| 1 |
Grammar rules have the following characteristics: |
| 2 |
|
| 3 |
itemization( |
| 4 |
it() The construction |
| 5 |
verb( |
| 6 |
exp: |
| 7 |
exp '+' exp |
| 8 |
; |
| 9 |
) |
| 10 |
is a recursive rule definition, stating that when two tt(exp) groupings, |
| 11 |
with a `+' token in between, have been recognized, these three elements |
| 12 |
themselves represent an tt(exp) gouping. |
| 13 |
|
| 14 |
it() Whitespace in rules is only used for separating symbols. |
| 15 |
|
| 16 |
it() em(Action blocks) determining the |
| 17 |
semantics of the rule can be inserted between the elements of production |
| 18 |
rules. An action block looks like this: |
| 19 |
verb( |
| 20 |
{ |
| 21 |
C++ statements |
| 22 |
} |
| 23 |
) |
| 24 |
Usually there is only one action block, defined at the end of production |
| 25 |
rules . See section ref(ACTIONS). |
| 26 |
|
| 27 |
it() Multiple production rules for the same grouping can be written |
| 28 |
separately or can be separated from each other using the vertical-bar |
| 29 |
character `|' as follows: |
| 30 |
verb( |
| 31 |
result: |
| 32 |
first-production-rule |
| 33 |
| |
| 34 |
second-production-rule |
| 35 |
| |
| 36 |
... |
| 37 |
; |
| 38 |
) |
| 39 |
|
| 40 |
it() Alternatively, multiple rules of the same grouping can be |
| 41 |
defined. E.g., the previous definition of tt(result:) could also have been |
| 42 |
defined as: |
| 43 |
verb( |
| 44 |
result: |
| 45 |
first-production-rule |
| 46 |
; |
| 47 |
|
| 48 |
result: |
| 49 |
second-production-rule |
| 50 |
; |
| 51 |
) |
| 52 |
However, this is a potentially dangerous practice, since one of the two |
| 53 |
tt(result) rules could also have used a misspelled rule-name (maybe the second |
| 54 |
tt(result)) should have been tt(results). Therefore, b() generates a warning |
| 55 |
if the same nonterminal is used repeatedly when defining production rules. |
| 56 |
|
| 57 |
it() Production rules can be empty. This means that em(result) can match |
| 58 |
any token, which then isn't processed by the production rule, although the |
| 59 |
rule's nonterminal has been recognized. Such a alternative is called an |
| 60 |
em(empty production rule). For example, here is how to define a |
| 61 |
comma-separated sequence of zero or more tt(exp) nonterminals: |
| 62 |
verb( |
| 63 |
expseq: |
| 64 |
expseq1 |
| 65 |
| |
| 66 |
// empty |
| 67 |
; |
| 68 |
|
| 69 |
expseq1: |
| 70 |
expseq1 ',' exp |
| 71 |
| |
| 72 |
exp |
| 73 |
; |
| 74 |
) |
| 75 |
By convention, to improve the visibility of an empty production rule, it |
| 76 |
contains the comment `tt(// empty)'. |
| 77 |
|
| 78 |
Empty production rules may contain action blocks. Statements in |
| 79 |
such action blocks are executed when the parser has recognized the empty |
| 80 |
production rule. In such cases the tt(// empty) comment is omitted. |
| 81 |
) |
| 82 |
|
| 83 |
|