implementing removal of KBlocks during unindex operation (#1926)

3 files changed, 441 insertions, 29 deletions

diff --git a/src/fs/fs_api.c b/src/fs/fs_api.c
index d718acb65..29539a498 100644
--- a/src/fs/fs_api.c
+++ b/src/fs/fs_api.c
@@ -1524,6 +1524,7 @@ void
 GNUNET_FS_unindex_sync_ (struct GNUNET_FS_UnindexContext *uc)
 {
   struct GNUNET_BIO_WriteHandle *wh;
+  char *uris;
 
   if (NULL == uc->serialization)
     uc->serialization =
@@ -1538,10 +1539,15 @@ GNUNET_FS_unindex_sync_ (struct GNUNET_FS_UnindexContext *uc)
     GNUNET_break (0);
     goto cleanup;
   }
+  uris = GNUNET_FS_uri_to_string (uc->ksk_uri);
   if ((GNUNET_OK != GNUNET_BIO_write_string (wh, uc->filename)) ||
       (GNUNET_OK != GNUNET_BIO_write_int64 (wh, uc->file_size)) ||
       (GNUNET_OK != write_start_time (wh, uc->start_time)) ||
       (GNUNET_OK != GNUNET_BIO_write_int32 (wh, (uint32_t) uc->state)) ||
+      (GNUNET_OK !=
+       GNUNET_BIO_write (wh, &uc->chk, sizeof (struct ContentHashKey))) ||
+      (GNUNET_OK != GNUNET_BIO_write_string (wh, uris)) ||
+      (GNUNET_OK != GNUNET_BIO_write_int32 (wh, (uint32_t) uc->ksk_offset)) ||
       ((uc->state == UNINDEX_STATE_FS_NOTIFY) &&
        (GNUNET_OK !=
         GNUNET_BIO_write (wh, &uc->file_id, sizeof (GNUNET_HashCode)))) ||
@@ -1964,6 +1970,7 @@ deserialize_unindex_file (void *cls, const char *filename)
   struct GNUNET_FS_UnindexContext *uc;
   struct GNUNET_FS_ProgressInfo pi;
   char *emsg;
+  char *uris;
   uint32_t state;
 
   uc = GNUNET_malloc (sizeof (struct GNUNET_FS_UnindexContext));
@@ -1979,11 +1986,31 @@ deserialize_unindex_file (void *cls, const char *filename)
        GNUNET_BIO_read_string (rh, "unindex-fn", &uc->filename, 10 * 1024)) ||
       (GNUNET_OK != GNUNET_BIO_read_int64 (rh, &uc->file_size)) ||
       (GNUNET_OK != read_start_time (rh, &uc->start_time)) ||
-      (GNUNET_OK != GNUNET_BIO_read_int32 (rh, &state)))
+      (GNUNET_OK != GNUNET_BIO_read_int32 (rh, &state)) ||
+      (GNUNET_OK != GNUNET_BIO_read (rh, "uri", &uc->chk, sizeof (struct ContentHashKey))) ||
+      (GNUNET_BIO_read_string (rh, "unindex-kskuri", &uris, 10 * 1024)) ||
+      (GNUNET_OK != GNUNET_BIO_read_int32 (rh, &uc->ksk_offset)) )
   {
     GNUNET_break (0);
     goto cleanup;
   }
+  if (NULL != uris)
+  {
+    uc->ksk_uri = GNUNET_FS_uri_parse (uris, &emsg);
+    GNUNET_free (uris);
+    if (NULL == uc->ksk_uri)
+    {
+      GNUNET_break (0);
+      goto cleanup;
+    }
+  }
+  if ( (uc->ksk_offset > 0) &&
+       ( (NULL == uc->ksk_uri) ||
+         (uc->ksk_offset > uc->ksk_uri->data.ksk.keywordCount) ) )
+  {
+    GNUNET_break (0);
+    goto cleanup;
+  }  
   uc->state = (enum UnindexState) state;
   switch (state)
   {
@@ -1999,6 +2026,8 @@ deserialize_unindex_file (void *cls, const char *filename)
     }
     break;
   case UNINDEX_STATE_DS_REMOVE:
+  case UNINDEX_STATE_EXTRACT_KEYWORDS:
+  case UNINDEX_STATE_DS_REMOVE_KBLOCKS:
     break;
   case UNINDEX_STATE_COMPLETE:
     break;
@@ -2035,6 +2064,12 @@ deserialize_unindex_file (void *cls, const char *filename)
   case UNINDEX_STATE_DS_REMOVE:
     GNUNET_FS_unindex_do_remove_ (uc);
     break;
+  case UNINDEX_STATE_EXTRACT_KEYWORDS:
+    GNUNET_FS_unindex_do_extract_keywords_ (uc);
+    break;
+  case UNINDEX_STATE_DS_REMOVE_KBLOCKS:
+    GNUNET_FS_unindex_do_remove_kblocks_ (uc);
+    break;
   case UNINDEX_STATE_COMPLETE:
   case UNINDEX_STATE_ERROR:
     /* no need to resume any operation, we were done */
diff --git a/src/fs/fs_api.h b/src/fs/fs_api.h
index 1d18e20bc..924911b63 100644
--- a/src/fs/fs_api.h
+++ b/src/fs/fs_api.h
@@ -734,6 +734,24 @@ GNUNET_FS_unindex_process_hash_ (void *cls, const GNUNET_HashCode * file_id);
 
 
 /**
+ * Extract the keywords for KBlock removal
+ *
+ * @param uc context for the unindex operation.
+ */
+void
+GNUNET_FS_unindex_do_extract_keywords_ (struct GNUNET_FS_UnindexContext *uc);
+
+
+/**
+ * If necessary, connect to the datastore and remove the KBlocks.
+ *
+ * @param uc context for the unindex operation.
+ */
+void
+GNUNET_FS_unindex_do_remove_kblocks_ (struct GNUNET_FS_UnindexContext *uc);
+
+
+/**
  * Fill in all of the generic fields for a publish event and call the
  * callback.
  *
@@ -1244,32 +1262,42 @@ struct GNUNET_FS_PublishContext
  */
 enum UnindexState
 {
-    /**
-     * We're currently hashing the file.
-     */
+  /**
+   * We're currently hashing the file.
+   */
   UNINDEX_STATE_HASHING = 0,
 
-    /**
-     * We're telling the datastore to delete
-     * the respective entries.
-     */
+  /**
+   * We're telling the datastore to delete
+   * the respective DBlocks and IBlocks.
+   */
   UNINDEX_STATE_DS_REMOVE = 1,
+  
+  /**
+   * Find out which keywords apply.
+   */
+  UNINDEX_STATE_EXTRACT_KEYWORDS = 2,
 
-    /**
-     * We're notifying the FS service about
-     * the unindexing.
-     */
-  UNINDEX_STATE_FS_NOTIFY = 2,
-
-    /**
-     * We're done.
-     */
-  UNINDEX_STATE_COMPLETE = 3,
+  /**
+   * We're telling the datastore to remove KBlocks.
+   */
+  UNINDEX_STATE_DS_REMOVE_KBLOCKS = 3,
 
-    /**
-     * We've encountered a fatal error.
-     */
-  UNINDEX_STATE_ERROR = 4
+  /**
+   * We're notifying the FS service about
+   * the unindexing.
+   */
+  UNINDEX_STATE_FS_NOTIFY = 4,
+  
+  /**
+   * We're done.
+   */
+  UNINDEX_STATE_COMPLETE = 5,
+  
+  /**
+   * We've encountered a fatal error.
+   */
+  UNINDEX_STATE_ERROR = 6
 };
 
 
@@ -1280,6 +1308,12 @@ struct GNUNET_FS_UnindexContext
 {
 
   /**
+   * The content hash key of the last block we processed, will in the
+   * end be set to the CHK from the URI.  Used to remove the KBlocks.
+   */
+  struct ContentHashKey chk; 
+
+  /**
    * Global FS context.
    */
   struct GNUNET_FS_Handle *h;
@@ -1290,6 +1324,21 @@ struct GNUNET_FS_UnindexContext
   struct TopLevelActivity *top;
 
   /**
+   * Directory scanner to find keywords (KBlock removal).
+   */
+  struct GNUNET_FS_DirScanner *dscan;
+
+  /**
+   * Keywords found (telling us which KBlocks to remove).
+   */
+  struct GNUNET_FS_Uri *ksk_uri;
+
+  /**
+   * Current offset in KSK removal.
+   */
+  uint32_t ksk_offset;
+
+  /**
    * Name of the file that we are unindexing.
    */
   char *filename;
@@ -1327,6 +1376,22 @@ struct GNUNET_FS_UnindexContext
   struct GNUNET_DISK_FileHandle *fh;
 
   /**
+   * Handle to datastore 'get_key' operation issued for
+   * obtaining KBlocks.
+   */
+  struct GNUNET_DATASTORE_QueueEntry *dqe;
+
+  /**
+   * Current query key of 'get_key' operation.
+   */
+  GNUNET_HashCode key;
+
+  /**
+   * First content UID, 0 for none.
+   */
+  uint64_t first_uid;
+
+  /**
    * Error message, NULL on success.
    */
   char *emsg;
@@ -1342,6 +1407,11 @@ struct GNUNET_FS_UnindexContext
   uint64_t file_size;
 
   /**
+   * Random offset given to 'GNUNET_DATASTORE_get_key'.
+   */
+  uint64_t roff;
+
+  /**
    * When did we start?
    */
   struct GNUNET_TIME_Absolute start_time;
diff --git a/src/fs/fs_unindex.c b/src/fs/fs_unindex.c
index ff1996a2f..07b9cccd3 100644
--- a/src/fs/fs_unindex.c
+++ b/src/fs/fs_unindex.c
@@ -30,6 +30,7 @@
 #include "gnunet_protocols.h"
 #include "fs_api.h"
 #include "fs_tree.h"
+#include "block_fs.h"
 
 
 /**
@@ -203,6 +204,7 @@ unindex_process (void *cls, const struct ContentHashKey *chk, uint64_t offset,
               "Sending REMOVE request to DATASTORE service\n");
   GNUNET_DATASTORE_remove (uc->dsh, &chk->query, size, data, -2, 1,
                            GNUNET_CONSTANTS_SERVICE_TIMEOUT, &process_cont, uc);
+  uc->chk = *chk;
 }
 
 
@@ -258,16 +260,15 @@ process_fs_response (void *cls, const struct GNUNET_MessageHeader *msg)
 
 
 /**
- * Function called when the tree encoder has
- * processed all blocks.  Clean up.
+ * Function called when we are done with removing KBlocks.
+ * Disconnect from datastore and notify FS service about
+ * the unindex event.
  *
- * @param cls our unindexing context
- * @param tc not used
+ * @param uc our unindexing context
  */
 static void
-unindex_finish (void *cls, const struct GNUNET_SCHEDULER_TaskContext *tc)
+unindex_finish (struct GNUNET_FS_UnindexContext *uc)
 {
-  struct GNUNET_FS_UnindexContext *uc = cls;
   char *emsg;
   struct GNUNET_FS_Uri *uri;
   struct UnindexMessage req;
@@ -310,6 +311,281 @@ unindex_finish (void *cls, const struct GNUNET_SCHEDULER_TaskContext *tc)
 }
 
 
+
+/**
+ * Function called by the directory scanner as we extract keywords
+ * that we will need to remove KBlocks.
+ *
+ * @param cls the 'struct GNUNET_FS_UnindexContext *'
+ * @param filename which file we are making progress on
+ * @param is_directory GNUNET_YES if this is a directory,
+ *                     GNUNET_NO if this is a file
+ *                     GNUNET_SYSERR if it is neither (or unknown)
+ * @param reason kind of progress we are making
+ */
+static void
+unindex_directory_scan_cb (void *cls, 
+                           const char *filename, 
+                           int is_directory,
+                           enum GNUNET_FS_DirScannerProgressUpdateReason reason)
+{
+  struct GNUNET_FS_UnindexContext *uc = cls;
+  static struct GNUNET_FS_ShareTreeItem * directory_scan_result;
+
+  switch (reason)
+  {
+  case GNUNET_FS_DIRSCANNER_FINISHED:
+    directory_scan_result = GNUNET_FS_directory_scan_get_result (uc->dscan);
+    uc->dscan = NULL;
+    if (NULL != directory_scan_result->ksk_uri)
+    {
+      uc->ksk_uri = GNUNET_FS_uri_dup (directory_scan_result->ksk_uri);
+      uc->state = UNINDEX_STATE_DS_REMOVE_KBLOCKS;
+      uc->emsg = GNUNET_strdup (_("Failed to connect to `datastore' service."));
+      GNUNET_FS_unindex_sync_ (uc);
+      GNUNET_FS_unindex_do_remove_kblocks_ (uc);
+    }
+    else
+    {
+      unindex_finish (uc);
+    }
+    GNUNET_FS_share_tree_free (directory_scan_result);
+    break;
+  case GNUNET_FS_DIRSCANNER_INTERNAL_ERROR:
+    GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
+                _("Internal error scanning `%s'.\n"),
+                uc->filename);
+    break;
+  default:
+    break;
+  }
+
+}
+
+
+/**
+ * If necessary, connect to the datastore and remove the KBlocks.
+ *
+ * @param uc context for the unindex operation.
+ */
+void
+GNUNET_FS_unindex_do_extract_keywords_ (struct GNUNET_FS_UnindexContext *uc)
+{
+  char *ex;
+
+  if (GNUNET_OK !=
+      GNUNET_CONFIGURATION_get_value_string (uc->h->cfg, "FS", "EXTRACTORS", &ex))
+    ex = NULL;
+  uc->dscan = GNUNET_FS_directory_scan_start (uc->filename,
+                                              GNUNET_NO, ex,
+                                              &unindex_directory_scan_cb, 
+                                              uc);
+  GNUNET_free_non_null (ex);
+}
+
+
+/**
+ * Continuation called to notify client about result of the remove
+ * operation for the KBlock.
+ *
+ * @param cls the 'struct GNUNET_FS_UnindexContext *' 
+ * @param success GNUNET_SYSERR on failure (including timeout/queue drop)
+ *                GNUNET_NO if content was already there
+ *                GNUNET_YES (or other positive value) on success
+ * @param min_expiration minimum expiration time required for 0-priority content to be stored
+ *                by the datacache at this time, zero for unknown, forever if we have no
+ *                space for 0-priority content
+ * @param msg NULL on success, otherwise an error message
+ */
+static void
+continue_after_remove (void *cls,
+                       int32_t success,
+                       struct GNUNET_TIME_Absolute min_expiration,
+                       const char *msg)
+{
+  struct GNUNET_FS_UnindexContext *uc = cls;
+
+  if (success != GNUNET_YES)  
+    GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
+                _("Failed to remove KBlock: %s\n"),
+                msg);  
+  uc->ksk_offset++;
+  GNUNET_FS_unindex_do_remove_kblocks_ (uc);
+}
+
+
+/**
+ * Function called from datastore with result from us looking for
+ * a KBlock.  There are four cases:
+ * 1) no result, means we move on to the next keyword
+ * 2) UID is the same as the first UID, means we move on to next keyword
+ * 3) KBlock for a different CHK, means we keep looking for more
+ * 4) KBlock is for our CHK, means we remove the block and then move
+ *           on to the next keyword
+ *
+ * @param cls the 'struct GNUNET_FS_UnindexContext *'
+ * @param key key for the content
+ * @param size number of bytes in data
+ * @param data content stored
+ * @param type type of the content
+ * @param priority priority of the content
+ * @param anonymity anonymity-level for the content
+ * @param expiration expiration time for the content
+ * @param uid unique identifier for the datum;
+ *        maybe 0 if no unique identifier is available
+ */
+static void
+process_kblock_for_unindex (void *cls,
+                            const GNUNET_HashCode * key,
+                            size_t size, const void *data,
+                            enum GNUNET_BLOCK_Type type,
+                            uint32_t priority,
+                            uint32_t anonymity,
+                            struct GNUNET_TIME_Absolute
+                            expiration, uint64_t uid)
+{
+  struct GNUNET_FS_UnindexContext *uc = cls;
+  const struct KBlock *kb;
+  const char *uris;
+  struct GNUNET_FS_Uri *chk_uri;
+
+  uc->dqe = NULL;
+  if (NULL == data)
+  {
+    /* no result */
+    uc->ksk_offset++;
+    GNUNET_FS_unindex_do_remove_kblocks_ (uc);
+    return;
+  }
+  if (0 == uc->first_uid)
+  {
+    /* remember UID of first result to detect cycles */
+    uc->first_uid = uid;    
+  }
+  else if (uid == uc->first_uid)
+  {
+    /* no more additional results */
+    uc->ksk_offset++;
+    GNUNET_FS_unindex_do_remove_kblocks_ (uc);
+    return;  
+  }
+  GNUNET_assert (GNUNET_BLOCK_TYPE_FS_KBLOCK == type);
+  if (size < sizeof (struct KBlock))
+  {
+    GNUNET_break (0);
+    goto get_next;
+  }
+  kb = data;
+  uris = (const char*) &kb[1];
+  if (NULL == memchr (uris, 0, size - sizeof (struct KBlock)))
+  {
+    GNUNET_break (0);
+    goto get_next;
+  }
+  chk_uri = GNUNET_FS_uri_parse (uris, NULL);
+  if (NULL == chk_uri)
+  {
+    GNUNET_break (0);
+    goto get_next;
+  }
+  if (0 != memcmp (&uc->chk,
+                   &chk_uri->data.chk.chk,
+                   sizeof (struct ContentHashKey)))
+  {
+    /* different CHK, ignore */
+    GNUNET_FS_uri_destroy (chk_uri);
+    goto get_next;
+  }
+  /* matches! */
+  uc->dqe = GNUNET_DATASTORE_remove (uc->dsh,
+                                     key, size, data,
+                                     0 /* priority */, 1 /* queue size */,
+                                     GNUNET_TIME_UNIT_FOREVER_REL,
+                                     &continue_after_remove,
+                                     uc);
+  return;
+ get_next:
+  uc->dqe = GNUNET_DATASTORE_get_key (uc->dsh,
+                                      uc->roff++,
+                                      &uc->key,
+                                      GNUNET_BLOCK_TYPE_FS_KBLOCK,
+                                      0 /* priority */, 1 /* queue size */,
+                                      GNUNET_TIME_UNIT_FOREVER_REL,
+                                      &process_kblock_for_unindex,
+                                      uc);
+}
+
+
+/**
+ * If necessary, connect to the datastore and remove the KBlocks.
+ *
+ * @param uc context for the unindex operation.
+ */
+void
+GNUNET_FS_unindex_do_remove_kblocks_ (struct GNUNET_FS_UnindexContext *uc)
+{
+  const char *keyword;
+  GNUNET_HashCode hc;
+  struct GNUNET_CRYPTO_RsaPublicKeyBinaryEncoded pub;
+  struct GNUNET_CRYPTO_RsaPrivateKey *pk;
+
+  if (NULL != uc->dsh)
+    uc->dsh = GNUNET_DATASTORE_connect (uc->h->cfg);
+  if (NULL == uc->dsh)
+  {
+    uc->state = UNINDEX_STATE_ERROR;
+    uc->emsg = GNUNET_strdup (_("Failed to connect to `datastore' service."));
+    GNUNET_FS_unindex_sync_ (uc);
+    signal_unindex_error (uc);
+    return;
+  }
+  if ( (NULL == uc->ksk_uri) ||
+       (uc->ksk_offset >= uc->ksk_uri->data.ksk.keywordCount) )
+  {
+    unindex_finish (uc);
+    return;
+  }
+  /* FIXME: code duplication with fs_search.c here... */
+  keyword = &uc->ksk_uri->data.ksk.keywords[uc->ksk_offset][1];
+  GNUNET_CRYPTO_hash (keyword, strlen (keyword), &hc);
+  pk = GNUNET_CRYPTO_rsa_key_create_from_hash (&hc);
+  GNUNET_assert (pk != NULL);
+  GNUNET_CRYPTO_rsa_key_get_public (pk, &pub);
+  GNUNET_CRYPTO_rsa_key_free (pk);
+  GNUNET_CRYPTO_hash (&pub,
+                      sizeof (struct
+                              GNUNET_CRYPTO_RsaPublicKeyBinaryEncoded),
+                      &uc->key);
+  uc->first_uid = 0;
+  uc->dqe = GNUNET_DATASTORE_get_key (uc->dsh,
+                                      uc->roff++,
+                                      &uc->key,
+                                      GNUNET_BLOCK_TYPE_FS_KBLOCK,
+                                      0 /* priority */, 1 /* queue size */,
+                                      GNUNET_TIME_UNIT_FOREVER_REL,
+                                      &process_kblock_for_unindex,
+                                      uc);
+}
+
+
+/**
+ * Function called when the tree encoder has
+ * processed all blocks.  Clean up.
+ *
+ * @param cls our unindexing context
+ * @param tc not used
+ */
+static void
+unindex_extract_keywords (void *cls, const struct GNUNET_SCHEDULER_TaskContext *tc)
+{
+  struct GNUNET_FS_UnindexContext *uc = cls;
+
+  uc->state = UNINDEX_STATE_EXTRACT_KEYWORDS;
+  GNUNET_FS_unindex_sync_ (uc);
+  GNUNET_FS_unindex_do_extract_keywords_ (uc);
+}
+
+
 /**
  * Connect to the datastore and remove the blocks.
  *
@@ -343,7 +619,7 @@ GNUNET_FS_unindex_do_remove_ (struct GNUNET_FS_UnindexContext *uc)
   uc->tc =
       GNUNET_FS_tree_encoder_create (uc->h, uc->file_size, uc, &unindex_reader,
                                      &unindex_process, &unindex_progress,
-                                     &unindex_finish);
+                                     &unindex_extract_keywords);
   GNUNET_FS_tree_encoder_next (uc->tc);
 }
 
@@ -393,11 +669,27 @@ GNUNET_FS_unindex_signal_suspend_ (void *cls)
   struct GNUNET_FS_UnindexContext *uc = cls;
   struct GNUNET_FS_ProgressInfo pi;
 
+  /* FIXME: lots of duplication with unindex_stop here! */
+  if (uc->dscan != NULL)
+  {
+    GNUNET_FS_directory_scan_abort (uc->dscan);
+    uc->dscan = NULL;
+  }
+  if (NULL != uc->dqe)
+  {
+    GNUNET_DATASTORE_cancel (uc->dqe);
+    uc->dqe = NULL;
+  }
   if (uc->fhc != NULL)
   {
     GNUNET_CRYPTO_hash_file_cancel (uc->fhc);
     uc->fhc = NULL;
   }
+  if (NULL != uc->ksk_uri)
+  {
+    GNUNET_FS_uri_destroy (uc->ksk_uri);
+    uc->ksk_uri = NULL;
+  }
   if (uc->client != NULL)
   {
     GNUNET_CLIENT_disconnect (uc->client, GNUNET_NO);
@@ -478,6 +770,16 @@ GNUNET_FS_unindex_stop (struct GNUNET_FS_UnindexContext *uc)
 {
   struct GNUNET_FS_ProgressInfo pi;
 
+  if (uc->dscan != NULL)
+  {
+    GNUNET_FS_directory_scan_abort (uc->dscan);
+    uc->dscan = NULL;
+  }
+  if (NULL != uc->dqe)
+  {
+    GNUNET_DATASTORE_cancel (uc->dqe);
+    uc->dqe = NULL;
+  }
   if (uc->fhc != NULL)
   {
     GNUNET_CRYPTO_hash_file_cancel (uc->fhc);
@@ -493,6 +795,11 @@ GNUNET_FS_unindex_stop (struct GNUNET_FS_UnindexContext *uc)
     GNUNET_DATASTORE_disconnect (uc->dsh, GNUNET_NO);
     uc->dsh = NULL;
   }
+  if (NULL != uc->ksk_uri)
+  {
+    GNUNET_FS_uri_destroy (uc->ksk_uri);
+    uc->ksk_uri = NULL;
+  }
   if (NULL != uc->tc)
   {
     GNUNET_FS_tree_encoder_finish (uc->tc, NULL, NULL);