 
&add_to_meta_counter, stack->metacounter);
 
-    if (stack->item->is_directory)
-    {
-      char *user = getenv ("USER");
-      if ((user == NULL) || (0 != strncasecmp (user, 
stack->item->short_filename, strlen(user))))
+      if (stack->item->is_directory)
       {
-        /* only use filename if it doesn't match $USER */
-        GNUNET_CONTAINER_meta_data_insert (stack->item->meta, "<libgnunetfs>",
-                                          EXTRACTOR_METATYPE_FILENAME,
-                                          EXTRACTOR_METAFORMAT_UTF8,
-                                          "text/plain", 
stack->item->short_filename,
-                                          strlen (stack->item->short_filename) 
+ 1);
-        GNUNET_CONTAINER_meta_data_insert (stack->item->meta, "<libgnunetfs>",
-                                          
EXTRACTOR_METATYPE_GNUNET_ORIGINAL_FILENAME,
-                                          EXTRACTOR_METAFORMAT_UTF8,
-                                          "text/plain", 
stack->item->short_filename,
-                                          strlen (stack->item->short_filename) 
+ 1);
+        char *user = getenv ("USER");
+        if ((user == NULL) || (0 != strncasecmp (user, 
stack->item->short_filename, strlen(user))))
+        {
+          /* only use filename if it doesn't match $USER */
+          GNUNET_CONTAINER_meta_data_insert (stack->item->meta, 
"<libgnunetfs>",
+                                            EXTRACTOR_METATYPE_FILENAME,
+                                            EXTRACTOR_METAFORMAT_UTF8,
+                                            "text/plain", 
stack->item->short_filename,
+                                            strlen 
(stack->item->short_filename) + 1);
+          GNUNET_CONTAINER_meta_data_insert (stack->item->meta, 
"<libgnunetfs>",
+                                            
EXTRACTOR_METATYPE_GNUNET_ORIGINAL_FILENAME,
+                                            EXTRACTOR_METAFORMAT_UTF8,
+                                            "text/plain", 
stack->item->short_filename,
+                                            strlen 
(stack->item->short_filename) + 1);
+        }
       }
     }
-
     stack->item->ksk_uri = GNUNET_FS_uri_ksk_create_from_meta_data 
(stack->item->meta);
-    GNUNET_FS_uri_ksk_get_keywords (stack->item->ksk_uri, 
&add_to_keyword_counter, stack->keywordcounter);
+    if (stack->ctx->toplevel->is_directory)
+    {
+      GNUNET_FS_uri_ksk_get_keywords (stack->item->ksk_uri, 
&add_to_keyword_counter, stack->keywordcounter);
+    }
     stack->item = stack->item->next;
   }
   /* Call this task again later, if there are more entries to process */
@@ -1270,8 +1277,13 @@
   ret->stack = GNUNET_malloc (sizeof (struct ProcessMetadataStackItem));
   ret->stack->ctx = ret;
   ret->stack->item = toplevel;
-  ret->stack->keywordcounter = GNUNET_CONTAINER_multihashmap_create (1024);
-  ret->stack->metacounter = GNUNET_CONTAINER_multihashmap_create (1024);
+
+  if (ret->stack->ctx->toplevel->is_directory)
+  {
+    ret->stack->keywordcounter = GNUNET_CONTAINER_multihashmap_create (1024);
+    ret->stack->metacounter = GNUNET_CONTAINER_multihashmap_create (1024);
+  }
+
   ret->stack->dir_entry_count = 0;
   ret->stack->end_directory = GNUNET_NO;
 

Modified: gnunet/src/fs/gnunet-publish.c
===================================================================
--- gnunet/src/fs/gnunet-publish.c      2012-01-15 12:16:42 UTC (rev 19156)
+++ gnunet/src/fs/gnunet-publish.c      2012-01-15 14:55:13 UTC (rev 19157)
@@ -66,7 +66,16 @@
 
 static GNUNET_SCHEDULER_TaskIdentifier kill_task;
 
+static struct GNUNET_FS_DirScanner *ds;
 
+static struct GNUNET_FS_ShareTreeItem * directory_scan_intermediary_result;
+
+static struct GNUNET_FS_ShareTreeItem * directory_scan_result;
+
+static struct GNUNET_FS_ProcessMetadataContext *pmc;
+
+static struct GNUNET_FS_Namespace *namespace;
+
 /**
  * FIXME: docu
  */
@@ -340,7 +349,151 @@
   ctx = NULL;
 }
 
+static struct GNUNET_FS_FileInformation *
+get_file_information (struct GNUNET_FS_ShareTreeItem *item)
+{
+  struct GNUNET_FS_FileInformation *fi;
+  struct GNUNET_FS_FileInformation *fic;
+  struct GNUNET_FS_ShareTreeItem *child;
 
+  if (item->is_directory)
+  {
+    GNUNET_CONTAINER_meta_data_delete (item->meta,
+        EXTRACTOR_METATYPE_MIMETYPE, NULL, 0);
+    GNUNET_FS_meta_data_make_directory (item->meta);
+    GNUNET_FS_uri_ksk_add_keyword (item->ksk_uri, GNUNET_FS_DIRECTORY_MIME,
+        GNUNET_NO);
+    fi = GNUNET_FS_file_information_create_empty_directory (
+        ctx, NULL, item->ksk_uri,
+        item->meta, &bo, item->filename);
+    for (child = item->children_head; child; child = child->next)
+    {
+      fic = get_file_information (child);
+      GNUNET_break (GNUNET_OK == GNUNET_FS_file_information_add (fi, fic));
+    }
+  }
+  else
+  {
+    fi = GNUNET_FS_file_information_create_from_file (
+        ctx, NULL, item->filename,
+        item->ksk_uri, item->meta, !do_insert,
+        &bo);
+  }
+  GNUNET_CONTAINER_meta_data_destroy (item->meta);
+  GNUNET_FS_uri_destroy (item->ksk_uri);
+  GNUNET_free (item->short_filename);
+  GNUNET_free (item->filename);
+  GNUNET_free (item);
+  return fi;
+}
+
+static void
+directory_trim_complete (void *cls,
+    const struct GNUNET_SCHEDULER_TaskContext *tc)
+{
+  struct GNUNET_FS_FileInformation *fi;
+  directory_scan_result = directory_scan_intermediary_result;
+  fi = get_file_information (directory_scan_result);
+  directory_scan_result = NULL;
+  if (fi == NULL)
+  {
+    FPRINTF (stderr, "%s", _("Could not publish\n"));
+    if (namespace != NULL)
+      GNUNET_FS_namespace_delete (namespace, GNUNET_NO);
+    GNUNET_FS_stop (ctx);
+    ret = 1;
+    return;
+  }
+  GNUNET_FS_file_information_inspect (fi, &publish_inspector, NULL);
+  if (extract_only)
+  {
+    if (namespace != NULL)
+      GNUNET_FS_namespace_delete (namespace, GNUNET_NO);
+    GNUNET_FS_file_information_destroy (fi, NULL, NULL);
+    GNUNET_FS_stop (ctx);
+    return;
+  }
+  pc = GNUNET_FS_publish_start (ctx, fi, namespace, this_id, next_id,
+                                (do_simulate) ?
+                                GNUNET_FS_PUBLISH_OPTION_SIMULATE_ONLY :
+                                GNUNET_FS_PUBLISH_OPTION_NONE);
+  if (NULL == pc)
+  {
+    FPRINTF (stderr, "%s",  _("Could not start publishing.\n"));
+    GNUNET_FS_stop (ctx);
+    ret = 1;
+    return;
+  }
+}
+
+static int
+directory_scan_cb (void *cls, struct GNUNET_FS_DirScanner *ds,
+    const char *filename, char is_directory,
+    enum GNUNET_FS_DirScannerProgressUpdateReason reason)
+{
+  switch (reason)
+  {
+    case GNUNET_DIR_SCANNER_NEW_FILE:
+      if (filename != NULL)
+      {
+        if (is_directory)
+          FPRINTF (stdout, _("Scanning directory `%s'.\n"), filename);
+        else
+          FPRINTF (stdout, _("Scanning file `%s'.\n"), filename);
+      }
+      break;
+    case GNUNET_DIR_SCANNER_DOES_NOT_EXIST:
+      if (filename != NULL)
+      {
+        FPRINTF (stdout, 
+            _("Failed to scan `%s', because it does not exist.\n"),
+            filename);
+      }
+      break;
+    case GNUNET_DIR_SCANNER_ASKED_TO_STOP:
+      if (filename != NULL)
+      {
+        FPRINTF (stdout, 
+            _("Scanner was about to scan `%s', but is now stopping.\n"),
+            filename);
+      }
+      else
+        FPRINTF (stdout, "%s", _("Scanner is stopping.\n"));
+      break;
+    case GNUNET_DIR_SCANNER_SHUTDOWN:
+      FPRINTF (stdout, "%s", _("Client is shutting down.\n"));
+      break;
+    case GNUNET_DIR_SCANNER_FINISHED:
+      FPRINTF (stdout, "%s", _("Scanner has finished.\n"));
+      break;
+    case GNUNET_DIR_SCANNER_PROTOCOL_ERROR:
+      FPRINTF (stdout, "%s", 
+          _("There was a failure communicating with the scanner.\n"));
+      break;
+    default:
+      FPRINTF (stdout, _("Got unknown scanner update with filename `%s'.\n"),
+          filename);
+      break;
+  }
+  if ((filename == NULL && GNUNET_DIR_SCANNER_FINISHED)
+      || reason == GNUNET_DIR_SCANNER_PROTOCOL_ERROR
+      || reason == GNUNET_DIR_SCANNER_SHUTDOWN)
+  {
+    /* Any of this causes us to try to clean up the scanner */
+    directory_scan_intermediary_result = GNUNET_FS_directory_scan_cleanup (ds);
+    pmc = GNUNET_FS_trim_share_tree (directory_scan_intermediary_result,
+      &directory_trim_complete, NULL);
+
+    ds = NULL;
+    /* FIXME: change the tree processor to be able to free untrimmed trees
+     * right here instead of waiting for trimming to complete, if we need to
+     * cancel everything.
+     */
+  }
+  return 0;
+}
+
+
 /**
  * Main function that will be run by the scheduler.
  *
@@ -353,11 +506,7 @@
 run (void *cls, char *const *args, const char *cfgfile,
      const struct GNUNET_CONFIGURATION_Handle *c)
 {
-  struct GNUNET_FS_FileInformation *fi;
-  struct GNUNET_FS_Namespace *namespace;
   struct EXTRACTOR_PluginList *plugins;
-  struct GNUNET_FS_Uri *keywords;
-  struct stat sbuf;
   char *ex;
   char *emsg;
 
@@ -451,6 +600,9 @@
       GNUNET_FS_namespace_delete (namespace, GNUNET_NO);
     return;
   }
+  ds = GNUNET_FS_directory_scan_start (args[0],
+      GNUNET_NO, NULL, directory_scan_cb, NULL);
+ 
   plugins = NULL;
   if (!disable_extractor)
   {
@@ -465,63 +617,6 @@
       GNUNET_free (ex);
     }
   }
-  emsg = NULL;
-  GNUNET_assert (NULL != args[0]);
-  if (0 != STAT (args[0], &sbuf))
-  {
-    GNUNET_asprintf (&emsg, _("Could not access file: %s\n"), STRERROR 
(errno));
-    fi = NULL;
-  }
-  else if (S_ISDIR (sbuf.st_mode))
-  {
-    fi = GNUNET_FS_file_information_create_from_directory (ctx, NULL, args[0],
-                                                           
&GNUNET_FS_directory_scanner_default,
-                                                           plugins, !do_insert,
-                                                           &bo, &emsg);
-  }
-  else
-  {
-    if (meta == NULL)
-      meta = GNUNET_CONTAINER_meta_data_create ();
-    GNUNET_FS_meta_data_extract_from_file (meta, args[0], plugins);
-    keywords = GNUNET_FS_uri_ksk_create_from_meta_data (meta);
-    fi = GNUNET_FS_file_information_create_from_file (ctx, NULL, args[0],
-                                                      keywords, NULL,
-                                                      !do_insert, &bo);
-    GNUNET_break (fi != NULL);
-    GNUNET_FS_uri_destroy (keywords);
-  }
-  EXTRACTOR_plugin_remove_all (plugins);
-  if (fi == NULL)
-  {
-    FPRINTF (stderr, _("Could not publish `%s': %s\n"), args[0], emsg);
-    GNUNET_free (emsg);
-    if (namespace != NULL)
-      GNUNET_FS_namespace_delete (namespace, GNUNET_NO);
-    GNUNET_FS_stop (ctx);
-    ret = 1;
-    return;
-  }
-  GNUNET_FS_file_information_inspect (fi, &publish_inspector, NULL);
-  if (extract_only)
-  {
-    if (namespace != NULL)
-      GNUNET_FS_namespace_delete (namespace, GNUNET_NO);
-    GNUNET_FS_file_information_destroy (fi, NULL, NULL);
-    GNUNET_FS_stop (ctx);
-    return;
-  }
-  pc = GNUNET_FS_publish_start (ctx, fi, namespace, this_id, next_id,
-                                (do_simulate) ?
-                                GNUNET_FS_PUBLISH_OPTION_SIMULATE_ONLY :
-                                GNUNET_FS_PUBLISH_OPTION_NONE);
-  if (NULL == pc)
-  {
-    FPRINTF (stderr, "%s",  _("Could not start publishing.\n"));
-    GNUNET_FS_stop (ctx);
-    ret = 1;
-    return;
-  }
   kill_task =
       GNUNET_SCHEDULER_add_delayed (GNUNET_TIME_UNIT_FOREVER_REL, 
&do_stop_task,
                                     NULL);