Future/bison
1
0
Fork 0
mirror of https://git.savannah.gnu.org/git/bison.git synced 2026-03-21 10:13:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

doc: formatting changes

Browse Source 
* doc/bison.texi: No visible changes.
This commit is contained in:
Akim Demaille
2019-11-12 18:44:22 +01:00
parent dbd6975b5c
commit 22ca07defa
1 changed files with 87 additions and 102 deletions
Download Patch File Download Diff File Expand all files Collapse all files

189
doc/bison.texi
View File

@@ -819,35 +819,32 @@ input. These are known respectively as @dfn{reduce/reduce} conflicts
(@pxref{Reduce/Reduce}), and @dfn{shift/reduce} conflicts (@pxref{Reduce/Reduce}), and @dfn{shift/reduce} conflicts
(@pxref{Shift/Reduce}). (@pxref{Shift/Reduce}).
To use a grammar that is not easily modified to be LR(1), a To use a grammar that is not easily modified to be LR(1), a more general
more general parsing algorithm is sometimes necessary. If you include parsing algorithm is sometimes necessary. If you include @code{%glr-parser}
@code{%glr-parser} among the Bison declarations in your file among the Bison declarations in your file (@pxref{Grammar Outline}), the
(@pxref{Grammar Outline}), the result is a Generalized LR result is a Generalized LR (GLR) parser. These parsers handle Bison
(GLR) parser. These parsers handle Bison grammars that grammars that contain no unresolved conflicts (i.e., after applying
contain no unresolved conflicts (i.e., after applying precedence precedence declarations) identically to deterministic parsers. However,
declarations) identically to deterministic parsers. However, when when faced with unresolved shift/reduce and reduce/reduce conflicts, GLR
faced with unresolved shift/reduce and reduce/reduce conflicts, parsers use the simple expedient of doing both, effectively cloning the
GLR parsers use the simple expedient of doing both, parser to follow both possibilities. Each of the resulting parsers can
effectively cloning the parser to follow both possibilities. Each of again split, so that at any given time, there can be any number of possible
the resulting parsers can again split, so that at any given time, there parses being explored. The parsers proceed in lockstep; that is, all of
can be any number of possible parses being explored. The parsers them consume (shift) a given input symbol before any of them proceed to the
proceed in lockstep; that is, all of them consume (shift) a given input next. Each of the cloned parsers eventually meets one of two possible
symbol before any of them proceed to the next. Each of the cloned fates: either it runs into a parsing error, in which case it simply
parsers eventually meets one of two possible fates: either it runs into vanishes, or it merges with another parser, because the two of them have
a parsing error, in which case it simply vanishes, or it merges with reduced the input to an identical set of symbols.
another parser, because the two of them have reduced the input to an
identical set of symbols.
During the time that there are multiple parsers, semantic actions are During the time that there are multiple parsers, semantic actions are
recorded, but not performed. When a parser disappears, its recorded recorded, but not performed. When a parser disappears, its recorded
semantic actions disappear as well, and are never performed. When a semantic actions disappear as well, and are never performed. When a
reduction makes two parsers identical, causing them to merge, Bison reduction makes two parsers identical, causing them to merge, Bison records
records both sets of semantic actions. Whenever the last two parsers both sets of semantic actions. Whenever the last two parsers merge,
merge, reverting to the single-parser case, Bison resolves all the reverting to the single-parser case, Bison resolves all the outstanding
outstanding actions either by precedences given to the grammar rules actions either by precedences given to the grammar rules involved, or by
involved, or by performing both actions, and then calling a designated performing both actions, and then calling a designated user-defined function
user-defined function on the resulting values to produce an arbitrary on the resulting values to produce an arbitrary merged result.
merged result.
@menu @menu
* Simple GLR Parsers:: Using GLR parsers on unambiguous grammars. * Simple GLR Parsers:: Using GLR parsers on unambiguous grammars.
@@ -881,13 +878,11 @@ type enum = (a, b, c);
@end example @end example
@noindent @noindent
The original language standard allows only numeric The original language standard allows only numeric literals and constant
literals and constant identifiers for the subrange bounds (@samp{lo} identifiers for the subrange bounds (@samp{lo} and @samp{hi}), but Extended
and @samp{hi}), but Extended Pascal (ISO/IEC Pascal (ISO/IEC 10206) and many other Pascal implementations allow arbitrary
10206) and many other expressions there. This gives rise to the following situation, containing a
Pascal implementations allow arbitrary expressions there. This gives superfluous pair of parentheses:
rise to the following situation, containing a superfluous pair of
parentheses:
@example @example
type subrange = (a) .. b; type subrange = (a) .. b;
@@ -902,62 +897,55 @@ type enum = (a);
@end example @end example
@noindent @noindent
(These declarations are contrived, but they are syntactically (These declarations are contrived, but they are syntactically valid, and
valid, and more-complicated cases can come up in practical programs.) more-complicated cases can come up in practical programs.)
These two declarations look identical until the @samp{..} token. These two declarations look identical until the @samp{..} token. With
With normal LR(1) one-token lookahead it is not normal LR(1) one-token lookahead it is not possible to decide between the
possible to decide between the two forms when the identifier two forms when the identifier @samp{a} is parsed. It is, however, desirable
@samp{a} is parsed. It is, however, desirable for a parser to decide this, since in the latter case @samp{a} must become a
for a parser to decide this, since in the latter case new identifier to represent the enumeration value, while in the former case
@samp{a} must become a new identifier to represent the enumeration @samp{a} must be evaluated with its current meaning, which may be a constant
value, while in the former case @samp{a} must be evaluated with its or even a function call.
current meaning, which may be a constant or even a function call.
You could parse @samp{(a)} as an ``unspecified identifier in parentheses'', You could parse @samp{(a)} as an ``unspecified identifier in parentheses'',
to be resolved later, but this typically requires substantial to be resolved later, but this typically requires substantial contortions in
contortions in both semantic actions and large parts of the both semantic actions and large parts of the grammar, where the parentheses
grammar, where the parentheses are nested in the recursive rules for are nested in the recursive rules for expressions.
expressions.
You might think of using the lexer to distinguish between the two You might think of using the lexer to distinguish between the two forms by
forms by returning different tokens for currently defined and returning different tokens for currently defined and undefined identifiers.
undefined identifiers. But if these declarations occur in a local But if these declarations occur in a local scope, and @samp{a} is defined in
scope, and @samp{a} is defined in an outer scope, then both forms an outer scope, then both forms are possible---either locally redefining
are possible---either locally redefining @samp{a}, or using the @samp{a}, or using the value of @samp{a} from the outer scope. So this
value of @samp{a} from the outer scope. So this approach cannot approach cannot work.
work.
A simple solution to this problem is to declare the parser to A simple solution to this problem is to declare the parser to use the GLR
use the GLR algorithm. algorithm. When the GLR parser reaches the critical state, it merely splits
When the GLR parser reaches the critical state, it into two branches and pursues both syntax rules simultaneously. Sooner or
merely splits into two branches and pursues both syntax rules later, one of them runs into a parsing error. If there is a @samp{..} token
simultaneously. Sooner or later, one of them runs into a parsing before the next @samp{;}, the rule for enumerated types fails since it
error. If there is a @samp{..} token before the next cannot accept @samp{..} anywhere; otherwise, the subrange type rule fails
@samp{;}, the rule for enumerated types fails since it cannot since it requires a @samp{..} token. So one of the branches fails silently,
accept @samp{..} anywhere; otherwise, the subrange type rule and the other one continues normally, performing all the intermediate
fails since it requires a @samp{..} token. So one of the branches actions that were postponed during the split.
fails silently, and the other one continues normally, performing
all the intermediate actions that were postponed during the split.
If the input is syntactically incorrect, both branches fail and the parser If the input is syntactically incorrect, both branches fail and the parser
reports a syntax error as usual. reports a syntax error as usual.
The effect of all this is that the parser seems to ``guess'' the The effect of all this is that the parser seems to ``guess'' the correct
correct branch to take, or in other words, it seems to use more branch to take, or in other words, it seems to use more lookahead than the
lookahead than the underlying LR(1) algorithm actually allows underlying LR(1) algorithm actually allows for. In this example, LR(2)
for. In this example, LR(2) would suffice, but also some cases would suffice, but also some cases that are not LR(@math{k}) for any
that are not LR(@math{k}) for any @math{k} can be handled this way. @math{k} can be handled this way.
In general, a GLR parser can take quadratic or cubic worst-case time, In general, a GLR parser can take quadratic or cubic worst-case time, and
and the current Bison parser even takes exponential time and space the current Bison parser even takes exponential time and space for some
for some grammars. In practice, this rarely happens, and for many grammars. In practice, this rarely happens, and for many grammars it is
grammars it is possible to prove that it cannot happen. possible to prove that it cannot happen. The present example contains only
The present example contains only one conflict between two one conflict between two rules, and the type-declaration context containing
rules, and the type-declaration context containing the conflict the conflict cannot be nested. So the number of branches that can exist at
cannot be nested. So the number of any time is limited by the constant 2, and the parsing time is still linear.
branches that can exist at any time is limited by the constant 2,
and the parsing time is still linear.
Here is a Bison grammar corresponding to the example above. It Here is a Bison grammar corresponding to the example above. It
parses a vastly simplified form of Pascal type declarations. parses a vastly simplified form of Pascal type declarations.
@@ -1020,32 +1008,29 @@ these two declarations to the Bison grammar file (before the first
@end example @end example
@noindent @noindent
No change in the grammar itself is required. Now the No change in the grammar itself is required. Now the parser recognizes all
parser recognizes all valid declarations, according to the valid declarations, according to the limited syntax above, transparently.
limited syntax above, transparently. In fact, the user does not even In fact, the user does not even notice when the parser splits.
notice when the parser splits.
So here we have a case where we can use the benefits of GLR, So here we have a case where we can use the benefits of GLR, almost without
almost without disadvantages. Even in simple cases like this, however, disadvantages. Even in simple cases like this, however, there are at least
there are at least two potential problems to beware. First, always two potential problems to beware. First, always analyze the conflicts
analyze the conflicts reported by Bison to make sure that GLR reported by Bison to make sure that GLR splitting is only done where it is
splitting is only done where it is intended. A GLR parser intended. A GLR parser splitting inadvertently may cause problems less
splitting inadvertently may cause problems less obvious than an obvious than an LR parser statically choosing the wrong alternative in a
LR parser statically choosing the wrong alternative in a
conflict. Second, consider interactions with the lexer (@pxref{Semantic conflict. Second, consider interactions with the lexer (@pxref{Semantic
Tokens}) with great care. Since a split parser consumes tokens without Tokens}) with great care. Since a split parser consumes tokens without
performing any actions during the split, the lexer cannot obtain performing any actions during the split, the lexer cannot obtain information
information via parser actions. Some cases of lexer interactions can be via parser actions. Some cases of lexer interactions can be eliminated by
eliminated by using GLR to shift the complications from the using GLR to shift the complications from the lexer to the parser. You must
lexer to the parser. You must check the remaining cases for check the remaining cases for correctness.
correctness.
In our example, it would be safe for the lexer to return tokens based on In our example, it would be safe for the lexer to return tokens based on
their current meanings in some symbol table, because no new symbols are their current meanings in some symbol table, because no new symbols are
defined in the middle of a type declaration. Though it is possible for defined in the middle of a type declaration. Though it is possible for a
a parser to define the enumeration constants as they are parsed, before parser to define the enumeration constants as they are parsed, before the
the type declaration is completed, it actually makes no difference since type declaration is completed, it actually makes no difference since they
they cannot be used within the same enumerated type declaration. cannot be used within the same enumerated type declaration.
@node Merging GLR Parses @node Merging GLR Parses
@subsection Using GLR to Resolve Ambiguities @subsection Using GLR to Resolve Ambiguities
@@ -7084,10 +7069,10 @@ If the grammar uses literal string tokens, there are two ways that
@itemize @bullet @itemize @bullet
@item @item
If the grammar defines symbolic token names as aliases for the If the grammar defines symbolic token names as aliases for the literal
literal string tokens, @code{yylex} can use these symbolic names like string tokens, @code{yylex} can use these symbolic names like all others.
all others. In this case, the use of the literal string tokens in In this case, the use of the literal string tokens in the grammar file has
the grammar file has no effect on @code{yylex}. no effect on @code{yylex}.
@item @item
@code{yylex} can find the multicharacter token in the @code{yytname} table. @code{yylex} can find the multicharacter token in the @code{yytname} table.