mtbl_merger - merge multiple MTBL data sources into a single output
Contents
Description
Multiple MTBL data sources may be merged together using the mtbl_merger interface, which reads key-value
entries from one or more sources and provides these entries in sorted order. The sorted entries may be
consumed via the mtbl_source(3) and mtbl_iter(3) interfaces.
Because the MTBL format does not allow duplicate keys, the caller must provide a function which will
accept a key and two conflicting values for that key and return a replacement value. This function may be
called multiple times for the same key if more than two sources are being merged.
mtbl_merger objects are created with the mtbl_merger_init() function, which requires a non-NULL mopt
argument which has been configured with a merge function fp.
One or more mtbl_reader objects must be provided as input to the mtbl_merger object by calling
mtbl_merger_add_source(). After the desired sources have been configured, mtbl_merger_source() should be
called in order to consume the merged output via the mtbl_source(3) interface.
Mergeroptionsmerge_func
This option specifies a merge function callback, consisting of a function pointer fp and a pointer to
user data clos which will be passed as the first argument to fp. The merge function callback will be
used during iteration over the mtbl_merger object to merge entries with duplicate keys in the input
sources.
The remaining arguments to the merge function are:
key — pointer to the key for which there exist duplicate values.
len_key — length of the key.
val0 — pointer to the first value.
len_val_0 — length of the first value.
val1 — pointer to the second value.
len_val_1 — length of the second value.
merged_val — pointer to where the callee should place its merged value.
len_merged_val — pointer to where the callee should place the length of its merged value.
merged_val must be allocated with the system allocator, and the mtbl_merger interface takes
responsibility for free()ing the value once it is no longer needed.
The callee may provide an empty value as the merged value, in which case merged_val must still
contain an allocated, non-NULL value and len_merged_val must contain the value 0.
The callee may indicate an error by returning NULL in the merged_val argument, which will abort
iteration over the mtbl_merger object.
dupsort_func
This option provides a comparison function for multiple data with the same key. It will be used to
sort the data during the merge process. If the merge_func option is set to or left NULL, the
dupsort_func will order the results returned by the merge process for data with identical keys.
Name
mtbl_merger - merge multiple MTBL data sources into a single output
Return Value
If the merge function callback is unable to provide a merged value (that is, it fails to return a
non-NULL value in its merged_val argument), the merge process will be aborted, and any iterators over the
mtbl_merger object (via the mtbl_source(3) interface) will return mtbl_res_failure.
03/28/2019 MTBL_MERGER(3)
Synopsis
#include<mtbl.h>
Merger objects:
structmtbl_merger*mtbl_merger_init(conststructmtbl_merger_options*mopt);voidmtbl_merger_destroy(structmtbl_merger**m);voidmtbl_merger_add_source(structmtbl_merger*m,conststructmtbl_source*s);conststructmtbl_source*mtbl_merger_source(structmtbl_merger*m);
Merger options:
structmtbl_merger_options*mtbl_merger_options_init(void);voidmtbl_merger_options_destroy(structmtbl_merger_options**mopt);voidmtbl_merger_options_set_merge_func(structmtbl_merger_options*mopt,mtbl_merge_funcfp,void*clos);voidmtbl_merger_options_set_dupsort_func(structmtbl_merger_options*mopt,mtbl_dupsort_funcfp,void*clos);typedefvoid(*mtbl_merge_func)(void*clos,constuint8_t*key,size_tlen_key,constuint8_t*val0,size_tlen_val0,constuint8_t*val1,size_tlen_val1,uint8_t**merged_val,size_t*len_merged_val);typedefint(*mtbl_dupsort_func)(void*clos,constuint8_t*key,size_tlen_key,constuint8_t*val0,size_tlen_val0,constuint8_t*val1,size_t'len_val1);