nix-super/src/libexpr/lexer.l

300 lines
7.9 KiB
Text
Raw Normal View History

%option reentrant bison-bridge bison-locations
%option noyywrap
%option never-interactive
%option stack
%option nodefault
%option nounput noyy_top_state
2016-06-14 18:42:46 +03:00
%s DEFAULT
%x STRING
%x IND_STRING
2021-07-29 19:03:07 +03:00
%x INPATH
%x INPATH_SLASH
%x PATH_START
%{
2020-12-02 15:33:20 +02:00
#ifdef __clang__
#pragma clang diagnostic ignored "-Wunneeded-internal-declaration"
2020-12-02 15:33:20 +02:00
#endif
libexpr: Use int64_t for NixInt Using a 64bit integer on 32bit systems will come with a bit of a performance overhead, but given that Nix doesn't use a lot of integers compared to other types, I think the overhead is negligible also considering that 32bit systems are in decline. The biggest advantage however is that when we use a consistent integer size across all platforms it's less likely that we miss things that we break due to that. One example would be: https://github.com/NixOS/nixpkgs/pull/44233 On Hydra it will evaluate, because the evaluator runs on a 64bit machine, but when evaluating the same on a 32bit machine it will fail, so using 64bit integers should make that consistent. While the change of the type in value.hh is rather easy to do, we have a few more options available for doing the conversion in the lexer: * Via an #ifdef on the architecture and using strtol() or strtoll() accordingly depending on which architecture we are. For the #ifdef we would need another AX_COMPILE_CHECK_SIZEOF in configure.ac. * Using istringstream, which would involve copying the value. * As we're already using boost, lexical_cast might be a good idea. Spoiler: I went for the latter, first of all because lexical_cast does have an overload for const char* and second of all, because it doesn't involve copying around the input string. Also, because istringstream seems to come with a bigger overhead than boost::lexical_cast: https://www.boost.org/doc/libs/release/doc/html/boost_lexical_cast/performance.html The first method (still using strtol/strtoll) also wasn't something I pursued further, because it is also locale-aware which I doubt is what we want, given that the regex for int is [0-9]+. Signed-off-by: aszlig <aszlig@nix.build> Fixes: #2339
2018-08-29 01:23:51 +03:00
#include <boost/lexical_cast.hpp>
#include "nixexpr.hh"
#include "parser-tab.hh"
using namespace nix;
namespace nix {
2021-09-30 02:37:51 +03:00
// backup to recover from yyless(0)
YYLTYPE prev_yylloc;
2013-09-02 17:29:15 +03:00
static void initLoc(YYLTYPE * loc)
{
loc->first_line = loc->last_line = 1;
loc->first_column = loc->last_column = 1;
}
2013-09-02 17:29:15 +03:00
static void adjustLoc(YYLTYPE * loc, const char * s, size_t len)
{
2021-09-30 02:37:51 +03:00
prev_yylloc = *loc;
loc->first_line = loc->last_line;
loc->first_column = loc->last_column;
for (size_t i = 0; i < len; i++) {
switch (*s++) {
case '\r':
if (*s == '\n') { /* cr/lf */
i++;
s++;
}
/* fall through */
2013-09-02 17:29:15 +03:00
case '\n':
++loc->last_line;
loc->last_column = 1;
break;
default:
++loc->last_column;
}
}
}
// we make use of the fact that the parser receives a private copy of the input
// string and can munge around in it.
2022-01-19 14:39:42 +02:00
static StringToken unescapeStr(SymbolTable & symbols, char * s, size_t length)
{
char * result = s;
char * t = s;
char c;
// the input string is terminated with *two* NULs, so we can safely take
// *one* character after the one being checked against.
while ((c = *s++)) {
if (c == '\\') {
c = *s++;
if (c == 'n') *t = '\n';
else if (c == 'r') *t = '\r';
else if (c == 't') *t = '\t';
else *t = c;
}
else if (c == '\r') {
/* Normalise CR and CR/LF into LF. */
*t = '\n';
if (*s == '\n') s++; /* cr/lf */
}
else *t = c;
t++;
}
2022-01-19 14:39:42 +02:00
return {result, size_t(t - result)};
}
2013-09-02 17:29:15 +03:00
}
#define YY_USER_INIT initLoc(yylloc)
#define YY_USER_ACTION adjustLoc(yylloc, yytext, yyleng);
#define PUSH_STATE(state) yy_push_state(state, yyscanner)
#define POP_STATE() yy_pop_state(yyscanner)
%}
ANY .|\n
ID [a-zA-Z\_][a-zA-Z0-9\_\'\-]*
INT [0-9]+
FLOAT (([1-9][0-9]*\.[0-9]*)|(0?\.[0-9]+))([Ee][+-]?[0-9]+)?
2021-07-29 19:03:07 +03:00
PATH_CHAR [a-zA-Z0-9\.\_\-\+]
PATH {PATH_CHAR}*(\/{PATH_CHAR}+)+\/?
PATH_SEG {PATH_CHAR}*\/
HPATH \~(\/{PATH_CHAR}+)+\/?
HPATH_START \~\/
SPATH \<{PATH_CHAR}+(\/{PATH_CHAR}+)*\>
URI [a-zA-Z][a-zA-Z0-9\+\-\.]*\:[a-zA-Z0-9\%\/\?\:\@\&\=\+\$\,\-\_\.\!\~\*\']+
%%
if { return IF; }
then { return THEN; }
else { return ELSE; }
assert { return ASSERT; }
with { return WITH; }
let { return LET; }
in { return IN; }
rec { return REC; }
inherit { return INHERIT; }
or { return OR_KW; }
\.\.\. { return ELLIPSIS; }
\=\= { return EQ; }
\!\= { return NEQ; }
\<\= { return LEQ; }
\>\= { return GEQ; }
\&\& { return AND; }
\|\| { return OR; }
\-\> { return IMPL; }
\/\/ { return UPDATE; }
\+\+ { return CONCAT; }
{ID} { yylval->id = {yytext, (size_t) yyleng}; return ID; }
{INT} { errno = 0;
libexpr: Use int64_t for NixInt Using a 64bit integer on 32bit systems will come with a bit of a performance overhead, but given that Nix doesn't use a lot of integers compared to other types, I think the overhead is negligible also considering that 32bit systems are in decline. The biggest advantage however is that when we use a consistent integer size across all platforms it's less likely that we miss things that we break due to that. One example would be: https://github.com/NixOS/nixpkgs/pull/44233 On Hydra it will evaluate, because the evaluator runs on a 64bit machine, but when evaluating the same on a 32bit machine it will fail, so using 64bit integers should make that consistent. While the change of the type in value.hh is rather easy to do, we have a few more options available for doing the conversion in the lexer: * Via an #ifdef on the architecture and using strtol() or strtoll() accordingly depending on which architecture we are. For the #ifdef we would need another AX_COMPILE_CHECK_SIZEOF in configure.ac. * Using istringstream, which would involve copying the value. * As we're already using boost, lexical_cast might be a good idea. Spoiler: I went for the latter, first of all because lexical_cast does have an overload for const char* and second of all, because it doesn't involve copying around the input string. Also, because istringstream seems to come with a bigger overhead than boost::lexical_cast: https://www.boost.org/doc/libs/release/doc/html/boost_lexical_cast/performance.html The first method (still using strtol/strtoll) also wasn't something I pursued further, because it is also locale-aware which I doubt is what we want, given that the regex for int is [0-9]+. Signed-off-by: aszlig <aszlig@nix.build> Fixes: #2339
2018-08-29 01:23:51 +03:00
try {
yylval->n = boost::lexical_cast<int64_t>(yytext);
} catch (const boost::bad_lexical_cast &) {
2020-04-23 00:00:11 +03:00
throw ParseError("invalid integer '%1%'", yytext);
libexpr: Use int64_t for NixInt Using a 64bit integer on 32bit systems will come with a bit of a performance overhead, but given that Nix doesn't use a lot of integers compared to other types, I think the overhead is negligible also considering that 32bit systems are in decline. The biggest advantage however is that when we use a consistent integer size across all platforms it's less likely that we miss things that we break due to that. One example would be: https://github.com/NixOS/nixpkgs/pull/44233 On Hydra it will evaluate, because the evaluator runs on a 64bit machine, but when evaluating the same on a 32bit machine it will fail, so using 64bit integers should make that consistent. While the change of the type in value.hh is rather easy to do, we have a few more options available for doing the conversion in the lexer: * Via an #ifdef on the architecture and using strtol() or strtoll() accordingly depending on which architecture we are. For the #ifdef we would need another AX_COMPILE_CHECK_SIZEOF in configure.ac. * Using istringstream, which would involve copying the value. * As we're already using boost, lexical_cast might be a good idea. Spoiler: I went for the latter, first of all because lexical_cast does have an overload for const char* and second of all, because it doesn't involve copying around the input string. Also, because istringstream seems to come with a bigger overhead than boost::lexical_cast: https://www.boost.org/doc/libs/release/doc/html/boost_lexical_cast/performance.html The first method (still using strtol/strtoll) also wasn't something I pursued further, because it is also locale-aware which I doubt is what we want, given that the regex for int is [0-9]+. Signed-off-by: aszlig <aszlig@nix.build> Fixes: #2339
2018-08-29 01:23:51 +03:00
}
return INT;
}
{FLOAT} { errno = 0;
2016-01-05 10:46:37 +02:00
yylval->nf = strtod(yytext, 0);
if (errno != 0)
2020-04-23 00:00:11 +03:00
throw ParseError("invalid float '%1%'", yytext);
return FLOAT;
}
2016-06-14 18:42:46 +03:00
\$\{ { PUSH_STATE(DEFAULT); return DOLLAR_CURLY; }
2016-06-14 18:42:46 +03:00
\} { /* State INITIAL only exists at the bottom of the stack and is
used as a marker. DEFAULT replaces it everywhere else.
Popping when in INITIAL state causes an empty stack exception,
so don't */
if (YYSTATE != INITIAL)
POP_STATE();
return '}';
}
\{ { PUSH_STATE(DEFAULT); return '{'; }
2016-06-14 18:42:46 +03:00
\" { PUSH_STATE(STRING); return '"'; }
<STRING>([^\$\"\\]|\$[^\{\"\\]|\\{ANY}|\$\\{ANY})*\$/\" |
<STRING>([^\$\"\\]|\$[^\{\"\\]|\\{ANY}|\$\\{ANY})+ {
/* It is impossible to match strings ending with '$' with one
regex because trailing contexts are only valid at the end
of a rule. (A sane but undocumented limitation.) */
2022-01-19 14:39:42 +02:00
yylval->str = unescapeStr(data->symbols, yytext, yyleng);
return STR;
}
2016-06-14 18:42:46 +03:00
<STRING>\$\{ { PUSH_STATE(DEFAULT); return DOLLAR_CURLY; }
<STRING>\" { POP_STATE(); return '"'; }
<STRING>\$|\\|\$\\ {
/* This can only occur when we reach EOF, otherwise the above
(...|\$[^\{\"\\]|\\.|\$\\.)+ would have triggered.
This is technically invalid, but we leave the problem to the
parser who fails with exact location. */
return STR;
}
2016-06-14 18:42:46 +03:00
\'\'(\ *\n)? { PUSH_STATE(IND_STRING); return IND_STRING_OPEN; }
<IND_STRING>([^\$\']|\$[^\{\']|\'[^\'\$])+ {
2022-01-19 14:39:42 +02:00
yylval->str = {yytext, (size_t) yyleng, true};
return IND_STR;
2007-12-06 12:20:58 +02:00
}
<IND_STRING>\'\'\$ |
<IND_STRING>\$ {
2022-01-19 14:39:42 +02:00
yylval->str = {"$", 1};
2007-12-06 12:20:58 +02:00
return IND_STR;
}
<IND_STRING>\'\'\' {
2022-01-19 14:39:42 +02:00
yylval->str = {"''", 2};
2007-12-06 12:20:58 +02:00
return IND_STR;
}
<IND_STRING>\'\'\\{ANY} {
2022-01-19 14:39:42 +02:00
yylval->str = unescapeStr(data->symbols, yytext + 2, yyleng - 2);
2007-12-06 12:20:58 +02:00
return IND_STR;
}
2016-06-14 18:42:46 +03:00
<IND_STRING>\$\{ { PUSH_STATE(DEFAULT); return DOLLAR_CURLY; }
<IND_STRING>\'\' { POP_STATE(); return IND_STRING_CLOSE; }
<IND_STRING>\' {
2022-01-19 14:39:42 +02:00
yylval->str = {"'", 1};
return IND_STR;
}
2021-07-29 19:03:07 +03:00
{PATH_SEG}\$\{ |
{HPATH_START}\$\{ {
PUSH_STATE(PATH_START);
yyless(0);
2021-09-30 02:37:51 +03:00
*yylloc = prev_yylloc;
2021-07-29 19:03:07 +03:00
}
<PATH_START>{PATH_SEG} {
POP_STATE();
PUSH_STATE(INPATH_SLASH);
yylval->path = {yytext, (size_t) yyleng};
2021-07-29 19:03:07 +03:00
return PATH;
}
<PATH_START>{HPATH_START} {
POP_STATE();
PUSH_STATE(INPATH_SLASH);
yylval->path = {yytext, (size_t) yyleng};
2021-07-29 19:03:07 +03:00
return HPATH;
}
{PATH} {
if (yytext[yyleng-1] == '/')
PUSH_STATE(INPATH_SLASH);
else
PUSH_STATE(INPATH);
yylval->path = {yytext, (size_t) yyleng};
2021-07-29 19:03:07 +03:00
return PATH;
}
{HPATH} {
if (yytext[yyleng-1] == '/')
PUSH_STATE(INPATH_SLASH);
else
PUSH_STATE(INPATH);
yylval->path = {yytext, (size_t) yyleng};
2021-07-29 19:03:07 +03:00
return HPATH;
}
<INPATH,INPATH_SLASH>\$\{ {
POP_STATE();
PUSH_STATE(INPATH);
PUSH_STATE(DEFAULT);
return DOLLAR_CURLY;
}
<INPATH,INPATH_SLASH>{PATH}|{PATH_SEG}|{PATH_CHAR}+ {
POP_STATE();
if (yytext[yyleng-1] == '/')
PUSH_STATE(INPATH_SLASH);
else
PUSH_STATE(INPATH);
2022-01-19 14:39:42 +02:00
yylval->str = {yytext, (size_t) yyleng};
2021-07-29 19:03:07 +03:00
return STR;
}
<INPATH>{ANY} |
<INPATH><<EOF>> {
/* if we encounter a non-path character we inform the parser that the path has
ended with a PATH_END token and re-parse this character in the default
context (it may be ')', ';', or something of that sort) */
POP_STATE();
yyless(0);
2021-09-30 02:37:51 +03:00
*yylloc = prev_yylloc;
2021-07-29 19:03:07 +03:00
return PATH_END;
}
<INPATH_SLASH>{ANY} |
<INPATH_SLASH><<EOF>> {
throw ParseError("path has a trailing slash");
}
{SPATH} { yylval->path = {yytext, (size_t) yyleng}; return SPATH; }
{URI} { yylval->uri = {yytext, (size_t) yyleng}; return URI; }
[ \t\r\n]+ /* eat up whitespace */
\#[^\r\n]* /* single-line comments */
\/\*([^*]|\*+[^*/])*\*+\/ /* long comments */
{ANY} {
/* Don't return a negative number, as this will cause
Bison to stop parsing without an error. */
return (unsigned char) yytext[0];
}
%%