Archives/gnosygnu_xowa
1
0
Fork 0
mirror of https://github.com/gnosygnu/xowa.git synced 2026-03-02 03:49:30 +00:00
Code Issues Projects Releases Wiki Activity

Source: Restore broken commit

Browse Source
This commit is contained in:
gnosygnu
2017-02-06 22:14:55 -05:00
parent 938beac9f9
commit 3bfeb94b43
4380 changed files with 328018 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

214
400_xowa/src/gplx/langs/gfs/Gfs_lxr.java Normal file
View File

@@ -0,0 +1,214 @@
/*
XOWA: the XOWA Offline Wiki Application
Copyright (C) 2012 gnosygnu@gmail.com
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx.langs.gfs; import gplx.*; import gplx.langs.*;
interface Gfs_lxr {
int Lxr_tid();
int Process(Gfs_parser_ctx ctx, int bgn, int end);
}
class Gfs_lxr_whitespace implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_whitespace;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
byte[] src = ctx.Src(); int src_len = ctx.Src_len();
int rv = Gfs_lxr_.Rv_eos, cur_pos;
for (cur_pos = end; cur_pos < src_len; cur_pos++) {
byte b = src[cur_pos];
Object o = ctx.Trie().Match_at_w_b0(ctx.Trie_rv(), b, src, cur_pos, src_len);
if (o == null) {
rv = Gfs_lxr_.Rv_null;
ctx.Process_null(cur_pos);
break;
}
else {
Gfs_lxr lxr = (Gfs_lxr)o;
if (lxr.Lxr_tid() == Gfs_lxr_.Tid_whitespace) {}
else {
rv = Gfs_lxr_.Rv_lxr;
ctx.Process_lxr(cur_pos, lxr);
break;
}
}
}
return rv;
}
public static final Gfs_lxr_whitespace Instance = new Gfs_lxr_whitespace(); Gfs_lxr_whitespace() {}
}
class Gfs_lxr_comment_flat implements Gfs_lxr {
public Gfs_lxr_comment_flat(byte[] bgn_bry, byte[] end_bry) {
this.bgn_bry = bgn_bry; this.bgn_bry_len = bgn_bry.length;
this.end_bry = end_bry; this.end_bry_len = end_bry.length;
} byte[] bgn_bry, end_bry; int bgn_bry_len, end_bry_len;
public int Lxr_tid() {return Gfs_lxr_.Tid_comment;}
public int Process(Gfs_parser_ctx ctx, int lxr_bgn, int lxr_end) {
byte[] src = ctx.Src(); int src_len = ctx.Src_len();
int end_pos = Bry_find_.Find_fwd(src, end_bry, lxr_end, src_len);
// if (end_pos == Bry_find_.Not_found) throw Err_.new_fmt_("comment is not closed: {0}", String_.new_u8(end_bry));
return (end_pos == Bry_find_.Not_found)
? src_len // allow eos to terminate flat comment; needed for "tidy-always-adds-nl-in-textarea" fix; NOTE: DATE:2014-06-21
: end_pos + end_bry_len; // position after end_bry
}
}
class Gfs_lxr_identifier implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_identifier;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
byte[] src = ctx.Src(); int src_len = ctx.Src_len();
int pos, rv = Gfs_lxr_.Rv_eos;
for (pos = end; pos < src_len; pos++) {
byte b = src[pos];
Object o = ctx.Trie().Match_at_w_b0(ctx.Trie_rv(), b, src, pos, src_len);
if (o == null) { // invalid char; stop;
rv = Gfs_lxr_.Rv_null;
ctx.Process_null(pos);
break;
}
else {
Gfs_lxr lxr = (Gfs_lxr)o;
if (lxr.Lxr_tid() == Gfs_lxr_.Tid_identifier) {} // still an identifier; continue
else { // new lxr (EX: "." in "abc."); (a) hold word of "abc"; mark "." as new lxr;
ctx.Hold_word(bgn, pos);
rv = Gfs_lxr_.Rv_lxr;
ctx.Process_lxr(pos, lxr);
break;
}
}
}
if (rv == Gfs_lxr_.Rv_eos) ctx.Process_eos(); // eos
return rv;
}
public static final Gfs_lxr_identifier Instance = new Gfs_lxr_identifier(); Gfs_lxr_identifier() {}
}
class Gfs_lxr_semic implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_semic;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
switch (ctx.Prv_lxr()) {
case Gfs_lxr_.Tid_identifier: ctx.Make_nde(bgn, end); ctx.Cur_nde_from_stack(); break; // a;
case Gfs_lxr_.Tid_quote:
case Gfs_lxr_.Tid_paren_end: ctx.Cur_nde_from_stack(); break; // a();
case Gfs_lxr_.Tid_semic: break; // a;; ignore;
default: ctx.Err_mgr().Fail_invalid_lxr(ctx, bgn, this.Lxr_tid(), Byte_ascii.Semic); break;
}
return end;
}
public static final Gfs_lxr_semic Instance = new Gfs_lxr_semic(); Gfs_lxr_semic() {}
}
class Gfs_lxr_dot implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_dot;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
switch (ctx.Prv_lxr()) {
case Gfs_lxr_.Tid_identifier: ctx.Make_nde(bgn, end); break; // a.
case Gfs_lxr_.Tid_paren_end: break; // a().
default: ctx.Err_mgr().Fail_invalid_lxr(ctx, bgn, this.Lxr_tid(), Byte_ascii.Dot); break;
}
return end;
}
public static final Gfs_lxr_dot Instance = new Gfs_lxr_dot(); Gfs_lxr_dot() {}
}
class Gfs_lxr_paren_bgn implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_paren_bgn;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
switch (ctx.Prv_lxr()) {
case Gfs_lxr_.Tid_identifier: ctx.Make_nde(bgn, end); break; // a(;
default: ctx.Err_mgr().Fail_invalid_lxr(ctx, bgn, this.Lxr_tid(), Byte_ascii.Paren_bgn); break;
}
return end;
}
public static final Gfs_lxr_paren_bgn Instance = new Gfs_lxr_paren_bgn(); Gfs_lxr_paren_bgn() {}
}
class Gfs_lxr_paren_end implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_paren_end;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
switch (ctx.Prv_lxr()) {
case Gfs_lxr_.Tid_paren_bgn:
case Gfs_lxr_.Tid_quote: break; // "))", "abc)", "'abc')"
case Gfs_lxr_.Tid_identifier: ctx.Make_atr_by_idf(); break; // 123)
default: ctx.Err_mgr().Fail_invalid_lxr(ctx, bgn, this.Lxr_tid(), Byte_ascii.Paren_end); break;
}
return end;
}
public static final Gfs_lxr_paren_end Instance = new Gfs_lxr_paren_end(); Gfs_lxr_paren_end() {}
}
class Gfs_lxr_quote implements Gfs_lxr {
public Gfs_lxr_quote(byte[] bgn_bry, byte[] end_bry) {
this.bgn_bry_len = bgn_bry.length;
this.end_bry = end_bry; this.end_bry_len = end_bry.length;
} private byte[] end_bry; private int bgn_bry_len, end_bry_len;
public int Lxr_tid() {return Gfs_lxr_.Tid_quote;}
public int Process(Gfs_parser_ctx ctx, int lxr_bgn, int lxr_end) {
byte[] src = ctx.Src(); int src_len = ctx.Src_len();
int end_pos = Bry_find_.Find_fwd(src, end_bry, lxr_end, src_len);
if (end_pos == Bry_find_.Not_found) throw Err_.new_wo_type("quote is not closed", "end", String_.new_u8(end_bry));
Bry_bfr bfr = ctx.Tmp_bfr().Clear();
int prv_pos = lxr_end;
int nxt_pos = end_pos + end_bry_len;
if (Bry_.Match(src, nxt_pos, nxt_pos + end_bry_len, end_bry)) { // end_bry is doubled; EX: end_bry = ' and raw = a''
while (true) {
bfr.Add_mid(src, prv_pos, end_pos); // add everything up to end_bry
bfr.Add(end_bry); // add end_bry
prv_pos = nxt_pos + end_bry_len; // set prv_pos to after doubled end_bry
end_pos = Bry_find_.Find_fwd(src, end_bry, prv_pos, src_len);
if (end_pos == Bry_find_.Not_found) throw Err_.new_wo_type("quote is not closed", "end", String_.new_u8(end_bry));
nxt_pos = end_pos + end_bry_len;
if (!Bry_.Match(src, nxt_pos, nxt_pos + end_bry_len, end_bry)) {
bfr.Add_mid(src, prv_pos, end_pos);
break;
}
}
ctx.Make_atr_by_bry(lxr_bgn + bgn_bry_len, end_pos, bfr.To_bry_and_clear());
}
else
ctx.Make_atr(lxr_bgn + bgn_bry_len, end_pos);
return end_pos + end_bry_len; // position after quote
}
}
class Gfs_lxr_curly_bgn implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_curly_bgn;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
switch (ctx.Prv_lxr()) {
case Gfs_lxr_.Tid_identifier: ctx.Make_nde(bgn, end); ctx.Stack_add(); break; // a{;
case Gfs_lxr_.Tid_paren_end: ctx.Stack_add(); break; // a(){; NOTE: node exists but needs to be pushed onto stack
default: ctx.Err_mgr().Fail_invalid_lxr(ctx, bgn, this.Lxr_tid(), Byte_ascii.Curly_bgn); break;
}
return end;
}
public static final Gfs_lxr_curly_bgn Instance = new Gfs_lxr_curly_bgn(); Gfs_lxr_curly_bgn() {}
}
class Gfs_lxr_curly_end implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_curly_end;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
ctx.Stack_pop(bgn);
return end;
}
public static final Gfs_lxr_curly_end Instance = new Gfs_lxr_curly_end(); Gfs_lxr_curly_end() {}
}
class Gfs_lxr_equal implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_eq;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
ctx.Make_nde(bgn, end).Op_tid_(Gfs_nde.Op_tid_assign);
return end;
}
public static final Gfs_lxr_equal Instance = new Gfs_lxr_equal(); Gfs_lxr_equal() {}
}
class Gfs_lxr_comma implements Gfs_lxr {
public int Lxr_tid() {return Gfs_lxr_.Tid_comma;}
public int Process(Gfs_parser_ctx ctx, int bgn, int end) {
switch (ctx.Prv_lxr()) {
case Gfs_lxr_.Tid_identifier: ctx.Make_atr_by_idf(); break; // 123,
}
return end;
}
public static final Gfs_lxr_comma Instance = new Gfs_lxr_comma(); Gfs_lxr_comma() {}
}

39
400_xowa/src/gplx/langs/gfs/Gfs_lxr_.java Normal file
View File

@@ -0,0 +1,39 @@
/*
XOWA: the XOWA Offline Wiki Application
Copyright (C) 2012 gnosygnu@gmail.com
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx.langs.gfs; import gplx.*; import gplx.langs.*;
class Gfs_lxr_ {
public static final int Rv_init = -1, Rv_null = -2, Rv_eos = -3, Rv_lxr = -4;
public static final int Tid_identifier = 1, Tid_dot = 2, Tid_semic = 3, Tid_paren_bgn = 4, Tid_paren_end = 5, Tid_curly_bgn = 6, Tid_curly_end = 7, Tid_quote = 8, Tid_comma = 9, Tid_whitespace = 10, Tid_comment = 11, Tid_eq = 12;
public static String Tid__name(int tid) {
switch (tid) {
case Tid_identifier: return "identifier";
case Tid_dot: return "dot";
case Tid_semic: return "semic";
case Tid_paren_bgn: return "paren_bgn";
case Tid_paren_end: return "paren_end";
case Tid_curly_bgn: return "curly_bgn";
case Tid_curly_end: return "curly_end";
case Tid_quote: return "quote";
case Tid_comma: return "comma";
case Tid_whitespace: return "whitespace";
case Tid_comment: return "comment";
case Tid_eq: return "eq";
default: throw Err_.new_unhandled(tid);
}
}
}

53
400_xowa/src/gplx/langs/gfs/Gfs_msg_bldr.java Normal file
View File

@@ -0,0 +1,53 @@
/*
XOWA: the XOWA Offline Wiki Application
Copyright (C) 2012 gnosygnu@gmail.com
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx.langs.gfs; import gplx.*; import gplx.langs.*;
import gplx.core.gfo_regys.*;
public class Gfs_msg_bldr implements GfoMsgParser {
private final Object thread_lock = new Object();
private final Gfs_parser parser = new Gfs_parser();
public GfoMsg ParseToMsg(String s) {return Bld(s);}
public GfoMsg Bld(String src) {return Bld(Bry_.new_u8(src));}
public GfoMsg Bld(byte[] src) {
synchronized (thread_lock) { // LOCK:Gfs_parser called when converting messages in Xow_msg_mgr; DATE:2016-10-18
Gfs_nde nde = parser.Parse(src);
return Bld_msg(src, nde);
}
}
private GfoMsg Bld_msg(byte[] src, Gfs_nde nde) {
boolean op_is_assign = (nde.Op_tid() == Gfs_nde.Op_tid_assign);
String name = String_.new_u8(nde.Name_bry(src));
if (op_is_assign) name += Tkn_mutator;
GfoMsg rv = GfoMsg_.new_parse_(name);
int len = nde.Atrs_len();
for (int i = 0; i < len; i++) {
Gfs_nde atr = nde.Atrs_get_at(i);
rv.Add("", String_.new_u8(atr.Name_bry(src)));
}
len = nde.Subs_len();
for (int i = 0; i < len; i++) {
Gfs_nde sub = nde.Subs_get_at(i);
if (op_is_assign) // NOTE: for now (a) assignss cannot be nested; EX: "a.b = c;" is okay but "a.b = c.d;" is not
rv.Add("", Bld_msg(src, sub).Key());
else
rv.Subs_add(Bld_msg(src, sub));
}
return rv;
}
public static final Gfs_msg_bldr Instance = new Gfs_msg_bldr(); Gfs_msg_bldr() {} // TS.static
public static final String Tkn_mutator = "_";
}

76
400_xowa/src/gplx/langs/gfs/Gfs_msg_bldr_tst.java Normal file
View File

@@ -0,0 +1,76 @@
/*
XOWA: the XOWA Offline Wiki Application
Copyright (C) 2012 gnosygnu@gmail.com
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx.langs.gfs; import gplx.*; import gplx.langs.*;
import org.junit.*; import gplx.core.strings.*;
public class Gfs_msg_bldr_tst {
@Before public void init() {fxt.Clear();} Gfs_msg_bldr_fxt fxt = new Gfs_msg_bldr_fxt();
@Test public void Basic() {
fxt.Test_build("a;", fxt.msg_("a"));
}
@Test public void Dot() {
fxt.Test_build("a.b.c;"
, fxt.msg_("a").Subs_
( fxt.msg_("b").Subs_
( fxt.msg_("c")
)));
}
@Test public void Args() {
fxt.Test_build("a('b', 'c');", fxt.msg_("a", fxt.kv_("", "b"), fxt.kv_("", "c")));
}
@Test public void Args_num() {
fxt.Test_build("a(1);", fxt.msg_("a", fxt.kv_("", "1")));
}
@Test public void Assign() {
fxt.Test_build("a = 'b';", fxt.msg_("a_", fxt.kv_("", "b")));
}
@Test public void Assign_num() {
fxt.Test_build("a = 1;", fxt.msg_("a_", fxt.kv_("", "1")));
}
}
class Gfs_msg_bldr_fxt {
public void Clear() {} String_bldr sb = String_bldr_.new_(); Gfs_msg_bldr msg_bldr = Gfs_msg_bldr.Instance;
public Keyval kv_(String key, String val) {return Keyval_.new_(key, val);}
public GfoMsg msg_(String key, Keyval... args) {
GfoMsg rv = GfoMsg_.new_parse_(key);
int len = args.length;
for (int i = 0; i < len; i++) {
Keyval kv = args[i];
rv.Add(kv.Key(), kv.Val());
}
return rv;
}
public void Test_build(String raw, GfoMsg... expd) {
GfoMsg root = msg_bldr.Bld(raw);
Tfds.Eq_str_lines(Xto_str(expd), Xto_str(To_ary(root)));
}
GfoMsg[] To_ary(GfoMsg msg) {
int len = msg.Subs_count();
GfoMsg[] rv = new GfoMsg[len];
for (int i = 0; i < len; i++)
rv[i] = msg.Subs_getAt(i);
return rv;
}
String Xto_str(GfoMsg[] ary) {
int len = ary.length;
for (int i = 0; i < len; i++) {
if (i != 0) sb.Add_char_crlf();
sb.Add(ary[i].To_str());
}
return sb.To_str_and_clear();
}
}

85
400_xowa/src/gplx/langs/gfs/Gfs_nde.java Normal file
View File

@@ -0,0 +1,85 @@
/*
XOWA: the XOWA Offline Wiki Application
Copyright (C) 2012 gnosygnu@gmail.com
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx.langs.gfs; import gplx.*; import gplx.langs.*;
public class Gfs_nde {
public byte[] Name_bry(byte[] src) {return name == null ? Bry_.Mid(src, name_bgn, name_end) : name;}
public byte[] Name() {return name;} public Gfs_nde Name_(byte[] v) {name = v; return this;} private byte[] name;
public int Name_bgn() {return name_bgn;} private int name_bgn = -1;
public int Name_end() {return name_end;} private int name_end = -1;
public Gfs_nde Name_rng_(int name_bgn, int name_end) {this.name_bgn = name_bgn; this.name_end = name_end; return this;}
public byte Op_tid() {return op_tid;} public Gfs_nde Op_tid_(byte v) {op_tid = v; return this;} private byte op_tid;
public void Subs_clear() {
for (int i = 0; i < subs_len; i++)
subs[i] = null;
subs_len = 0;
}
public int Subs_len() {return subs_len;} private int subs_len;
public Gfs_nde Subs_add_many(Gfs_nde... ary) {
int len = ary.length;
for (int i = 0; i < len; i++)
Subs_add(ary[i]);
return this;
}
public Gfs_nde Subs_add(Gfs_nde nde) {
int new_len = subs_len + 1;
if (new_len > subs_max) { // ary too small >>> expand
subs_max = new_len * 2;
Gfs_nde[] new_subs = new Gfs_nde[subs_max];
Array_.Copy_to(subs, 0, new_subs, 0, subs_len);
subs = new_subs;
}
subs[subs_len] = nde;
subs_len = new_len;
return this;
} Gfs_nde[] subs = Gfs_nde.Ary_empty; int subs_max; int[] subs_pos_ary = Int_.Ary_empty;
public Gfs_nde Subs_get_at(int i) {return subs[i];}
public Gfs_nde[] Subs_to_ary() {
Gfs_nde[] rv = new Gfs_nde[subs_len];
for (int i = 0; i < subs_len; i++)
rv[i] = subs[i];
return rv;
}
public int Atrs_len() {return args_len;} private int args_len;
public Gfs_nde Atrs_get_at(int i) {return args[i];}
public Gfs_nde Atrs_add_many(Gfs_nde... ary) {
int len = ary.length;
for (int i = 0; i < len; i++)
Atrs_add(ary[i]);
return this;
}
public Gfs_nde Atrs_add(Gfs_nde nde) {
int new_len = args_len + 1;
if (new_len > args_max) { // ary too small >>> expand
args_max = new_len * 2;
Gfs_nde[] new_args = new Gfs_nde[args_max];
Array_.Copy_to(args, 0, new_args, 0, args_len);
args = new_args;
}
args[args_len] = nde;
args_len = new_len;
return this;
} Gfs_nde[] args = Gfs_nde.Ary_empty; int args_max; int[] args_pos_ary = Int_.Ary_empty;
public Gfs_nde[] Atrs_to_ary() {
Gfs_nde[] rv = new Gfs_nde[args_len];
for (int i = 0; i < args_len; i++)
rv[i] = args[i];
return rv;
}
public static final Gfs_nde[] Ary_empty = new Gfs_nde[0];
public static final byte Op_tid_null = 0, Op_tid_assign = 1;
}

105
400_xowa/src/gplx/langs/gfs/Gfs_parser.java Normal file
View File

@@ -0,0 +1,105 @@
/*
XOWA: the XOWA Offline Wiki Application
Copyright (C) 2012 gnosygnu@gmail.com
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx.langs.gfs; import gplx.*; import gplx.langs.*;
import gplx.core.btries.*;
public class Gfs_parser {
private final Btrie_fast_mgr trie = Gfs_parser_.trie_();
private final Gfs_parser_ctx ctx = new Gfs_parser_ctx();
public Gfs_nde Parse(byte[] src) {
ctx.Root().Subs_clear();
int src_len = src.length; if (src_len == 0) return ctx.Root();
ctx.Init(trie, src, src_len);
int pos = 0;
while (pos < src_len) {
byte b = src[pos];
Object o = trie.Match_at_w_b0(ctx.Trie_rv(), b, src, pos, src_len);
if (o == null)
ctx.Err_mgr().Fail_unknown_char(ctx, pos, b);
else {
Gfs_lxr lxr = (Gfs_lxr)o;
while (lxr != null) {
int rslt = lxr.Process(ctx, pos, ctx.Trie_rv().Pos());
switch (lxr.Lxr_tid()) {
case Gfs_lxr_.Tid_whitespace: break;
case Gfs_lxr_.Tid_comment: break;
default: ctx.Prv_lxr_(lxr.Lxr_tid()); break;
}
switch (rslt) {
case Gfs_lxr_.Rv_lxr:
pos = ctx.Nxt_pos();
lxr = ctx.Nxt_lxr();
break;
case Gfs_lxr_.Rv_eos:
pos = src_len;
lxr = null;
break;
default:
pos = rslt;
lxr = null;
break;
}
}
}
}
switch (ctx.Prv_lxr()) {
case Gfs_lxr_.Tid_curly_end:
case Gfs_lxr_.Tid_semic: break;
default: ctx.Err_mgr().Fail_eos(ctx); break;
}
return ctx.Root();
}
}
class Gfs_parser_ {
public static Btrie_fast_mgr trie_() {
Btrie_fast_mgr rv = Btrie_fast_mgr.ci_a7(); // NOTE:ci.ascii:gfs;letters/symbols only;
Gfs_lxr_identifier word_lxr = Gfs_lxr_identifier.Instance;
trie_add_rng(rv, word_lxr, Byte_ascii.Ltr_a, Byte_ascii.Ltr_z);
trie_add_rng(rv, word_lxr, Byte_ascii.Ltr_A, Byte_ascii.Ltr_Z);
trie_add_rng(rv, word_lxr, Byte_ascii.Num_0, Byte_ascii.Num_9);
rv.Add(Byte_ascii.Underline, word_lxr);
trie_add_many(rv, Gfs_lxr_whitespace.Instance, Byte_ascii.Space, Byte_ascii.Nl, Byte_ascii.Cr, Byte_ascii.Tab);
trie_add_quote(rv, new byte[] {Byte_ascii.Apos});
trie_add_quote(rv, new byte[] {Byte_ascii.Quote});
trie_add_quote(rv, Bry_.new_a7("<:[\"\n"), Bry_.new_a7("\n\"]:>"));
trie_add_quote(rv, Bry_.new_a7("<:['\n"), Bry_.new_a7("\n']:>"));
trie_add_quote(rv, Bry_.new_a7("<:{'"), Bry_.new_a7("'}:>"));
trie_add_comment(rv, new byte[] {Byte_ascii.Slash, Byte_ascii.Slash}, new byte[] {Byte_ascii.Nl});
trie_add_comment(rv, new byte[] {Byte_ascii.Slash, Byte_ascii.Star}, new byte[] {Byte_ascii.Star, Byte_ascii.Slash});
rv.Add(Byte_ascii.Semic, Gfs_lxr_semic.Instance);
rv.Add(Byte_ascii.Paren_bgn, Gfs_lxr_paren_bgn.Instance);
rv.Add(Byte_ascii.Paren_end, Gfs_lxr_paren_end.Instance);
rv.Add(Byte_ascii.Curly_bgn, Gfs_lxr_curly_bgn.Instance);
rv.Add(Byte_ascii.Curly_end, Gfs_lxr_curly_end.Instance);
rv.Add(Byte_ascii.Dot, Gfs_lxr_dot.Instance);
rv.Add(Byte_ascii.Comma, Gfs_lxr_comma.Instance);
rv.Add(Byte_ascii.Eq, Gfs_lxr_equal.Instance);
return rv;
}
private static void trie_add_rng(Btrie_fast_mgr trie, Gfs_lxr lxr, byte bgn, byte end) {
for (byte b = bgn; b <= end; b++)
trie.Add(b, lxr);
}
private static void trie_add_many(Btrie_fast_mgr trie, Gfs_lxr lxr, byte... ary) {
int len = ary.length;
for (int i = 0; i < len; i++)
trie.Add(ary[i], lxr);
}
private static void trie_add_quote(Btrie_fast_mgr trie, byte[] bgn) {trie_add_quote(trie, bgn, bgn);}
private static void trie_add_quote(Btrie_fast_mgr trie, byte[] bgn, byte[] end) {trie.Add(bgn, new Gfs_lxr_quote(bgn, end));}
private static void trie_add_comment(Btrie_fast_mgr trie, byte[] bgn, byte[] end) {trie.Add(bgn, new Gfs_lxr_comment_flat(bgn, end));}
}

127
400_xowa/src/gplx/langs/gfs/Gfs_parser_ctx.java Normal file
View File

@@ -0,0 +1,127 @@
/*
XOWA: the XOWA Offline Wiki Application
Copyright (C) 2012 gnosygnu@gmail.com
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx.langs.gfs; import gplx.*; import gplx.langs.*;
import gplx.core.btries.*;
class Gfs_parser_ctx {
public Btrie_fast_mgr Trie() {return trie;} Btrie_fast_mgr trie;
public Btrie_rv Trie_rv() {return trie_rv;} private final Btrie_rv trie_rv = new Btrie_rv();
public Gfs_nde Root() {return root;} Gfs_nde root = new Gfs_nde();
public byte[] Src() {return src;} private byte[] src;
public int Src_len() {return src_len;} private int src_len;
public int Prv_lxr() {return prv_lxr;} public Gfs_parser_ctx Prv_lxr_(int v) {prv_lxr = v; return this;} private int prv_lxr;
public Gfs_nde Cur_nde() {return cur_nde;} Gfs_nde cur_nde;
public int Nxt_pos() {return nxt_pos;} private int nxt_pos;
public Gfs_lxr Nxt_lxr() {return nxt_lxr;} Gfs_lxr nxt_lxr;
public Bry_bfr Tmp_bfr() {return tmp_bfr;} private Bry_bfr tmp_bfr = Bry_bfr_.New();
public void Process_eos() {}
public void Process_lxr(int nxt_pos, Gfs_lxr nxt_lxr) {this.nxt_pos = nxt_pos; this.nxt_lxr = nxt_lxr;}
public void Process_null(int cur_pos) {this.nxt_pos = cur_pos; this.nxt_lxr = null;}
public void Init(Btrie_fast_mgr trie, byte[] src, int src_len) {
this.trie = trie; this.src = src; this.src_len = src_len;
cur_nde = root;
Stack_add();
}
public void Hold_word(int bgn, int end) {
cur_idf_bgn = bgn;
cur_idf_end = end;
} int cur_idf_bgn = -1, cur_idf_end = -1;
private void Held_word_clear() {cur_idf_bgn = -1; cur_idf_end = -1;}
public Gfs_nde Make_nde(int tkn_bgn, int tkn_end) { // "abc."; "abc("; "abc;"; "abc{"
Gfs_nde nde = new Gfs_nde().Name_rng_(cur_idf_bgn, cur_idf_end);
this.Held_word_clear();
cur_nde.Subs_add(nde);
cur_nde = nde;
return nde;
}
public void Make_atr_by_idf() {Make_atr(cur_idf_bgn, cur_idf_end); Held_word_clear();}
public void Make_atr_by_bry(int bgn, int end, byte[] bry) {Make_atr(bgn, end).Name_(bry);}
public Gfs_nde Make_atr(int bgn, int end) {
Gfs_nde nde = new Gfs_nde().Name_rng_(bgn, end);
cur_nde.Atrs_add(nde);
return nde;
}
public void Cur_nde_from_stack() {cur_nde = (Gfs_nde)nodes.Get_at_last();}
public void Stack_add() {nodes.Add(cur_nde);} List_adp nodes = List_adp_.New();
public void Stack_pop(int pos) {
if (nodes.Count() < 2) err_mgr.Fail_nde_stack_empty(this, pos); // NOTE: need at least 2 items; 1 to pop and 1 to set as current
List_adp_.Del_at_last(nodes);
Cur_nde_from_stack();
}
public Gfs_err_mgr Err_mgr() {return err_mgr;} Gfs_err_mgr err_mgr = new Gfs_err_mgr();
}
class Gfs_err_mgr {
public void Fail_eos(Gfs_parser_ctx ctx) {Fail(ctx, Fail_msg_eos, ctx.Src_len());}
public void Fail_unknown_char(Gfs_parser_ctx ctx, int pos, byte c) {Fail(ctx, Fail_msg_unknown_char, pos, Keyval_.new_("char", Char_.To_str((char)c)));}
public void Fail_nde_stack_empty(Gfs_parser_ctx ctx, int pos) {Fail(ctx, Fail_msg_nde_stack_empty, pos);}
public void Fail_invalid_lxr(Gfs_parser_ctx ctx, int pos, int lxr_tid, byte c) {
Fail(ctx, Fail_msg_invalid_lxr, pos, Keyval_.new_("char", Char_.To_str((char)c)), Keyval_.new_("cur_lxr", Gfs_lxr_.Tid__name(lxr_tid)), Keyval_.new_("prv_lxr", Gfs_lxr_.Tid__name(ctx.Prv_lxr())));
}
private void Fail(Gfs_parser_ctx ctx, String msg, int pos, Keyval... args) {
byte[] src = ctx.Src(); int src_len = ctx.Src_len();
Fail_args_standard(src, src_len, pos);
int len = args.length;
for (int i = 0; i < len; i++) {
Keyval arg = args[i];
tmp_fail_args.Add(arg.Key(), arg.Val_to_str_or_empty());
}
throw Err_.new_wo_type(Fail_msg(msg, tmp_fail_args));
}
private void Fail_args_standard(byte[] src, int src_len, int pos) {
tmp_fail_args.Add("excerpt_bgn", Fail_excerpt_bgn(src, src_len, pos));
tmp_fail_args.Add("excerpt_end", Fail_excerpt_end(src, src_len, pos));
tmp_fail_args.Add("pos" , pos);
}
public static final String Fail_msg_invalid_lxr = "invalid character", Fail_msg_unknown_char = "unknown char", Fail_msg_eos = "end of stream", Fail_msg_nde_stack_empty = "node stack empty";
String Fail_msg(String type, Keyval_list fail_args) {
tmp_fail_bfr.Add_str_u8(type).Add_byte(Byte_ascii.Colon);
int len = fail_args.Count();
for (int i = 0; i < len; i++) {
tmp_fail_bfr.Add_byte(Byte_ascii.Space);
Keyval kv = fail_args.Get_at(i);
tmp_fail_bfr.Add_str_u8(kv.Key());
tmp_fail_bfr.Add_byte(Byte_ascii.Eq).Add_byte(Byte_ascii.Apos);
tmp_fail_bfr.Add_str_u8(kv.Val_to_str_or_empty()).Add_byte(Byte_ascii.Apos);
}
return tmp_fail_bfr.To_str_and_clear();
}
Bry_bfr tmp_fail_bfr = Bry_bfr_.Reset(255);
Keyval_list tmp_fail_args = new Keyval_list();
private static int excerpt_len = 50;
String Fail_excerpt_bgn(byte[] src, int src_len, int pos) {
int bgn = pos - excerpt_len; if (bgn < 0) bgn = 0;
Fail_excerpt_rng(tmp_fail_bfr, src, bgn, pos);
return tmp_fail_bfr.To_str_and_clear();
}
String Fail_excerpt_end(byte[] src, int src_len, int pos) {
int end = pos + excerpt_len; if (end > src_len) end = src_len;
Fail_excerpt_rng(tmp_fail_bfr, src, pos, end);
return tmp_fail_bfr.To_str_and_clear();
}
private static void Fail_excerpt_rng(Bry_bfr bfr, byte[] src, int bgn, int end) {
for (int i = bgn; i < end; i++) {
byte b = src[i];
switch (b) {
case Byte_ascii.Tab: bfr.Add(Esc_tab); break;
case Byte_ascii.Nl: bfr.Add(Esc_nl); break;
case Byte_ascii.Cr: bfr.Add(Esc_cr); break;
default: bfr.Add_byte(b); break;
}
}
}
private static final byte[] Esc_nl = Bry_.new_a7("\\n"), Esc_cr = Bry_.new_a7("\\r"), Esc_tab = Bry_.new_a7("\\t");
}

196
400_xowa/src/gplx/langs/gfs/Gfs_parser_tst.java Normal file
View File

@@ -0,0 +1,196 @@
/*
XOWA: the XOWA Offline Wiki Application
Copyright (C) 2012 gnosygnu@gmail.com
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx.langs.gfs; import gplx.*; import gplx.langs.*;
import org.junit.*;
public class Gfs_parser_tst {
@Before public void init() {fxt.Clear();} Gfs_parser_fxt fxt = new Gfs_parser_fxt();
@Test public void Semicolon() {
fxt .Test_parse("a;", fxt.nde_("a"));
fxt .Test_parse("a;b;c;", fxt.nde_("a"), fxt.nde_("b"), fxt.nde_("c"));
fxt .Test_parse("a_0;", fxt.nde_("a_0"));
}
@Test public void Dot() {
fxt .Test_parse("a.b;", fxt.nde_("a").Subs_add(fxt.nde_("b")));
fxt .Test_parse("a.b;c.d;", fxt.nde_("a").Subs_add(fxt.nde_("b")), fxt.nde_("c").Subs_add(fxt.nde_("d")));
}
@Test public void Parens() {
fxt .Test_parse("a();b();", fxt.nde_("a"), fxt.nde_("b"));
fxt .Test_parse("a().b();c().d();", fxt.nde_("a").Subs_add(fxt.nde_("b")), fxt.nde_("c").Subs_add(fxt.nde_("d")));
}
@Test public void Num() {
fxt .Test_parse("a(1,2);", fxt.nde_("a").Atrs_add_many(fxt.val_("1"), fxt.val_("2")));
}
@Test public void Quote() {
fxt .Test_parse("a('b');", fxt.nde_("a").Atrs_add(fxt.val_("b")));
}
@Test public void Quote_escaped() {
fxt .Test_parse("a('b''c''d');", fxt.nde_("a").Atrs_add(fxt.val_("b'c'd")));
}
@Test public void Quote_escaped_2() {
fxt .Test_parse("a('a''''b');", fxt.nde_("a").Atrs_add(fxt.val_("a''b")));
}
@Test public void Quote_mixed() {
fxt .Test_parse("a('b\"c');", fxt.nde_("a").Atrs_add(fxt.val_("b\"c")));
}
@Test public void Comma() {
fxt .Test_parse("a('b','c','d');", fxt.nde_("a").Atrs_add_many(fxt.val_("b"), fxt.val_("c"), fxt.val_("d")));
}
@Test public void Ws() {
fxt .Test_parse(" a ( 'b' , 'c' ) ; ", fxt.nde_("a").Atrs_add_many(fxt.val_("b"), fxt.val_("c")));
}
@Test public void Comment_slash_slash() {
fxt .Test_parse("//z\na;//y\n", fxt.nde_("a"));
}
@Test public void Comment_slash_star() {
fxt .Test_parse("/*z*/a;/*y*/", fxt.nde_("a"));
}
@Test public void Curly() {
fxt .Test_parse("a{b;}", fxt.nde_("a").Subs_add(fxt.nde_("b")));
}
@Test public void Curly_nest() {
fxt .Test_parse("a{b{c{d;}}}"
, fxt.nde_("a").Subs_add
( fxt.nde_("b").Subs_add
( fxt.nde_("c").Subs_add
( fxt.nde_("d")
))));
}
@Test public void Curly_nest_peers() {
fxt .Test_parse(String_.Concat_lines_nl
( "a{"
, " a0{"
, " a00{"
, " a000;"
, " }"
, " a01;"
, " }"
, " a1;"
, "}"
)
, fxt.nde_("a").Subs_add_many
( fxt.nde_("a0").Subs_add_many
( fxt.nde_("a00").Subs_add
( fxt.nde_("a000")
)
, fxt.nde_("a01")
)
, fxt.nde_("a1")
));
}
@Test public void Curly_dot() {
fxt .Test_parse("a{a0.a00;a1.a10;}"
, fxt.nde_("a").Subs_add_many
( fxt.nde_("a0").Subs_add_many(fxt.nde_("a00"))
, fxt.nde_("a1").Subs_add_many(fxt.nde_("a10"))
));
}
@Test public void Eq() {
fxt .Test_parse("a='b';", fxt.nde_("a").Atrs_add(fxt.val_("b")));
fxt .Test_parse("a.b.c='d';"
, fxt.nde_("a").Subs_add
( fxt.nde_("b").Subs_add_many
( fxt.nde_("c").Atrs_add(fxt.val_("d"))
)));
fxt .Test_parse("a.b{c='d'; e='f'}"
, fxt.nde_("a").Subs_add
( fxt.nde_("b").Subs_add_many
( fxt.nde_("c").Atrs_add(fxt.val_("d"))
, fxt.nde_("e").Atrs_add(fxt.val_("f"))
)));
}
@Test public void Curly_nest_peers2() {
fxt .Test_parse(String_.Concat_lines_nl
( "a() {"
, " k1 = 'v1';"
, "}"
)
, fxt.nde_("a").Subs_add_many
( fxt.nde_("k1").Atrs_add(fxt.val_("v1"))
)
);
}
@Test public void Fail() {
fxt .Test_parse_fail("a(.);", Gfs_err_mgr.Fail_msg_invalid_lxr); // (.)
fxt .Test_parse_fail("a..b;", Gfs_err_mgr.Fail_msg_invalid_lxr); // ..
fxt .Test_parse_fail("a.;", Gfs_err_mgr.Fail_msg_invalid_lxr); // .;
fxt .Test_parse_fail("a", Gfs_err_mgr.Fail_msg_eos); // eos
fxt .Test_parse_fail("a;~;", Gfs_err_mgr.Fail_msg_unknown_char); // ~
}
}
class Gfs_parser_fxt {
public void Clear() {}
public Gfs_nde nde_(String v) {return new Gfs_nde().Name_(Bry_.new_a7(v));}
public Gfs_nde val_(String v) {return new Gfs_nde().Name_(Bry_.new_a7(v));}
public void Test_parse(String src_str, Gfs_nde... expd) {
byte[] src_bry = Bry_.new_u8(src_str);
Gfs_nde root = parser.Parse(src_bry);
Tfds.Eq_str_lines(To_str(null, expd), To_str(src_bry, root.Subs_to_ary()));
} private Bry_bfr tmp_bfr = Bry_bfr_.New(), path_bfr = Bry_bfr_.New(); Gfs_parser parser = new Gfs_parser();
public void Test_parse_fail(String src_str, String expd_err) {
byte[] src_bry = Bry_.new_u8(src_str);
try {parser.Parse(src_bry);}
catch (Exception e) {
String actl_err = Err_.Message_gplx_full(e);
actl_err = String_.GetStrBefore(actl_err, ":");
boolean match = String_.Has(actl_err, expd_err);
if (!match) Tfds.Fail("expecting '" + expd_err + "' got '" + actl_err + "'");
return;
}
Tfds.Fail("expected to fail with " + expd_err);
}
public String To_str(byte[] src, Gfs_nde[] expd) {
int subs_len = expd.length;
for (int i = 0; i < subs_len; i++) {
path_bfr.Clear().Add_int_variable(i);
To_str(tmp_bfr, path_bfr, src, expd[i]);
}
return tmp_bfr.To_str_and_clear();
}
public void To_str(Bry_bfr bfr, Bry_bfr path, byte[] src, Gfs_nde nde) {
To_str_atr(bfr, path, src, Atr_name, nde.Name(), nde.Name_bgn(), nde.Name_end());
int atrs_len = nde.Atrs_len();
for (int i = 0; i < atrs_len; i++) {
Gfs_nde atr = nde.Atrs_get_at(i);
int path_len_old = path.Len();
path.Add_byte(Byte_ascii.Dot).Add_byte((byte)(Byte_ascii.Ltr_a + i));
int path_len_new = path.Len();
To_str(bfr, path, src, atr);
path.Del_by(path_len_new - path_len_old);
}
int subs_len = nde.Subs_len();
for (int i = 0; i < subs_len; i++) {
Gfs_nde sub = nde.Subs_get_at(i);
int path_len_old = path.Len();
path.Add_byte(Byte_ascii.Dot).Add_int_variable(i);
int path_len_new = path.Len();
To_str(bfr, path, src, sub);
path.Del_by(path_len_new - path_len_old);
}
}
private void To_str_atr(Bry_bfr bfr, Bry_bfr path_bfr, byte[] src, byte[] name, byte[] val, int val_bgn, int val_end) {
if (val == null && val_bgn == -1 && val_end == -1) return;
bfr.Add_bfr_and_preserve(path_bfr).Add_byte(Byte_ascii.Colon);
bfr.Add(name);
if (val == null)
bfr.Add_mid(src, val_bgn, val_end);
else
bfr.Add(val);
bfr.Add_byte_nl();
}
private static final byte[] Atr_name = Bry_.new_a7("name=");
}

46
400_xowa/src/gplx/langs/gfs/Gfs_wtr.java Normal file
View File

@@ -0,0 +1,46 @@
/*
XOWA: the XOWA Offline Wiki Application
Copyright (C) 2012 gnosygnu@gmail.com
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
package gplx.langs.gfs; import gplx.*; import gplx.langs.*;
public class Gfs_wtr {
public byte Quote_char() {return quote_char;} public Gfs_wtr Quote_char_(byte v) {quote_char = v; return this;} private byte quote_char = Byte_ascii.Apos;
public Bry_bfr Bfr() {return bfr;} private Bry_bfr bfr = Bry_bfr_.Reset(255);
public void Add_grp_bgn(byte[] key) {
bfr.Add(key); // key
bfr.Add_byte(Byte_ascii.Curly_bgn); // {
}
public void Add_grp_end(byte[] key) {
bfr.Add_byte(Byte_ascii.Curly_end); // }
}
public void Add_set_eq(byte[] key, byte[] val) {
bfr.Add(key); // key
bfr.Add_byte_eq(); // =
bfr.Add_byte(quote_char); // '
Write_val(val);
bfr.Add_byte(quote_char); // '
bfr.Add_byte(Byte_ascii.Semic); // ;
}
private void Write_val(byte[] bry) {
int bry_len = bry.length;
for (int i = 0; i < bry_len; i++) {
byte b = bry[i];
if (b == quote_char) // byte is quote
bfr.Add_byte(b); // double up
bfr.Add_byte(b);
}
}
}