| 1 |
B() implements a simple error recovery mechanism. When the tt(lookup()) |
| 2 |
function cannot find an action for the current token in the current state it |
| 3 |
throws an tt(UNEXPECTED_TOKEN_) exception. |
| 4 |
|
| 5 |
This exception is caught by the parsing function, calling the |
| 6 |
tt(errorRecovery()) member function. By default, this member function |
| 7 |
terminates the parsing process. The non-default recovery procedure is |
| 8 |
available once an tt(error) token is used in a production rule. When the |
| 9 |
parsing process throws bf(UNEXPECTED_TOKEN_) the recovery procedure is |
| 10 |
started (i.e., it is started whenever a syntactical error is encountered or |
| 11 |
tt(ERROR()) is called). |
| 12 |
|
| 13 |
The recovery procedure consists of |
| 14 |
itemization( |
| 15 |
it() looking for the first state on the state-stack having an |
| 16 |
error-production, followed by: |
| 17 |
it() handling all state transitions that are possible without retrieving a |
| 18 |
terminal token. |
| 19 |
it() then, in the state requiring a terminal token and starting |
| 20 |
with the initial unexpected token (3) all subsequent terminal tokens are |
| 21 |
ignored until a token is retrieved which is a continuation token in that |
| 22 |
state. |
| 23 |
) |
| 24 |
|
| 25 |
If the error recovery procedure fails (i.e., if no acceptable token is |
| 26 |
ever encountered) error recovery falls back to the default recovery |
| 27 |
mode (i.e., the parsing process is terminated). |
| 28 |
|
| 29 |
Not all syntactic errors are always reported: the option |
| 30 |
link(--required-tokens)(REQUIRED) can be used to specify the minimum number of |
| 31 |
tokens that must have been successfully processed before another syntactic |
| 32 |
error is reported (and counted). |
| 33 |
|
| 34 |
The option link(--error-verbose)(ERRORVERBOSE) may be specified to obtain the |
| 35 |
contents of the state stack when a syntactic error is reported. |
| 36 |
|
| 37 |
The example grammar may be provided with an tt(error) production rule: |
| 38 |
verb( |
| 39 |
%token NR |
| 40 |
|
| 41 |
%left '+' |
| 42 |
|
| 43 |
%% |
| 44 |
|
| 45 |
start: |
| 46 |
start expr |
| 47 |
| |
| 48 |
// empty |
| 49 |
; |
| 50 |
|
| 51 |
expr: |
| 52 |
error |
| 53 |
| |
| 54 |
NR |
| 55 |
| |
| 56 |
expr '+' expr |
| 57 |
; |
| 58 |
) |
| 59 |
The resulting grammar has one additional state (handling the error |
| 60 |
production) and one state in which the tt(ERR_ITEM) flag has been set. When |
| 61 |
and error is encountered, this state obtains tokens until a token having a |
| 62 |
valid continuation was received, after which normal processing continues. |
| 63 |
|
| 64 |
In the parser's verbose output (using the tt(-V) option) the various grammar |
| 65 |
rules and state transition tables are shown. The debug output below refers to |
| 66 |
this information. |
| 67 |
|
| 68 |
The rules are: |
| 69 |
verb( |
| 70 |
1: start -> start expr |
| 71 |
2: start -> <empty> |
| 72 |
3: expr (errTok_) -> errTok_ |
| 73 |
4: expr (NR) -> NR |
| 74 |
5: expr ('+') -> expr '+' expr |
| 75 |
6: start_$ -> start |
| 76 |
) |
| 77 |
|
| 78 |
The state-transitions are: |
| 79 |
verb( |
| 80 |
State 0: |
| 81 |
6: start_$ -> . start |
| 82 |
0: On start to state 1 |
| 83 |
Reduce by 2: start -> . |
| 84 |
|
| 85 |
|
| 86 |
State 1: |
| 87 |
6: start_$ -> start . |
| 88 |
1: start -> start . expr |
| 89 |
0: On expr to state 2 |
| 90 |
1: On errTok_ to state 3 |
| 91 |
2: On NR to state 4 |
| 92 |
|
| 93 |
|
| 94 |
State 2: |
| 95 |
1: start -> start expr . |
| 96 |
5: expr -> expr . '+' expr |
| 97 |
0: On '+' to state 5 |
| 98 |
Reduce by 1: start -> start expr . |
| 99 |
|
| 100 |
|
| 101 |
State 3: |
| 102 |
3: expr -> errTok_ . |
| 103 |
Reduce by 3: expr -> errTok_ . |
| 104 |
|
| 105 |
|
| 106 |
State 4: |
| 107 |
4: expr -> NR . |
| 108 |
Reduce by 4: expr -> NR . |
| 109 |
|
| 110 |
|
| 111 |
State 5: |
| 112 |
5: expr -> expr '+' . expr |
| 113 |
0: On expr to state 6 |
| 114 |
1: On errTok_ to state 3 |
| 115 |
2: On NR to state 4 |
| 116 |
|
| 117 |
|
| 118 |
State 6: |
| 119 |
5: expr -> expr '+' expr . |
| 120 |
5: expr -> expr . '+' expr |
| 121 |
0 (removed by precedence): On '+' to state 5 |
| 122 |
Reduce by 5: expr -> expr '+' expr . |
| 123 |
) |
| 124 |
|
| 125 |
|
| 126 |
The following output from the tt(parse()) function, generated by b() using the |
| 127 |
tt(--debug) option illustrates error recovery for the above grammar, entering |
| 128 |
the input |
| 129 |
verb( |
| 130 |
a |
| 131 |
3+a |
| 132 |
) |
| 133 |
results in: |
| 134 |
verb( |
| 135 |
parse(): Parsing starts |
| 136 |
|
| 137 |
PUSH 0 (initializing the state stack) |
| 138 |
|
| 139 |
|
| 140 |
LOOKUP: [0, 'Reserved_::UNDETERMINED_'] -> default reduce using rule 2 |
| 141 |
|
| 142 |
REDUCE: rule 2 |
| 143 |
execute action 2 ... |
| 144 |
... completed |
| 145 |
rule 2: pop 0 elements. Next will be: [0, 'start'] |
| 146 |
pop 0 elements from the stack having size 1 |
| 147 |
next: [0, 'start'] |
| 148 |
|
| 149 |
|
| 150 |
PUSH: [0, 'start'] -> 1 |
| 151 |
|
| 152 |
scanner token `a' (97) |
| 153 |
|
| 154 |
ERROR: [1, `a' (97)] -> ??. Errors: 1 |
| 155 |
Syntax error |
| 156 |
Reached ERROR state 1 |
| 157 |
|
| 158 |
PUSH: [1, 'errTok_'] -> 3 |
| 159 |
|
| 160 |
|
| 161 |
LOOKUP: [3, `a' (97)] -> default reduce using rule 3 |
| 162 |
|
| 163 |
REDUCE: rule 3 |
| 164 |
rule 3: pop 1 elements. Next will be: [1, 'expr'] |
| 165 |
pop 1 elements from the stack having size 3 |
| 166 |
next: [1, 'expr'] |
| 167 |
|
| 168 |
available token 'expr' |
| 169 |
|
| 170 |
PUSH: [1, 'expr'] -> 2 |
| 171 |
|
| 172 |
available token `a' (97) |
| 173 |
|
| 174 |
LOOKUP: [2, `a' (97)] -> default reduce using rule 1 |
| 175 |
|
| 176 |
REDUCE: rule 1 |
| 177 |
rule 1: pop 2 elements. Next will be: [0, 'start'] |
| 178 |
pop 2 elements from the stack having size 3 |
| 179 |
next: [0, 'start'] |
| 180 |
|
| 181 |
|
| 182 |
PUSH: [0, 'start'] -> 1 |
| 183 |
|
| 184 |
available token `a' (97) |
| 185 |
|
| 186 |
scanner token 'NR' |
| 187 |
|
| 188 |
PUSH: [1, 'NR'] -> 4 |
| 189 |
ERROR RECOVERED: next state 4 |
| 190 |
|
| 191 |
|
| 192 |
LOOKUP: [4, 'Reserved_::UNDETERMINED_'] -> default reduce using rule 4 |
| 193 |
|
| 194 |
REDUCE: rule 4 |
| 195 |
execute action 4 ... |
| 196 |
... completed |
| 197 |
rule 4: pop 1 elements. Next will be: [1, 'expr'] |
| 198 |
pop 1 elements from the stack having size 3 |
| 199 |
next: [1, 'expr'] |
| 200 |
|
| 201 |
available token 'expr' |
| 202 |
|
| 203 |
PUSH: [1, 'expr'] -> 2 |
| 204 |
|
| 205 |
scanner token 'Reserved_::EOF_' |
| 206 |
|
| 207 |
LOOKUP: [2, 'Reserved_::EOF_'] -> default reduce using rule 1 |
| 208 |
|
| 209 |
REDUCE: rule 1 |
| 210 |
execute action 1 ... |
| 211 |
... completed |
| 212 |
rule 1: pop 2 elements. Next will be: [0, 'start'] |
| 213 |
pop 2 elements from the stack having size 3 |
| 214 |
next: [0, 'start'] |
| 215 |
|
| 216 |
|
| 217 |
PUSH: [0, 'start'] -> 1 |
| 218 |
|
| 219 |
available token 'Reserved_::EOF_' |
| 220 |
ACCEPT(): Parsing successful |
| 221 |
parse(): returns 0 or 1 |
| 222 |
) |
| 223 |
|
| 224 |
The final debug message should be interpreted as a bf(C++) expression: the |
| 225 |
0 indicates that tt(Parse::ACCEPT) rather than tt(Parser::ABORT) was called, |
| 226 |
the 1 shows the value of tt(d_nErrors_). Consequently, tt(parse()) returns 1 |
| 227 |
(i.e., tt(`0 or 1')). |
| 228 |
|
| 229 |
|
| 230 |
|
| 231 |
|
| 232 |
|