]>
Commit | Line | Data |
---|---|---|
d0fb370f RS |
1 | /* Type definitions for nondeterministic finite state machine for bison, |
2 | Copyright (C) 1984, 1989 Free Software Foundation, Inc. | |
3 | ||
4 | This file is part of Bison, the GNU Compiler Compiler. | |
5 | ||
6 | Bison is free software; you can redistribute it and/or modify | |
7 | it under the terms of the GNU General Public License as published by | |
8 | the Free Software Foundation; either version 2, or (at your option) | |
9 | any later version. | |
10 | ||
11 | Bison is distributed in the hope that it will be useful, | |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
14 | GNU General Public License for more details. | |
15 | ||
16 | You should have received a copy of the GNU General Public License | |
17 | along with Bison; see the file COPYING. If not, write to | |
c49a8e71 JT |
18 | the Free Software Foundation, Inc., 59 Temple Place - Suite 330, |
19 | Boston, MA 02111-1307, USA. */ | |
d0fb370f RS |
20 | |
21 | ||
22 | /* These type definitions are used to represent a nondeterministic | |
23 | finite state machine that parses the specified grammar. | |
24 | This information is generated by the function generate_states | |
25 | in the file LR0. | |
26 | ||
27 | Each state of the machine is described by a set of items -- | |
28 | particular positions in particular rules -- that are the possible | |
29 | places where parsing could continue when the machine is in this state. | |
30 | These symbols at these items are the allowable inputs that can follow now. | |
31 | ||
32 | A core represents one state. States are numbered in the number field. | |
33 | When generate_states is finished, the starting state is state 0 | |
34 | and nstates is the number of states. (A transition to a state | |
35 | whose state number is nstates indicates termination.) All the cores | |
36 | are chained together and first_state points to the first one (state 0). | |
37 | ||
38 | For each state there is a particular symbol which must have been the | |
39 | last thing accepted to reach that state. It is the accessing_symbol | |
40 | of the core. | |
41 | ||
42 | Each core contains a vector of nitems items which are the indices | |
43 | in the ritems vector of the items that are selected in this state. | |
44 | ||
45 | The link field is used for chaining buckets that hash states by | |
46 | their itemsets. This is for recognizing equivalent states and | |
47 | combining them when the states are generated. | |
48 | ||
49 | The two types of transitions are shifts (push the lookahead token | |
50 | and read another) and reductions (combine the last n things on the | |
51 | stack via a rule, replace them with the symbol that the rule derives, | |
52 | and leave the lookahead token alone). When the states are generated, | |
53 | these transitions are represented in two other lists. | |
54 | ||
55 | Each shifts structure describes the possible shift transitions out | |
56 | of one state, the state whose number is in the number field. | |
57 | The shifts structures are linked through next and first_shift points to them. | |
58 | Each contains a vector of numbers of the states that shift transitions | |
59 | can go to. The accessing_symbol fields of those states' cores say what kind | |
60 | of input leads to them. | |
61 | ||
62 | A shift to state zero should be ignored. Conflict resolution | |
63 | deletes shifts by changing them to zero. | |
64 | ||
65 | Each reductions structure describes the possible reductions at the state | |
66 | whose number is in the number field. The data is a list of nreds rules, | |
67 | represented by their rule numbers. first_reduction points to the list | |
68 | of these structures. | |
69 | ||
70 | Conflict resolution can decide that certain tokens in certain | |
71 | states should explicitly be errors (for implementing %nonassoc). | |
72 | For each state, the tokens that are errors for this reason | |
73 | are recorded in an errs structure, which has the state number | |
74 | in its number field. The rest of the errs structure is full | |
75 | of token numbers. | |
76 | ||
77 | There is at least one shift transition present in state zero. | |
78 | It leads to a next-to-final state whose accessing_symbol is | |
79 | the grammar's start symbol. The next-to-final state has one shift | |
80 | to the final state, whose accessing_symbol is zero (end of input). | |
81 | The final state has one shift, which goes to the termination state | |
82 | (whose number is nstates-1). | |
83 | The reason for the extra state at the end is to placate the parser's | |
84 | strategy of making all decisions one token ahead of its actions. */ | |
85 | ||
86 | ||
87 | typedef | |
88 | struct core | |
89 | { | |
90 | struct core *next; | |
91 | struct core *link; | |
92 | short number; | |
93 | short accessing_symbol; | |
94 | short nitems; | |
95 | short items[1]; | |
96 | } | |
97 | core; | |
98 | ||
99 | ||
100 | ||
101 | typedef | |
102 | struct shifts | |
103 | { | |
104 | struct shifts *next; | |
105 | short number; | |
106 | short nshifts; | |
107 | short shifts[1]; | |
108 | } | |
109 | shifts; | |
110 | ||
111 | ||
112 | ||
113 | typedef | |
114 | struct errs | |
115 | { | |
116 | short nerrs; | |
117 | short errs[1]; | |
118 | } | |
119 | errs; | |
120 | ||
121 | ||
122 | ||
123 | typedef | |
124 | struct reductions | |
125 | { | |
126 | struct reductions *next; | |
127 | short number; | |
128 | short nreds; | |
129 | short rules[1]; | |
130 | } | |
131 | reductions; | |
132 | ||
133 | ||
134 | ||
135 | extern int nstates; | |
136 | extern core *first_state; | |
137 | extern shifts *first_shift; | |
138 | extern reductions *first_reduction; |