|
| 1 | +// Copyright 2012 The Go Authors. All rights reserved. |
| 2 | +// Use of this source code is governed by a BSD-style |
| 3 | +// license that can be found in the LICENSE file. |
| 4 | + |
| 5 | +// Package atom provides integer codes (also known as atoms) for a fixed set of |
| 6 | +// frequently occurring HTML strings: tag names and attribute keys such as "p" |
| 7 | +// and "id". |
| 8 | +// |
| 9 | +// Sharing an atom's name between all elements with the same tag can result in |
| 10 | +// fewer string allocations when tokenizing and parsing HTML. Integer |
| 11 | +// comparisons are also generally faster than string comparisons. |
| 12 | +// |
| 13 | +// The value of an atom's particular code is not guaranteed to stay the same |
| 14 | +// between versions of this package. Neither is any ordering guaranteed: |
| 15 | +// whether atom.H1 < atom.H2 may also change. The codes are not guaranteed to |
| 16 | +// be dense. The only guarantees are that e.g. looking up "div" will yield |
| 17 | +// atom.Div, calling atom.Div.String will return "div", and atom.Div != 0. |
| 18 | +package atom |
| 19 | + |
| 20 | +// Atom is an integer code for a string. The zero value maps to "". |
| 21 | +type Atom uint32 |
| 22 | + |
| 23 | +// String returns the atom's name. |
| 24 | +func (a Atom) String() string { |
| 25 | + start := uint32(a >> 8) |
| 26 | + n := uint32(a & 0xff) |
| 27 | + if start+n > uint32(len(atomText)) { |
| 28 | + return "" |
| 29 | + } |
| 30 | + return atomText[start : start+n] |
| 31 | +} |
| 32 | + |
| 33 | +func (a Atom) string() string { |
| 34 | + return atomText[a>>8 : a>>8+a&0xff] |
| 35 | +} |
| 36 | + |
| 37 | +// fnv computes the FNV hash with an arbitrary starting value h. |
| 38 | +func fnv(h uint32, s []byte) uint32 { |
| 39 | + for i := range s { |
| 40 | + h ^= uint32(s[i]) |
| 41 | + h *= 16777619 |
| 42 | + } |
| 43 | + return h |
| 44 | +} |
| 45 | + |
| 46 | +func match(s string, t []byte) bool { |
| 47 | + for i, c := range t { |
| 48 | + if s[i] != c { |
| 49 | + return false |
| 50 | + } |
| 51 | + } |
| 52 | + return true |
| 53 | +} |
| 54 | + |
| 55 | +// Lookup returns the atom whose name is s. It returns zero if there is no |
| 56 | +// such atom. The lookup is case sensitive. |
| 57 | +func Lookup(s []byte) Atom { |
| 58 | + if len(s) == 0 || len(s) > maxAtomLen { |
| 59 | + return 0 |
| 60 | + } |
| 61 | + h := fnv(hash0, s) |
| 62 | + if a := table[h&uint32(len(table)-1)]; int(a&0xff) == len(s) && match(a.string(), s) { |
| 63 | + return a |
| 64 | + } |
| 65 | + if a := table[(h>>16)&uint32(len(table)-1)]; int(a&0xff) == len(s) && match(a.string(), s) { |
| 66 | + return a |
| 67 | + } |
| 68 | + return 0 |
| 69 | +} |
| 70 | + |
| 71 | +// String returns a string whose contents are equal to s. In that sense, it is |
| 72 | +// equivalent to string(s) but may be more efficient. |
| 73 | +func String(s []byte) string { |
| 74 | + if a := Lookup(s); a != 0 { |
| 75 | + return a.String() |
| 76 | + } |
| 77 | + return string(s) |
| 78 | +} |
0 commit comments