rewrite in XS
Lukas Mai [Sun, 17 Jun 2012 00:54:41 +0000 (02:54 +0200)]
GNUmakefile [new file with mode: 0644]
MANIFEST
Makefile.PL
Parameters.xs [new file with mode: 0644]
README
ignore.txt [deleted file]
lib/Function/Parameters.pm
t/elsewhere.t
t/lineno.t
toke_on_crack.c.inc [new file with mode: 0644]

diff --git a/GNUmakefile b/GNUmakefile
new file mode 100644 (file)
index 0000000..6b3454d
--- /dev/null
@@ -0,0 +1,3 @@
+include ./Makefile
+
+CCFLAGS := -DDEVEL $(CCFLAGS)
index c113acc..91bb5ac 100644 (file)
--- a/MANIFEST
+++ b/MANIFEST
@@ -1,17 +1,19 @@
 Changes
-MANIFEST
+lib/Function/Parameters.pm
 Makefile.PL
+MANIFEST
+MANIFEST.SKIP
+MYMETA.json
+MYMETA.yml
+Parameters.xs
 README
-lib/Function/Parameters.pm
-t/pod.t
 t/00-load.t
 t/01-compiles.t
 t/02-compiles.t
-t/elsewhere.t
-t/rename.t
-t/eating_strict_error.t
 t/eating_strict_error.fail
+t/eating_strict_error.t
 t/eating_strict_error_2.fail
+t/elsewhere.t
 t/lineno-torture.t
 t/lineno.t
 t/named.t
@@ -19,3 +21,11 @@ t/named_1.fail
 t/named_2.fail
 t/named_3.fail
 t/named_4.fail
+t/pod.t
+t/rename.t
+t/strict.t
+t/strict_1.fail
+t/strict_2.fail
+t/strict_3.fail
+t/strict_4.fail
+toke_on_crack.c.inc
index ea52471..b9fdbfa 100644 (file)
@@ -17,7 +17,10 @@ WriteMakefile(
         'Test::More' => 0,
     },
     PREREQ_PM => {
+       'Carp' => 0,
+       'B::Hooks::EndOfScope' => 0,
        'XSLoader' => 0,
+       'bytes' => 0,
         'warnings' => 0,
         'strict' => 0,
     },
diff --git a/Parameters.xs b/Parameters.xs
new file mode 100644 (file)
index 0000000..240a859
--- /dev/null
@@ -0,0 +1,348 @@
+/*
+Copyright 2012 Lukas Mai.
+
+This program is free software; you can redistribute it and/or modify it
+under the terms of either: the GNU General Public License as published
+by the Free Software Foundation; or the Artistic License.
+
+See http://dev.perl.org/licenses/ for more information.
+ */
+
+#ifdef __GNUC__
+ #if (__GNUC__ == 4 && __GNUC_MINOR__ >= 6) || __GNUC__ >= 5
+  #define PRAGMA_GCC_(X) _Pragma(#X)
+  #define PRAGMA_GCC(X) PRAGMA_GCC_(GCC X)
+ #endif
+#endif
+
+#ifndef PRAGMA_GCC
+ #define PRAGMA_GCC(X)
+#endif
+
+#ifdef DEVEL
+ #define WARNINGS_RESET PRAGMA_GCC(diagnostic pop)
+ #define WARNINGS_ENABLEW(X) PRAGMA_GCC(diagnostic warning #X)
+ #define WARNINGS_ENABLE \
+       WARNINGS_ENABLEW(-Wall) \
+       WARNINGS_ENABLEW(-Wextra) \
+       WARNINGS_ENABLEW(-Wundef) \
+       WARNINGS_ENABLEW(-Wshadow) \
+       WARNINGS_ENABLEW(-Wbad-function-cast) \
+       WARNINGS_ENABLEW(-Wcast-align) \
+       WARNINGS_ENABLEW(-Wwrite-strings) \
+       /* WARNINGS_ENABLEW(-Wnested-externs) wtf? */ \
+       WARNINGS_ENABLEW(-Wstrict-prototypes) \
+       WARNINGS_ENABLEW(-Wmissing-prototypes) \
+       WARNINGS_ENABLEW(-Winline) \
+       WARNINGS_ENABLEW(-Wdisabled-optimization)
+
+#else
+ #define WARNINGS_RESET
+ #define WARNINGS_ENABLE(X)
+#endif
+
+
+#include "EXTERN.h"
+#include "perl.h"
+#include "XSUB.h"
+
+#include <string.h>
+
+WARNINGS_ENABLE
+
+#define MY_PKG "Function::Parameters"
+
+#define HINTK_KEYWORDS MY_PKG "/keywords"
+#define HINTK_NAME_    MY_PKG "/name:"
+#define HINTK_SHIFT_   MY_PKG "/shift:"
+
+typedef struct {
+       enum {
+               FLAG_NAME_OPTIONAL = 1,
+               FLAG_NAME_REQUIRED,
+               FLAG_NAME_PROHIBITED
+       } name;
+       char shift[256];
+} Spec;
+
+static int (*next_keyword_plugin)(pTHX_ char *, STRLEN, OP **);
+
+static int kw_flags(const char *kw_ptr, STRLEN kw_len, Spec *spec) {
+       HV *hints;
+       SV *sv, **psv;
+       const char *p, *kw_active;
+       STRLEN kw_active_len;
+
+       spec->name = 0;
+       spec->shift[0] = '\0';
+
+       if (!(hints = GvHV(PL_hintgv))) {
+               return FALSE;
+       }
+       if (!(psv = hv_fetchs(hints, HINTK_KEYWORDS, 0))) {
+               return FALSE;
+       }
+       sv = *psv;
+       kw_active = SvPV(sv, kw_active_len);
+       if (kw_active_len <= kw_len) {
+               return FALSE;
+       }
+       for (p = kw_active; p < kw_active + kw_active_len - kw_len; p++) {
+               if (
+                       (p == kw_active || p[-1] == ' ') &&
+                       p[kw_len] == ' ' &&
+                       memcmp(kw_ptr, p, kw_len) == 0
+               ) {
+                       const char *kf_ptr;
+                       STRLEN kf_len;
+                       SV *kf_sv;
+
+                       kf_sv = sv_2mortal(newSVpvs(HINTK_NAME_));
+                       sv_catpvn(kf_sv, kw_ptr, kw_len);
+                       kf_ptr = SvPV(kf_sv, kf_len);
+                       if (!(psv = hv_fetch(hints, kf_ptr, kf_len, 0))) {
+                               croak("%s: internal error: $^H{'%.*s'} not set", MY_PKG, (int)kf_len, kf_ptr);
+                       }
+                       spec->name = SvIV(*psv);
+
+                       kf_sv = sv_2mortal(newSVpvs(HINTK_SHIFT_));
+                       sv_catpvn(kf_sv, kw_ptr, kw_len);
+                       kf_ptr = SvPV(kf_sv, kf_len);
+                       if (!(psv = hv_fetch(hints, kf_ptr, kf_len, 0))) {
+                               croak("%s: internal error: $^H{'%.*s'} not set", MY_PKG, (int)kf_len, kf_ptr);
+                       }
+                       my_sprintf(spec->shift, "%.*s", (int)(sizeof spec->shift - 1), SvPV_nolen(*psv));
+
+                       return TRUE;
+               }
+       }
+       return FALSE;
+}
+
+
+#include "toke_on_crack.c.inc"
+
+
+static int parse_fun(OP **pop, const char *keyword_ptr, STRLEN keyword_len, const Spec *spec) {
+       SV *gen, *declarator, *params, *sv;
+       line_t line_start;
+       int saw_name, saw_colon;
+       STRLEN len;
+       char *s;
+       I32 c;
+
+       gen = sv_2mortal(newSVpvs("sub"));
+       declarator = sv_2mortal(newSVpvn(keyword_ptr, keyword_len));
+       params = sv_2mortal(newSVpvs(""));
+
+       line_start = CopLINE(PL_curcop);
+       lex_read_space(0);
+
+       /* function name */
+       saw_name = 0;
+       s = PL_parser->bufptr;
+       if (spec->name != FLAG_NAME_PROHIBITED && (len = S_scan_word(s, TRUE))) {
+               sv_catpvs(gen, " ");
+               sv_catpvn(gen, s, len);
+               sv_catpvs(declarator, " ");
+               sv_catpvn(declarator, s, len);
+               lex_read_to(s + len);
+               lex_read_space(0);
+               saw_name = 1;
+       } else if (spec->name == FLAG_NAME_REQUIRED) {
+               croak("I was expecting a function name, not \"%.*s\"", (int)(PL_parser->bufend - s), s);
+       } else {
+               sv_catpvs(declarator, " (anon)");
+       }
+
+       /* parameters */
+       c = lex_peek_unichar(0);
+       if (c == '(') {
+               SV *saw_slurpy = NULL;
+
+               lex_read_unichar(0);
+               lex_read_space(0);
+
+               for (;;) {
+                       c = lex_peek_unichar(0);
+                       if (c && strchr("$@%", c)) {
+                               sv_catpvf(params, "%c", (int)c);
+                               lex_read_unichar(0);
+                               lex_read_space(0);
+
+                               s = PL_parser->bufptr;
+                               if (!(len = S_scan_word(s, FALSE))) {
+                                       croak("In %.*s: missing identifier", (int)SvCUR(declarator), SvPV_nolen(declarator));
+                               }
+                               if (saw_slurpy) {
+                                       croak("In %.*s: I was expecting \")\" after \"%s\", not \"%c%.*s\"", (int)SvCUR(declarator), SvPV_nolen(declarator), SvPV_nolen(saw_slurpy), (int)c, (int)len, s);
+                               }
+                               if (c != '$') {
+                                       saw_slurpy = sv_2mortal(newSVpvf("%c%.*s", (int)c, (int)len, s));
+                               }
+                               sv_catpvn(params, s, len);
+                               sv_catpvs(params, ",");
+                               lex_read_to(s + len);
+                               lex_read_space(0);
+
+                               c = lex_peek_unichar(0);
+                               if (c == ',') {
+                                       lex_read_unichar(0);
+                                       lex_read_space(0);
+                                       continue;
+                               }
+                       }
+
+                       if (c == ')') {
+                               lex_read_unichar(0);
+                               lex_read_space(0);
+                               break;
+                       }
+
+                       if (c == -1) {
+                               croak("In %.*s: unexpected EOF in parameter list", (int)SvCUR(declarator), SvPV_nolen(declarator));
+                       }
+                       croak("In %.*s: unexpected '%c' in parameter list", (int)SvCUR(declarator), SvPV_nolen(declarator), (int)c);
+               }
+       }
+
+       /* prototype */
+       saw_colon = 0;
+       c = lex_peek_unichar(0);
+       if (c == ':') {
+               lex_read_unichar(0);
+               lex_read_space(0);
+
+               c = lex_peek_unichar(0);
+               if (c != '(') {
+                       saw_colon = 1;
+               } else {
+                       sv = sv_2mortal(newSVpvs(""));
+                       if (!S_scan_str(sv, TRUE, TRUE)) {
+                               croak("In %.*s: malformed prototype", (int)SvCUR(declarator), SvPV_nolen(declarator));
+                       }
+                       sv_catsv(gen, sv);
+                       lex_read_space(0);
+               }
+       }
+
+       if (saw_name) {
+               len = SvCUR(gen);
+               s = SvGROW(gen, (len + 1) * 2);
+               sv_catpvs(gen, ";");
+               sv_catpvn(gen, s, len);
+       }
+
+       /* attributes */
+       if (!saw_colon) {
+               c = lex_peek_unichar(0);
+               if (c == ':') {
+                       saw_colon = 1;
+                       lex_read_unichar(0);
+                       lex_read_space(0);
+               }
+       }
+       if (saw_colon) {
+               for (;;) {
+                       s = PL_parser->bufptr;
+                       if (!(len = S_scan_word(s, FALSE))) {
+                               break;
+                       }
+                       sv_catpvs(gen, ":");
+                       sv_catpvn(gen, s, len);
+                       lex_read_to(s + len);
+                       lex_read_space(0);
+                       c = lex_peek_unichar(0);
+                       if (c == '(') {
+                               sv = sv_2mortal(newSVpvs(""));
+                               if (!S_scan_str(sv, TRUE, TRUE)) {
+                                       croak("In %.*s: malformed attribute argument list", (int)SvCUR(declarator), SvPV_nolen(declarator));
+                               }
+                               sv_catsv(gen, sv);
+                               lex_read_space(0);
+                               c = lex_peek_unichar(0);
+                       }
+                       if (c == ':') {
+                               lex_read_unichar(0);
+                               lex_read_space(0);
+                       }
+               }
+       }
+
+       /* body */
+       c = lex_peek_unichar(0);
+       if (c != '{') {
+               croak("In %.*s: I was expecting a function body, not \"%c\"", (int)SvCUR(declarator), SvPV_nolen(declarator), (int)c);
+       }
+       lex_read_unichar(0);
+       sv_catpvs(gen, "{");
+       if (spec->shift[0]) {
+               sv_catpvf(gen, "my%s=shift;", spec->shift);
+       }
+       if (SvCUR(params)) {
+               sv_catpvs(gen, "my(");
+               sv_catsv(gen, params);
+               sv_catpvs(gen, ")=@_;");
+       }
+
+       /* fprintf(stderr, "! [%.*s]\n", (int)(PL_bufend - PL_bufptr), PL_bufptr); */
+
+       /* named sub */
+       if (saw_name) {
+               lex_stuff_sv(gen, SvUTF8(gen));
+               *pop = parse_barestmt(0);
+               return KEYWORD_PLUGIN_STMT;
+       }
+
+       /* anon sub */
+       sv_catpvs(gen, "BEGIN{" MY_PKG "::_fini}");
+       lex_stuff_sv(gen, SvUTF8(gen));
+       *pop = parse_arithexpr(0);
+       s = PL_parser->bufptr;
+       if (*s != '}') {
+               croak("%s: internal error: expected '}', found '%c'", MY_PKG, *s);
+       }
+       lex_unstuff(s + 1);
+       return KEYWORD_PLUGIN_EXPR;
+}
+
+static int my_keyword_plugin(pTHX_ char *keyword_ptr, STRLEN keyword_len, OP **op_ptr) {
+       Spec spec;
+       int ret;
+
+       SAVETMPS;
+
+       if (kw_flags(keyword_ptr, keyword_len, &spec)) {
+               ret = parse_fun(op_ptr, keyword_ptr, keyword_len, &spec);
+       } else {
+               ret = next_keyword_plugin(keyword_ptr, keyword_len, op_ptr);
+       }
+
+       FREETMPS;
+
+       return ret;
+}
+
+WARNINGS_RESET
+
+MODULE = Function::Parameters   PACKAGE = Function::Parameters
+PROTOTYPES: ENABLE
+
+BOOT:
+WARNINGS_ENABLE {
+       HV *const stash = gv_stashpvs(MY_PKG, GV_ADD);
+       newCONSTSUB(stash, "FLAG_NAME_OPTIONAL", newSViv(FLAG_NAME_OPTIONAL));
+       newCONSTSUB(stash, "FLAG_NAME_REQUIRED", newSViv(FLAG_NAME_REQUIRED));
+       newCONSTSUB(stash, "FLAG_NAME_PROHIBITED", newSViv(FLAG_NAME_PROHIBITED));
+       newCONSTSUB(stash, "HINTK_KEYWORDS", newSVpvs(HINTK_KEYWORDS));
+       newCONSTSUB(stash, "HINTK_NAME_", newSVpvs(HINTK_NAME_));
+       newCONSTSUB(stash, "HINTK_SHIFT_", newSVpvs(HINTK_SHIFT_));
+       newCONSTSUB(stash, "SHIFT_NAME_LIMIT", newSViv(sizeof ((Spec *)NULL)->shift));
+       next_keyword_plugin = PL_keyword_plugin;
+       PL_keyword_plugin = my_keyword_plugin;
+} WARNINGS_RESET
+
+void
+xs_fini()
+       CODE:
+       lex_stuff_pvn("}", 1, 0);
diff --git a/README b/README
index b5e7069..0dba1d0 100644 (file)
--- a/README
+++ b/README
@@ -36,7 +36,7 @@ You can also look for information at:
 
 COPYRIGHT AND LICENCE
 
-Copyright (C) 2009 Lukas Mai
+Copyright (C) 2009-2012 Lukas Mai
 
 This program is free software; you can redistribute it and/or modify it
 under the terms of either: the GNU General Public License as published
diff --git a/ignore.txt b/ignore.txt
deleted file mode 100644 (file)
index 319a58f..0000000
+++ /dev/null
@@ -1,10 +0,0 @@
-blib*
-Makefile
-Makefile.old
-Build
-_build*
-pm_to_blib*
-*.tar.gz
-.lwpcookies
-Function-Parameters-*
-cover_db
index 69aa627..50fb348 100644 (file)
@@ -3,37 +3,15 @@ package Function::Parameters;
 use strict;
 use warnings;
 
-our $VERSION = '0.05';
-
-use Carp qw(croak confess);
-use Devel::Declare;
-use B::Hooks::EndOfScope;
-
-our @CARP_NOT = qw(Devel::Declare);
-
-
-# Make our import chainable so a wrapper module that wants to turn on F:P
-# for its users can just say
-#    sub import { Function::Parameters->import; }
-#
-# To make that possible we skip all subs named 'import' in our search for the
-# target package.
-#
-sub guess_caller {
-       my ($start) = @_;
-       $start ||= 1;
-
-       my $defcaller = (caller $start)[0];
-       my $caller = $defcaller;
-
-       for (my $level = $start; ; ++$level) {
-               my ($pkg, $function) = (caller $level)[0, 3] or last;
-               $function =~ /::import\z/ or return $caller;
-               $caller = $pkg;
-       }
-       $defcaller
+use XSLoader;
+BEGIN {
+       our $VERSION = '0.05_01';
+       XSLoader::load;
 }
 
+use B::Hooks::EndOfScope qw(on_scope_end);
+use Carp qw(confess);
+use bytes ();
 
 sub _assert_valid_identifier {
        my ($name, $with_dollar) = @_;
@@ -42,16 +20,14 @@ sub _assert_valid_identifier {
                or confess qq{"$name" doesn't look like a valid identifier};
 }
 
-# Parse import spec and make shit happen.
-#
 my @bare_arms = qw(function method);
 my %type_map = (
        function => { name => 'optional' },
        method   => { name => 'optional', shift => '$self' },
 );
 
-sub import_into {
-       my $victim = shift;
+sub import {
+       my $class = shift;
 
        @_ or @_ = ('fun', 'method');
        if (@_ == 1 && ref($_[0]) eq 'HASH') {
@@ -78,272 +54,48 @@ sub import_into {
                $type->{name} ||= 'optional';
                $type->{name} =~ /^(?:optional|required|prohibited)\z/
                        or confess qq["$type->{name}" doesn't look like a valid name attribute (one of optional, required, prohibited)];
-               $type->{shift}
-                       and _assert_valid_identifier $type->{shift}, 1;
+               if ($type->{shift}) {
+                       _assert_valid_identifier $type->{shift}, 1;
+                       bytes::length($type->{shift}) < SHIFT_NAME_LIMIT
+                               or confess qq["$type->{shift}" is longer than I can handle];
+               }
                
-               $spec{$name} = {const => mk_parse($type)};
+               $spec{$name} = $type;
        }
        
-       Devel::Declare->setup_for($victim, \%spec);
-       for my $name (keys %spec) {
-               no strict 'refs';
-               *{$victim . '::' . $name} = \&_declarator;
+       for my $kw (keys %spec) {
+               my $type = $spec{$kw};
+
+               $^H{HINTK_SHIFT_ . $kw} = $type->{shift} || '';
+               $^H{HINTK_NAME_ . $kw} =
+                       $type->{name} eq 'prohibited' ? FLAG_NAME_PROHIBITED :
+                       $type->{name} eq 'required' ? FLAG_NAME_REQUIRED :
+                       FLAG_NAME_OPTIONAL
+               ;
+               $^H{+HINTK_KEYWORDS} .= "$kw ";
        }
 }
 
-sub import {
+sub unimport {
        my $class = shift;
-       my $caller = guess_caller;
-       import_into $caller, @_;
-}
-
-sub _declarator {
-       $_[0]
-}
-
-
-# Wrapper around substr where param 3 is an end offset, not a length.
-#
-sub _substring {
-       @_ >= 4
-       ? substr $_[0], $_[1], $_[2] - $_[1], $_[3]
-       : substr $_[0], $_[1], $_[2] - $_[1]
-}
-
-sub _skip_space {
-       my ($ctx, $key) = @_;
-       my $cur = my $start = $ctx->{offset};
-       while (my $d = Devel::Declare::toke_skipspace $cur) {
-               $cur += $d;
-       }
-       $ctx->{space}{$key} .= _substring Devel::Declare::get_linestr, $start, $cur if $key;
-       $ctx->{offset} = $cur;
-}
-
-sub _grab_name {
-       my ($ctx) = @_;
-       my $p = $ctx->{offset};
-       my $namlen = Devel::Declare::toke_scan_word $p, !!'handle_package'
-               or return;
-       my $str = Devel::Declare::get_linestr;
-       $ctx->{name} = substr $str, $p, $namlen;
-       $ctx->{offset} += $namlen;
-       _skip_space $ctx, 'name';
-}
-
-sub _grab_params {
-       my ($ctx) = @_;
-       substr(Devel::Declare::get_linestr, $ctx->{offset}, 1) eq '('
-               or return;
-       $ctx->{offset}++;
-       _skip_space $ctx, 'params';
-
-       my $pcount = 0;
-
-       LOOP: {
-               my $c = substr Devel::Declare::get_linestr, $ctx->{offset}, 1;
-
-               if ($c =~ /^[\$\@%]\z/) {
-                       $ctx->{offset}++;
-                       _skip_space $ctx, "params_$pcount";
-                       my $namlen = Devel::Declare::toke_scan_word $ctx->{offset}, !'handle_package'
-                               or croak "Missing identifier";
-                       my $name = substr Devel::Declare::get_linestr, $ctx->{offset}, $namlen;
-                       $ctx->{params} .= $c . $name . ',';
-                       $ctx->{offset} += $namlen;
-                       _skip_space $ctx, "params_$pcount";
-
-                       $c = substr Devel::Declare::get_linestr, $ctx->{offset}, 1;
-                       if ($c eq ',') {
-                               $ctx->{offset}++;
-                               _skip_space $ctx, "params_$pcount";
-                               $pcount++;
-                               redo LOOP;
-                       }
-               }
-
-               if ($c eq ')') {
-                       $ctx->{offset}++;
-                       _skip_space $ctx, 'params';
-                       return;
-               }
-
-               if ($c eq '') {
-                       croak "Unexpected EOF in parameter list";
-               }
-
-               croak "Unexpected '$c' in parameter list";
-       }
-}
-
-sub _parse_parens {
-       my ($ctx) = @_;
-
-       my $strlen = Devel::Declare::toke_scan_str $ctx->{offset};
-       $strlen == 0 || $strlen == -1 and return;
-
-       $strlen < 0 and confess "Devel::Declare::toke_scan_str done fucked up ($strlen); see https://rt.cpan.org/Ticket/Display.html?id=51679";
-
-       my $str = Devel::Declare::get_lex_stuff;
-       Devel::Declare::clear_lex_stuff;
-
-       $ctx->{offset} += $strlen;
-
-       $str
-}
-
-sub _grab_proto {
-       my ($ctx) = @_;
-
-       my $savepos = $ctx->{offset};
 
-       substr(Devel::Declare::get_linestr, $ctx->{offset}, 1) eq ':'
-               or return;
-       $ctx->{offset}++;
-       _skip_space $ctx, 'proto_tmp';
-
-       unless (substr(Devel::Declare::get_linestr, $ctx->{offset}, 1) eq '(') {
-               $ctx->{offset} = $savepos;
-               delete $ctx->{space}{proto_tmp};
-               return;
-       }
-       $_->{proto} .= delete $_->{proto_tmp} for $ctx->{space};
-
-       defined(my $str = _parse_parens $ctx)
-               or croak "Malformed prototype";
-       $ctx->{proto} = $str;
-
-       _skip_space $ctx, 'proto';
-}
-
-sub _grab_attr {
-       my ($ctx) = @_;
-
-       my $pcount = 0;
-
-       if (substr(Devel::Declare::get_linestr, $ctx->{offset}, 1) eq ':') {
-               $ctx->{offset}++;
-               _skip_space $ctx, "attr_$pcount";
-       } elsif (!defined $ctx->{proto}) {
+       if (!@_) {
+               delete $^H{+HINTK_KEYWORDS};
                return;
        }
 
-       while () {
-               my $namlen = Devel::Declare::toke_scan_word $ctx->{offset}, !'handle_package'
-                       or return;
-               $ctx->{attr} .= substr Devel::Declare::get_linestr, $ctx->{offset}, $namlen;
-               $ctx->{offset} += $namlen;
-               _skip_space $ctx, "attr_$pcount";
-               if (substr(Devel::Declare::get_linestr, $ctx->{offset}, 1) eq '(') {
-                       defined(my $str = _parse_parens $ctx)
-                               or croak "Malformed attribute argument list";
-                       $ctx->{attr} .= "($str)";
-                       _skip_space $ctx, "attr_$pcount";
-               }
-               $pcount++;
-
-               if (substr(Devel::Declare::get_linestr, $ctx->{offset}, 1) eq ':') {
-                       $ctx->{offset}++;
-                       _skip_space $ctx, "attr_$pcount";
-               }
-       }
-}
-
-# IN:
-#  fun name (params) :(proto) :attr { ... }
-# OUT:
-#  fun (do { sub                        (proto) :attr { self? my (params) = @_; ... } })
-#  fun (do { sub name (proto); sub name (proto) :attr { self? my (params) = @_; ... } });
-#
-sub _generate {
-       my ($ctx, $declarator, $shift) = @_;
-
-       my $gen = '(do{sub';
-
-       my $skipped = join '', values %{$ctx->{space}};
-       my $lines = $skipped =~ tr/\n//;
-       $gen .= "\n" x $lines;
-
-       my $proto = defined $ctx->{proto} ? "($ctx->{proto})" : '';
-
-       my $is_stmt = 0;
-       if (defined(my $name = $ctx->{name})) {
-               $is_stmt = 1;
-               $gen .= " $name$proto;";
-               $gen .= "sub $name";
-       }
-
-       $gen .= $proto;
-
-       if (defined $ctx->{attr}) {
-               $gen .= ":$ctx->{attr}";
-       }
-
-       $gen .= '{';
-       $gen .= "BEGIN{${\__PACKAGE__}::_fini($is_stmt)}";
-
-       if ($shift) {
-               _assert_valid_identifier $shift, 1;
-               $gen .= "my$shift=shift;";
-       }
-       if (defined $ctx->{params}) {
-               $gen .= "my($ctx->{params})=\@_;";
-       }
-       $gen
-}
-
-sub mk_parse {
-       my ($spec) = @_;
-
-       sub {
-               my ($declarator, $offset_orig) = @_;
-               my $ctx = {
-                       offset => $offset_orig,
-                       space => {},
-               };
-
-               $ctx->{offset} += Devel::Declare::toke_move_past_token($ctx->{offset});
-               _skip_space $ctx;
-
-               my $start = $ctx->{offset};
-
-               _grab_name $ctx unless $spec->{name} eq 'prohibited';
-               $ctx->{name} or croak qq[I was expecting a function name, not "${\substr Devel::Declare::get_linestr, $ctx->{offset}}"] if $spec->{name} eq 'required';
-               my $fname = $ctx->{name} || '(anon)';
-               _grab_params $ctx;
-               if ($ctx->{params} && $ctx->{params} =~ /([\@%]\w+),([\$\@%]\w+)/) {
-                       my ($slurpy, $after) = ($1, $2);
-                       croak qq[In $declarator $fname: I was expecting ")" after "$slurpy", not "$after"];
-               }
-               _grab_proto $ctx;
-               _grab_attr $ctx;
-
-               my $offset = $ctx->{offset};
-
-               my $linestr = Devel::Declare::get_linestr;
-               substr($linestr, $offset, 1) eq '{'
-                       or croak qq[In $declarator $fname: I was expecting a function body, not "${\substr $linestr, $offset}"];
-
-               my $gen = _generate $ctx, $declarator, $spec->{shift};
-               my $oldlen = $offset + 1 - $start;
-               _substring $linestr, $start, $offset + 1, (' ' x $oldlen) . $gen;
-               Devel::Declare::set_linestr $linestr;
+       for my $kw (@_) {
+               $^H{+HINTK_KEYWORDS} =~ s/(?<![^ ])\Q$kw\E //g;
        }
 }
 
-# Patch in the end of our synthetic 'do' block, close argument list, and
-# optionally terminate the statement.
-#
 sub _fini {
-       my ($stmt) = @_;
        on_scope_end {
-               my $off = Devel::Declare::get_linestr_offset;
-               my $str = Devel::Declare::get_linestr;
-               substr $str, $off, 0, '})' . ($stmt ? ';' : '');
-               Devel::Declare::set_linestr $str;
+               xs_fini;
        };
 }
 
+
 'ok'
 
 __END__
@@ -388,9 +140,9 @@ Function::Parameters - subroutine definitions with parameter lists
 =head1 DESCRIPTION
 
 This module lets you use parameter lists in your subroutines. Thanks to
-L<Devel::Declare> it works without source filters.
+L<perlapi/PL_keyword_plugin> it works without source filters.
 
-WARNING: This is my first attempt at using L<Devel::Declare> and I have
+WARNING: This is my first attempt at writing L<XS code|perlxs> and I have
 almost no experience with perl's internals. So while this module might
 appear to work, it could also conceivably make your programs segfault.
 Consider this module alpha quality.
@@ -494,27 +246,24 @@ by L<prototypes|perlsub/"Prototypes">. However, you can include a prototype by
 specifying it as the first attribute (this is syntactically unambiguous
 because normal attributes have to start with a letter).
 
-If you want to wrap C<Function::Parameters>, you may find C<import_into>
-helpful. It lets you specify a target package for the syntax magic, as in:
+If you want to wrap L<Function::Parameters>, you just have to call its
+C<import> method. It always applies to the file that is currently being parsed
+and its effects are lexical (i.e. it works like L<warnings> or L<strict>);
 
   package Some::Wrapper;
   use Function::Parameters ();
   sub import {
-    my $caller = caller;
-    Function::Parameters::import_into $caller;
-    # or Function::Parameters::import_into $caller, @other_import_args;
+    Function::Parameters->import;
+    # or Function::Parameters->import(@other_import_args);
   }
 
-C<import_into> is not exported by this module, so you have to use a fully
-qualified name to call it.
-
 =head1 AUTHOR
 
 Lukas Mai, C<< <l.mai at web.de> >>
 
 =head1 COPYRIGHT & LICENSE
 
-Copyright 2010, 2011 Lukas Mai.
+Copyright 2010, 2011, 2012 Lukas Mai.
 
 This program is free software; you can redistribute it and/or modify it
 under the terms of either: the GNU General Public License as published
index 432a1fc..a25a602 100644 (file)
@@ -2,16 +2,19 @@ use strict;
 use warnings;
 use Test::More;
 
-use Function::Parameters ();
+{
+       package Wrapper;
+       use Function::Parameters ();
+       sub shazam { Function::Parameters->import(@_); }
+}
 
-BEGIN { Function::Parameters::import_into __PACKAGE__; }
+BEGIN { Wrapper::shazam; }
 
 ok fun ($x) { $x }->(1);
 
-BEGIN { Function::Parameters::import_into 'Cu::Ba', 'gorn'; }
-
 {
        package Cu::Ba;
+       BEGIN { Wrapper::shazam 'gorn'; }
 
        gorn wooden ($gorn) { !$gorn }
 }
index 40ac0be..dabddf5 100644 (file)
@@ -1,7 +1,7 @@
 use warnings;
 use strict;
 
-use Test::More tests => 8;
+use Test::More tests => 10;
 
 use Function::Parameters;
 
@@ -60,4 +60,17 @@ TODO: {
        test_loc 'LX -- 4';
 }
 
+TODO: {
+       local $TODO = 'newlines in prototype/attributes';
+
+       fun wtf :(
+
+       )
+       :
+       { test_loc 'LX -- 5 (inner)' }
+
+       wtf;
+       test_loc 'LX -- 5 (outer)';
+}
+
 __DATA__
diff --git a/toke_on_crack.c.inc b/toke_on_crack.c.inc
new file mode 100644 (file)
index 0000000..905c85f
--- /dev/null
@@ -0,0 +1,289 @@
+/*
+ * This code was copied from perl/toke.c and subsequently butchered
+ * by Lukas Mai (2012).
+ */
+
+/* vvvvvvvvvvvvvvvvvvvvv I HAVE NO IDEA WHAT I'M DOING vvvvvvvvvvvvvvvvvvvv */
+#define PL_linestr              (PL_parser->linestr)
+#define PL_copline              (PL_parser->copline)
+#define PL_bufptr               (PL_parser->bufptr)
+#define PL_bufend               (PL_parser->bufend)
+#define PL_multi_start          (PL_parser->multi_start)
+#define PL_multi_open           (PL_parser->multi_open)
+#define PL_multi_close          (PL_parser->multi_close)
+#define PL_multi_end            (PL_parser->multi_end)
+#define PL_rsfp                 (PL_parser->rsfp)
+
+#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
+
+#ifdef USE_UTF8_SCRIPTS
+#   define UTF (!IN_BYTES)
+#else
+#   define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
+#endif
+
+static STRLEN S_scan_word(const char *start, int allow_package) {
+       const char *s = start;
+       for (;;) {
+               if (isALNUM(*s) || (!UTF && isALNUMC_L1(*s))) {  /* UTF handled below */
+                       s++;
+               } else if (allow_package && s > start && *s == '\'' && isIDFIRST_lazy_if(s+1, UTF)) {
+                       s++;
+               } else if (allow_package && s[0] == ':' && s[1] == ':' && isIDFIRST_lazy_if(s+2, UTF)) {
+                       s += 2;
+               } else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
+                       do {
+                               s += UTF8SKIP(s);
+                       } while (UTF8_IS_CONTINUED(*s) && is_utf8_mark((U8*)s));
+               } else {
+                       return s - start;
+               }
+       }
+}
+
+static char *S_scan_str(pTHX_ SV *sv, int keep_quoted, int keep_delims) {
+       dVAR;
+       char *start = PL_bufptr;
+       const char *tmps;                   /* temp string, used for delimiter matching */
+       char *s = start;                    /* current position in the buffer */
+       char term;                          /* terminating character */
+       char *to;                           /* current position in the sv's data */
+       I32 brackets = 1;                   /* bracket nesting level */
+       bool has_utf8 = FALSE;              /* is there any utf8 content? */
+       I32 termcode;                       /* terminating char. code */
+       U8 termstr[UTF8_MAXBYTES];          /* terminating string */
+       STRLEN termlen;                     /* length of terminating string */
+       int last_off = 0;                   /* last position for nesting bracket */
+
+       /* XXX ATTENTION: we don't skip whitespace! */
+
+       /* mark where we are, in case we need to report errors */
+       CLINE;
+
+       /* after skipping whitespace, the next character is the terminator */
+       term = *s;
+       if (!UTF) {
+               termcode = termstr[0] = term;
+               termlen = 1;
+       }
+       else {
+               termcode = utf8_to_uvchr_buf((U8*)s, (U8*)PL_bufend, &termlen);
+               Copy(s, termstr, termlen, U8);
+               if (!UTF8_IS_INVARIANT(term))
+                       has_utf8 = TRUE;
+       }
+
+       /* mark where we are */
+       PL_multi_start = CopLINE(PL_curcop);
+       PL_multi_open = term;
+
+       /* find corresponding closing delimiter */
+       if (term && (tmps = strchr("([{< )]}> )]}>",term)))
+               termcode = termstr[0] = term = tmps[5];
+
+       PL_multi_close = term;
+
+       {
+               STRLEN dummy;
+               SvPV_force(sv, dummy);
+               sv_setpvs(sv, "");
+               SvGROW(sv, 80);
+       }
+
+       /* move past delimiter and try to read a complete string */
+       if (keep_delims)
+               sv_catpvn(sv, s, termlen);
+       s += termlen;
+       for (;;) {
+               if (PL_encoding && !UTF) {
+                       bool cont = TRUE;
+
+                       while (cont) {
+                               int offset = s - SvPVX_const(PL_linestr);
+                               const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
+                                                                &offset, (char*)termstr, termlen);
+                               const char * const ns = SvPVX_const(PL_linestr) + offset;
+                               char * const svlast = SvEND(sv) - 1;
+
+                               for (; s < ns; s++) {
+                                       if (*s == '\n' && !PL_rsfp && !PL_parser->filtered)
+                                               CopLINE_inc(PL_curcop);
+                               }
+                               if (!found)
+                                       goto read_more_line;
+                               else {
+                                       /* handle quoted delimiters */
+                                       if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
+                                               const char *t;
+                                               for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
+                                                       t--;
+                                               if ((svlast-1 - t) % 2) {
+                                                       if (!keep_quoted) {
+                                                               *(svlast-1) = term;
+                                                               *svlast = '\0';
+                                                               SvCUR_set(sv, SvCUR(sv) - 1);
+                                                       }
+                                                       continue;
+                                               }
+                                       }
+                                       if (PL_multi_open == PL_multi_close) {
+                                               cont = FALSE;
+                                       }
+                                       else {
+                                               const char *t;
+                                               char *w;
+                                               for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
+                                                       /* At here, all closes are "was quoted" one,
+                                                          so we don't check PL_multi_close. */
+                                                       if (*t == '\\') {
+                                                               if (!keep_quoted && *(t+1) == PL_multi_open)
+                                                                       t++;
+                                                               else
+                                                                       *w++ = *t++;
+                                                       }
+                                                       else if (*t == PL_multi_open)
+                                                               brackets++;
+
+                                                       *w = *t;
+                                               }
+                                               if (w < t) {
+                                                       *w++ = term;
+                                                       *w = '\0';
+                                                       SvCUR_set(sv, w - SvPVX_const(sv));
+                                               }
+                                               last_off = w - SvPVX(sv);
+                                               if (--brackets <= 0)
+                                                       cont = FALSE;
+                                       }
+                               }
+                       }
+                       if (!keep_delims) {
+                               SvCUR_set(sv, SvCUR(sv) - 1);
+                               *SvEND(sv) = '\0';
+                       }
+                       break;
+               }
+
+               /* extend sv if need be */
+               SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
+               /* set 'to' to the next character in the sv's string */
+               to = SvPVX(sv)+SvCUR(sv);
+
+               /* if open delimiter is the close delimiter read unbridle */
+               if (PL_multi_open == PL_multi_close) {
+                       for (; s < PL_bufend; s++,to++) {
+                               /* embedded newlines increment the current line number */
+                               if (*s == '\n' && !PL_rsfp && !PL_parser->filtered)
+                                       CopLINE_inc(PL_curcop);
+                               /* handle quoted delimiters */
+                               if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
+                                       if (!keep_quoted && s[1] == term)
+                                               s++;
+                               /* any other quotes are simply copied straight through */
+                                       else
+                                               *to++ = *s++;
+                               }
+                               /* terminate when run out of buffer (the for() condition), or
+                                  have found the terminator */
+                               else if (*s == term) {
+                                       if (termlen == 1)
+                                               break;
+                                       if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
+                                               break;
+                               }
+                               else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
+                                       has_utf8 = TRUE;
+                               *to = *s;
+                       }
+               }
+
+               /* if the terminator isn't the same as the start character (e.g.,
+                  matched brackets), we have to allow more in the quoting, and
+                  be prepared for nested brackets.
+               */
+               else {
+                       /* read until we run out of string, or we find the terminator */
+                       for (; s < PL_bufend; s++,to++) {
+                               /* embedded newlines increment the line count */
+                               if (*s == '\n' && !PL_rsfp && !PL_parser->filtered)
+                                       CopLINE_inc(PL_curcop);
+                               /* backslashes can escape the open or closing characters */
+                               if (*s == '\\' && s+1 < PL_bufend) {
+                                       if (!keep_quoted &&
+                                               ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
+                                               s++;
+                                       else
+                                               *to++ = *s++;
+                               }
+                               /* allow nested opens and closes */
+                               else if (*s == PL_multi_close && --brackets <= 0)
+                                       break;
+                               else if (*s == PL_multi_open)
+                                       brackets++;
+                               else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
+                                       has_utf8 = TRUE;
+                               *to = *s;
+                       }
+               }
+               /* terminate the copied string and update the sv's end-of-string */
+               *to = '\0';
+               SvCUR_set(sv, to - SvPVX_const(sv));
+
+               /*
+                * this next chunk reads more into the buffer if we're not done yet
+                */
+
+               if (s < PL_bufend)
+                       break;              /* handle case where we are done yet :-) */
+
+#ifndef PERL_STRICT_CR
+               if (to - SvPVX_const(sv) >= 2) {
+                       if ((to[-2] == '\r' && to[-1] == '\n') ||
+                           (to[-2] == '\n' && to[-1] == '\r'))
+                       {
+                               to[-2] = '\n';
+                               to--;
+                               SvCUR_set(sv, to - SvPVX_const(sv));
+                       }
+                       else if (to[-1] == '\r')
+                               to[-1] = '\n';
+               }
+               else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
+                       to[-1] = '\n';
+#endif
+
+        read_more_line:
+               /* if we're out of file, or a read fails, bail and reset the current
+                  line marker so we can report where the unterminated string began
+               */
+               CopLINE_inc(PL_curcop);
+               PL_bufptr = PL_bufend;
+               if (!lex_next_chunk(0)) {
+                       CopLINE_set(PL_curcop, (line_t)PL_multi_start);
+                       return NULL;
+               }
+               s = PL_bufptr;
+       }
+
+       /* at this point, we have successfully read the delimited string */
+
+       if (!PL_encoding || UTF) {
+               if (keep_delims)
+                       sv_catpvn(sv, s, termlen);
+               s += termlen;
+       }
+       if (has_utf8 || PL_encoding)
+               SvUTF8_on(sv);
+
+       PL_multi_end = CopLINE(PL_curcop);
+
+       /* if we allocated too much space, give some back */
+       if (SvCUR(sv) + 5 < SvLEN(sv)) {
+               SvLEN_set(sv, SvCUR(sv) + 1);
+               SvPV_renew(sv, SvLEN(sv));
+       }
+
+       PL_bufptr = s;
+       return s;
+}
+/* ^^^^^^^^^^^^^^^^^^^^^ I HAVE NO IDEA WHAT I'M DOING ^^^^^^^^^^^^^^^^^^^^ */