diff options
author | Pedro Vicente Nunes <pvn@hdfgroup.org> | 2006-11-08 21:34:21 (GMT) |
---|---|---|
committer | Pedro Vicente Nunes <pvn@hdfgroup.org> | 2006-11-08 21:34:21 (GMT) |
commit | 50c658d66d015d3630e46b6b30bcf5c67613cf8a (patch) | |
tree | 26a6f86fda7cd5fedd80833fcf62d518e5f470af /tools/h5repack/h5repack_copy.c | |
parent | 89cca34b9763e12a73a86164306d5ffbede28f64 (diff) | |
download | hdf5-50c658d66d015d3630e46b6b30bcf5c67613cf8a.zip hdf5-50c658d66d015d3630e46b6b30bcf5c67613cf8a.tar.gz hdf5-50c658d66d015d3630e46b6b30bcf5c67613cf8a.tar.bz2 |
[svn-r12884]
h5repack support for H5Ocopy in the copy of objects. The old method
for recreating references was dropped (references recreated in a second
traversal of the file)
The logic for using H5Ocopy or not is
if the input DCPL has filters or non default layout OR these are
requested by the user THEN
use the old h5repack read / write
ELSE
use H5Ocopy
Diffstat (limited to 'tools/h5repack/h5repack_copy.c')
-rw-r--r-- | tools/h5repack/h5repack_copy.c | 604 |
1 files changed, 368 insertions, 236 deletions
diff --git a/tools/h5repack/h5repack_copy.c b/tools/h5repack/h5repack_copy.c index 03edd01..6b3e4d0 100644 --- a/tools/h5repack/h5repack_copy.c +++ b/tools/h5repack/h5repack_copy.c @@ -33,6 +33,8 @@ extern char *progname; #define FORMAT_OBJ_ATTR " %-21s %s\n" /* obj type, name */ +static +int do_hardlinks(hid_t fidout,trav_table_t *travt); /*------------------------------------------------------------------------- * Function: print_dataset_info @@ -148,12 +150,8 @@ static void print_dataset_info(hid_t dcpl_id, strcat(str,temp); printf(FORMAT_OBJ,str,objname); } - - - } - /*------------------------------------------------------------------------- * Function: copy_objects * @@ -209,11 +207,11 @@ int copy_objects(const char* fnamein, } /*------------------------------------------------------------------------- - * do the copy of referenced objects - * and create hard links + * create hard links *------------------------------------------------------------------------- */ - if(do_copy_refobjs(fidin,fidout,travt,options)<0) { + + if(do_hardlinks(fidout,travt)<0) { error_msg(progname, "<%s>: Could not copy data to: %s\n", fnamein, fnameout); goto out; } @@ -299,6 +297,8 @@ int do_copy_objects(hid_t fidin, void *buf=NULL; /* buffer for raw data */ void *sm_buf=NULL; /* buffer for raw data */ int i, j; + int nfilters; /* number of filters in DCPL */ + H5D_layout_t layout; /*------------------------------------------------------------------------- * copy the suppplied object list @@ -351,268 +351,337 @@ int do_copy_objects(hid_t fidin, *------------------------------------------------------------------------- */ case H5G_DATASET: - + +/*------------------------------------------------------------------------- + * early check for filters or a non default layout + *------------------------------------------------------------------------- + */ if ((dset_in=H5Dopen(fidin,travt->objs[i].name))<0) goto error; - if ((f_space_id=H5Dget_space(dset_in))<0) - goto error; - if ((ftype_id=H5Dget_type (dset_in))<0) - goto error; if ((dcpl_id=H5Dget_create_plist(dset_in))<0) + goto error; + + /* get information about input filters */ + if ((nfilters = H5Pget_nfilters(dcpl_id))<0) goto error; - if ((dcpl_out = H5Pcopy (dcpl_id))<0) - goto error; - if ( (rank=H5Sget_simple_extent_ndims(f_space_id))<0) - goto error; - HDmemset(dims, 0, sizeof dims); - if ( H5Sget_simple_extent_dims(f_space_id,dims,NULL)<0) - goto error; - nelmts=1; - for (j=0; j<rank; j++) - nelmts*=dims[j]; - - if (options->use_native==1) - wtype_id = h5tools_get_native_type(ftype_id); - else - wtype_id = H5Tcopy(ftype_id); - - if ((msize=H5Tget_size(wtype_id))==0) - goto error; - -/*------------------------------------------------------------------------- - * check for external files - *------------------------------------------------------------------------- - */ - if ((next=H5Pget_external_count (dcpl_id))<0) + if ((layout = H5Pget_layout(dcpl_id))<0) goto error; - if (next) - fprintf(stderr," <warning: %s has external files, ignoring read...>\n", - travt->objs[i].name ); - -/*------------------------------------------------------------------------- - * check if the dataset creation property list has filters that - * are not registered in the current configuration - * 1) the external filters GZIP and SZIP might not be available - * 2) the internal filters might be turned off - *------------------------------------------------------------------------- - */ - if (next==0 && h5tools_canreadf((travt->objs[i].name),dcpl_id)==1) - { - apply_s=1; - apply_f=1; - + if (H5Dclose(dset_in)<0) + goto error; + if (H5Pclose(dcpl_id)<0) + goto error; + /*------------------------------------------------------------------------- - * references are a special case - * we cannot just copy the buffers, but instead we recreate the reference - * in a second traversal of the output file + * check if we should use H5Ocopy or not + * if the DCPL has filters or a non default layout, we read/write the object + * otherwise we do a copy using H5Ocopy *------------------------------------------------------------------------- */ - if ( (H5T_REFERENCE!=H5Tget_class(wtype_id))) + + if (nfilters || layout!=H5D_CONTIGUOUS + /* does the dcpl has filters or non default layout ? */ + || + options->op_tbl->nelems + /* do we have input request for filter/chunking */ + || + options->all_filter==1 || options->all_layout==1 + ) { - /* get the storage size of the input dataset */ - dsize_in=H5Dget_storage_size(dset_in); - - /* check for datasets too small */ - if (nelmts*msize < options->threshold ) - apply_s=0; + if ((dset_in=H5Dopen(fidin,travt->objs[i].name))<0) + goto error; + if ((f_space_id=H5Dget_space(dset_in))<0) + goto error; + if ((ftype_id=H5Dget_type (dset_in))<0) + goto error; + if ((dcpl_id=H5Dget_create_plist(dset_in))<0) + goto error; + if ((dcpl_out = H5Pcopy (dcpl_id))<0) + goto error; + if ( (rank=H5Sget_simple_extent_ndims(f_space_id))<0) + goto error; + HDmemset(dims, 0, sizeof dims); + if ( H5Sget_simple_extent_dims(f_space_id,dims,NULL)<0) + goto error; + nelmts=1; + for (j=0; j<rank; j++) + nelmts*=dims[j]; - /* apply the filter */ - if (apply_s){ - if (apply_filters(travt->objs[i].name,rank,dims,dcpl_out,options)<0) - goto error; - } + if (options->use_native==1) + wtype_id = h5tools_get_native_type(ftype_id); + else + wtype_id = H5Tcopy(ftype_id); - /*------------------------------------------------------------------------- - * create the output dataset; - * disable error checking in case the dataset cannot be created with the - * modified dcpl; in that case use the original instead - *------------------------------------------------------------------------- + if ((msize=H5Tget_size(wtype_id))==0) + goto error; + + /*------------------------------------------------------------------------- + * check for external files + *------------------------------------------------------------------------- + */ + if ((next=H5Pget_external_count (dcpl_id))<0) + goto error; + + if (next) + fprintf(stderr," <warning: %s has external files, ignoring read...>\n", + travt->objs[i].name ); + + /*------------------------------------------------------------------------- + * check if the dataset creation property list has filters that + * are not registered in the current configuration + * 1) the external filters GZIP and SZIP might not be available + * 2) the internal filters might be turned off + *------------------------------------------------------------------------- */ - H5E_BEGIN_TRY { - dset_out=H5Dcreate(fidout,travt->objs[i].name,wtype_id,f_space_id,dcpl_out); - } H5E_END_TRY; - if (dset_out==FAIL) + if (next==0 && h5tools_canreadf((travt->objs[i].name),dcpl_id)==1) { - if ((dset_out=H5Dcreate(fidout,travt->objs[i].name,wtype_id,f_space_id,dcpl_id))<0) - goto error; - apply_f=0; - } - - /*------------------------------------------------------------------------- - * read/write - *------------------------------------------------------------------------- - */ - if (nelmts) - { - size_t need = (size_t)(nelmts*msize); /* bytes needed */ - if ( need < H5TOOLS_MALLOCSIZE ) - buf = HDmalloc(need); - - if (buf != NULL ) - { - if (H5Dread(dset_in,wtype_id,H5S_ALL,H5S_ALL,H5P_DEFAULT,buf)<0) - goto error; - if (H5Dwrite(dset_out,wtype_id,H5S_ALL,H5S_ALL,H5P_DEFAULT,buf)<0) - goto error; - } + apply_s=1; + apply_f=1; - else /* possibly not enough memory, read/write by hyperslabs */ - + /*------------------------------------------------------------------------- + * references are a special case + * we cannot just copy the buffers, but instead we recreate the reference + * in a second traversal of the output file + *------------------------------------------------------------------------- + */ + if ( (H5T_REFERENCE!=H5Tget_class(wtype_id))) { - size_t p_type_nbytes = msize; /*size of memory type */ - hsize_t p_nelmts = nelmts; /*total selected elmts */ - hsize_t elmtno; /*counter */ - int carry; /*counter carry value */ - unsigned int vl_data = 0; /*contains VL datatypes */ + /* get the storage size of the input dataset */ + dsize_in=H5Dget_storage_size(dset_in); - /* stripmine info */ - hsize_t sm_size[H5S_MAX_RANK]; /*stripmine size */ - hsize_t sm_nbytes; /*bytes per stripmine */ - hsize_t sm_nelmts; /*elements per stripmine*/ - hid_t sm_space; /*stripmine data space */ + /* check for datasets too small */ + if (nelmts*msize < options->threshold ) + apply_s=0; - /* hyperslab info */ - hsize_t hs_offset[H5S_MAX_RANK];/*starting offset */ - hsize_t hs_size[H5S_MAX_RANK]; /*size this pass */ - hsize_t hs_nelmts; /*elements in request */ - hsize_t zero[8]; /*vector of zeros */ + /* apply the filter */ + if (apply_s){ + if (apply_filters(travt->objs[i].name,rank,dims,dcpl_out,options)<0) + goto error; + } - /* check if we have VL data in the dataset's datatype */ - if (H5Tdetect_class(wtype_id, H5T_VLEN) == TRUE) - vl_data = TRUE; - - /* - * determine the strip mine size and allocate a buffer. The strip mine is - * a hyperslab whose size is manageable. + /*------------------------------------------------------------------------- + * create the output dataset; + * disable error checking in case the dataset cannot be created with the + * modified dcpl; in that case use the original instead + *------------------------------------------------------------------------- */ - sm_nbytes = p_type_nbytes; - - for (i = rank; i > 0; --i) { - sm_size[i - 1] = MIN(dims[i - 1], H5TOOLS_BUFSIZE / sm_nbytes); - sm_nbytes *= sm_size[i - 1]; - assert(sm_nbytes > 0); + H5E_BEGIN_TRY { + dset_out=H5Dcreate(fidout,travt->objs[i].name,wtype_id,f_space_id,dcpl_out); + } H5E_END_TRY; + if (dset_out==FAIL) + { + if ((dset_out=H5Dcreate(fidout,travt->objs[i].name,wtype_id,f_space_id,dcpl_id))<0) + goto error; + apply_f=0; } - sm_buf = HDmalloc((size_t)sm_nbytes); - - sm_nelmts = sm_nbytes / p_type_nbytes; - sm_space = H5Screate_simple(1, &sm_nelmts, NULL); - /* the stripmine loop */ - memset(hs_offset, 0, sizeof hs_offset); - memset(zero, 0, sizeof zero); - - for (elmtno = 0; elmtno < p_nelmts; elmtno += hs_nelmts) + /*------------------------------------------------------------------------- + * read/write + *------------------------------------------------------------------------- + */ + if (nelmts) { - /* calculate the hyperslab size */ - if (rank > 0) + size_t need = (size_t)(nelmts*msize); /* bytes needed */ + if ( need < H5TOOLS_MALLOCSIZE ) + buf = HDmalloc(need); + + if (buf != NULL ) { - for (i = 0, hs_nelmts = 1; i < rank; i++) - { - hs_size[i] = MIN(dims[i] - hs_offset[i], sm_size[i]); - hs_nelmts *= hs_size[i]; - } - - if (H5Sselect_hyperslab(f_space_id, H5S_SELECT_SET, hs_offset, NULL, hs_size, NULL)<0) + if (H5Dread(dset_in,wtype_id,H5S_ALL,H5S_ALL,H5P_DEFAULT,buf)<0) goto error; - if (H5Sselect_hyperslab(sm_space, H5S_SELECT_SET, zero, NULL, &hs_nelmts, NULL)<0) + if (H5Dwrite(dset_out,wtype_id,H5S_ALL,H5S_ALL,H5P_DEFAULT,buf)<0) goto error; - } - else - { - H5Sselect_all(f_space_id); - H5Sselect_all(sm_space); - hs_nelmts = 1; - } /* rank */ - - /* read/write */ - if (H5Dread(dset_in, wtype_id, sm_space, f_space_id, H5P_DEFAULT, sm_buf) < 0) - goto error; - if (H5Dwrite(dset_out, wtype_id, sm_space, f_space_id, H5P_DEFAULT, sm_buf) < 0) - goto error; + } - /* reclaim any VL memory, if necessary */ - if(vl_data) - H5Dvlen_reclaim(wtype_id, sm_space, H5P_DEFAULT, sm_buf); - - /* calculate the next hyperslab offset */ - for (i = rank, carry = 1; i > 0 && carry; --i) + else /* possibly not enough memory, read/write by hyperslabs */ + { - hs_offset[i - 1] += hs_size[i - 1]; - if (hs_offset[i - 1] == dims[i - 1]) - hs_offset[i - 1] = 0; - else - carry = 0; - } /* i */ - } /* elmtno */ - - H5Sclose(sm_space); - /* free */ - if (sm_buf!=NULL) - { - HDfree(sm_buf); - sm_buf=NULL; - } - } /* hyperslab read */ + size_t p_type_nbytes = msize; /*size of memory type */ + hsize_t p_nelmts = nelmts; /*total selected elmts */ + hsize_t elmtno; /*counter */ + int carry; /*counter carry value */ + unsigned int vl_data = 0; /*contains VL datatypes */ + + /* stripmine info */ + hsize_t sm_size[H5S_MAX_RANK]; /*stripmine size */ + hsize_t sm_nbytes; /*bytes per stripmine */ + hsize_t sm_nelmts; /*elements per stripmine*/ + hid_t sm_space; /*stripmine data space */ + + /* hyperslab info */ + hsize_t hs_offset[H5S_MAX_RANK];/*starting offset */ + hsize_t hs_size[H5S_MAX_RANK]; /*size this pass */ + hsize_t hs_nelmts; /*elements in request */ + hsize_t zero[8]; /*vector of zeros */ + + /* check if we have VL data in the dataset's datatype */ + if (H5Tdetect_class(wtype_id, H5T_VLEN) == TRUE) + vl_data = TRUE; + + /* + * determine the strip mine size and allocate a buffer. The strip mine is + * a hyperslab whose size is manageable. + */ + sm_nbytes = p_type_nbytes; + + for (i = rank; i > 0; --i) { + sm_size[i - 1] = MIN(dims[i - 1], H5TOOLS_BUFSIZE / sm_nbytes); + sm_nbytes *= sm_size[i - 1]; + assert(sm_nbytes > 0); + } + sm_buf = HDmalloc((size_t)sm_nbytes); + + sm_nelmts = sm_nbytes / p_type_nbytes; + sm_space = H5Screate_simple(1, &sm_nelmts, NULL); + + /* the stripmine loop */ + memset(hs_offset, 0, sizeof hs_offset); + memset(zero, 0, sizeof zero); + + for (elmtno = 0; elmtno < p_nelmts; elmtno += hs_nelmts) + { + /* calculate the hyperslab size */ + if (rank > 0) + { + for (i = 0, hs_nelmts = 1; i < rank; i++) + { + hs_size[i] = MIN(dims[i] - hs_offset[i], sm_size[i]); + hs_nelmts *= hs_size[i]; + } + + if (H5Sselect_hyperslab(f_space_id, H5S_SELECT_SET, hs_offset, NULL, hs_size, NULL)<0) + goto error; + if (H5Sselect_hyperslab(sm_space, H5S_SELECT_SET, zero, NULL, &hs_nelmts, NULL)<0) + goto error; + } + else + { + H5Sselect_all(f_space_id); + H5Sselect_all(sm_space); + hs_nelmts = 1; + } /* rank */ + + /* read/write */ + if (H5Dread(dset_in, wtype_id, sm_space, f_space_id, H5P_DEFAULT, sm_buf) < 0) + goto error; + if (H5Dwrite(dset_out, wtype_id, sm_space, f_space_id, H5P_DEFAULT, sm_buf) < 0) + goto error; + + /* reclaim any VL memory, if necessary */ + if(vl_data) + H5Dvlen_reclaim(wtype_id, sm_space, H5P_DEFAULT, sm_buf); + + /* calculate the next hyperslab offset */ + for (i = rank, carry = 1; i > 0 && carry; --i) + { + hs_offset[i - 1] += hs_size[i - 1]; + if (hs_offset[i - 1] == dims[i - 1]) + hs_offset[i - 1] = 0; + else + carry = 0; + } /* i */ + } /* elmtno */ + + H5Sclose(sm_space); + /* free */ + if (sm_buf!=NULL) + { + HDfree(sm_buf); + sm_buf=NULL; + } + } /* hyperslab read */ }/*nelmts*/ - - /*------------------------------------------------------------------------- - * amount of compression used - *------------------------------------------------------------------------- - */ - if (options->verbose) - { - if (apply_s && apply_f) + + /*------------------------------------------------------------------------- + * amount of compression used + *------------------------------------------------------------------------- + */ + if (options->verbose) { - /* get the storage size of the input dataset */ - dsize_out=H5Dget_storage_size(dset_out); - PER((hssize_t)dsize_in,(hssize_t)dsize_out); - print_dataset_info(dcpl_out,travt->objs[i].name,per*(float)100.0); + if (apply_s && apply_f) + { + /* get the storage size of the input dataset */ + dsize_out=H5Dget_storage_size(dset_out); + PER((hssize_t)dsize_in,(hssize_t)dsize_out); + print_dataset_info(dcpl_out,travt->objs[i].name,per*(float)100.0); + } + else + print_dataset_info(dcpl_id,travt->objs[i].name,0); } - else - print_dataset_info(dcpl_id,travt->objs[i].name,0); - } - - if (apply_s==0 && options->verbose) - printf(" <warning: filter not applied to %s. dataset smaller than %d bytes>\n", - travt->objs[i].name, - (int)options->threshold); - - if (apply_f==0 && options->verbose) + + if (apply_s==0 && options->verbose) + printf(" <warning: filter not applied to %s. dataset smaller than %d bytes>\n", + travt->objs[i].name, + (int)options->threshold); + + if (apply_f==0 && options->verbose) printf(" <warning: could not apply the filter to %s>\n", travt->objs[i].name); - - /*------------------------------------------------------------------------- - * copy attrs - *------------------------------------------------------------------------- - */ - if (copy_attr(dset_in,dset_out,options)<0) - goto error; - - /*close */ - if (H5Dclose(dset_out)<0) - goto error; - + + /*------------------------------------------------------------------------- + * copy attrs + *------------------------------------------------------------------------- + */ + if (copy_attr(dset_in,dset_out,options)<0) + goto error; + + /*close */ + if (H5Dclose(dset_out)<0) + goto error; + }/*H5T_STD_REF_OBJ*/ }/*can_read*/ + + + /*------------------------------------------------------------------------- + * close + *------------------------------------------------------------------------- + */ + if (H5Tclose(ftype_id)<0) + goto error; + if (H5Tclose(wtype_id)<0) + goto error; + if (H5Pclose(dcpl_id)<0) + goto error; + if (H5Pclose(dcpl_out)<0) + goto error; + if (H5Sclose(f_space_id)<0) + goto error; + if (H5Dclose(dset_in)<0) + goto error; + } + /*------------------------------------------------------------------------- + * we do not have request for filter/chunking use H5Ocopy instead + *------------------------------------------------------------------------- + */ + else + { + hid_t pid; + + /* create property to pass copy options */ + if ( (pid = H5Pcreate(H5P_OBJECT_COPY)) < 0) + goto error; + + /*------------------------------------------------------------------------- + * do the copy + *------------------------------------------------------------------------- + */ + + if (H5Ocopy(fidin, /* Source file or group identifier */ + travt->objs[i].name, /* Name of the source object to be copied */ + fidout, /* Destination file or group identifier */ + travt->objs[i].name, /* Name of the destination object */ + pid, /* Properties which apply to the copy */ + H5P_DEFAULT)<0) /* Properties which apply to the new hard link */ + goto error; + + /* close property */ + if (H5Pclose(pid)<0) + goto error; + + } /* end do we have request for filter/chunking */ -/*------------------------------------------------------------------------- - * close - *------------------------------------------------------------------------- - */ - if (H5Tclose(ftype_id)<0) - goto error; - if (H5Tclose(wtype_id)<0) - goto error; - if (H5Pclose(dcpl_id)<0) - goto error; - if (H5Pclose(dcpl_out)<0) - goto error; - if (H5Sclose(f_space_id)<0) - goto error; - if (H5Dclose(dset_in)<0) - goto error; break; @@ -686,8 +755,6 @@ int do_copy_objects(hid_t fidin, /*------------------------------------------------------------------------- * the root is a special case, we get an ID for the root group * and copy its attributes using that ID - * it must be done last, because the attributes might contain references to - * objects in the object list *------------------------------------------------------------------------- */ @@ -827,8 +894,6 @@ int copy_attr(hid_t loc_in, */ if ( ! H5Tequal(wtype_id, H5T_STD_REF_OBJ)) { - - /*------------------------------------------------------------------------- * read to memory *------------------------------------------------------------------------- @@ -894,6 +959,73 @@ error: return -1; } +/*------------------------------------------------------------------------- + * Function: do_hardlinks + * + * Purpose: duplicate hard links + * + * Return: 0, ok, -1 no + * + * Programmer: Pedro Vicente, pvn@ncsa.uiuc.edu + * + * Date: December, 10, 2003 + * + *------------------------------------------------------------------------- + */ + +static +int do_hardlinks(hid_t fidout,trav_table_t *travt) +{ + int i, j; + + for ( i = 0; i < travt->nobjs; i++) + { + switch ( travt->objs[i].type ) + { + case H5G_GROUP: + + if (travt->objs[i].nlinks) + { + for ( j=0; j<travt->objs[i].nlinks; j++) + { + if (H5Glink(fidout, + H5L_TYPE_HARD, + travt->objs[i].name, + travt->objs[i].links[j].new_name)<0) + return -1; + } + } + + break; + + case H5G_DATASET: + if (travt->objs[i].nlinks) + { + for ( j=0; j<travt->objs[i].nlinks; j++){ + if (H5Glink(fidout, + H5L_TYPE_HARD, + travt->objs[i].name, + travt->objs[i].links[j].new_name)<0) + return -1; + } + } + break; + case H5G_TYPE: + case H5G_LINK: + case H5G_UDLINK: + + /*nothing to do */ + break; + + default: + + break; + } + } + + return 0; + +} |