1 Brief explanation of the hyphenation algorithm herein.[1] 2 3 Raph Levien <raph (a] acm.org> 4 4 Aug 1998 5 6 The hyphenation algorithm is basically the same as Knuth's TeX 7 algorithm. However, the implementation is quite a bit faster. 8 9 The hyphenation files from TeX can almost be used directly. There 10 is a preprocessing step, however. If you don't do the preprocessing 11 step, you'll get bad hyphenations (i.e. a silent failure). 12 13 Start with a file such as hyphen.us. This is the TeX ushyph1.tex 14 file, with the exception dictionary encoded using the same rules as 15 the main portion of the file. Any line beginning with % is a comment. 16 Each other line should contain exactly one rule. 17 18 Then, do the preprocessing - "perl substrings.pl hyphen.us". The 19 resulting file is hyphen.mashed. It's in Perl, and it's fairly slow 20 (it uses brute force algorithms; about 17 seconds on a P100), but it 21 could probably be redone in C with clever algorithms. This would be 22 valuable, for example, if it was handle user-supplied exception 23 dictionaries by integrating them into the rule table.[2] 24 25 Once the rules are preprocessed, loading them is quite quick - 26 about 200ms on a P100. It then hyphenates at about 40,000 words per 27 second on a P100. I haven't benchmarked it against other 28 implementations (both TeX and groff contain essentially the same 29 algorithm), but expect that it runs quite a bit faster than any of 30 them. 31 32 Knuth's algorithm 33 34 This section contains a brief explanation of Knuth's algorithm, in 35 case you missed it from the TeX books. We'll use the semi-word 36 "example" as our running example. 37 38 Since the beginning and end of a word are special, the algorithm is 39 actually run over the prepared word (prep_word in the source) 40 ".example.". Knuths algorithm basically just does pattern matches from 41 the rule set, then applies the matches. The patterns in this case that 42 match are "xa", "xam", "mp", and "pl". These are actually stored as 43 "x1a", "xam3", "4m1p", and "1p2l2". Whenever numbers appear between 44 the letters, they are added in. If two (or more) patterns have numbers 45 in the same place, the highest number wins. Here's the example: 46 47 . e x a m p l e . 48 x1a 49 x a m3 50 4m1p 51 1p2l2 52 ----------------- 53 . e x1a4m3p2l2e . 54 55 Finally, hyphens are placed wherever odd numbers appear. They are, 56 however, suppressed after the first letter and before the last letter 57 of the word (TeX actually suppresses them before the next-to-last, as 58 well). So, it's "ex-am-ple", which is correct. 59 60 Knuth uses a trie to implement this. I.e. he stores each rule in a 61 trie structure. For each position in the word, he searches the trie, 62 searching for a match. Most patterns are short, so efficiency should 63 be quite good. 64 65 Theory of the algorithm 66 67 The algorithm works as a slightly modified finite state machine. 68 There are two kinds of transitions: those that consume one letter of 69 input (which work just like your regular finite state machine), and 70 "fallback" transitions, which don't consume any input. If no 71 transition matching the next letter is found, the fallback is used. 72 One way of looking at this is a form of compression of the transition 73 tables - i.e. it behaves the same as a completely vanilla state 74 machine in which the actual transition table of a node is made up of 75 the union of transition tables of the node itself, plus its fallbacks. 76 77 Each state is represented by a string. Thus, if the current state 78 is "am" and the next letter is "p", then the next state is "amp". 79 Fallback transitions go to states which chop off one or (sometimes) 80 more letters from the beginning. For example, if none of the 81 transitions from "amp" match the next letter, then it will fall back 82 to "mp". Similarly, if none of the transitions from "mp" match the 83 next letter, it will fall back to "m". 84 85 Each state is also associated with a (possibly null) "match" 86 string. This represents the union of all patterns which are 87 right-justified substrings of the match string. I.e. the pattern "mp" 88 is a right-justified substring of the state "amp", so it's numbers get 89 added in. The actual calculation of this union is done by the 90 Perl preprocessing script, but could probably be done in C just about 91 as easily. 92 93 Because each state transition either consumes one input character 94 or shortens the state string by one character, the total number of 95 state transitions is linear in the length of the word. 96 97 [1] Documentations: 98 99 Franklin M. Liang: Word Hy-phen-a-tion by Com-put-er. 100 Stanford University, 1983. http://www.tug.org/docs/liang. 101 102 Lszl Nmeth: Automatic non-standard hyphenation in OpenOffice.org, 103 TUGboat (27), 2006. No. 2., http://hunspell.sourceforge.net/tb87nemeth.pdf 104 105 [2] There is the C version of pattern converter "substrings.c" 106 in the distribution written by Nanning Buitenhuis. Unfortunatelly, 107 this version hasn't handled the non standard extension of the 108 algorithm, yet. 109