1binary(3) Erlang Module Definition binary(3)
2
6 binary - Library for handling binary data.
7
9 This module contains functions for manipulating byte-oriented binaries.
10 Although the majority of functions could be provided using bit-syntax,
11 the functions in this library are highly optimized and are expected to
12 either execute faster or consume less memory, or both, than a counter‐
13 part written in pure Erlang.
14 The module is provided according to Erlang Enhancement Proposal (EEP)
16 31.
17 Note:
19 The library handles byte-oriented data. For bitstrings that are not
20 binaries (does not contain whole octets of bits) a badarg exception is
21 thrown from any of the functions in this module.
22
25 cp()
26 Opaque data type representing a compiled search pattern. Guaran‐
28 teed to be a tuple() to allow programs to distinguish it from
29 non-precompiled search patterns.
30 part() = {Start :: integer() >= 0, Length :: integer()}
32 A representaion of a part (or range) in a binary. Start is a
34 zero-based offset into a binary() and Length is the length of
35 that part. As input to functions in this module, a reverse part
36 specification is allowed, constructed with a negative Length, so
37 that the part of the binary begins at Start + Length and is
38 -Length long. This is useful for referencing the last N bytes of
39 a binary as {size(Binary), -N}. The functions in this module
40 always return part()s with positive Length.
41
43 at(Subject, Pos) -> byte()
44 Types:
46 Subject = binary()
48 Pos = integer() >= 0
49 Returns the byte at position Pos (zero-based) in binary Subject
51 as an integer. If Pos >= byte_size(Subject), a badarg exception
52 is raised.
53 bin_to_list(Subject) -> [byte()]
55 Types:
57 Subject = binary()
59 Same as bin_to_list(Subject, {0,byte_size(Subject)}).
61 bin_to_list(Subject, PosLen) -> [byte()]
63 Types:
65 Subject = binary()
67 PosLen = part()
68 Converts Subject to a list of byte()s, each representing the
70 value of one byte. part() denotes which part of the binary() to
71 convert.
72 Example:
74 1> binary:bin_to_list(<<"erlang">>, {1,3}).
76 "rla"
77 %% or [114,108,97] in list notation.
78 If PosLen in any way references outside the binary, a badarg
80 exception is raised.
81 bin_to_list(Subject, Pos, Len) -> [byte()]
83 Types:
85 Subject = binary()
87 Pos = integer() >= 0
88 Len = integer()
89 Same as bin_to_list(Subject, {Pos, Len}).
91 compile_pattern(Pattern) -> cp()
93 Types:
95 Pattern = binary() | [binary()]
97 Builds an internal structure representing a compilation of a
99 search pattern, later to be used in functions match/3,
100 matches/3, split/3, or replace/4. The cp() returned is guaran‐
101 teed to be a tuple() to allow programs to distinguish it from
102 non-precompiled search patterns.
103 When a list of binaries is specified, it denotes a set of alter‐
105 native binaries to search for. For example, if [<<"func‐
106 tional">>,<<"programming">>] is specified as Pattern, this means
107 either <<"functional">> or <<"programming">>". The pattern is a
108 set of alternatives; when only a single binary is specified, the
109 set has only one element. The order of alternatives in a pattern
110 is not significant.
111 The list of binaries used for search alternatives must be flat
113 and proper.
114 If Pattern is not a binary or a flat proper list of binaries
116 with length > 0, a badarg exception is raised.
117 copy(Subject) -> binary()
119 Types:
121 Subject = binary()
123 Same as copy(Subject, 1).
125 copy(Subject, N) -> binary()
127 Types:
129 Subject = binary()
131 N = integer() >= 0
132 Creates a binary with the content of Subject duplicated N times.
134 This function always creates a new binary, even if N = 1. By
136 using copy/1 on a binary referencing a larger binary, one can
137 free up the larger binary for garbage collection.
138 Note:
140 By deliberately copying a single binary to avoid referencing a
141 larger binary, one can, instead of freeing up the larger binary
142 for later garbage collection, create much more binary data than
143 needed. Sharing binary data is usually good. Only in special
144 cases, when small parts reference large binaries and the large
145 binaries are no longer used in any process, deliberate copying
146 can be a good idea.
147 If N < 0, a badarg exception is raised.
150 decode_unsigned(Subject) -> Unsigned
152 Types:
154 Subject = binary()
156 Unsigned = integer() >= 0
157 Same as decode_unsigned(Subject, big).
159 decode_unsigned(Subject, Endianness) -> Unsigned
161 Types:
163 Subject = binary()
165 Endianness = big | little
166 Unsigned = integer() >= 0
167 Converts the binary digit representation, in big endian or lit‐
169 tle endian, of a positive integer in Subject to an Erlang inte‐
170 ger().
171 Example:
173 1> binary:decode_unsigned(<<169,138,199>>,big).
175 11111111
176 encode_unsigned(Unsigned) -> binary()
178 Types:
180 Unsigned = integer() >= 0
182 Same as encode_unsigned(Unsigned, big).
184 encode_unsigned(Unsigned, Endianness) -> binary()
186 Types:
188 Unsigned = integer() >= 0
190 Endianness = big | little
191 Converts a positive integer to the smallest possible representa‐
193 tion in a binary digit representation, either big endian or lit‐
194 tle endian.
195 Example:
197 1> binary:encode_unsigned(11111111, big).
199 <<169,138,199>>
200 first(Subject) -> byte()
202 Types:
204 Subject = binary()
206 Returns the first byte of binary Subject as an integer. If the
208 size of Subject is zero, a badarg exception is raised.
209 last(Subject) -> byte()
211 Types:
213 Subject = binary()
215 Returns the last byte of binary Subject as an integer. If the
217 size of Subject is zero, a badarg exception is raised.
218 list_to_bin(ByteList) -> binary()
220 Types:
222 ByteList = iodata()
224 Works exactly as erlang:list_to_binary/1, added for complete‐
226 ness.
227 longest_common_prefix(Binaries) -> integer() >= 0
229 Types:
231 Binaries = [binary()]
233 Returns the length of the longest common prefix of the binaries
235 in list Binaries.
236 Example:
238 1> binary:longest_common_prefix([<<"erlang">>, <<"ergonomy">>]).
240 2
241 2> binary:longest_common_prefix([<<"erlang">>, <<"perl">>]).
242 0
243 If Binaries is not a flat list of binaries, a badarg exception
245 is raised.
246 longest_common_suffix(Binaries) -> integer() >= 0
248 Types:
250 Binaries = [binary()]
252 Returns the length of the longest common suffix of the binaries
254 in list Binaries.
255 Example:
257 1> binary:longest_common_suffix([<<"erlang">>, <<"fang">>]).
259 3
260 2> binary:longest_common_suffix([<<"erlang">>, <<"perl">>]).
261 0
262 If Binaries is not a flat list of binaries, a badarg exception
264 is raised.
265 match(Subject, Pattern) -> Found | nomatch
267 Types:
269 Subject = binary()
271 Pattern = binary() | [binary()] | cp()
272 Found = part()
273 Same as match(Subject, Pattern, []).
275 match(Subject, Pattern, Options) -> Found | nomatch
277 Types:
279 Subject = binary()
281 Pattern = binary() | [binary()] | cp()
282 Found = part()
283 Options = [Option]
284 Option = {scope, part()}
285 part() = {Start :: integer() >= 0, Length :: integer()}
286 Searches for the first occurrence of Pattern in Subject and
288 returns the position and length.
289 The function returns {Pos, Length} for the binary in Pattern,
291 starting at the lowest position in Subject.
292 Example:
294 1> binary:match(<<"abcde">>, [<<"bcde">>, <<"cd">>],[]).
296 {1,4}
297 Even though <<"cd">> ends before <<"bcde">>, <<"bcde">> begins
299 first and is therefore the first match. If two overlapping
300 matches begin at the same position, the longest is returned.
301 Summary of the options:
303 {scope, {Start, Length}}:
305 Only the specified part is searched. Return values still
306 have offsets from the beginning of Subject. A negative
307 Length is allowed as described in section Data Types in this
308 manual.
309 If none of the strings in Pattern is found, the atom nomatch is
311 returned.
312 For a description of Pattern, see function compile_pattern/1.
314 If {scope, {Start,Length}} is specified in the options such that
316 Start > size of Subject, Start + Length < 0 or Start + Length >
317 size of Subject, a badarg exception is raised.
318 matches(Subject, Pattern) -> Found
320 Types:
322 Subject = binary()
324 Pattern = binary() | [binary()] | cp()
325 Found = [part()]
326 Same as matches(Subject, Pattern, []).
328 matches(Subject, Pattern, Options) -> Found
330 Types:
332 Subject = binary()
334 Pattern = binary() | [binary()] | cp()
335 Found = [part()]
336 Options = [Option]
337 Option = {scope, part()}
338 part() = {Start :: integer() >= 0, Length :: integer()}
339 As match/2, but Subject is searched until exhausted and a list
341 of all non-overlapping parts matching Pattern is returned (in
342 order).
343 The first and longest match is preferred to a shorter, which is
345 illustrated by the following example:
346 1> binary:matches(<<"abcde">>,
348 [<<"bcde">>,<<"bc">>,<<"de">>],[]).
349 [{1,4}]
350 The result shows that <<"bcde">> is selected instead of the
352 shorter match <<"bc">> (which would have given raise to one more
353 match, <<"de">>). This corresponds to the behavior of POSIX reg‐
354 ular expressions (and programs like awk), but is not consistent
355 with alternative matches in re (and Perl), where instead lexical
356 ordering in the search pattern selects which string matches.
357 If none of the strings in a pattern is found, an empty list is
359 returned.
360 For a description of Pattern, see compile_pattern/1. For a
362 description of available options, see match/3.
363 If {scope, {Start,Length}} is specified in the options such that
365 Start > size of Subject, Start + Length < 0 or Start + Length is
366 > size of Subject, a badarg exception is raised.
367 part(Subject, PosLen) -> binary()
369 Types:
371 Subject = binary()
373 PosLen = part()
374 Extracts the part of binary Subject described by PosLen.
376 A negative length can be used to extract bytes at the end of a
378 binary:
379 1> Bin = <<1,2,3,4,5,6,7,8,9,10>>.
381 2> binary:part(Bin, {byte_size(Bin), -5}).
382 <<6,7,8,9,10>>
383 Note:
385 part/2 and part/3 are also available in the erlang module under
386 the names binary_part/2 and binary_part/3. Those BIFs are
387 allowed in guard tests.
388 If PosLen in any way references outside the binary, a badarg
391 exception is raised.
392 part(Subject, Pos, Len) -> binary()
394 Types:
396 Subject = binary()
398 Pos = integer() >= 0
399 Len = integer()
400 Same as part(Subject, {Pos, Len}).
402 referenced_byte_size(Binary) -> integer() >= 0
404 Types:
406 Binary = binary()
408 If a binary references a larger binary (often described as being
410 a subbinary), it can be useful to get the size of the referenced
411 binary. This function can be used in a program to trigger the
412 use of copy/1. By copying a binary, one can dereference the
413 original, possibly large, binary that a smaller binary is a ref‐
414 erence to.
415 Example:
417 store(Binary, GBSet) ->
419 NewBin =
420 case binary:referenced_byte_size(Binary) of
421 Large when Large > 2 * byte_size(Binary) ->
422 binary:copy(Binary);
423 _ ->
424 Binary
425 end,
426 gb_sets:insert(NewBin,GBSet).
427 In this example, we chose to copy the binary content before
429 inserting it in gb_sets:set() if it references a binary more
430 than twice the data size we want to keep. Of course, different
431 rules apply when copying to different programs.
432 Binary sharing occurs whenever binaries are taken apart. This is
434 the fundamental reason why binaries are fast, decomposition can
435 always be done with O(1) complexity. In rare circumstances this
436 data sharing is however undesirable, why this function together
437 with copy/1 can be useful when optimizing for memory use.
438 Example of binary sharing:
440 1> A = binary:copy(<<1>>, 100).
442 <<1,1,1,1,1 ...
443 2> byte_size(A).
444 100
445 3> binary:referenced_byte_size(A)
446 100
447 4> <<_:10/binary,B:10/binary,_/binary>> = A.
448 <<1,1,1,1,1 ...
449 5> byte_size(B).
450 10
451 6> binary:referenced_byte_size(B)
452 100
453 Note:
455 Binary data is shared among processes. If another process still
456 references the larger binary, copying the part this process uses
457 only consumes more memory and does not free up the larger binary
458 for garbage collection. Use this kind of intrusive functions
459 with extreme care and only if a real problem is detected.
460 replace(Subject, Pattern, Replacement) -> Result
463 Types:
465 Subject = binary()
467 Pattern = binary() | [binary()] | cp()
468 Replacement = Result = binary()
469 Same as replace(Subject, Pattern, Replacement,[]).
471 replace(Subject, Pattern, Replacement, Options) -> Result
473 Types:
475 Subject = binary()
477 Pattern = binary() | [binary()] | cp()
478 Replacement = binary()
479 Options = [Option]
480 Option = global | {scope, part()} | {insert_replaced, InsPos}
481 InsPos = OnePos | [OnePos]
482 OnePos = integer() >= 0
483 An integer() =< byte_size(Replacement)
484 Result = binary()
485 Constructs a new binary by replacing the parts in Subject match‐
487 ing Pattern with the content of Replacement.
488 If the matching subpart of Subject giving raise to the replace‐
490 ment is to be inserted in the result, option {insert_replaced,
491 InsPos} inserts the matching part into Replacement at the speci‐
492 fied position (or positions) before inserting Replacement into
493 Subject.
494 Example:
496 1> binary:replace(<<"abcde">>,<<"b">>,<<"[]">>, [{insert_replaced,1}]).
498 <<"a[b]cde">>
499 2> binary:replace(<<"abcde">>,[<<"b">>,<<"d">>],<<"[]">>,[global,{insert_replaced,1}]).
500 <<"a[b]c[d]e">>
501 3> binary:replace(<<"abcde">>,[<<"b">>,<<"d">>],<<"[]">>,[global,{insert_replaced,[1,1]}]).
502 <<"a[bb]c[dd]e">>
503 4> binary:replace(<<"abcde">>,[<<"b">>,<<"d">>],<<"[-]">>,[global,{insert_replaced,[1,2]}]).
504 <<"a[b-b]c[d-d]e">>
505 If any position specified in InsPos > size of the replacement
507 binary, a badarg exception is raised.
508 Options global and {scope, part()} work as for split/3. The
510 return type is always a binary().
511 For a description of Pattern, see compile_pattern/1.
513 split(Subject, Pattern) -> Parts
515 Types:
517 Subject = binary()
519 Pattern = binary() | [binary()] | cp()
520 Parts = [binary()]
521 Same as split(Subject, Pattern, []).
523 split(Subject, Pattern, Options) -> Parts
525 Types:
527 Subject = binary()
529 Pattern = binary() | [binary()] | cp()
530 Options = [Option]
531 Option = {scope, part()} | trim | global | trim_all
532 Parts = [binary()]
533 Splits Subject into a list of binaries based on Pattern. If
535 option global is not specified, only the first occurrence of
536 Pattern in Subject gives rise to a split.
537 The parts of Pattern found in Subject are not included in the
539 result.
540 Example:
542 1> binary:split(<<1,255,4,0,0,0,2,3>>, [<<0,0,0>>,<<2>>],[]).
544 [<<1,255,4>>, <<2,3>>]
545 2> binary:split(<<0,1,0,0,4,255,255,9>>, [<<0,0>>, <<255,255>>],[global]).
546 [<<0,1>>,<<4>>,<<9>>]
547 Summary of options:
549 {scope, part()}:
551 Works as in match/3 and matches/3. Notice that this only
552 defines the scope of the search for matching strings, it
553 does not cut the binary before splitting. The bytes before
554 and after the scope are kept in the result. See the example
555 below.
556 trim:
558 Removes trailing empty parts of the result (as does trim in
559 re:split/3.
560 trim_all:
562 Removes all empty parts of the result.
563 global:
565 Repeats the split until Subject is exhausted. Conceptually
566 option global makes split work on the positions returned by
567 matches/3, while it normally works on the position returned
568 by match/3.
569 Example of the difference between a scope and taking the binary
571 apart before splitting:
572 1> binary:split(<<"banana">>, [<<"a">>],[{scope,{2,3}}]).
574 [<<"ban">>,<<"na">>]
575 2> binary:split(binary:part(<<"banana">>,{2,3}), [<<"a">>],[]).
576 [<<"n">>,<<"n">>]
577 The return type is always a list of binaries that are all refer‐
579 encing Subject. This means that the data in Subject is not
580 copied to new binaries, and that Subject cannot be garbage col‐
581 lected until the results of the split are no longer referenced.
582 For a description of Pattern, see compile_pattern/1.
584Ericsson AB stdlib 3.8.2.1 binary(3)