7
Network Working Group T. Berners-Lee
8
Request for Comments: 3986 W3C/MIT
10
Updates: 1738 Day Software
11
Obsoletes: 2732, 2396, 1808 L. Masinter
12
Category: Standards Track Adobe Systems
16
Uniform Resource Identifier (URI): Generic Syntax
20
This document specifies an Internet standards track protocol for the
21
Internet community, and requests discussion and suggestions for
22
improvements. Please refer to the current edition of the "Internet
23
Official Protocol Standards" (STD 1) for the standardization state
24
and status of this protocol. Distribution of this memo is unlimited.
28
Copyright (C) The Internet Society (2005).
32
A Uniform Resource Identifier (URI) is a compact sequence of
33
characters that identifies an abstract or physical resource. This
34
specification defines the generic URI syntax and a process for
35
resolving URI references that might be in relative form, along with
36
guidelines and security considerations for the use of URIs on the
37
Internet. The URI syntax defines a grammar that is a superset of all
38
valid URIs, allowing an implementation to parse the common components
39
of a URI reference without knowing the scheme-specific requirements
40
of every possible identifier. This specification does not define a
41
generative grammar for URIs; that task is performed by the individual
42
specifications of each URI scheme.
58
Berners-Lee, et al. Standards Track [Page 1]
60
RFC 3986 URI Generic Syntax January 2005
65
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4
66
1.1. Overview of URIs . . . . . . . . . . . . . . . . . . . . 4
67
1.1.1. Generic Syntax . . . . . . . . . . . . . . . . . 6
68
1.1.2. Examples . . . . . . . . . . . . . . . . . . . . 7
69
1.1.3. URI, URL, and URN . . . . . . . . . . . . . . . 7
70
1.2. Design Considerations . . . . . . . . . . . . . . . . . 8
71
1.2.1. Transcription . . . . . . . . . . . . . . . . . 8
72
1.2.2. Separating Identification from Interaction . . . 9
73
1.2.3. Hierarchical Identifiers . . . . . . . . . . . . 10
74
1.3. Syntax Notation . . . . . . . . . . . . . . . . . . . . 11
75
2. Characters . . . . . . . . . . . . . . . . . . . . . . . . . . 11
76
2.1. Percent-Encoding . . . . . . . . . . . . . . . . . . . . 12
77
2.2. Reserved Characters . . . . . . . . . . . . . . . . . . 12
78
2.3. Unreserved Characters . . . . . . . . . . . . . . . . . 13
79
2.4. When to Encode or Decode . . . . . . . . . . . . . . . . 14
80
2.5. Identifying Data . . . . . . . . . . . . . . . . . . . . 14
81
3. Syntax Components . . . . . . . . . . . . . . . . . . . . . . 16
82
3.1. Scheme . . . . . . . . . . . . . . . . . . . . . . . . . 17
83
3.2. Authority . . . . . . . . . . . . . . . . . . . . . . . 17
84
3.2.1. User Information . . . . . . . . . . . . . . . . 18
85
3.2.2. Host . . . . . . . . . . . . . . . . . . . . . . 18
86
3.2.3. Port . . . . . . . . . . . . . . . . . . . . . . 22
87
3.3. Path . . . . . . . . . . . . . . . . . . . . . . . . . . 22
88
3.4. Query . . . . . . . . . . . . . . . . . . . . . . . . . 23
89
3.5. Fragment . . . . . . . . . . . . . . . . . . . . . . . . 24
90
4. Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
91
4.1. URI Reference . . . . . . . . . . . . . . . . . . . . . 25
92
4.2. Relative Reference . . . . . . . . . . . . . . . . . . . 26
93
4.3. Absolute URI . . . . . . . . . . . . . . . . . . . . . . 27
94
4.4. Same-Document Reference . . . . . . . . . . . . . . . . 27
95
4.5. Suffix Reference . . . . . . . . . . . . . . . . . . . . 27
96
5. Reference Resolution . . . . . . . . . . . . . . . . . . . . . 28
97
5.1. Establishing a Base URI . . . . . . . . . . . . . . . . 28
98
5.1.1. Base URI Embedded in Content . . . . . . . . . . 29
99
5.1.2. Base URI from the Encapsulating Entity . . . . . 29
100
5.1.3. Base URI from the Retrieval URI . . . . . . . . 30
101
5.1.4. Default Base URI . . . . . . . . . . . . . . . . 30
102
5.2. Relative Resolution . . . . . . . . . . . . . . . . . . 30
103
5.2.1. Pre-parse the Base URI . . . . . . . . . . . . . 31
104
5.2.2. Transform References . . . . . . . . . . . . . . 31
105
5.2.3. Merge Paths . . . . . . . . . . . . . . . . . . 32
106
5.2.4. Remove Dot Segments . . . . . . . . . . . . . . 33
107
5.3. Component Recomposition . . . . . . . . . . . . . . . . 35
108
5.4. Reference Resolution Examples . . . . . . . . . . . . . 35
109
5.4.1. Normal Examples . . . . . . . . . . . . . . . . 36
110
5.4.2. Abnormal Examples . . . . . . . . . . . . . . . 36
114
Berners-Lee, et al. Standards Track [Page 2]
116
RFC 3986 URI Generic Syntax January 2005
119
6. Normalization and Comparison . . . . . . . . . . . . . . . . . 38
120
6.1. Equivalence . . . . . . . . . . . . . . . . . . . . . . 38
121
6.2. Comparison Ladder . . . . . . . . . . . . . . . . . . . 39
122
6.2.1. Simple String Comparison . . . . . . . . . . . . 39
123
6.2.2. Syntax-Based Normalization . . . . . . . . . . . 40
124
6.2.3. Scheme-Based Normalization . . . . . . . . . . . 41
125
6.2.4. Protocol-Based Normalization . . . . . . . . . . 42
126
7. Security Considerations . . . . . . . . . . . . . . . . . . . 43
127
7.1. Reliability and Consistency . . . . . . . . . . . . . . 43
128
7.2. Malicious Construction . . . . . . . . . . . . . . . . . 43
129
7.3. Back-End Transcoding . . . . . . . . . . . . . . . . . . 44
130
7.4. Rare IP Address Formats . . . . . . . . . . . . . . . . 45
131
7.5. Sensitive Information . . . . . . . . . . . . . . . . . 45
132
7.6. Semantic Attacks . . . . . . . . . . . . . . . . . . . . 45
133
8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 46
134
9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 46
135
10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 46
136
10.1. Normative References . . . . . . . . . . . . . . . . . . 46
137
10.2. Informative References . . . . . . . . . . . . . . . . . 47
138
A. Collected ABNF for URI . . . . . . . . . . . . . . . . . . . . 49
139
B. Parsing a URI Reference with a Regular Expression . . . . . . 50
140
C. Delimiting a URI in Context . . . . . . . . . . . . . . . . . 51
141
D. Changes from RFC 2396 . . . . . . . . . . . . . . . . . . . . 53
142
D.1. Additions . . . . . . . . . . . . . . . . . . . . . . . 53
143
D.2. Modifications . . . . . . . . . . . . . . . . . . . . . 53
144
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
145
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 60
146
Full Copyright Statement . . . . . . . . . . . . . . . . . . . . . 61
170
Berners-Lee, et al. Standards Track [Page 3]
172
RFC 3986 URI Generic Syntax January 2005
177
A Uniform Resource Identifier (URI) provides a simple and extensible
178
means for identifying a resource. This specification of URI syntax
179
and semantics is derived from concepts introduced by the World Wide
180
Web global information initiative, whose use of these identifiers
181
dates from 1990 and is described in "Universal Resource Identifiers
182
in WWW" [RFC1630]. The syntax is designed to meet the
183
recommendations laid out in "Functional Recommendations for Internet
184
Resource Locators" [RFC1736] and "Functional Requirements for Uniform
185
Resource Names" [RFC1737].
187
This document obsoletes [RFC2396], which merged "Uniform Resource
188
Locators" [RFC1738] and "Relative Uniform Resource Locators"
189
[RFC1808] in order to define a single, generic syntax for all URIs.
190
It obsoletes [RFC2732], which introduced syntax for an IPv6 address.
191
It excludes portions of RFC 1738 that defined the specific syntax of
192
individual URI schemes; those portions will be updated as separate
193
documents. The process for registration of new URI schemes is
194
defined separately by [BCP35]. Advice for designers of new URI
195
schemes can be found in [RFC2718]. All significant changes from RFC
196
2396 are noted in Appendix D.
198
This specification uses the terms "character" and "coded character
199
set" in accordance with the definitions provided in [BCP19], and
200
"character encoding" in place of what [BCP19] refers to as a
203
1.1. Overview of URIs
205
URIs are characterized as follows:
209
Uniformity provides several benefits. It allows different types
210
of resource identifiers to be used in the same context, even when
211
the mechanisms used to access those resources may differ. It
212
allows uniform semantic interpretation of common syntactic
213
conventions across different types of resource identifiers. It
214
allows introduction of new types of resource identifiers without
215
interfering with the way that existing identifiers are used. It
216
allows the identifiers to be reused in many different contexts,
217
thus permitting new applications or protocols to leverage a pre-
218
existing, large, and widely used set of resource identifiers.
226
Berners-Lee, et al. Standards Track [Page 4]
228
RFC 3986 URI Generic Syntax January 2005
233
This specification does not limit the scope of what might be a
234
resource; rather, the term "resource" is used in a general sense
235
for whatever might be identified by a URI. Familiar examples
236
include an electronic document, an image, a source of information
237
with a consistent purpose (e.g., "today's weather report for Los
238
Angeles"), a service (e.g., an HTTP-to-SMS gateway), and a
239
collection of other resources. A resource is not necessarily
240
accessible via the Internet; e.g., human beings, corporations, and
241
bound books in a library can also be resources. Likewise,
242
abstract concepts can be resources, such as the operators and
243
operands of a mathematical equation, the types of a relationship
244
(e.g., "parent" or "employee"), or numeric values (e.g., zero,
249
An identifier embodies the information required to distinguish
250
what is being identified from all other things within its scope of
251
identification. Our use of the terms "identify" and "identifying"
252
refer to this purpose of distinguishing one resource from all
253
other resources, regardless of how that purpose is accomplished
254
(e.g., by name, address, or context). These terms should not be
255
mistaken as an assumption that an identifier defines or embodies
256
the identity of what is referenced, though that may be the case
257
for some identifiers. Nor should it be assumed that a system
258
using URIs will access the resource identified: in many cases,
259
URIs are used to denote resources without any intention that they
260
be accessed. Likewise, the "one" resource identified might not be
261
singular in nature (e.g., a resource might be a named set or a
262
mapping that varies over time).
264
A URI is an identifier consisting of a sequence of characters
265
matching the syntax rule named <URI> in Section 3. It enables
266
uniform identification of resources via a separately defined
267
extensible set of naming schemes (Section 3.1). How that
268
identification is accomplished, assigned, or enabled is delegated to
269
each scheme specification.
271
This specification does not place any limits on the nature of a
272
resource, the reasons why an application might seek to refer to a
273
resource, or the kinds of systems that might use URIs for the sake of
274
identifying resources. This specification does not require that a
275
URI persists in identifying the same resource over time, though that
276
is a common goal of all URI schemes. Nevertheless, nothing in this
282
Berners-Lee, et al. Standards Track [Page 5]
284
RFC 3986 URI Generic Syntax January 2005
287
specification prevents an application from limiting itself to
288
particular types of resources, or to a subset of URIs that maintains
289
characteristics desired by that application.
291
URIs have a global scope and are interpreted consistently regardless
292
of context, though the result of that interpretation may be in
293
relation to the end-user's context. For example, "http://localhost/"
294
has the same interpretation for every user of that reference, even
295
though the network interface corresponding to "localhost" may be
296
different for each end-user: interpretation is independent of access.
297
However, an action made on the basis of that reference will take
298
place in relation to the end-user's context, which implies that an
299
action intended to refer to a globally unique thing must use a URI
300
that distinguishes that resource from all other things. URIs that
301
identify in relation to the end-user's local context should only be
302
used when the context itself is a defining aspect of the resource,
303
such as when an on-line help manual refers to a file on the end-
304
user's file system (e.g., "file:///etc/hosts").
306
1.1.1. Generic Syntax
308
Each URI begins with a scheme name, as defined in Section 3.1, that
309
refers to a specification for assigning identifiers within that
310
scheme. As such, the URI syntax is a federated and extensible naming
311
system wherein each scheme's specification may further restrict the
312
syntax and semantics of identifiers using that scheme.
314
This specification defines those elements of the URI syntax that are
315
required of all URI schemes or are common to many URI schemes. It
316
thus defines the syntax and semantics needed to implement a scheme-
317
independent parsing mechanism for URI references, by which the
318
scheme-dependent handling of a URI can be postponed until the
319
scheme-dependent semantics are needed. Likewise, protocols and data
320
formats that make use of URI references can refer to this
321
specification as a definition for the range of syntax allowed for all
322
URIs, including those schemes that have yet to be defined. This
323
decouples the evolution of identification schemes from the evolution
324
of protocols, data formats, and implementations that make use of
327
A parser of the generic URI syntax can parse any URI reference into
328
its major components. Once the scheme is determined, further
329
scheme-specific parsing can be performed on the components. In other
330
words, the URI generic syntax is a superset of the syntax of all URI
338
Berners-Lee, et al. Standards Track [Page 6]
340
RFC 3986 URI Generic Syntax January 2005
345
The following example URIs illustrate several URI schemes and
346
variations in their common syntax components:
348
ftp://ftp.is.co.za/rfc/rfc1808.txt
350
http://www.ietf.org/rfc/rfc2396.txt
352
ldap://[2001:db8::7]/c=GB?objectClass?one
354
mailto:John.Doe@example.com
356
news:comp.infosystems.www.servers.unix
360
telnet://192.0.2.16:80/
362
urn:oasis:names:specification:docbook:dtd:xml:4.1.2
365
1.1.3. URI, URL, and URN
367
A URI can be further classified as a locator, a name, or both. The
368
term "Uniform Resource Locator" (URL) refers to the subset of URIs
369
that, in addition to identifying a resource, provide a means of
370
locating the resource by describing its primary access mechanism
371
(e.g., its network "location"). The term "Uniform Resource Name"
372
(URN) has been used historically to refer to both URIs under the
373
"urn" scheme [RFC2141], which are required to remain globally unique
374
and persistent even when the resource ceases to exist or becomes
375
unavailable, and to any other URI with the properties of a name.
377
An individual scheme does not have to be classified as being just one
378
of "name" or "locator". Instances of URIs from any given scheme may
379
have the characteristics of names or locators or both, often
380
depending on the persistence and care in the assignment of
381
identifiers by the naming authority, rather than on any quality of
382
the scheme. Future specifications and related documentation should
383
use the general term "URI" rather than the more restrictive terms
384
"URL" and "URN" [RFC3305].
394
Berners-Lee, et al. Standards Track [Page 7]
396
RFC 3986 URI Generic Syntax January 2005
399
1.2. Design Considerations
403
The URI syntax has been designed with global transcription as one of
404
its main considerations. A URI is a sequence of characters from a
405
very limited set: the letters of the basic Latin alphabet, digits,
406
and a few special characters. A URI may be represented in a variety
407
of ways; e.g., ink on paper, pixels on a screen, or a sequence of
408
character encoding octets. The interpretation of a URI depends only
409
on the characters used and not on how those characters are
410
represented in a network protocol.
412
The goal of transcription can be described by a simple scenario.
413
Imagine two colleagues, Sam and Kim, sitting in a pub at an
414
international conference and exchanging research ideas. Sam asks Kim
415
for a location to get more information, so Kim writes the URI for the
416
research site on a napkin. Upon returning home, Sam takes out the
417
napkin and types the URI into a computer, which then retrieves the
418
information to which Kim referred.
420
There are several design considerations revealed by the scenario:
422
o A URI is a sequence of characters that is not always represented
423
as a sequence of octets.
425
o A URI might be transcribed from a non-network source and thus
426
should consist of characters that are most likely able to be
427
entered into a computer, within the constraints imposed by
428
keyboards (and related input devices) across languages and
431
o A URI often has to be remembered by people, and it is easier for
432
people to remember a URI when it consists of meaningful or
435
These design considerations are not always in alignment. For
436
example, it is often the case that the most meaningful name for a URI
437
component would require characters that cannot be typed into some
438
systems. The ability to transcribe a resource identifier from one
439
medium to another has been considered more important than having a
440
URI consist of the most meaningful of components.
442
In local or regional contexts and with improving technology, users
443
might benefit from being able to use a wider range of characters;
444
such use is not defined by this specification. Percent-encoded
445
octets (Section 2.1) may be used within a URI to represent characters
446
outside the range of the US-ASCII coded character set if this
450
Berners-Lee, et al. Standards Track [Page 8]
452
RFC 3986 URI Generic Syntax January 2005
455
representation is allowed by the scheme or by the protocol element in
456
which the URI is referenced. Such a definition should specify the
457
character encoding used to map those characters to octets prior to
458
being percent-encoded for the URI.
460
1.2.2. Separating Identification from Interaction
462
A common misunderstanding of URIs is that they are only used to refer
463
to accessible resources. The URI itself only provides
464
identification; access to the resource is neither guaranteed nor
465
implied by the presence of a URI. Instead, any operation associated
466
with a URI reference is defined by the protocol element, data format
467
attribute, or natural language text in which it appears.
469
Given a URI, a system may attempt to perform a variety of operations
470
on the resource, as might be characterized by words such as "access",
471
"update", "replace", or "find attributes". Such operations are
472
defined by the protocols that make use of URIs, not by this
473
specification. However, we do use a few general terms for describing
474
common operations on URIs. URI "resolution" is the process of
475
determining an access mechanism and the appropriate parameters
476
necessary to dereference a URI; this resolution may require several
477
iterations. To use that access mechanism to perform an action on the
478
URI's resource is to "dereference" the URI.
480
When URIs are used within information retrieval systems to identify
481
sources of information, the most common form of URI dereference is
482
"retrieval": making use of a URI in order to retrieve a
483
representation of its associated resource. A "representation" is a
484
sequence of octets, along with representation metadata describing
485
those octets, that constitutes a record of the state of the resource
486
at the time when the representation is generated. Retrieval is
487
achieved by a process that might include using the URI as a cache key
488
to check for a locally cached representation, resolution of the URI
489
to determine an appropriate access mechanism (if any), and
490
dereference of the URI for the sake of applying a retrieval
491
operation. Depending on the protocols used to perform the retrieval,
492
additional information might be supplied about the resource (resource
493
metadata) and its relation to other resources.
495
URI references in information retrieval systems are designed to be
496
late-binding: the result of an access is generally determined when it
497
is accessed and may vary over time or due to other aspects of the
498
interaction. These references are created in order to be used in the
499
future: what is being identified is not some specific result that was
500
obtained in the past, but rather some characteristic that is expected
501
to be true for future results. In such cases, the resource referred
502
to by the URI is actually a sameness of characteristics as observed
506
Berners-Lee, et al. Standards Track [Page 9]
508
RFC 3986 URI Generic Syntax January 2005
511
over time, perhaps elucidated by additional comments or assertions
512
made by the resource provider.
514
Although many URI schemes are named after protocols, this does not
515
imply that use of these URIs will result in access to the resource
516
via the named protocol. URIs are often used simply for the sake of
517
identification. Even when a URI is used to retrieve a representation
518
of a resource, that access might be through gateways, proxies,
519
caches, and name resolution services that are independent of the
520
protocol associated with the scheme name. The resolution of some
521
URIs may require the use of more than one protocol (e.g., both DNS
522
and HTTP are typically used to access an "http" URI's origin server
523
when a representation isn't found in a local cache).
525
1.2.3. Hierarchical Identifiers
527
The URI syntax is organized hierarchically, with components listed in
528
order of decreasing significance from left to right. For some URI
529
schemes, the visible hierarchy is limited to the scheme itself:
530
everything after the scheme component delimiter (":") is considered
531
opaque to URI processing. Other URI schemes make the hierarchy
532
explicit and visible to generic parsing algorithms.
534
The generic syntax uses the slash ("/"), question mark ("?"), and
535
number sign ("#") characters to delimit components that are
536
significant to the generic parser's hierarchical interpretation of an
537
identifier. In addition to aiding the readability of such
538
identifiers through the consistent use of familiar syntax, this
539
uniform representation of hierarchy across naming schemes allows
540
scheme-independent references to be made relative to that hierarchy.
542
It is often the case that a group or "tree" of documents has been
543
constructed to serve a common purpose, wherein the vast majority of
544
URI references in these documents point to resources within the tree
545
rather than outside it. Similarly, documents located at a particular
546
site are much more likely to refer to other resources at that site
547
than to resources at remote sites. Relative referencing of URIs
548
allows document trees to be partially independent of their location
549
and access scheme. For instance, it is possible for a single set of
550
hypertext documents to be simultaneously accessible and traversable
551
via each of the "file", "http", and "ftp" schemes if the documents
552
refer to each other with relative references. Furthermore, such
553
document trees can be moved, as a whole, without changing any of the
556
A relative reference (Section 4.2) refers to a resource by describing
557
the difference within a hierarchical name space between the reference
558
context and the target URI. The reference resolution algorithm,
562
Berners-Lee, et al. Standards Track [Page 10]
564
RFC 3986 URI Generic Syntax January 2005
567
presented in Section 5, defines how such a reference is transformed
568
to the target URI. As relative references can only be used within
569
the context of a hierarchical URI, designers of new URI schemes
570
should use a syntax consistent with the generic syntax's hierarchical
571
components unless there are compelling reasons to forbid relative
572
referencing within that scheme.
574
NOTE: Previous specifications used the terms "partial URI" and
575
"relative URI" to denote a relative reference to a URI. As some
576
readers misunderstood those terms to mean that relative URIs are a
577
subset of URIs rather than a method of referencing URIs, this
578
specification simply refers to them as relative references.
580
All URI references are parsed by generic syntax parsers when used.
581
However, because hierarchical processing has no effect on an absolute
582
URI used in a reference unless it contains one or more dot-segments
583
(complete path segments of "." or "..", as described in Section 3.3),
584
URI scheme specifications can define opaque identifiers by
585
disallowing use of slash characters, question mark characters, and
586
the URIs "scheme:." and "scheme:..".
590
This specification uses the Augmented Backus-Naur Form (ABNF)
591
notation of [RFC2234], including the following core ABNF syntax rules
592
defined by that specification: ALPHA (letters), CR (carriage return),
593
DIGIT (decimal digits), DQUOTE (double quote), HEXDIG (hexadecimal
594
digits), LF (line feed), and SP (space). The complete URI syntax is
595
collected in Appendix A.
599
The URI syntax provides a method of encoding data, presumably for the
600
sake of identifying a resource, as a sequence of characters. The URI
601
characters are, in turn, frequently encoded as octets for transport
602
or presentation. This specification does not mandate any particular
603
character encoding for mapping between URI characters and the octets
604
used to store or transmit those characters. When a URI appears in a
605
protocol element, the character encoding is defined by that protocol;
606
without such a definition, a URI is assumed to be in the same
607
character encoding as the surrounding text.
609
The ABNF notation defines its terminal values to be non-negative
610
integers (codepoints) based on the US-ASCII coded character set
611
[ASCII]. Because a URI is a sequence of characters, we must invert
612
that relation in order to understand the URI syntax. Therefore, the
618
Berners-Lee, et al. Standards Track [Page 11]
620
RFC 3986 URI Generic Syntax January 2005
623
integer values used by the ABNF must be mapped back to their
624
corresponding characters via US-ASCII in order to complete the syntax
627
A URI is composed from a limited set of characters consisting of
628
digits, letters, and a few graphic symbols. A reserved subset of
629
those characters may be used to delimit syntax components within a
630
URI while the remaining characters, including both the unreserved set
631
and those reserved characters not acting as delimiters, define each
632
component's identifying data.
634
2.1. Percent-Encoding
636
A percent-encoding mechanism is used to represent a data octet in a
637
component when that octet's corresponding character is outside the
638
allowed set or is being used as a delimiter of, or within, the
639
component. A percent-encoded octet is encoded as a character
640
triplet, consisting of the percent character "%" followed by the two
641
hexadecimal digits representing that octet's numeric value. For
642
example, "%20" is the percent-encoding for the binary octet
643
"00100000" (ABNF: %x20), which in US-ASCII corresponds to the space
644
character (SP). Section 2.4 describes when percent-encoding and
647
pct-encoded = "%" HEXDIG HEXDIG
649
The uppercase hexadecimal digits 'A' through 'F' are equivalent to
650
the lowercase digits 'a' through 'f', respectively. If two URIs
651
differ only in the case of hexadecimal digits used in percent-encoded
652
octets, they are equivalent. For consistency, URI producers and
653
normalizers should use uppercase hexadecimal digits for all percent-
656
2.2. Reserved Characters
658
URIs include components and subcomponents that are delimited by
659
characters in the "reserved" set. These characters are called
660
"reserved" because they may (or may not) be defined as delimiters by
661
the generic syntax, by each scheme-specific syntax, or by the
662
implementation-specific syntax of a URI's dereferencing algorithm.
663
If data for a URI component would conflict with a reserved
664
character's purpose as a delimiter, then the conflicting data must be
665
percent-encoded before the URI is formed.
674
Berners-Lee, et al. Standards Track [Page 12]
676
RFC 3986 URI Generic Syntax January 2005
679
reserved = gen-delims / sub-delims
681
gen-delims = ":" / "/" / "?" / "#" / "[" / "]" / "@"
683
sub-delims = "!" / "$" / "&" / "'" / "(" / ")"
684
/ "*" / "+" / "," / ";" / "="
686
The purpose of reserved characters is to provide a set of delimiting
687
characters that are distinguishable from other data within a URI.
688
URIs that differ in the replacement of a reserved character with its
689
corresponding percent-encoded octet are not equivalent. Percent-
690
encoding a reserved character, or decoding a percent-encoded octet
691
that corresponds to a reserved character, will change how the URI is
692
interpreted by most applications. Thus, characters in the reserved
693
set are protected from normalization and are therefore safe to be
694
used by scheme-specific and producer-specific algorithms for
695
delimiting data subcomponents within a URI.
697
A subset of the reserved characters (gen-delims) is used as
698
delimiters of the generic URI components described in Section 3. A
699
component's ABNF syntax rule will not use the reserved or gen-delims
700
rule names directly; instead, each syntax rule lists the characters
701
allowed within that component (i.e., not delimiting it), and any of
702
those characters that are also in the reserved set are "reserved" for
703
use as subcomponent delimiters within the component. Only the most
704
common subcomponents are defined by this specification; other
705
subcomponents may be defined by a URI scheme's specification, or by
706
the implementation-specific syntax of a URI's dereferencing
707
algorithm, provided that such subcomponents are delimited by
708
characters in the reserved set allowed within that component.
710
URI producing applications should percent-encode data octets that
711
correspond to characters in the reserved set unless these characters
712
are specifically allowed by the URI scheme to represent data in that
713
component. If a reserved character is found in a URI component and
714
no delimiting role is known for that character, then it must be
715
interpreted as representing the data octet corresponding to that
716
character's encoding in US-ASCII.
718
2.3. Unreserved Characters
720
Characters that are allowed in a URI but do not have a reserved
721
purpose are called unreserved. These include uppercase and lowercase
722
letters, decimal digits, hyphen, period, underscore, and tilde.
724
unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~"
730
Berners-Lee, et al. Standards Track [Page 13]
732
RFC 3986 URI Generic Syntax January 2005
735
URIs that differ in the replacement of an unreserved character with
736
its corresponding percent-encoded US-ASCII octet are equivalent: they
737
identify the same resource. However, URI comparison implementations
738
do not always perform normalization prior to comparison (see Section
739
6). For consistency, percent-encoded octets in the ranges of ALPHA
740
(%41-%5A and %61-%7A), DIGIT (%30-%39), hyphen (%2D), period (%2E),
741
underscore (%5F), or tilde (%7E) should not be created by URI
742
producers and, when found in a URI, should be decoded to their
743
corresponding unreserved characters by URI normalizers.
745
2.4. When to Encode or Decode
747
Under normal circumstances, the only time when octets within a URI
748
are percent-encoded is during the process of producing the URI from
749
its component parts. This is when an implementation determines which
750
of the reserved characters are to be used as subcomponent delimiters
751
and which can be safely used as data. Once produced, a URI is always
752
in its percent-encoded form.
754
When a URI is dereferenced, the components and subcomponents
755
significant to the scheme-specific dereferencing process (if any)
756
must be parsed and separated before the percent-encoded octets within
757
those components can be safely decoded, as otherwise the data may be
758
mistaken for component delimiters. The only exception is for
759
percent-encoded octets corresponding to characters in the unreserved
760
set, which can be decoded at any time. For example, the octet
761
corresponding to the tilde ("~") character is often encoded as "%7E"
762
by older URI processing implementations; the "%7E" can be replaced by
763
"~" without changing its interpretation.
765
Because the percent ("%") character serves as the indicator for
766
percent-encoded octets, it must be percent-encoded as "%25" for that
767
octet to be used as data within a URI. Implementations must not
768
percent-encode or decode the same string more than once, as decoding
769
an already decoded string might lead to misinterpreting a percent
770
data octet as the beginning of a percent-encoding, or vice versa in
771
the case of percent-encoding an already percent-encoded string.
773
2.5. Identifying Data
775
URI characters provide identifying data for each of the URI
776
components, serving as an external interface for identification
777
between systems. Although the presence and nature of the URI
778
production interface is hidden from clients that use its URIs (and is
779
thus beyond the scope of the interoperability requirements defined by
780
this specification), it is a frequent source of confusion and errors
781
in the interpretation of URI character issues. Implementers have to
782
be aware that there are multiple character encodings involved in the
786
Berners-Lee, et al. Standards Track [Page 14]
788
RFC 3986 URI Generic Syntax January 2005
791
production and transmission of URIs: local name and data encoding,
792
public interface encoding, URI character encoding, data format
793
encoding, and protocol encoding.
795
Local names, such as file system names, are stored with a local
796
character encoding. URI producing applications (e.g., origin
797
servers) will typically use the local encoding as the basis for
798
producing meaningful names. The URI producer will transform the
799
local encoding to one that is suitable for a public interface and
800
then transform the public interface encoding into the restricted set
801
of URI characters (reserved, unreserved, and percent-encodings).
802
Those characters are, in turn, encoded as octets to be used as a
803
reference within a data format (e.g., a document charset), and such
804
data formats are often subsequently encoded for transmission over
807
For most systems, an unreserved character appearing within a URI
808
component is interpreted as representing the data octet corresponding
809
to that character's encoding in US-ASCII. Consumers of URIs assume
810
that the letter "X" corresponds to the octet "01011000", and even
811
when that assumption is incorrect, there is no harm in making it. A
812
system that internally provides identifiers in the form of a
813
different character encoding, such as EBCDIC, will generally perform
814
character translation of textual identifiers to UTF-8 [STD63] (or
815
some other superset of the US-ASCII character encoding) at an
816
internal interface, thereby providing more meaningful identifiers
817
than those resulting from simply percent-encoding the original
820
For example, consider an information service that provides data,
821
stored locally using an EBCDIC-based file system, to clients on the
822
Internet through an HTTP server. When an author creates a file with
823
the name "Laguna Beach" on that file system, the "http" URI
824
corresponding to that resource is expected to contain the meaningful
825
string "Laguna%20Beach". If, however, that server produces URIs by
826
using an overly simplistic raw octet mapping, then the result would
827
be a URI containing "%D3%81%87%A4%95%81@%C2%85%81%83%88". An
828
internal transcoding interface fixes this problem by transcoding the
829
local name to a superset of US-ASCII prior to producing the URI.
830
Naturally, proper interpretation of an incoming URI on such an
831
interface requires that percent-encoded octets be decoded (e.g.,
832
"%20" to SP) before the reverse transcoding is applied to obtain the
835
In some cases, the internal interface between a URI component and the
836
identifying data that it has been crafted to represent is much less
837
direct than a character encoding translation. For example, portions
838
of a URI might reflect a query on non-ASCII data, or numeric
842
Berners-Lee, et al. Standards Track [Page 15]
844
RFC 3986 URI Generic Syntax January 2005
847
coordinates on a map. Likewise, a URI scheme may define components
848
with additional encoding requirements that are applied prior to
849
forming the component and producing the URI.
851
When a new URI scheme defines a component that represents textual
852
data consisting of characters from the Universal Character Set [UCS],
853
the data should first be encoded as octets according to the UTF-8
854
character encoding [STD63]; then only those octets that do not
855
correspond to characters in the unreserved set should be percent-
856
encoded. For example, the character A would be represented as "A",
857
the character LATIN CAPITAL LETTER A WITH GRAVE would be represented
858
as "%C3%80", and the character KATAKANA LETTER A would be represented
863
The generic URI syntax consists of a hierarchical sequence of
864
components referred to as the scheme, authority, path, query, and
867
URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ]
869
hier-part = "//" authority path-abempty
874
The scheme and path components are required, though the path may be
875
empty (no characters). When authority is present, the path must
876
either be empty or begin with a slash ("/") character. When
877
authority is not present, the path cannot begin with two slash
878
characters ("//"). These restrictions result in five different ABNF
879
rules for a path (Section 3.3), only one of which will match any
882
The following are two example URIs and their component parts:
884
foo://example.com:8042/over/there?name=ferret#nose
885
\_/ \______________/\_________/ \_________/ \__/
887
scheme authority path query fragment
888
| _____________________|__
890
urn:example:animal:ferret:nose
898
Berners-Lee, et al. Standards Track [Page 16]
900
RFC 3986 URI Generic Syntax January 2005
905
Each URI begins with a scheme name that refers to a specification for
906
assigning identifiers within that scheme. As such, the URI syntax is
907
a federated and extensible naming system wherein each scheme's
908
specification may further restrict the syntax and semantics of
909
identifiers using that scheme.
911
Scheme names consist of a sequence of characters beginning with a
912
letter and followed by any combination of letters, digits, plus
913
("+"), period ("."), or hyphen ("-"). Although schemes are case-
914
insensitive, the canonical form is lowercase and documents that
915
specify schemes must do so with lowercase letters. An implementation
916
should accept uppercase letters as equivalent to lowercase in scheme
917
names (e.g., allow "HTTP" as well as "http") for the sake of
918
robustness but should only produce lowercase scheme names for
921
scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
923
Individual schemes are not specified by this document. The process
924
for registration of new URI schemes is defined separately by [BCP35].
925
The scheme registry maintains the mapping between scheme names and
926
their specifications. Advice for designers of new URI schemes can be
927
found in [RFC2718]. URI scheme specifications must define their own
928
syntax so that all strings matching their scheme-specific syntax will
929
also match the <absolute-URI> grammar, as described in Section 4.3.
931
When presented with a URI that violates one or more scheme-specific
932
restrictions, the scheme-specific resolution process should flag the
933
reference as an error rather than ignore the unused parts; doing so
934
reduces the number of equivalent URIs and helps detect abuses of the
935
generic syntax, which might indicate that the URI has been
936
constructed to mislead the user (Section 7.6).
940
Many URI schemes include a hierarchical element for a naming
941
authority so that governance of the name space defined by the
942
remainder of the URI is delegated to that authority (which may, in
943
turn, delegate it further). The generic syntax provides a common
944
means for distinguishing an authority based on a registered name or
945
server address, along with optional port and user information.
947
The authority component is preceded by a double slash ("//") and is
948
terminated by the next slash ("/"), question mark ("?"), or number
949
sign ("#") character, or by the end of the URI.
954
Berners-Lee, et al. Standards Track [Page 17]
956
RFC 3986 URI Generic Syntax January 2005
959
authority = [ userinfo "@" ] host [ ":" port ]
961
URI producers and normalizers should omit the ":" delimiter that
962
separates host from port if the port component is empty. Some
963
schemes do not allow the userinfo and/or port subcomponents.
965
If a URI contains an authority component, then the path component
966
must either be empty or begin with a slash ("/") character. Non-
967
validating parsers (those that merely separate a URI reference into
968
its major components) will often ignore the subcomponent structure of
969
authority, treating it as an opaque string from the double-slash to
970
the first terminating delimiter, until such time as the URI is
973
3.2.1. User Information
975
The userinfo subcomponent may consist of a user name and, optionally,
976
scheme-specific information about how to gain authorization to access
977
the resource. The user information, if present, is followed by a
978
commercial at-sign ("@") that delimits it from the host.
980
userinfo = *( unreserved / pct-encoded / sub-delims / ":" )
982
Use of the format "user:password" in the userinfo field is
983
deprecated. Applications should not render as clear text any data
984
after the first colon (":") character found within a userinfo
985
subcomponent unless the data after the colon is the empty string
986
(indicating no password). Applications may choose to ignore or
987
reject such data when it is received as part of a reference and
988
should reject the storage of such data in unencrypted form. The
989
passing of authentication information in clear text has proven to be
990
a security risk in almost every case where it has been used.
992
Applications that render a URI for the sake of user feedback, such as
993
in graphical hypertext browsing, should render userinfo in a way that
994
is distinguished from the rest of a URI, when feasible. Such
995
rendering will assist the user in cases where the userinfo has been
996
misleadingly crafted to look like a trusted domain name
1001
The host subcomponent of authority is identified by an IP literal
1002
encapsulated within square brackets, an IPv4 address in dotted-
1003
decimal form, or a registered name. The host subcomponent is case-
1004
insensitive. The presence of a host subcomponent within a URI does
1005
not imply that the scheme requires access to the given host on the
1006
Internet. In many cases, the host syntax is used only for the sake
1010
Berners-Lee, et al. Standards Track [Page 18]
1012
RFC 3986 URI Generic Syntax January 2005
1015
of reusing the existing registration process created and deployed for
1016
DNS, thus obtaining a globally unique name without the cost of
1017
deploying another registry. However, such use comes with its own
1018
costs: domain name ownership may change over time for reasons not
1019
anticipated by the URI producer. In other cases, the data within the
1020
host component identifies a registered name that has nothing to do
1021
with an Internet host. We use the name "host" for the ABNF rule
1022
because that is its most common purpose, not its only purpose.
1024
host = IP-literal / IPv4address / reg-name
1026
The syntax rule for host is ambiguous because it does not completely
1027
distinguish between an IPv4address and a reg-name. In order to
1028
disambiguate the syntax, we apply the "first-match-wins" algorithm:
1029
If host matches the rule for IPv4address, then it should be
1030
considered an IPv4 address literal and not a reg-name. Although host
1031
is case-insensitive, producers and normalizers should use lowercase
1032
for registered names and hexadecimal addresses for the sake of
1033
uniformity, while only using uppercase letters for percent-encodings.
1035
A host identified by an Internet Protocol literal address, version 6
1036
[RFC3513] or later, is distinguished by enclosing the IP literal
1037
within square brackets ("[" and "]"). This is the only place where
1038
square bracket characters are allowed in the URI syntax. In
1039
anticipation of future, as-yet-undefined IP literal address formats,
1040
an implementation may use an optional version flag to indicate such a
1041
format explicitly rather than rely on heuristic determination.
1043
IP-literal = "[" ( IPv6address / IPvFuture ) "]"
1045
IPvFuture = "v" 1*HEXDIG "." 1*( unreserved / sub-delims / ":" )
1047
The version flag does not indicate the IP version; rather, it
1048
indicates future versions of the literal format. As such,
1049
implementations must not provide the version flag for the existing
1050
IPv4 and IPv6 literal address forms described below. If a URI
1051
containing an IP-literal that starts with "v" (case-insensitive),
1052
indicating that the version flag is present, is dereferenced by an
1053
application that does not know the meaning of that version flag, then
1054
the application should return an appropriate error for "address
1055
mechanism not supported".
1057
A host identified by an IPv6 literal address is represented inside
1058
the square brackets without a preceding version flag. The ABNF
1059
provided here is a translation of the text definition of an IPv6
1060
literal address provided in [RFC3513]. This syntax does not support
1061
IPv6 scoped addressing zone identifiers.
1066
Berners-Lee, et al. Standards Track [Page 19]
1068
RFC 3986 URI Generic Syntax January 2005
1071
A 128-bit IPv6 address is divided into eight 16-bit pieces. Each
1072
piece is represented numerically in case-insensitive hexadecimal,
1073
using one to four hexadecimal digits (leading zeroes are permitted).
1074
The eight encoded pieces are given most-significant first, separated
1075
by colon characters. Optionally, the least-significant two pieces
1076
may instead be represented in IPv4 address textual format. A
1077
sequence of one or more consecutive zero-valued 16-bit pieces within
1078
the address may be elided, omitting all their digits and leaving
1079
exactly two consecutive colons in their place to mark the elision.
1081
IPv6address = 6( h16 ":" ) ls32
1082
/ "::" 5( h16 ":" ) ls32
1083
/ [ h16 ] "::" 4( h16 ":" ) ls32
1084
/ [ *1( h16 ":" ) h16 ] "::" 3( h16 ":" ) ls32
1085
/ [ *2( h16 ":" ) h16 ] "::" 2( h16 ":" ) ls32
1086
/ [ *3( h16 ":" ) h16 ] "::" h16 ":" ls32
1087
/ [ *4( h16 ":" ) h16 ] "::" ls32
1088
/ [ *5( h16 ":" ) h16 ] "::" h16
1089
/ [ *6( h16 ":" ) h16 ] "::"
1091
ls32 = ( h16 ":" h16 ) / IPv4address
1092
; least-significant 32 bits of address
1095
; 16 bits of address represented in hexadecimal
1097
A host identified by an IPv4 literal address is represented in
1098
dotted-decimal notation (a sequence of four decimal numbers in the
1099
range 0 to 255, separated by "."), as described in [RFC1123] by
1100
reference to [RFC0952]. Note that other forms of dotted notation may
1101
be interpreted on some platforms, as described in Section 7.4, but
1102
only the dotted-decimal form of four octets is allowed by this
1105
IPv4address = dec-octet "." dec-octet "." dec-octet "." dec-octet
1107
dec-octet = DIGIT ; 0-9
1108
/ %x31-39 DIGIT ; 10-99
1109
/ "1" 2DIGIT ; 100-199
1110
/ "2" %x30-34 DIGIT ; 200-249
1111
/ "25" %x30-35 ; 250-255
1113
A host identified by a registered name is a sequence of characters
1114
usually intended for lookup within a locally defined host or service
1115
name registry, though the URI's scheme-specific semantics may require
1116
that a specific registry (or fixed name table) be used instead. The
1117
most common name registry mechanism is the Domain Name System (DNS).
1118
A registered name intended for lookup in the DNS uses the syntax
1122
Berners-Lee, et al. Standards Track [Page 20]
1124
RFC 3986 URI Generic Syntax January 2005
1127
defined in Section 3.5 of [RFC1034] and Section 2.1 of [RFC1123].
1128
Such a name consists of a sequence of domain labels separated by ".",
1129
each domain label starting and ending with an alphanumeric character
1130
and possibly also containing "-" characters. The rightmost domain
1131
label of a fully qualified domain name in DNS may be followed by a
1132
single "." and should be if it is necessary to distinguish between
1133
the complete domain name and some local domain.
1135
reg-name = *( unreserved / pct-encoded / sub-delims )
1137
If the URI scheme defines a default for host, then that default
1138
applies when the host subcomponent is undefined or when the
1139
registered name is empty (zero length). For example, the "file" URI
1140
scheme is defined so that no authority, an empty host, and
1141
"localhost" all mean the end-user's machine, whereas the "http"
1142
scheme considers a missing authority or empty host invalid.
1144
This specification does not mandate a particular registered name
1145
lookup technology and therefore does not restrict the syntax of reg-
1146
name beyond what is necessary for interoperability. Instead, it
1147
delegates the issue of registered name syntax conformance to the
1148
operating system of each application performing URI resolution, and
1149
that operating system decides what it will allow for the purpose of
1150
host identification. A URI resolution implementation might use DNS,
1151
host tables, yellow pages, NetInfo, WINS, or any other system for
1152
lookup of registered names. However, a globally scoped naming
1153
system, such as DNS fully qualified domain names, is necessary for
1154
URIs intended to have global scope. URI producers should use names
1155
that conform to the DNS syntax, even when use of DNS is not
1156
immediately apparent, and should limit these names to no more than
1157
255 characters in length.
1159
The reg-name syntax allows percent-encoded octets in order to
1160
represent non-ASCII registered names in a uniform way that is
1161
independent of the underlying name resolution technology. Non-ASCII
1162
characters must first be encoded according to UTF-8 [STD63], and then
1163
each octet of the corresponding UTF-8 sequence must be percent-
1164
encoded to be represented as URI characters. URI producing
1165
applications must not use percent-encoding in host unless it is used
1166
to represent a UTF-8 character sequence. When a non-ASCII registered
1167
name represents an internationalized domain name intended for
1168
resolution via the DNS, the name must be transformed to the IDNA
1169
encoding [RFC3490] prior to name lookup. URI producers should
1170
provide these registered names in the IDNA encoding, rather than a
1171
percent-encoding, if they wish to maximize interoperability with
1172
legacy URI resolvers.
1178
Berners-Lee, et al. Standards Track [Page 21]
1180
RFC 3986 URI Generic Syntax January 2005
1185
The port subcomponent of authority is designated by an optional port
1186
number in decimal following the host and delimited from it by a
1187
single colon (":") character.
1191
A scheme may define a default port. For example, the "http" scheme
1192
defines a default port of "80", corresponding to its reserved TCP
1193
port number. The type of port designated by the port number (e.g.,
1194
TCP, UDP, SCTP) is defined by the URI scheme. URI producers and
1195
normalizers should omit the port component and its ":" delimiter if
1196
port is empty or if its value would be the same as that of the
1201
The path component contains data, usually organized in hierarchical
1202
form, that, along with data in the non-hierarchical query component
1203
(Section 3.4), serves to identify a resource within the scope of the
1204
URI's scheme and naming authority (if any). The path is terminated
1205
by the first question mark ("?") or number sign ("#") character, or
1206
by the end of the URI.
1208
If a URI contains an authority component, then the path component
1209
must either be empty or begin with a slash ("/") character. If a URI
1210
does not contain an authority component, then the path cannot begin
1211
with two slash characters ("//"). In addition, a URI reference
1212
(Section 4.1) may be a relative-path reference, in which case the
1213
first path segment cannot contain a colon (":") character. The ABNF
1214
requires five separate rules to disambiguate these cases, only one of
1215
which will match the path substring within a given URI reference. We
1216
use the generic term "path component" to describe the URI substring
1217
matched by the parser to one of these rules.
1219
path = path-abempty ; begins with "/" or is empty
1220
/ path-absolute ; begins with "/" but not "//"
1221
/ path-noscheme ; begins with a non-colon segment
1222
/ path-rootless ; begins with a segment
1223
/ path-empty ; zero characters
1225
path-abempty = *( "/" segment )
1226
path-absolute = "/" [ segment-nz *( "/" segment ) ]
1227
path-noscheme = segment-nz-nc *( "/" segment )
1228
path-rootless = segment-nz *( "/" segment )
1229
path-empty = 0<pchar>
1234
Berners-Lee, et al. Standards Track [Page 22]
1236
RFC 3986 URI Generic Syntax January 2005
1240
segment-nz = 1*pchar
1241
segment-nz-nc = 1*( unreserved / pct-encoded / sub-delims / "@" )
1242
; non-zero-length segment without any colon ":"
1244
pchar = unreserved / pct-encoded / sub-delims / ":" / "@"
1246
A path consists of a sequence of path segments separated by a slash
1247
("/") character. A path is always defined for a URI, though the
1248
defined path may be empty (zero length). Use of the slash character
1249
to indicate hierarchy is only required when a URI will be used as the
1250
context for relative references. For example, the URI
1251
<mailto:fred@example.com> has a path of "fred@example.com", whereas
1252
the URI <foo://info.example.com?fred> has an empty path.
1254
The path segments "." and "..", also known as dot-segments, are
1255
defined for relative reference within the path name hierarchy. They
1256
are intended for use at the beginning of a relative-path reference
1257
(Section 4.2) to indicate relative position within the hierarchical
1258
tree of names. This is similar to their role within some operating
1259
systems' file directory structures to indicate the current directory
1260
and parent directory, respectively. However, unlike in a file
1261
system, these dot-segments are only interpreted within the URI path
1262
hierarchy and are removed as part of the resolution process (Section
1265
Aside from dot-segments in hierarchical paths, a path segment is
1266
considered opaque by the generic syntax. URI producing applications
1267
often use the reserved characters allowed in a segment to delimit
1268
scheme-specific or dereference-handler-specific subcomponents. For
1269
example, the semicolon (";") and equals ("=") reserved characters are
1270
often used to delimit parameters and parameter values applicable to
1271
that segment. The comma (",") reserved character is often used for
1272
similar purposes. For example, one URI producer might use a segment
1273
such as "name;v=1.1" to indicate a reference to version 1.1 of
1274
"name", whereas another might use a segment such as "name,1.1" to
1275
indicate the same. Parameter types may be defined by scheme-specific
1276
semantics, but in most cases the syntax of a parameter is specific to
1277
the implementation of the URI's dereferencing algorithm.
1281
The query component contains non-hierarchical data that, along with
1282
data in the path component (Section 3.3), serves to identify a
1283
resource within the scope of the URI's scheme and naming authority
1284
(if any). The query component is indicated by the first question
1285
mark ("?") character and terminated by a number sign ("#") character
1286
or by the end of the URI.
1290
Berners-Lee, et al. Standards Track [Page 23]
1292
RFC 3986 URI Generic Syntax January 2005
1295
query = *( pchar / "/" / "?" )
1297
The characters slash ("/") and question mark ("?") may represent data
1298
within the query component. Beware that some older, erroneous
1299
implementations may not handle such data correctly when it is used as
1300
the base URI for relative references (Section 5.1), apparently
1301
because they fail to distinguish query data from path data when
1302
looking for hierarchical separators. However, as query components
1303
are often used to carry identifying information in the form of
1304
"key=value" pairs and one frequently used value is a reference to
1305
another URI, it is sometimes better for usability to avoid percent-
1306
encoding those characters.
1310
The fragment identifier component of a URI allows indirect
1311
identification of a secondary resource by reference to a primary
1312
resource and additional identifying information. The identified
1313
secondary resource may be some portion or subset of the primary
1314
resource, some view on representations of the primary resource, or
1315
some other resource defined or described by those representations. A
1316
fragment identifier component is indicated by the presence of a
1317
number sign ("#") character and terminated by the end of the URI.
1319
fragment = *( pchar / "/" / "?" )
1321
The semantics of a fragment identifier are defined by the set of
1322
representations that might result from a retrieval action on the
1323
primary resource. The fragment's format and resolution is therefore
1324
dependent on the media type [RFC2046] of a potentially retrieved
1325
representation, even though such a retrieval is only performed if the
1326
URI is dereferenced. If no such representation exists, then the
1327
semantics of the fragment are considered unknown and are effectively
1328
unconstrained. Fragment identifier semantics are independent of the
1329
URI scheme and thus cannot be redefined by scheme specifications.
1331
Individual media types may define their own restrictions on or
1332
structures within the fragment identifier syntax for specifying
1333
different types of subsets, views, or external references that are
1334
identifiable as secondary resources by that media type. If the
1335
primary resource has multiple representations, as is often the case
1336
for resources whose representation is selected based on attributes of
1337
the retrieval request (a.k.a., content negotiation), then whatever is
1338
identified by the fragment should be consistent across all of those
1339
representations. Each representation should either define the
1340
fragment so that it corresponds to the same secondary resource,
1341
regardless of how it is represented, or should leave the fragment
1342
undefined (i.e., not found).
1346
Berners-Lee, et al. Standards Track [Page 24]
1348
RFC 3986 URI Generic Syntax January 2005
1351
As with any URI, use of a fragment identifier component does not
1352
imply that a retrieval action will take place. A URI with a fragment
1353
identifier may be used to refer to the secondary resource without any
1354
implication that the primary resource is accessible or will ever be
1357
Fragment identifiers have a special role in information retrieval
1358
systems as the primary form of client-side indirect referencing,
1359
allowing an author to specifically identify aspects of an existing
1360
resource that are only indirectly provided by the resource owner. As
1361
such, the fragment identifier is not used in the scheme-specific
1362
processing of a URI; instead, the fragment identifier is separated
1363
from the rest of the URI prior to a dereference, and thus the
1364
identifying information within the fragment itself is dereferenced
1365
solely by the user agent, regardless of the URI scheme. Although
1366
this separate handling is often perceived to be a loss of
1367
information, particularly for accurate redirection of references as
1368
resources move over time, it also serves to prevent information
1369
providers from denying reference authors the right to refer to
1370
information within a resource selectively. Indirect referencing also
1371
provides additional flexibility and extensibility to systems that use
1372
URIs, as new media types are easier to define and deploy than new
1373
schemes of identification.
1375
The characters slash ("/") and question mark ("?") are allowed to
1376
represent data within the fragment identifier. Beware that some
1377
older, erroneous implementations may not handle this data correctly
1378
when it is used as the base URI for relative references (Section
1383
When applications make reference to a URI, they do not always use the
1384
full form of reference defined by the "URI" syntax rule. To save
1385
space and take advantage of hierarchical locality, many Internet
1386
protocol elements and media type formats allow an abbreviation of a
1387
URI, whereas others restrict the syntax to a particular form of URI.
1388
We define the most common forms of reference syntax in this
1389
specification because they impact and depend upon the design of the
1390
generic syntax, requiring a uniform parsing algorithm in order to be
1391
interpreted consistently.
1395
URI-reference is used to denote the most common usage of a resource
1398
URI-reference = URI / relative-ref
1402
Berners-Lee, et al. Standards Track [Page 25]
1404
RFC 3986 URI Generic Syntax January 2005
1407
A URI-reference is either a URI or a relative reference. If the
1408
URI-reference's prefix does not match the syntax of a scheme followed
1409
by its colon separator, then the URI-reference is a relative
1412
A URI-reference is typically parsed first into the five URI
1413
components, in order to determine what components are present and
1414
whether the reference is relative. Then, each component is parsed
1415
for its subparts and their validation. The ABNF of URI-reference,
1416
along with the "first-match-wins" disambiguation rule, is sufficient
1417
to define a validating parser for the generic syntax. Readers
1418
familiar with regular expressions should see Appendix B for an
1419
example of a non-validating URI-reference parser that will take any
1420
given string and extract the URI components.
1422
4.2. Relative Reference
1424
A relative reference takes advantage of the hierarchical syntax
1425
(Section 1.2.3) to express a URI reference relative to the name space
1426
of another hierarchical URI.
1428
relative-ref = relative-part [ "?" query ] [ "#" fragment ]
1430
relative-part = "//" authority path-abempty
1435
The URI referred to by a relative reference, also known as the target
1436
URI, is obtained by applying the reference resolution algorithm of
1439
A relative reference that begins with two slash characters is termed
1440
a network-path reference; such references are rarely used. A
1441
relative reference that begins with a single slash character is
1442
termed an absolute-path reference. A relative reference that does
1443
not begin with a slash character is termed a relative-path reference.
1445
A path segment that contains a colon character (e.g., "this:that")
1446
cannot be used as the first segment of a relative-path reference, as
1447
it would be mistaken for a scheme name. Such a segment must be
1448
preceded by a dot-segment (e.g., "./this:that") to make a relative-
1458
Berners-Lee, et al. Standards Track [Page 26]
1460
RFC 3986 URI Generic Syntax January 2005
1465
Some protocol elements allow only the absolute form of a URI without
1466
a fragment identifier. For example, defining a base URI for later
1467
use by relative references calls for an absolute-URI syntax rule that
1468
does not allow a fragment.
1470
absolute-URI = scheme ":" hier-part [ "?" query ]
1472
URI scheme specifications must define their own syntax so that all
1473
strings matching their scheme-specific syntax will also match the
1474
<absolute-URI> grammar. Scheme specifications will not define
1475
fragment identifier syntax or usage, regardless of its applicability
1476
to resources identifiable via that scheme, as fragment identification
1477
is orthogonal to scheme definition. However, scheme specifications
1478
are encouraged to include a wide range of examples, including
1479
examples that show use of the scheme's URIs with fragment identifiers
1480
when such usage is appropriate.
1482
4.4. Same-Document Reference
1484
When a URI reference refers to a URI that is, aside from its fragment
1485
component (if any), identical to the base URI (Section 5.1), that
1486
reference is called a "same-document" reference. The most frequent
1487
examples of same-document references are relative references that are
1488
empty or include only the number sign ("#") separator followed by a
1489
fragment identifier.
1491
When a same-document reference is dereferenced for a retrieval
1492
action, the target of that reference is defined to be within the same
1493
entity (representation, document, or message) as the reference;
1494
therefore, a dereference should not result in a new retrieval action.
1496
Normalization of the base and target URIs prior to their comparison,
1497
as described in Sections 6.2.2 and 6.2.3, is allowed but rarely
1498
performed in practice. Normalization may increase the set of same-
1499
document references, which may be of benefit to some caching
1500
applications. As such, reference authors should not assume that a
1501
slightly different, though equivalent, reference URI will (or will
1502
not) be interpreted as a same-document reference by any given
1505
4.5. Suffix Reference
1507
The URI syntax is designed for unambiguous reference to resources and
1508
extensibility via the URI scheme. However, as URI identification and
1509
usage have become commonplace, traditional media (television, radio,
1510
newspapers, billboards, etc.) have increasingly used a suffix of the
1514
Berners-Lee, et al. Standards Track [Page 27]
1516
RFC 3986 URI Generic Syntax January 2005
1519
URI as a reference, consisting of only the authority and path
1520
portions of the URI, such as
1522
www.w3.org/Addressing/
1524
or simply a DNS registered name on its own. Such references are
1525
primarily intended for human interpretation rather than for machines,
1526
with the assumption that context-based heuristics are sufficient to
1527
complete the URI (e.g., most registered names beginning with "www"
1528
are likely to have a URI prefix of "http://"). Although there is no
1529
standard set of heuristics for disambiguating a URI suffix, many
1530
client implementations allow them to be entered by the user and
1531
heuristically resolved.
1533
Although this practice of using suffix references is common, it
1534
should be avoided whenever possible and should never be used in
1535
situations where long-term references are expected. The heuristics
1536
noted above will change over time, particularly when a new URI scheme
1537
becomes popular, and are often incorrect when used out of context.
1538
Furthermore, they can lead to security issues along the lines of
1539
those described in [RFC1535].
1541
As a URI suffix has the same syntax as a relative-path reference, a
1542
suffix reference cannot be used in contexts where a relative
1543
reference is expected. As a result, suffix references are limited to
1544
places where there is no defined base URI, such as dialog boxes and
1545
off-line advertisements.
1547
5. Reference Resolution
1549
This section defines the process of resolving a URI reference within
1550
a context that allows relative references so that the result is a
1551
string matching the <URI> syntax rule of Section 3.
1553
5.1. Establishing a Base URI
1555
The term "relative" implies that a "base URI" exists against which
1556
the relative reference is applied. Aside from fragment-only
1557
references (Section 4.4), relative references are only usable when a
1558
base URI is known. A base URI must be established by the parser
1559
prior to parsing URI references that might be relative. A base URI
1560
must conform to the <absolute-URI> syntax rule (Section 4.3). If the
1561
base URI is obtained from a URI reference, then that reference must
1562
be converted to absolute form and stripped of any fragment component
1563
prior to its use as a base URI.
1570
Berners-Lee, et al. Standards Track [Page 28]
1572
RFC 3986 URI Generic Syntax January 2005
1575
The base URI of a reference can be established in one of four ways,
1576
discussed below in order of precedence. The order of precedence can
1577
be thought of in terms of layers, where the innermost defined base
1578
URI has the highest precedence. This can be visualized graphically
1581
.----------------------------------------------------------.
1582
| .----------------------------------------------------. |
1583
| | .----------------------------------------------. | |
1584
| | | .----------------------------------------. | | |
1585
| | | | .----------------------------------. | | | |
1586
| | | | | <relative-reference> | | | | |
1587
| | | | `----------------------------------' | | | |
1588
| | | | (5.1.1) Base URI embedded in content | | | |
1589
| | | `----------------------------------------' | | |
1590
| | | (5.1.2) Base URI of the encapsulating entity | | |
1591
| | | (message, representation, or none) | | |
1592
| | `----------------------------------------------' | |
1593
| | (5.1.3) URI used to retrieve the entity | |
1594
| `----------------------------------------------------' |
1595
| (5.1.4) Default Base URI (application-dependent) |
1596
`----------------------------------------------------------'
1598
5.1.1. Base URI Embedded in Content
1600
Within certain media types, a base URI for relative references can be
1601
embedded within the content itself so that it can be readily obtained
1602
by a parser. This can be useful for descriptive documents, such as
1603
tables of contents, which may be transmitted to others through
1604
protocols other than their usual retrieval context (e.g., email or
1607
It is beyond the scope of this specification to specify how, for each
1608
media type, a base URI can be embedded. The appropriate syntax, when
1609
available, is described by the data format specification associated
1610
with each media type.
1612
5.1.2. Base URI from the Encapsulating Entity
1614
If no base URI is embedded, the base URI is defined by the
1615
representation's retrieval context. For a document that is enclosed
1616
within another entity, such as a message or archive, the retrieval
1617
context is that entity. Thus, the default base URI of a
1618
representation is the base URI of the entity in which the
1619
representation is encapsulated.
1626
Berners-Lee, et al. Standards Track [Page 29]
1628
RFC 3986 URI Generic Syntax January 2005
1631
A mechanism for embedding a base URI within MIME container types
1632
(e.g., the message and multipart types) is defined by MHTML
1633
[RFC2557]. Protocols that do not use the MIME message header syntax,
1634
but that do allow some form of tagged metadata to be included within
1635
messages, may define their own syntax for defining a base URI as part
1638
5.1.3. Base URI from the Retrieval URI
1640
If no base URI is embedded and the representation is not encapsulated
1641
within some other entity, then, if a URI was used to retrieve the
1642
representation, that URI shall be considered the base URI. Note that
1643
if the retrieval was the result of a redirected request, the last URI
1644
used (i.e., the URI that resulted in the actual retrieval of the
1645
representation) is the base URI.
1647
5.1.4. Default Base URI
1649
If none of the conditions described above apply, then the base URI is
1650
defined by the context of the application. As this definition is
1651
necessarily application-dependent, failing to define a base URI by
1652
using one of the other methods may result in the same content being
1653
interpreted differently by different types of applications.
1655
A sender of a representation containing relative references is
1656
responsible for ensuring that a base URI for those references can be
1657
established. Aside from fragment-only references, relative
1658
references can only be used reliably in situations where the base URI
1661
5.2. Relative Resolution
1663
This section describes an algorithm for converting a URI reference
1664
that might be relative to a given base URI into the parsed components
1665
of the reference's target. The components can then be recomposed, as
1666
described in Section 5.3, to form the target URI. This algorithm
1667
provides definitive results that can be used to test the output of
1668
other implementations. Applications may implement relative reference
1669
resolution by using some other algorithm, provided that the results
1670
match what would be given by this one.
1682
Berners-Lee, et al. Standards Track [Page 30]
1684
RFC 3986 URI Generic Syntax January 2005
1687
5.2.1. Pre-parse the Base URI
1689
The base URI (Base) is established according to the procedure of
1690
Section 5.1 and parsed into the five main components described in
1691
Section 3. Note that only the scheme component is required to be
1692
present in a base URI; the other components may be empty or
1693
undefined. A component is undefined if its associated delimiter does
1694
not appear in the URI reference; the path component is never
1695
undefined, though it may be empty.
1697
Normalization of the base URI, as described in Sections 6.2.2 and
1698
6.2.3, is optional. A URI reference must be transformed to its
1699
target URI before it can be normalized.
1701
5.2.2. Transform References
1703
For each URI reference (R), the following pseudocode describes an
1704
algorithm for transforming R into its target URI (T):
1706
-- The URI reference is parsed into the five URI components
1708
(R.scheme, R.authority, R.path, R.query, R.fragment) = parse(R);
1710
-- A non-strict parser may ignore a scheme in the reference
1711
-- if it is identical to the base URI's scheme.
1713
if ((not strict) and (R.scheme == Base.scheme)) then
1738
Berners-Lee, et al. Standards Track [Page 31]
1740
RFC 3986 URI Generic Syntax January 2005
1743
if defined(R.scheme) then
1744
T.scheme = R.scheme;
1745
T.authority = R.authority;
1746
T.path = remove_dot_segments(R.path);
1749
if defined(R.authority) then
1750
T.authority = R.authority;
1751
T.path = remove_dot_segments(R.path);
1754
if (R.path == "") then
1756
if defined(R.query) then
1759
T.query = Base.query;
1762
if (R.path starts-with "/") then
1763
T.path = remove_dot_segments(R.path);
1765
T.path = merge(Base.path, R.path);
1766
T.path = remove_dot_segments(T.path);
1770
T.authority = Base.authority;
1772
T.scheme = Base.scheme;
1775
T.fragment = R.fragment;
1779
The pseudocode above refers to a "merge" routine for merging a
1780
relative-path reference with the path of the base URI. This is
1781
accomplished as follows:
1783
o If the base URI has a defined authority component and an empty
1784
path, then return a string consisting of "/" concatenated with the
1785
reference's path; otherwise,
1794
Berners-Lee, et al. Standards Track [Page 32]
1796
RFC 3986 URI Generic Syntax January 2005
1799
o return a string consisting of the reference's path component
1800
appended to all but the last segment of the base URI's path (i.e.,
1801
excluding any characters after the right-most "/" in the base URI
1802
path, or excluding the entire base URI path if it does not contain
1803
any "/" characters).
1805
5.2.4. Remove Dot Segments
1807
The pseudocode also refers to a "remove_dot_segments" routine for
1808
interpreting and removing the special "." and ".." complete path
1809
segments from a referenced path. This is done after the path is
1810
extracted from a reference, whether or not the path was relative, in
1811
order to remove any invalid or extraneous dot-segments prior to
1812
forming the target URI. Although there are many ways to accomplish
1813
this removal process, we describe a simple method using two string
1816
1. The input buffer is initialized with the now-appended path
1817
components and the output buffer is initialized to the empty
1820
2. While the input buffer is not empty, loop as follows:
1822
A. If the input buffer begins with a prefix of "../" or "./",
1823
then remove that prefix from the input buffer; otherwise,
1825
B. if the input buffer begins with a prefix of "/./" or "/.",
1826
where "." is a complete path segment, then replace that
1827
prefix with "/" in the input buffer; otherwise,
1829
C. if the input buffer begins with a prefix of "/../" or "/..",
1830
where ".." is a complete path segment, then replace that
1831
prefix with "/" in the input buffer and remove the last
1832
segment and its preceding "/" (if any) from the output
1835
D. if the input buffer consists only of "." or "..", then remove
1836
that from the input buffer; otherwise,
1838
E. move the first path segment in the input buffer to the end of
1839
the output buffer, including the initial "/" character (if
1840
any) and any subsequent characters up to, but not including,
1841
the next "/" character or the end of the input buffer.
1843
3. Finally, the output buffer is returned as the result of
1844
remove_dot_segments.
1850
Berners-Lee, et al. Standards Track [Page 33]
1852
RFC 3986 URI Generic Syntax January 2005
1855
Note that dot-segments are intended for use in URI references to
1856
express an identifier relative to the hierarchy of names in the base
1857
URI. The remove_dot_segments algorithm respects that hierarchy by
1858
removing extra dot-segments rather than treat them as an error or
1859
leaving them to be misinterpreted by dereference implementations.
1861
The following illustrates how the above steps are applied for two
1862
examples of merged paths, showing the state of the two buffers after
1865
STEP OUTPUT BUFFER INPUT BUFFER
1867
1 : /a/b/c/./../../g
1868
2E: /a /b/c/./../../g
1869
2E: /a/b /c/./../../g
1870
2E: /a/b/c /./../../g
1876
STEP OUTPUT BUFFER INPUT BUFFER
1878
1 : mid/content=5/../6
1879
2E: mid /content=5/../6
1880
2E: mid/content=5 /../6
1884
Some applications may find it more efficient to implement the
1885
remove_dot_segments algorithm by using two segment stacks rather than
1888
Note: Beware that some older, erroneous implementations will fail
1889
to separate a reference's query component from its path component
1890
prior to merging the base and reference paths, resulting in an
1891
interoperability failure if the query component contains the
1892
strings "/../" or "/./".
1906
Berners-Lee, et al. Standards Track [Page 34]
1908
RFC 3986 URI Generic Syntax January 2005
1911
5.3. Component Recomposition
1913
Parsed URI components can be recomposed to obtain the corresponding
1914
URI reference string. Using pseudocode, this would be:
1918
if defined(scheme) then
1919
append scheme to result;
1920
append ":" to result;
1923
if defined(authority) then
1924
append "//" to result;
1925
append authority to result;
1928
append path to result;
1930
if defined(query) then
1931
append "?" to result;
1932
append query to result;
1935
if defined(fragment) then
1936
append "#" to result;
1937
append fragment to result;
1942
Note that we are careful to preserve the distinction between a
1943
component that is undefined, meaning that its separator was not
1944
present in the reference, and a component that is empty, meaning that
1945
the separator was present and was immediately followed by the next
1946
component separator or the end of the reference.
1948
5.4. Reference Resolution Examples
1950
Within a representation with a well defined base URI of
1954
a relative reference is transformed to its target URI as follows.
1962
Berners-Lee, et al. Standards Track [Page 35]
1964
RFC 3986 URI Generic Syntax January 2005
1967
5.4.1. Normal Examples
1970
"g" = "http://a/b/c/g"
1971
"./g" = "http://a/b/c/g"
1972
"g/" = "http://a/b/c/g/"
1975
"?y" = "http://a/b/c/d;p?y"
1976
"g?y" = "http://a/b/c/g?y"
1977
"#s" = "http://a/b/c/d;p?q#s"
1978
"g#s" = "http://a/b/c/g#s"
1979
"g?y#s" = "http://a/b/c/g?y#s"
1980
";x" = "http://a/b/c/;x"
1981
"g;x" = "http://a/b/c/g;x"
1982
"g;x?y#s" = "http://a/b/c/g;x?y#s"
1983
"" = "http://a/b/c/d;p?q"
1984
"." = "http://a/b/c/"
1985
"./" = "http://a/b/c/"
1986
".." = "http://a/b/"
1987
"../" = "http://a/b/"
1988
"../g" = "http://a/b/g"
1989
"../.." = "http://a/"
1990
"../../" = "http://a/"
1991
"../../g" = "http://a/g"
1993
5.4.2. Abnormal Examples
1995
Although the following abnormal examples are unlikely to occur in
1996
normal practice, all URI parsers should be capable of resolving them
1997
consistently. Each example uses the same base as that above.
1999
Parsers must be careful in handling cases where there are more ".."
2000
segments in a relative-path reference than there are hierarchical
2001
levels in the base URI's path. Note that the ".." syntax cannot be
2002
used to change the authority component of a URI.
2004
"../../../g" = "http://a/g"
2005
"../../../../g" = "http://a/g"
2018
Berners-Lee, et al. Standards Track [Page 36]
2020
RFC 3986 URI Generic Syntax January 2005
2023
Similarly, parsers must remove the dot-segments "." and ".." when
2024
they are complete components of a path, but not when they are only
2027
"/./g" = "http://a/g"
2028
"/../g" = "http://a/g"
2029
"g." = "http://a/b/c/g."
2030
".g" = "http://a/b/c/.g"
2031
"g.." = "http://a/b/c/g.."
2032
"..g" = "http://a/b/c/..g"
2034
Less likely are cases where the relative reference uses unnecessary
2035
or nonsensical forms of the "." and ".." complete path segments.
2037
"./../g" = "http://a/b/g"
2038
"./g/." = "http://a/b/c/g/"
2039
"g/./h" = "http://a/b/c/g/h"
2040
"g/../h" = "http://a/b/c/h"
2041
"g;x=1/./y" = "http://a/b/c/g;x=1/y"
2042
"g;x=1/../y" = "http://a/b/c/y"
2044
Some applications fail to separate the reference's query and/or
2045
fragment components from the path component before merging it with
2046
the base path and removing dot-segments. This error is rarely
2047
noticed, as typical usage of a fragment never includes the hierarchy
2048
("/") character and the query component is not normally used within
2049
relative references.
2051
"g?y/./x" = "http://a/b/c/g?y/./x"
2052
"g?y/../x" = "http://a/b/c/g?y/../x"
2053
"g#s/./x" = "http://a/b/c/g#s/./x"
2054
"g#s/../x" = "http://a/b/c/g#s/../x"
2056
Some parsers allow the scheme name to be present in a relative
2057
reference if it is the same as the base URI scheme. This is
2058
considered to be a loophole in prior specifications of partial URI
2059
[RFC1630]. Its use should be avoided but is allowed for backward
2062
"http:g" = "http:g" ; for strict parsers
2063
/ "http://a/b/c/g" ; for backward compatibility
2074
Berners-Lee, et al. Standards Track [Page 37]
2076
RFC 3986 URI Generic Syntax January 2005
2079
6. Normalization and Comparison
2081
One of the most common operations on URIs is simple comparison:
2082
determining whether two URIs are equivalent without using the URIs to
2083
access their respective resource(s). A comparison is performed every
2084
time a response cache is accessed, a browser checks its history to
2085
color a link, or an XML parser processes tags within a namespace.
2086
Extensive normalization prior to comparison of URIs is often used by
2087
spiders and indexing engines to prune a search space or to reduce
2088
duplication of request actions and response storage.
2090
URI comparison is performed for some particular purpose. Protocols
2091
or implementations that compare URIs for different purposes will
2092
often be subject to differing design trade-offs in regards to how
2093
much effort should be spent in reducing aliased identifiers. This
2094
section describes various methods that may be used to compare URIs,
2095
the trade-offs between them, and the types of applications that might
2100
Because URIs exist to identify resources, presumably they should be
2101
considered equivalent when they identify the same resource. However,
2102
this definition of equivalence is not of much practical use, as there
2103
is no way for an implementation to compare two resources unless it
2104
has full knowledge or control of them. For this reason,
2105
determination of equivalence or difference of URIs is based on string
2106
comparison, perhaps augmented by reference to additional rules
2107
provided by URI scheme definitions. We use the terms "different" and
2108
"equivalent" to describe the possible outcomes of such comparisons,
2109
but there are many application-dependent versions of equivalence.
2111
Even though it is possible to determine that two URIs are equivalent,
2112
URI comparison is not sufficient to determine whether two URIs
2113
identify different resources. For example, an owner of two different
2114
domain names could decide to serve the same resource from both,
2115
resulting in two different URIs. Therefore, comparison methods are
2116
designed to minimize false negatives while strictly avoiding false
2119
In testing for equivalence, applications should not directly compare
2120
relative references; the references should be converted to their
2121
respective target URIs before comparison. When URIs are compared to
2122
select (or avoid) a network action, such as retrieval of a
2123
representation, fragment components (if any) should be excluded from
2130
Berners-Lee, et al. Standards Track [Page 38]
2132
RFC 3986 URI Generic Syntax January 2005
2135
6.2. Comparison Ladder
2137
A variety of methods are used in practice to test URI equivalence.
2138
These methods fall into a range, distinguished by the amount of
2139
processing required and the degree to which the probability of false
2140
negatives is reduced. As noted above, false negatives cannot be
2141
eliminated. In practice, their probability can be reduced, but this
2142
reduction requires more processing and is not cost-effective for all
2145
If this range of comparison practices is considered as a ladder, the
2146
following discussion will climb the ladder, starting with practices
2147
that are cheap but have a relatively higher chance of producing false
2148
negatives, and proceeding to those that have higher computational
2149
cost and lower risk of false negatives.
2151
6.2.1. Simple String Comparison
2153
If two URIs, when considered as character strings, are identical,
2154
then it is safe to conclude that they are equivalent. This type of
2155
equivalence test has very low computational cost and is in wide use
2156
in a variety of applications, particularly in the domain of parsing.
2158
Testing strings for equivalence requires some basic precautions.
2159
This procedure is often referred to as "bit-for-bit" or
2160
"byte-for-byte" comparison, which is potentially misleading. Testing
2161
strings for equality is normally based on pair comparison of the
2162
characters that make up the strings, starting from the first and
2163
proceeding until both strings are exhausted and all characters are
2164
found to be equal, until a pair of characters compares unequal, or
2165
until one of the strings is exhausted before the other.
2167
This character comparison requires that each pair of characters be
2168
put in comparable form. For example, should one URI be stored in a
2169
byte array in EBCDIC encoding and the second in a Java String object
2170
(UTF-16), bit-for-bit comparisons applied naively will produce
2171
errors. It is better to speak of equality on a character-for-
2172
character basis rather than on a byte-for-byte or bit-for-bit basis.
2173
In practical terms, character-by-character comparisons should be done
2174
codepoint-by-codepoint after conversion to a common character
2177
False negatives are caused by the production and use of URI aliases.
2178
Unnecessary aliases can be reduced, regardless of the comparison
2179
method, by consistently providing URI references in an already-
2180
normalized form (i.e., a form identical to what would be produced
2181
after normalization is applied, as described below).
2186
Berners-Lee, et al. Standards Track [Page 39]
2188
RFC 3986 URI Generic Syntax January 2005
2191
Protocols and data formats often limit some URI comparisons to simple
2192
string comparison, based on the theory that people and
2193
implementations will, in their own best interest, be consistent in
2194
providing URI references, or at least consistent enough to negate any
2195
efficiency that might be obtained from further normalization.
2197
6.2.2. Syntax-Based Normalization
2199
Implementations may use logic based on the definitions provided by
2200
this specification to reduce the probability of false negatives.
2201
This processing is moderately higher in cost than character-for-
2202
character string comparison. For example, an application using this
2203
approach could reasonably consider the following two URIs equivalent:
2205
example://a/b/c/%7Bfoo%7D
2206
eXAMPLE://a/./b/../b/%63/%7bfoo%7d
2208
Web user agents, such as browsers, typically apply this type of URI
2209
normalization when determining whether a cached response is
2210
available. Syntax-based normalization includes such techniques as
2211
case normalization, percent-encoding normalization, and removal of
2214
6.2.2.1. Case Normalization
2216
For all URIs, the hexadecimal digits within a percent-encoding
2217
triplet (e.g., "%3a" versus "%3A") are case-insensitive and therefore
2218
should be normalized to use uppercase letters for the digits A-F.
2220
When a URI uses components of the generic syntax, the component
2221
syntax equivalence rules always apply; namely, that the scheme and
2222
host are case-insensitive and therefore should be normalized to
2223
lowercase. For example, the URI <HTTP://www.EXAMPLE.com/> is
2224
equivalent to <http://www.example.com/>. The other generic syntax
2225
components are assumed to be case-sensitive unless specifically
2226
defined otherwise by the scheme (see Section 6.2.3).
2228
6.2.2.2. Percent-Encoding Normalization
2230
The percent-encoding mechanism (Section 2.1) is a frequent source of
2231
variance among otherwise identical URIs. In addition to the case
2232
normalization issue noted above, some URI producers percent-encode
2233
octets that do not require percent-encoding, resulting in URIs that
2234
are equivalent to their non-encoded counterparts. These URIs should
2235
be normalized by decoding any percent-encoded octet that corresponds
2236
to an unreserved character, as described in Section 2.3.
2242
Berners-Lee, et al. Standards Track [Page 40]
2244
RFC 3986 URI Generic Syntax January 2005
2247
6.2.2.3. Path Segment Normalization
2249
The complete path segments "." and ".." are intended only for use
2250
within relative references (Section 4.1) and are removed as part of
2251
the reference resolution process (Section 5.2). However, some
2252
deployed implementations incorrectly assume that reference resolution
2253
is not necessary when the reference is already a URI and thus fail to
2254
remove dot-segments when they occur in non-relative paths. URI
2255
normalizers should remove dot-segments by applying the
2256
remove_dot_segments algorithm to the path, as described in
2259
6.2.3. Scheme-Based Normalization
2261
The syntax and semantics of URIs vary from scheme to scheme, as
2262
described by the defining specification for each scheme.
2263
Implementations may use scheme-specific rules, at further processing
2264
cost, to reduce the probability of false negatives. For example,
2265
because the "http" scheme makes use of an authority component, has a
2266
default port of "80", and defines an empty path to be equivalent to
2267
"/", the following four URIs are equivalent:
2271
http://example.com:/
2272
http://example.com:80/
2274
In general, a URI that uses the generic syntax for authority with an
2275
empty path should be normalized to a path of "/". Likewise, an
2276
explicit ":port", for which the port is empty or the default for the
2277
scheme, is equivalent to one where the port and its ":" delimiter are
2278
elided and thus should be removed by scheme-based normalization. For
2279
example, the second URI above is the normal form for the "http"
2282
Another case where normalization varies by scheme is in the handling
2283
of an empty authority component or empty host subcomponent. For many
2284
scheme specifications, an empty authority or host is considered an
2285
error; for others, it is considered equivalent to "localhost" or the
2286
end-user's host. When a scheme defines a default for authority and a
2287
URI reference to that default is desired, the reference should be
2288
normalized to an empty authority for the sake of uniformity, brevity,
2289
and internationalization. If, however, either the userinfo or port
2290
subcomponents are non-empty, then the host should be given explicitly
2291
even if it matches the default.
2293
Normalization should not remove delimiters when their associated
2294
component is empty unless licensed to do so by the scheme
2298
Berners-Lee, et al. Standards Track [Page 41]
2300
RFC 3986 URI Generic Syntax January 2005
2303
specification. For example, the URI "http://example.com/?" cannot be
2304
assumed to be equivalent to any of the examples above. Likewise, the
2305
presence or absence of delimiters within a userinfo subcomponent is
2306
usually significant to its interpretation. The fragment component is
2307
not subject to any scheme-based normalization; thus, two URIs that
2308
differ only by the suffix "#" are considered different regardless of
2311
Some schemes define additional subcomponents that consist of case-
2312
insensitive data, giving an implicit license to normalizers to
2313
convert this data to a common case (e.g., all lowercase). For
2314
example, URI schemes that define a subcomponent of path to contain an
2315
Internet hostname, such as the "mailto" URI scheme, cause that
2316
subcomponent to be case-insensitive and thus subject to case
2317
normalization (e.g., "mailto:Joe@Example.COM" is equivalent to
2318
"mailto:Joe@example.com", even though the generic syntax considers
2319
the path component to be case-sensitive).
2321
Other scheme-specific normalizations are possible.
2323
6.2.4. Protocol-Based Normalization
2325
Substantial effort to reduce the incidence of false negatives is
2326
often cost-effective for web spiders. Therefore, they implement even
2327
more aggressive techniques in URI comparison. For example, if they
2328
observe that a URI such as
2330
http://example.com/data
2332
redirects to a URI differing only in the trailing slash
2334
http://example.com/data/
2336
they will likely regard the two as equivalent in the future. This
2337
kind of technique is only appropriate when equivalence is clearly
2338
indicated by both the result of accessing the resources and the
2339
common conventions of their scheme's dereference algorithm (in this
2340
case, use of redirection by HTTP origin servers to avoid problems
2341
with relative references).
2354
Berners-Lee, et al. Standards Track [Page 42]
2356
RFC 3986 URI Generic Syntax January 2005
2359
7. Security Considerations
2361
A URI does not in itself pose a security threat. However, as URIs
2362
are often used to provide a compact set of instructions for access to
2363
network resources, care must be taken to properly interpret the data
2364
within a URI, to prevent that data from causing unintended access,
2365
and to avoid including data that should not be revealed in plain
2368
7.1. Reliability and Consistency
2370
There is no guarantee that once a URI has been used to retrieve
2371
information, the same information will be retrievable by that URI in
2372
the future. Nor is there any guarantee that the information
2373
retrievable via that URI in the future will be observably similar to
2374
that retrieved in the past. The URI syntax does not constrain how a
2375
given scheme or authority apportions its namespace or maintains it
2376
over time. Such guarantees can only be obtained from the person(s)
2377
controlling that namespace and the resource in question. A specific
2378
URI scheme may define additional semantics, such as name persistence,
2379
if those semantics are required of all naming authorities for that
2382
7.2. Malicious Construction
2384
It is sometimes possible to construct a URI so that an attempt to
2385
perform a seemingly harmless, idempotent operation, such as the
2386
retrieval of a representation, will in fact cause a possibly damaging
2387
remote operation. The unsafe URI is typically constructed by
2388
specifying a port number other than that reserved for the network
2389
protocol in question. The client unwittingly contacts a site running
2390
a different protocol service, and data within the URI contains
2391
instructions that, when interpreted according to this other protocol,
2392
cause an unexpected operation. A frequent example of such abuse has
2393
been the use of a protocol-based scheme with a port component of
2394
"25", thereby fooling user agent software into sending an unintended
2395
or impersonating message via an SMTP server.
2397
Applications should prevent dereference of a URI that specifies a TCP
2398
port number within the "well-known port" range (0 - 1023) unless the
2399
protocol being used to dereference that URI is compatible with the
2400
protocol expected on that well-known port. Although IANA maintains a
2401
registry of well-known ports, applications should make such
2402
restrictions user-configurable to avoid preventing the deployment of
2410
Berners-Lee, et al. Standards Track [Page 43]
2412
RFC 3986 URI Generic Syntax January 2005
2415
When a URI contains percent-encoded octets that match the delimiters
2416
for a given resolution or dereference protocol (for example, CR and
2417
LF characters for the TELNET protocol), these percent-encodings must
2418
not be decoded before transmission across that protocol. Transfer of
2419
the percent-encoding, which might violate the protocol, is less
2420
harmful than allowing decoded octets to be interpreted as additional
2421
operations or parameters, perhaps triggering an unexpected and
2422
possibly harmful remote operation.
2424
7.3. Back-End Transcoding
2426
When a URI is dereferenced, the data within it is often parsed by
2427
both the user agent and one or more servers. In HTTP, for example, a
2428
typical user agent will parse a URI into its five major components,
2429
access the authority's server, and send it the data within the
2430
authority, path, and query components. A typical server will take
2431
that information, parse the path into segments and the query into
2432
key/value pairs, and then invoke implementation-specific handlers to
2433
respond to the request. As a result, a common security concern for
2434
server implementations that handle a URI, either as a whole or split
2435
into separate components, is proper interpretation of the octet data
2436
represented by the characters and percent-encodings within that URI.
2438
Percent-encoded octets must be decoded at some point during the
2439
dereference process. Applications must split the URI into its
2440
components and subcomponents prior to decoding the octets, as
2441
otherwise the decoded octets might be mistaken for delimiters.
2442
Security checks of the data within a URI should be applied after
2443
decoding the octets. Note, however, that the "%00" percent-encoding
2444
(NUL) may require special handling and should be rejected if the
2445
application is not expecting to receive raw data within a component.
2447
Special care should be taken when the URI path interpretation process
2448
involves the use of a back-end file system or related system
2449
functions. File systems typically assign an operational meaning to
2450
special characters, such as the "/", "\", ":", "[", and "]"
2451
characters, and to special device names like ".", "..", "...", "aux",
2452
"lpt", etc. In some cases, merely testing for the existence of such
2453
a name will cause the operating system to pause or invoke unrelated
2454
system calls, leading to significant security concerns regarding
2455
denial of service and unintended data transfer. It would be
2456
impossible for this specification to list all such significant
2457
characters and device names. Implementers should research the
2458
reserved names and characters for the types of storage device that
2459
may be attached to their applications and restrict the use of data
2460
obtained from URI components accordingly.
2466
Berners-Lee, et al. Standards Track [Page 44]
2468
RFC 3986 URI Generic Syntax January 2005
2471
7.4. Rare IP Address Formats
2473
Although the URI syntax for IPv4address only allows the common
2474
dotted-decimal form of IPv4 address literal, many implementations
2475
that process URIs make use of platform-dependent system routines,
2476
such as gethostbyname() and inet_aton(), to translate the string
2477
literal to an actual IP address. Unfortunately, such system routines
2478
often allow and process a much larger set of formats than those
2479
described in Section 3.2.2.
2481
For example, many implementations allow dotted forms of three
2482
numbers, wherein the last part is interpreted as a 16-bit quantity
2483
and placed in the right-most two bytes of the network address (e.g.,
2484
a Class B network). Likewise, a dotted form of two numbers means
2485
that the last part is interpreted as a 24-bit quantity and placed in
2486
the right-most three bytes of the network address (Class A), and a
2487
single number (without dots) is interpreted as a 32-bit quantity and
2488
stored directly in the network address. Adding further to the
2489
confusion, some implementations allow each dotted part to be
2490
interpreted as decimal, octal, or hexadecimal, as specified in the C
2491
language (i.e., a leading 0x or 0X implies hexadecimal; a leading 0
2492
implies octal; otherwise, the number is interpreted as decimal).
2494
These additional IP address formats are not allowed in the URI syntax
2495
due to differences between platform implementations. However, they
2496
can become a security concern if an application attempts to filter
2497
access to resources based on the IP address in string literal format.
2498
If this filtering is performed, literals should be converted to
2499
numeric form and filtered based on the numeric value, and not on a
2500
prefix or suffix of the string form.
2502
7.5. Sensitive Information
2504
URI producers should not provide a URI that contains a username or
2505
password that is intended to be secret. URIs are frequently
2506
displayed by browsers, stored in clear text bookmarks, and logged by
2507
user agent history and intermediary applications (proxies). A
2508
password appearing within the userinfo component is deprecated and
2509
should be considered an error (or simply ignored) except in those
2510
rare cases where the 'password' parameter is intended to be public.
2512
7.6. Semantic Attacks
2514
Because the userinfo subcomponent is rarely used and appears before
2515
the host in the authority component, it can be used to construct a
2516
URI intended to mislead a human user by appearing to identify one
2517
(trusted) naming authority while actually identifying a different
2518
authority hidden behind the noise. For example
2522
Berners-Lee, et al. Standards Track [Page 45]
2524
RFC 3986 URI Generic Syntax January 2005
2527
ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm
2529
might lead a human user to assume that the host is 'cnn.example.com',
2530
whereas it is actually '10.0.0.1'. Note that a misleading userinfo
2531
subcomponent could be much longer than the example above.
2533
A misleading URI, such as that above, is an attack on the user's
2534
preconceived notions about the meaning of a URI rather than an attack
2535
on the software itself. User agents may be able to reduce the impact
2536
of such attacks by distinguishing the various components of the URI
2537
when they are rendered, such as by using a different color or tone to
2538
render userinfo if any is present, though there is no panacea. More
2539
information on URI-based semantic attacks can be found in [Siedzik].
2541
8. IANA Considerations
2543
URI scheme names, as defined by <scheme> in Section 3.1, form a
2544
registered namespace that is managed by IANA according to the
2545
procedures defined in [BCP35]. No IANA actions are required by this
2550
This specification is derived from RFC 2396 [RFC2396], RFC 1808
2551
[RFC1808], and RFC 1738 [RFC1738]; the acknowledgements in those
2552
documents still apply. It also incorporates the update (with
2553
corrections) for IPv6 literals in the host syntax, as defined by
2554
Robert M. Hinden, Brian E. Carpenter, and Larry Masinter in
2555
[RFC2732]. In addition, contributions by Gisle Aas, Reese Anschultz,
2556
Daniel Barclay, Tim Bray, Mike Brown, Rob Cameron, Jeremy Carroll,
2557
Dan Connolly, Adam M. Costello, John Cowan, Jason Diamond, Martin
2558
Duerst, Stefan Eissing, Clive D.W. Feather, Al Gilman, Tony Hammond,
2559
Elliotte Harold, Pat Hayes, Henry Holtzman, Ian B. Jacobs, Michael
2560
Kay, John C. Klensin, Graham Klyne, Dan Kohn, Bruce Lilly, Andrew
2561
Main, Dave McAlpin, Ira McDonald, Michael Mealling, Ray Merkert,
2562
Stephen Pollei, Julian Reschke, Tomas Rokicki, Miles Sabin, Kai
2563
Schaetzl, Mark Thomson, Ronald Tschalaer, Norm Walsh, Marc Warne,
2564
Stuart Williams, and Henry Zongaro are gratefully acknowledged.
2568
10.1. Normative References
2570
[ASCII] American National Standards Institute, "Coded Character
2571
Set -- 7-bit American Standard Code for Information
2572
Interchange", ANSI X3.4, 1986.
2578
Berners-Lee, et al. Standards Track [Page 46]
2580
RFC 3986 URI Generic Syntax January 2005
2583
[RFC2234] Crocker, D. and P. Overell, "Augmented BNF for Syntax
2584
Specifications: ABNF", RFC 2234, November 1997.
2586
[STD63] Yergeau, F., "UTF-8, a transformation format of
2587
ISO 10646", STD 63, RFC 3629, November 2003.
2589
[UCS] International Organization for Standardization,
2590
"Information Technology - Universal Multiple-Octet Coded
2591
Character Set (UCS)", ISO/IEC 10646:2003, December 2003.
2593
10.2. Informative References
2595
[BCP19] Freed, N. and J. Postel, "IANA Charset Registration
2596
Procedures", BCP 19, RFC 2978, October 2000.
2598
[BCP35] Petke, R. and I. King, "Registration Procedures for URL
2599
Scheme Names", BCP 35, RFC 2717, November 1999.
2601
[RFC0952] Harrenstien, K., Stahl, M., and E. Feinler, "DoD Internet
2602
host table specification", RFC 952, October 1985.
2604
[RFC1034] Mockapetris, P., "Domain names - concepts and facilities",
2605
STD 13, RFC 1034, November 1987.
2607
[RFC1123] Braden, R., "Requirements for Internet Hosts - Application
2608
and Support", STD 3, RFC 1123, October 1989.
2610
[RFC1535] Gavron, E., "A Security Problem and Proposed Correction
2611
With Widely Deployed DNS Software", RFC 1535,
2614
[RFC1630] Berners-Lee, T., "Universal Resource Identifiers in WWW: A
2615
Unifying Syntax for the Expression of Names and Addresses
2616
of Objects on the Network as used in the World-Wide Web",
2617
RFC 1630, June 1994.
2619
[RFC1736] Kunze, J., "Functional Recommendations for Internet
2620
Resource Locators", RFC 1736, February 1995.
2622
[RFC1737] Sollins, K. and L. Masinter, "Functional Requirements for
2623
Uniform Resource Names", RFC 1737, December 1994.
2625
[RFC1738] Berners-Lee, T., Masinter, L., and M. McCahill, "Uniform
2626
Resource Locators (URL)", RFC 1738, December 1994.
2628
[RFC1808] Fielding, R., "Relative Uniform Resource Locators",
2629
RFC 1808, June 1995.
2634
Berners-Lee, et al. Standards Track [Page 47]
2636
RFC 3986 URI Generic Syntax January 2005
2639
[RFC2046] Freed, N. and N. Borenstein, "Multipurpose Internet Mail
2640
Extensions (MIME) Part Two: Media Types", RFC 2046,
2643
[RFC2141] Moats, R., "URN Syntax", RFC 2141, May 1997.
2645
[RFC2396] Berners-Lee, T., Fielding, R., and L. Masinter, "Uniform
2646
Resource Identifiers (URI): Generic Syntax", RFC 2396,
2649
[RFC2518] Goland, Y., Whitehead, E., Faizi, A., Carter, S., and D.
2650
Jensen, "HTTP Extensions for Distributed Authoring --
2651
WEBDAV", RFC 2518, February 1999.
2653
[RFC2557] Palme, J., Hopmann, A., and N. Shelness, "MIME
2654
Encapsulation of Aggregate Documents, such as HTML
2655
(MHTML)", RFC 2557, March 1999.
2657
[RFC2718] Masinter, L., Alvestrand, H., Zigmond, D., and R. Petke,
2658
"Guidelines for new URL Schemes", RFC 2718, November 1999.
2660
[RFC2732] Hinden, R., Carpenter, B., and L. Masinter, "Format for
2661
Literal IPv6 Addresses in URL's", RFC 2732, December 1999.
2663
[RFC3305] Mealling, M. and R. Denenberg, "Report from the Joint
2664
W3C/IETF URI Planning Interest Group: Uniform Resource
2665
Identifiers (URIs), URLs, and Uniform Resource Names
2666
(URNs): Clarifications and Recommendations", RFC 3305,
2669
[RFC3490] Faltstrom, P., Hoffman, P., and A. Costello,
2670
"Internationalizing Domain Names in Applications (IDNA)",
2671
RFC 3490, March 2003.
2673
[RFC3513] Hinden, R. and S. Deering, "Internet Protocol Version 6
2674
(IPv6) Addressing Architecture", RFC 3513, April 2003.
2676
[Siedzik] Siedzik, R., "Semantic Attacks: What's in a URL?",
2677
April 2001, <http://www.giac.org/practical/gsec/
2678
Richard_Siedzik_GSEC.pdf>.
2690
Berners-Lee, et al. Standards Track [Page 48]
2692
RFC 3986 URI Generic Syntax January 2005
2695
Appendix A. Collected ABNF for URI
2697
URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ]
2699
hier-part = "//" authority path-abempty
2704
URI-reference = URI / relative-ref
2706
absolute-URI = scheme ":" hier-part [ "?" query ]
2708
relative-ref = relative-part [ "?" query ] [ "#" fragment ]
2710
relative-part = "//" authority path-abempty
2715
scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
2717
authority = [ userinfo "@" ] host [ ":" port ]
2718
userinfo = *( unreserved / pct-encoded / sub-delims / ":" )
2719
host = IP-literal / IPv4address / reg-name
2722
IP-literal = "[" ( IPv6address / IPvFuture ) "]"
2724
IPvFuture = "v" 1*HEXDIG "." 1*( unreserved / sub-delims / ":" )
2726
IPv6address = 6( h16 ":" ) ls32
2727
/ "::" 5( h16 ":" ) ls32
2728
/ [ h16 ] "::" 4( h16 ":" ) ls32
2729
/ [ *1( h16 ":" ) h16 ] "::" 3( h16 ":" ) ls32
2730
/ [ *2( h16 ":" ) h16 ] "::" 2( h16 ":" ) ls32
2731
/ [ *3( h16 ":" ) h16 ] "::" h16 ":" ls32
2732
/ [ *4( h16 ":" ) h16 ] "::" ls32
2733
/ [ *5( h16 ":" ) h16 ] "::" h16
2734
/ [ *6( h16 ":" ) h16 ] "::"
2737
ls32 = ( h16 ":" h16 ) / IPv4address
2738
IPv4address = dec-octet "." dec-octet "." dec-octet "." dec-octet
2746
Berners-Lee, et al. Standards Track [Page 49]
2748
RFC 3986 URI Generic Syntax January 2005
2751
dec-octet = DIGIT ; 0-9
2752
/ %x31-39 DIGIT ; 10-99
2753
/ "1" 2DIGIT ; 100-199
2754
/ "2" %x30-34 DIGIT ; 200-249
2755
/ "25" %x30-35 ; 250-255
2757
reg-name = *( unreserved / pct-encoded / sub-delims )
2759
path = path-abempty ; begins with "/" or is empty
2760
/ path-absolute ; begins with "/" but not "//"
2761
/ path-noscheme ; begins with a non-colon segment
2762
/ path-rootless ; begins with a segment
2763
/ path-empty ; zero characters
2765
path-abempty = *( "/" segment )
2766
path-absolute = "/" [ segment-nz *( "/" segment ) ]
2767
path-noscheme = segment-nz-nc *( "/" segment )
2768
path-rootless = segment-nz *( "/" segment )
2769
path-empty = 0<pchar>
2772
segment-nz = 1*pchar
2773
segment-nz-nc = 1*( unreserved / pct-encoded / sub-delims / "@" )
2774
; non-zero-length segment without any colon ":"
2776
pchar = unreserved / pct-encoded / sub-delims / ":" / "@"
2778
query = *( pchar / "/" / "?" )
2780
fragment = *( pchar / "/" / "?" )
2782
pct-encoded = "%" HEXDIG HEXDIG
2784
unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~"
2785
reserved = gen-delims / sub-delims
2786
gen-delims = ":" / "/" / "?" / "#" / "[" / "]" / "@"
2787
sub-delims = "!" / "$" / "&" / "'" / "(" / ")"
2788
/ "*" / "+" / "," / ";" / "="
2790
Appendix B. Parsing a URI Reference with a Regular Expression
2792
As the "first-match-wins" algorithm is identical to the "greedy"
2793
disambiguation method used by POSIX regular expressions, it is
2794
natural and commonplace to use a regular expression for parsing the
2795
potential five components of a URI reference.
2797
The following line is the regular expression for breaking-down a
2798
well-formed URI reference into its components.
2802
Berners-Lee, et al. Standards Track [Page 50]
2804
RFC 3986 URI Generic Syntax January 2005
2807
^(([^:/?#]+):)?(//([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?
2810
The numbers in the second line above are only to assist readability;
2811
they indicate the reference points for each subexpression (i.e., each
2812
paired parenthesis). We refer to the value matched for subexpression
2813
<n> as $<n>. For example, matching the above expression to
2815
http://www.ics.uci.edu/pub/ietf/uri/#Related
2817
results in the following subexpression matches:
2821
$3 = //www.ics.uci.edu
2822
$4 = www.ics.uci.edu
2829
where <undefined> indicates that the component is not present, as is
2830
the case for the query component in the above example. Therefore, we
2831
can determine the value of the five components as
2839
Going in the opposite direction, we can recreate a URI reference from
2840
its components by using the algorithm of Section 5.3.
2842
Appendix C. Delimiting a URI in Context
2844
URIs are often transmitted through formats that do not provide a
2845
clear context for their interpretation. For example, there are many
2846
occasions when a URI is included in plain text; examples include text
2847
sent in email, USENET news, and on printed paper. In such cases, it
2848
is important to be able to delimit the URI from the rest of the text,
2849
and in particular from punctuation marks that might be mistaken for
2852
In practice, URIs are delimited in a variety of ways, but usually
2853
within double-quotes "http://example.com/", angle brackets
2854
<http://example.com/>, or just by using whitespace:
2858
Berners-Lee, et al. Standards Track [Page 51]
2860
RFC 3986 URI Generic Syntax January 2005
2865
These wrappers do not form part of the URI.
2867
In some cases, extra whitespace (spaces, line-breaks, tabs, etc.) may
2868
have to be added to break a long URI across lines. The whitespace
2869
should be ignored when the URI is extracted.
2871
No whitespace should be introduced after a hyphen ("-") character.
2872
Because some typesetters and printers may (erroneously) introduce a
2873
hyphen at the end of line when breaking it, the interpreter of a URI
2874
containing a line break immediately after a hyphen should ignore all
2875
whitespace around the line break and should be aware that the hyphen
2876
may or may not actually be part of the URI.
2878
Using <> angle brackets around each URI is especially recommended as
2879
a delimiting style for a reference that contains embedded whitespace.
2881
The prefix "URL:" (with or without a trailing space) was formerly
2882
recommended as a way to help distinguish a URI from other bracketed
2883
designators, though it is not commonly used in practice and is no
2886
For robustness, software that accepts user-typed URI should attempt
2887
to recognize and strip both delimiters and embedded whitespace.
2889
For example, the text
2891
Yes, Jim, I found it under "http://www.w3.org/Addressing/",
2892
but you can probably pick it up from <ftp://foo.example.
2893
com/rfc/>. Note the warning in <http://www.ics.uci.edu/pub/
2894
ietf/uri/historical.html#WARNING>.
2896
contains the URI references
2898
http://www.w3.org/Addressing/
2899
ftp://foo.example.com/rfc/
2900
http://www.ics.uci.edu/pub/ietf/uri/historical.html#WARNING
2914
Berners-Lee, et al. Standards Track [Page 52]
2916
RFC 3986 URI Generic Syntax January 2005
2919
Appendix D. Changes from RFC 2396
2923
An ABNF rule for URI has been introduced to correspond to one common
2924
usage of the term: an absolute URI with optional fragment.
2926
IPv6 (and later) literals have been added to the list of possible
2927
identifiers for the host portion of an authority component, as
2928
described by [RFC2732], with the addition of "[" and "]" to the
2929
reserved set and a version flag to anticipate future versions of IP
2930
literals. Square brackets are now specified as reserved within the
2931
authority component and are not allowed outside their use as
2932
delimiters for an IP literal within host. In order to make this
2933
change without changing the technical definition of the path, query,
2934
and fragment components, those rules were redefined to directly
2935
specify the characters allowed.
2937
As [RFC2732] defers to [RFC3513] for definition of an IPv6 literal
2938
address, which, unfortunately, lacks an ABNF description of
2939
IPv6address, we created a new ABNF rule for IPv6address that matches
2940
the text representations defined by Section 2.2 of [RFC3513].
2941
Likewise, the definition of IPv4address has been improved in order to
2942
limit each decimal octet to the range 0-255.
2944
Section 6, on URI normalization and comparison, has been completely
2945
rewritten and extended by using input from Tim Bray and discussion
2946
within the W3C Technical Architecture Group.
2950
The ad-hoc BNF syntax of RFC 2396 has been replaced with the ABNF of
2951
[RFC2234]. This change required all rule names that formerly
2952
included underscore characters to be renamed with a dash instead. In
2953
addition, a number of syntax rules have been eliminated or simplified
2954
to make the overall grammar more comprehensible. Specifications that
2955
refer to the obsolete grammar rules may be understood by replacing
2956
those rules according to the following table:
2970
Berners-Lee, et al. Standards Track [Page 53]
2972
RFC 3986 URI Generic Syntax January 2005
2975
+----------------+--------------------------------------------------+
2976
| obsolete rule | translation |
2977
+----------------+--------------------------------------------------+
2978
| absoluteURI | absolute-URI |
2979
| relativeURI | relative-part [ "?" query ] |
2980
| hier_part | ( "//" authority path-abempty / |
2981
| | path-absolute ) [ "?" query ] |
2983
| opaque_part | path-rootless [ "?" query ] |
2984
| net_path | "//" authority path-abempty |
2985
| abs_path | path-absolute |
2986
| rel_path | path-rootless |
2987
| rel_segment | segment-nz-nc |
2988
| reg_name | reg-name |
2989
| server | authority |
2990
| hostport | host [ ":" port ] |
2991
| hostname | reg-name |
2992
| path_segments | path-abempty |
2993
| param | *<pchar excluding ";"> |
2995
| uric | unreserved / pct-encoded / ";" / "?" / ":" |
2996
| | / "@" / "&" / "=" / "+" / "$" / "," / "/" |
2998
| uric_no_slash | unreserved / pct-encoded / ";" / "?" / ":" |
2999
| | / "@" / "&" / "=" / "+" / "$" / "," |
3001
| mark | "-" / "_" / "." / "!" / "~" / "*" / "'" |
3004
| escaped | pct-encoded |
3006
| alphanum | ALPHA / DIGIT |
3007
+----------------+--------------------------------------------------+
3009
Use of the above obsolete rules for the definition of scheme-specific
3010
syntax is deprecated.
3012
Section 2, on characters, has been rewritten to explain what
3013
characters are reserved, when they are reserved, and why they are
3014
reserved, even when they are not used as delimiters by the generic
3015
syntax. The mark characters that are typically unsafe to decode,
3016
including the exclamation mark ("!"), asterisk ("*"), single-quote
3017
("'"), and open and close parentheses ("(" and ")"), have been moved
3018
to the reserved set in order to clarify the distinction between
3019
reserved and unreserved and, hopefully, to answer the most common
3020
question of scheme designers. Likewise, the section on
3021
percent-encoded characters has been rewritten, and URI normalizers
3022
are now given license to decode any percent-encoded octets
3026
Berners-Lee, et al. Standards Track [Page 54]
3028
RFC 3986 URI Generic Syntax January 2005
3031
corresponding to unreserved characters. In general, the terms
3032
"escaped" and "unescaped" have been replaced with "percent-encoded"
3033
and "decoded", respectively, to reduce confusion with other forms of
3036
The ABNF for URI and URI-reference has been redesigned to make them
3037
more friendly to LALR parsers and to reduce complexity. As a result,
3038
the layout form of syntax description has been removed, along with
3039
the uric, uric_no_slash, opaque_part, net_path, abs_path, rel_path,
3040
path_segments, rel_segment, and mark rules. All references to
3041
"opaque" URIs have been replaced with a better description of how the
3042
path component may be opaque to hierarchy. The relativeURI rule has
3043
been replaced with relative-ref to avoid unnecessary confusion over
3044
whether they are a subset of URI. The ambiguity regarding the
3045
parsing of URI-reference as a URI or a relative-ref with a colon in
3046
the first segment has been eliminated through the use of five
3047
separate path matching rules.
3049
The fragment identifier has been moved back into the section on
3050
generic syntax components and within the URI and relative-ref rules,
3051
though it remains excluded from absolute-URI. The number sign ("#")
3052
character has been moved back to the reserved set as a result of
3053
reintegrating the fragment syntax.
3055
The ABNF has been corrected to allow the path component to be empty.
3056
This also allows an absolute-URI to consist of nothing after the
3057
"scheme:", as is present in practice with the "dav:" namespace
3058
[RFC2518] and with the "about:" scheme used internally by many WWW
3059
browser implementations. The ambiguity regarding the boundary
3060
between authority and path has been eliminated through the use of
3061
five separate path matching rules.
3063
Registry-based naming authorities that use the generic syntax are now
3064
defined within the host rule. This change allows current
3065
implementations, where whatever name provided is simply fed to the
3066
local name resolution mechanism, to be consistent with the
3067
specification. It also removes the need to re-specify DNS name
3068
formats here. Furthermore, it allows the host component to contain
3069
percent-encoded octets, which is necessary to enable
3070
internationalized domain names to be provided in URIs, processed in
3071
their native character encodings at the application layers above URI
3072
processing, and passed to an IDNA library as a registered name in the
3073
UTF-8 character encoding. The server, hostport, hostname,
3074
domainlabel, toplabel, and alphanum rules have been removed.
3076
The resolving relative references algorithm of [RFC2396] has been
3077
rewritten with pseudocode for this revision to improve clarity and
3078
fix the following issues:
3082
Berners-Lee, et al. Standards Track [Page 55]
3084
RFC 3986 URI Generic Syntax January 2005
3087
o [RFC2396] section 5.2, step 6a, failed to account for a base URI
3090
o Restored the behavior of [RFC1808] where, if the reference
3091
contains an empty path and a defined query component, the target
3092
URI inherits the base URI's path component.
3094
o The determination of whether a URI reference is a same-document
3095
reference has been decoupled from the URI parser, simplifying the
3096
URI processing interface within applications in a way consistent
3097
with the internal architecture of deployed URI processing
3098
implementations. The determination is now based on comparison to
3099
the base URI after transforming a reference to absolute form,
3100
rather than on the format of the reference itself. This change
3101
may result in more references being considered "same-document"
3102
under this specification than there would be under the rules given
3103
in RFC 2396, especially when normalization is used to reduce
3104
aliases. However, it does not change the status of existing
3105
same-document references.
3107
o Separated the path merge routine into two routines: merge, for
3108
describing combination of the base URI path with a relative-path
3109
reference, and remove_dot_segments, for describing how to remove
3110
the special "." and ".." segments from a composed path. The
3111
remove_dot_segments algorithm is now applied to all URI reference
3112
paths in order to match common implementations and to improve the
3113
normalization of URIs in practice. This change only impacts the
3114
parsing of abnormal references and same-scheme references wherein
3115
the base URI has a non-hierarchical path.
3131
character encoding 4
3134
coded character set 4
3138
Berners-Lee, et al. Standards Track [Page 56]
3140
RFC 3986 URI Generic Syntax January 2005
3188
path-abempty 16, 22, 26
3189
path-absolute 16, 22, 26
3190
path-empty 16, 22, 26
3194
Berners-Lee, et al. Standards Track [Page 57]
3196
RFC 3986 URI Generic Syntax January 2005
3199
path-rootless 16, 22
3214
remove_dot_segments 33
3250
Berners-Lee, et al. Standards Track [Page 58]
3252
RFC 3986 URI Generic Syntax January 2005
3306
Berners-Lee, et al. Standards Track [Page 59]
3308
RFC 3986 URI Generic Syntax January 2005
3314
World Wide Web Consortium
3315
Massachusetts Institute of Technology
3316
77 Massachusetts Avenue
3320
Phone: +1-617-253-5702
3321
Fax: +1-617-258-5999
3323
URI: http://www.w3.org/People/Berners-Lee/
3328
5251 California Ave., Suite 110
3332
Phone: +1-949-679-2960
3333
Fax: +1-949-679-2972
3334
EMail: fielding@gbiv.com
3335
URI: http://roy.gbiv.com/
3339
Adobe Systems Incorporated
3344
Phone: +1-408-536-3024
3346
URI: http://larry.masinter.net/
3362
Berners-Lee, et al. Standards Track [Page 60]
3364
RFC 3986 URI Generic Syntax January 2005
3367
Full Copyright Statement
3369
Copyright (C) The Internet Society (2005).
3371
This document is subject to the rights, licenses and restrictions
3372
contained in BCP 78, and except as set forth therein, the authors
3373
retain all their rights.
3375
This document and the information contained herein are provided on an
3376
"AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
3377
OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET
3378
ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED,
3379
INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE
3380
INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
3381
WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
3383
Intellectual Property
3385
The IETF takes no position regarding the validity or scope of any
3386
Intellectual Property Rights or other rights that might be claimed to
3387
pertain to the implementation or use of the technology described in
3388
this document or the extent to which any license under such rights
3389
might or might not be available; nor does it represent that it has
3390
made any independent effort to identify any such rights. Information
3391
on the IETF's procedures with respect to rights in IETF Documents can
3392
be found in BCP 78 and BCP 79.
3394
Copies of IPR disclosures made to the IETF Secretariat and any
3395
assurances of licenses to be made available, or the result of an
3396
attempt made to obtain a general license or permission for the use of
3397
such proprietary rights by implementers or users of this
3398
specification can be obtained from the IETF on-line IPR repository at
3399
http://www.ietf.org/ipr.
3401
The IETF invites any interested party to bring to its attention any
3402
copyrights, patents or patent applications, or other proprietary
3403
rights that may cover technology that may be required to implement
3404
this standard. Please address the information to the IETF at ietf-
3410
Funding for the RFC Editor function is currently provided by the
3418
Berners-Lee, et al. Standards Track [Page 61]