Add H5Dchunk_iter method for iterating over chunks (#6)

* Add H5Dchunk_iter method for iterating over chunks

This method iterates over all chunks in dataset, calling a user-supplied
callback with the chunk information and optional user supplied data.

The iterator is stopped when ITER_STOP is returned by the user-supplied
callback or the iterator is exhausted.

Existing methods to get chunk_info performs an iteration each time, so
to get many or all chunks causes SUM(i) for i = 0 -> N operations for N
chunks, as opposed to N operations when using this iterator for this use case.

* H5Dchunk_iter: test iterating all chunks, some chunks and failing iteration.

* H5D: move H5Dchunk_iter private methods to specific

* trace: add H5D_chunk_iter_op_t and trace H5D.c

* chunks-iter: document chunk_iter

* chunk-iter: chunk add FUNC_ENTER/FUNC_LEAVE macros

* Committing clang-format changes

Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>

1 files changed, 98 insertions, 0 deletions

diff --git a/src/H5Dchunk.c b/src/H5Dchunk.c
index 25fb2f8..6be29fd 100644
--- a/src/H5Dchunk.c
+++ b/src/H5Dchunk.c
@@ -245,6 +245,11 @@ typedef struct H5D_chunk_coll_info_t {
 } H5D_chunk_coll_info_t;
 #endif /* H5_HAVE_PARALLEL */
 
+typedef struct H5D_chunk_iter_cb_data_t {
+    H5D_chunk_iter_op_t cb;      /* User defined callback */
+    void *              op_data; /* User data for user defined callback */
+} H5D_chunk_iter_cb_data_t;
+
 /********************/
 /* Local Prototypes */
 /********************/
@@ -269,6 +274,7 @@ static herr_t H5D__chunk_dest(H5D_t *dset);
 static int H5D__get_num_chunks_cb(const H5D_chunk_rec_t *chunk_rec, void *_udata);
 static int H5D__get_chunk_info_cb(const H5D_chunk_rec_t *chunk_rec, void *_udata);
 static int H5D__get_chunk_info_by_coord_cb(const H5D_chunk_rec_t *chunk_rec, void *_udata);
+static int H5D__chunk_iter_cb(const H5D_chunk_rec_t *chunk_rec, void *udata);
 
 /* "Nonexistent" layout operation callback */
 static ssize_t H5D__nonexistent_readvv(const H5D_io_info_t *io_info, size_t chunk_max_nseq,
@@ -7451,3 +7457,95 @@ H5D__get_chunk_info_by_coord(const H5D_t *dset, const hsize_t *offset, unsigned
 done:
     FUNC_LEAVE_NOAPI_TAG(ret_value)
 } /* end H5D__get_chunk_info_by_coord() */
+
+/*-------------------------------------------------------------------------
+ * Function:    H5D__chunk_iter
+ *
+ * Purpose:     Iterate over all the chunks in the dataset with given callbak.
+ *
+ * Return:      Success:        Non-negative
+ *              Failure:        Negative
+ *
+ * Programmer:  Gaute Hope
+ *              August 2020
+ *
+ *-------------------------------------------------------------------------
+ */
+herr_t
+H5D__chunk_iter(const H5D_t *dset, H5D_chunk_iter_op_t cb, void *op_data)
+{
+    const H5O_layout_t *layout = NULL;       /* Dataset layout */
+    const H5D_rdcc_t *  rdcc   = NULL;       /* Raw data chunk cache */
+    H5D_rdcc_ent_t *    ent;                 /* Cache entry index */
+    H5D_chk_idx_info_t  idx_info;            /* Chunked index info */
+    herr_t              ret_value = SUCCEED; /* Return value */
+
+    FUNC_ENTER_PACKAGE_TAG(dset->oloc.addr)
+
+    /* Check args */
+    HDassert(dset);
+    HDassert(dset->shared);
+
+    /* Get dataset layout and raw data chunk cache */
+    layout = &(dset->shared->layout);
+    rdcc   = &(dset->shared->cache.chunk);
+    HDassert(layout);
+    HDassert(rdcc);
+    HDassert(H5D_CHUNKED == layout->type);
+
+    /* Search for cached chunks that haven't been written out */
+    for (ent = rdcc->head; ent; ent = ent->next)
+        /* Flush the chunk out to disk, to make certain the size is correct later */
+        if (H5D__chunk_flush_entry(dset, ent, FALSE) < 0)
+            HGOTO_ERROR(H5E_IO, H5E_WRITEERROR, FAIL, "cannot flush indexed storage buffer")
+
+    /* Compose chunked index info struct */
+    idx_info.f       = dset->oloc.file;
+    idx_info.pline   = &dset->shared->dcpl_cache.pline;
+    idx_info.layout  = &dset->shared->layout.u.chunk;
+    idx_info.storage = &dset->shared->layout.storage.u.chunk;
+
+    /* If the dataset is not written, return without errors */
+    if (H5F_addr_defined(idx_info.storage->idx_addr)) {
+        H5D_chunk_iter_cb_data_t data;
+        data.cb      = cb;
+        data.op_data = op_data;
+
+        /* Iterate over the allocated chunks calling the iterator callback */
+        if ((dset->shared->layout.storage.u.chunk.ops->iterate)(&idx_info, H5D__chunk_iter_cb, &data) < 0)
+            HGOTO_ERROR(H5E_DATASET, H5E_CANTGET, FAIL, "unable to iterate over chunks.")
+    } /* end if H5F_addr_defined */
+
+done:
+    FUNC_LEAVE_NOAPI_TAG(ret_value)
+} /* end H5D__chunk_iter() */
+
+/*-------------------------------------------------------------------------
+ * Function:    H5D__chunk_iter_cb
+ *
+ * Purpose:     Call the user-defined function with the chunk data. The iterator continues if
+ *              the user-defined function returns H5_ITER_CONT, and stops if H5_ITER_STOP is
+ *              returned.
+ *
+ * Return:      Success:    H5_ITER_CONT or H5_ITER_STOP
+ *              Failure:    Negative (H5_ITER_ERROR)
+ *
+ * Programmer:  Gaute Hope
+ *              August 2020
+ *
+ *-------------------------------------------------------------------------
+ */
+static int
+H5D__chunk_iter_cb(const H5D_chunk_rec_t *chunk_rec, void *udata)
+{
+    int ret_value = 0;
+
+    FUNC_ENTER_STATIC_NOERR
+
+    const H5D_chunk_iter_cb_data_t *data = (H5D_chunk_iter_cb_data_t *)udata;
+
+    ret_value = (data->cb)(chunk_rec->scaled, chunk_rec->filter_mask, chunk_rec->chunk_addr,
+                           chunk_rec->nbytes, data->op_data);
+
+    FUNC_LEAVE_NOAPI(ret_value)
+} /* end H5D__chunk_iter_cb */