mirror of
https://github.com/gnosygnu/xowa.git
synced 2025-06-13 12:54:14 +00:00
Scribunto: Allow int for find parameter [#802]
This commit is contained in:
gnosygnu
parent
59d8a42b22
commit
c801e3a20b
@ -1,6 +1,6 @@
|
|||||||
/*
|
/*
|
||||||
XOWA: the XOWA Offline Wiki Application
|
XOWA: the XOWA Offline Wiki Application
|
||||||
Copyright (C) 2012-2017 gnosygnu@gmail.com
|
Copyright (C) 2012-2020 gnosygnu@gmail.com
|
||||||
|
|
||||||
XOWA is licensed under the terms of the General Public License (GPL) Version 3,
|
XOWA is licensed under the terms of the General Public License (GPL) Version 3,
|
||||||
or alternatively under the terms of the Apache License Version 2.0.
|
or alternatively under the terms of the Apache License Version 2.0.
|
||||||
@ -13,202 +13,224 @@ The terms of each license can be found in the source code repository:
|
|||||||
GPLv3 License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-GPLv3.txt
|
GPLv3 License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-GPLv3.txt
|
||||||
Apache License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-APACHE2.txt
|
Apache License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-APACHE2.txt
|
||||||
*/
|
*/
|
||||||
package gplx.xowa.xtns.scribunto.libs; import gplx.*; import gplx.xowa.*; import gplx.xowa.xtns.*; import gplx.xowa.xtns.scribunto.*;
|
package gplx.xowa.xtns.scribunto.libs;
|
||||||
import gplx.objects.strings.unicodes.*;
|
|
||||||
import gplx.core.intls.*; import gplx.langs.regxs.*;
|
import gplx.Bool_;
|
||||||
import gplx.xowa.parsers.*;
|
import gplx.Err_;
|
||||||
import gplx.xowa.xtns.scribunto.procs.*;
|
import gplx.Gfo_usr_dlg_;
|
||||||
import gplx.xowa.xtns.scribunto.libs.patterns.*;
|
import gplx.Io_url;
|
||||||
public class Scrib_lib_ustring implements Scrib_lib {
|
import gplx.Keyval;
|
||||||
public Scrib_lib_ustring(Scrib_core core) {this.core = core;} private Scrib_core core;
|
import gplx.Keyval_;
|
||||||
public String Key() {return "mw.ustring";}
|
import gplx.List_adp;
|
||||||
public Scrib_lua_mod Mod() {return mod;} private Scrib_lua_mod mod;
|
import gplx.List_adp_;
|
||||||
public int String_len_max() {return string_len_max;} public Scrib_lib_ustring String_len_max_(int v) {string_len_max = v; return this;} private int string_len_max = Xoa_page_.Page_len_max;
|
import gplx.String_;
|
||||||
public int Pattern_len_max() {return pattern_len_max;} public Scrib_lib_ustring Pattern_len_max_(int v) {pattern_len_max = v; return this;} private int pattern_len_max = 10000;
|
import gplx.langs.regxs.Regx_adp;
|
||||||
public Scrib_lib Init() {procs.Init_by_lib(this, Proc_names); return this;}
|
import gplx.langs.regxs.Regx_adp_;
|
||||||
public Scrib_lib Clone_lib(Scrib_core core) {return new Scrib_lib_ustring(core);}
|
import gplx.langs.regxs.Regx_group;
|
||||||
public Scrib_lua_mod Register(Scrib_core core, Io_url script_dir) {
|
import gplx.langs.regxs.Regx_match;
|
||||||
Init();
|
import gplx.objects.strings.unicodes.Ustring;
|
||||||
mod = core.RegisterInterface(this, script_dir.GenSubFil("mw.ustring.lua")
|
import gplx.objects.strings.unicodes.Ustring_;
|
||||||
, Keyval_.new_("stringLengthLimit", string_len_max)
|
import gplx.xowa.Xoa_page_;
|
||||||
, Keyval_.new_("patternLengthLimit", pattern_len_max)
|
import gplx.xowa.xtns.scribunto.Scrib_core;
|
||||||
);
|
import gplx.xowa.xtns.scribunto.Scrib_kv_utl_;
|
||||||
return mod;
|
import gplx.xowa.xtns.scribunto.Scrib_lib;
|
||||||
}
|
import gplx.xowa.xtns.scribunto.Scrib_lua_mod;
|
||||||
public Scrib_proc_mgr Procs() {return procs;} private Scrib_proc_mgr procs = new Scrib_proc_mgr();
|
import gplx.xowa.xtns.scribunto.libs.patterns.Scrib_pattern_matcher;
|
||||||
public boolean Procs_exec(int key, Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
import gplx.xowa.xtns.scribunto.procs.Scrib_proc_args;
|
||||||
switch (key) {
|
import gplx.xowa.xtns.scribunto.procs.Scrib_proc_mgr;
|
||||||
case Proc_find: return Find(args, rslt);
|
import gplx.xowa.xtns.scribunto.procs.Scrib_proc_rslt;
|
||||||
case Proc_match: return Match(args, rslt);
|
|
||||||
case Proc_gmatch_init: return Gmatch_init(args, rslt);
|
public class Scrib_lib_ustring implements Scrib_lib {
|
||||||
case Proc_gmatch_callback: return Gmatch_callback(args, rslt);
|
public Scrib_lib_ustring(Scrib_core core) {this.core = core;} private Scrib_core core;
|
||||||
case Proc_gsub: return Gsub(args, rslt);
|
public String Key() {return "mw.ustring";}
|
||||||
default: throw Err_.new_unhandled(key);
|
public Scrib_lua_mod Mod() {return mod;} private Scrib_lua_mod mod;
|
||||||
}
|
public int String_len_max() {return string_len_max;} public Scrib_lib_ustring String_len_max_(int v) {string_len_max = v; return this;} private int string_len_max = Xoa_page_.Page_len_max;
|
||||||
}
|
public int Pattern_len_max() {return pattern_len_max;} public Scrib_lib_ustring Pattern_len_max_(int v) {pattern_len_max = v; return this;} private int pattern_len_max = 10000;
|
||||||
private static final int Proc_find = 0, Proc_match = 1, Proc_gmatch_init = 2, Proc_gmatch_callback = 3, Proc_gsub = 4;
|
public Scrib_lib Init() {procs.Init_by_lib(this, Proc_names); return this;}
|
||||||
public static final String Invk_find = "find", Invk_match = "match", Invk_gmatch_init = "gmatch_init", Invk_gmatch_callback = "gmatch_callback", Invk_gsub = "gsub";
|
public Scrib_lib Clone_lib(Scrib_core core) {return new Scrib_lib_ustring(core);}
|
||||||
private static final String[] Proc_names = String_.Ary(Invk_find, Invk_match, Invk_gmatch_init, Invk_gmatch_callback, Invk_gsub);
|
public Scrib_lua_mod Register(Scrib_core core, Io_url script_dir) {
|
||||||
public boolean Find(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
Init();
|
||||||
// get args
|
mod = core.RegisterInterface(this, script_dir.GenSubFil("mw.ustring.lua")
|
||||||
String text_str = args.Xstr_str_or_null(0);
|
, Keyval_.new_("stringLengthLimit", string_len_max)
|
||||||
String find_str = args.Pull_str(1);
|
, Keyval_.new_("patternLengthLimit", pattern_len_max)
|
||||||
int bgn_as_codes_base1 = args.Cast_int_or(2, 1);
|
);
|
||||||
boolean plain = args.Cast_bool_or_n(3);
|
return mod;
|
||||||
|
}
|
||||||
// init text vars
|
public Scrib_proc_mgr Procs() {return procs;} private Scrib_proc_mgr procs = new Scrib_proc_mgr();
|
||||||
Ustring text_ucs = Ustring_.New_codepoints(text_str); // NOTE: must count codes for supplementaries; PAGE:en.d:iglesia DATE:2017-04-23
|
public boolean Procs_exec(int key, Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
||||||
|
switch (key) {
|
||||||
// convert bgn from base_1 to base_0
|
case Proc_find: return Find(args, rslt);
|
||||||
int bgn_as_codes = To_java_by_lua(bgn_as_codes_base1, text_ucs.Len_in_data());
|
case Proc_match: return Match(args, rslt);
|
||||||
|
case Proc_gmatch_init: return Gmatch_init(args, rslt);
|
||||||
/*
|
case Proc_gmatch_callback: return Gmatch_callback(args, rslt);
|
||||||
int offset = 0;
|
case Proc_gsub: return Gsub(args, rslt);
|
||||||
if (bgn_as_codes > 0) { // NOTE: MW.BASE
|
default: throw Err_.new_unhandled(key);
|
||||||
// $offset = strlen( mb_substr( $s, 0, $init - 1, 'UTF-8' ) );
|
}
|
||||||
}
|
}
|
||||||
else {
|
private static final int Proc_find = 0, Proc_match = 1, Proc_gmatch_init = 2, Proc_gmatch_callback = 3, Proc_gsub = 4;
|
||||||
bgn_as_codes_base1 = 0; // NOTE: MW.BASE1
|
public static final String Invk_find = "find", Invk_match = "match", Invk_gmatch_init = "gmatch_init", Invk_gmatch_callback = "gmatch_callback", Invk_gsub = "gsub";
|
||||||
offset = 0; // -1?
|
private static final String[] Proc_names = String_.Ary(Invk_find, Invk_match, Invk_gmatch_init, Invk_gmatch_callback, Invk_gsub);
|
||||||
}
|
public boolean Find(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
||||||
*/
|
// get args
|
||||||
|
String text_str = args.Xstr_str_or_null(0);
|
||||||
// find_str of "" should return (bgn, bgn - 1) regardless of whether plain is true or false;
|
String find_str = args.Pull_str(1);
|
||||||
// NOTE: do not include surrogate calc; PAGE:en.d:佻 DATE:2017-04-24
|
int bgn_as_codes_base1 = args.Cast_int_or(2, 1);
|
||||||
// NOTE: not in MW; is this needed? DATE:2019-02-24
|
boolean plain = args.Cast_bool_or_n(3);
|
||||||
if (String_.Len_eq_0(find_str))
|
|
||||||
return rslt.Init_many_objs(bgn_as_codes_base1, bgn_as_codes_base1 - 1);
|
// init text vars
|
||||||
|
Ustring text_ucs = Ustring_.New_codepoints(text_str); // NOTE: must count codes for supplementaries; PAGE:en.d:iglesia DATE:2017-04-23
|
||||||
// if plain, just do literal match of find and exit
|
|
||||||
if (plain) {
|
// convert bgn from base_1 to base_0
|
||||||
// find pos by literal match
|
int bgn_as_codes = To_java_by_lua(bgn_as_codes_base1, text_ucs.Len_in_data());
|
||||||
Ustring find_ucs = Ustring_.New_codepoints(find_str);
|
|
||||||
int pos = text_ucs.Index_of(find_ucs, bgn_as_codes);
|
/*
|
||||||
|
int offset = 0;
|
||||||
// if nothing found, return empty
|
if (bgn_as_codes > 0) { // NOTE: MW.BASE
|
||||||
if (pos == String_.Find_none)
|
// $offset = strlen( mb_substr( $s, 0, $init - 1, 'UTF-8' ) );
|
||||||
return rslt.Init_ary_empty();
|
}
|
||||||
|
else {
|
||||||
// bgn: adjust for base1
|
bgn_as_codes_base1 = 0; // NOTE: MW.BASE1
|
||||||
int bgn = pos + Base1;
|
offset = 0; // -1?
|
||||||
|
}
|
||||||
// end: add find.Len_in_codes and adjust end for PHP/LUA
|
*/
|
||||||
int end = bgn + find_ucs.Len_in_data() - End_adj;
|
|
||||||
|
// find_str of "" should return (bgn, bgn - 1) regardless of whether plain is true or false;
|
||||||
return rslt.Init_many_objs(bgn, end);
|
// NOTE: do not include surrogate calc; PAGE:en.d:佻 DATE:2017-04-24
|
||||||
}
|
// NOTE: not in MW; is this needed? DATE:2019-02-24
|
||||||
|
if (String_.Len_eq_0(find_str))
|
||||||
// run regex; NOTE: take only 1st result; DATE:2014-08-27
|
return rslt.Init_many_objs(bgn_as_codes_base1, bgn_as_codes_base1 - 1);
|
||||||
Scrib_pattern_matcher matcher = Scrib_pattern_matcher.New(core.Page_url());
|
|
||||||
Regx_match match = matcher.Match_one(text_ucs, find_str, bgn_as_codes, true);
|
// if plain, just do literal match of find and exit
|
||||||
if (match.Rslt_none()) return rslt.Init_null(); // null verified on MW; EX: =mw.ustring.find("abc", "z"); DATE:2019-04-11
|
if (plain) {
|
||||||
|
// find pos by literal match
|
||||||
// add to tmp_list
|
Ustring find_ucs = Ustring_.New_codepoints(find_str);
|
||||||
List_adp tmp_list = List_adp_.New();
|
int pos = text_ucs.Index_of(find_ucs, bgn_as_codes);
|
||||||
tmp_list.Add(text_ucs.Map_char_to_data(match.Find_bgn()) + Scrib_lib_ustring.Base1);
|
|
||||||
tmp_list.Add(text_ucs.Map_char_to_data(match.Find_end()) + Scrib_lib_ustring.Base1 - Scrib_lib_ustring.End_adj);
|
// if nothing found, return empty
|
||||||
AddCapturesFromMatch(tmp_list, match, text_str, matcher.Capt_ary(), false);
|
if (pos == String_.Find_none)
|
||||||
return rslt.Init_many_list(tmp_list);
|
return rslt.Init_ary_empty();
|
||||||
}
|
|
||||||
public boolean Match(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
// bgn: adjust for base1
|
||||||
// get args
|
int bgn = pos + Base1;
|
||||||
String text_str = args.Xstr_str_or_null(0); // Module can pass raw ints; PAGE:en.w:Budget_of_the_European_Union; DATE:2015-01-22
|
|
||||||
String find_str = args.Cast_str_or_null(1);
|
// end: add find.Len_in_codes and adjust end for PHP/LUA
|
||||||
int bgn_as_codes_base1 = args.Cast_int_or(2, 1);
|
int end = bgn + find_ucs.Len_in_data() - End_adj;
|
||||||
|
|
||||||
// validate / adjust
|
return rslt.Init_many_objs(bgn, end);
|
||||||
if (text_str == null) // if no text_str is passed, do not fail; return empty; EX:d:changed; DATE:2014-02-06
|
}
|
||||||
return rslt.Init_many_list(List_adp_.Noop);
|
|
||||||
Ustring text_ucs = Ustring_.New_codepoints(text_str); // NOTE: must count codes for supplementaries; PAGE:en.d:iglesia DATE:2017-04-23
|
// run regex; NOTE: take only 1st result; DATE:2014-08-27
|
||||||
int bgn_as_codes = To_java_by_lua(bgn_as_codes_base1, text_ucs.Len_in_data());
|
Scrib_pattern_matcher matcher = Scrib_pattern_matcher.New(core.Page_url());
|
||||||
|
Regx_match match = matcher.Match_one(text_ucs, find_str, bgn_as_codes, true);
|
||||||
// run regex; NOTE add 1st match only; do not add all; PAGE:en.d:действительное_причастие_настоящего_времени DATE:2017-04-23
|
if (match.Rslt_none()) return rslt.Init_null(); // null verified on MW; EX: =mw.ustring.find("abc", "z"); DATE:2019-04-11
|
||||||
Scrib_pattern_matcher matcher = Scrib_pattern_matcher.New(core.Page_url());
|
|
||||||
Regx_match match = matcher.Match_one(text_ucs, find_str, bgn_as_codes, true);
|
// add to tmp_list
|
||||||
if (match.Rslt_none()) return rslt.Init_null(); // return null if no matches found; EX:w:Mount_Gambier_(volcano); DATE:2014-04-02; confirmed with en.d:民; DATE:2015-01-30
|
List_adp tmp_list = List_adp_.New();
|
||||||
|
tmp_list.Add(text_ucs.Map_char_to_data(match.Find_bgn()) + Scrib_lib_ustring.Base1);
|
||||||
List_adp tmp_list = List_adp_.New();
|
tmp_list.Add(text_ucs.Map_char_to_data(match.Find_end()) + Scrib_lib_ustring.Base1 - Scrib_lib_ustring.End_adj);
|
||||||
AddCapturesFromMatch(tmp_list, match, text_str, matcher.Capt_ary(), true);
|
AddCapturesFromMatch(tmp_list, match, text_str, matcher.Capt_ary(), false);
|
||||||
return rslt.Init_many_list(tmp_list);
|
return rslt.Init_many_list(tmp_list);
|
||||||
}
|
}
|
||||||
public boolean Gsub(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
public boolean Match(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
||||||
Scrib_lib_ustring_gsub_mgr gsub_mgr = new Scrib_lib_ustring_gsub_mgr(core);
|
// get args
|
||||||
return gsub_mgr.Exec(args, rslt);
|
String text_str = args.Xstr_str_or_null(0); // Module can pass raw ints; PAGE:en.w:Budget_of_the_European_Union; DATE:2015-01-22
|
||||||
}
|
// 2019-20-01|ISSUE#:802|passing integer should return NULL, not throw error
|
||||||
public boolean Gmatch_init(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
String find_str = args.Xstr_str_or_null(1);
|
||||||
// String text = Scrib_kv_utl_.Val_to_str(values, 0);
|
int bgn_as_codes_base1 = args.Cast_int_or(2, 1);
|
||||||
String regx = args.Pull_str(1);
|
|
||||||
Scrib_regx_converter regx_converter = new Scrib_regx_converter();
|
// validate / adjust
|
||||||
if (Scrib_pattern_matcher.Mode_is_xowa())
|
if (text_str == null) // if no text_str is passed, do not fail; return empty; EX:d:changed; DATE:2014-02-06
|
||||||
regx_converter.patternToRegex(regx, Scrib_regx_converter.Anchor_null, true);
|
return rslt.Init_many_list(List_adp_.Noop);
|
||||||
else
|
Ustring text_ucs = Ustring_.New_codepoints(text_str); // NOTE: must count codes for supplementaries; PAGE:en.d:iglesia DATE:2017-04-23
|
||||||
regx = regx_converter.patternToRegex(regx, Scrib_regx_converter.Anchor_null, true);
|
int bgn_as_codes = To_java_by_lua(bgn_as_codes_base1, text_ucs.Len_in_data());
|
||||||
return rslt.Init_many_objs(regx, regx_converter.Capt_ary());
|
|
||||||
}
|
// run regex; NOTE add 1st match only; do not add all; PAGE:en.d:действительное_причастие_настоящего_времени DATE:2017-04-23
|
||||||
public boolean Gmatch_callback(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
Scrib_pattern_matcher matcher = Scrib_pattern_matcher.New(core.Page_url());
|
||||||
String text = args.Xstr_str_or_null(0); // NOTE: UstringLibrary.php!ustringGmatchCallback calls preg_match directly; $s can be any type, and php casts automatically;
|
Regx_match match = matcher.Match_one(text_ucs, find_str, bgn_as_codes, true);
|
||||||
String regx = args.Pull_str(1);
|
if (match.Rslt_none()) return rslt.Init_null(); // return null if no matches found; EX:w:Mount_Gambier_(volcano); DATE:2014-04-02; confirmed with en.d:民; DATE:2015-01-30
|
||||||
Keyval[] capt = args.Cast_kv_ary_or_null(2);
|
|
||||||
int pos = args.Pull_int(3);
|
List_adp tmp_list = List_adp_.New();
|
||||||
|
AddCapturesFromMatch(tmp_list, match, text_str, matcher.Capt_ary(), true);
|
||||||
Ustring text_ucs = Ustring_.New_codepoints(text);
|
return rslt.Init_many_list(tmp_list);
|
||||||
// int pos_as_codes = To_java_by_lua(pos, text_ucs.Len_in_data());
|
}
|
||||||
Regx_match match = Scrib_pattern_matcher.New(core.Page_url()).Match_one(text_ucs, regx, pos, false);
|
public boolean Gsub(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
||||||
if (match.Rslt_none()) return rslt.Init_many_objs(pos, Keyval_.Ary_empty);
|
Scrib_lib_ustring_gsub_mgr gsub_mgr = new Scrib_lib_ustring_gsub_mgr(core);
|
||||||
List_adp tmp_list = List_adp_.New();
|
return gsub_mgr.Exec(args, rslt);
|
||||||
AddCapturesFromMatch(tmp_list, match, text, capt, true); // NOTE: was incorrectly set as false; DATE:2014-04-23
|
}
|
||||||
return rslt.Init_many_objs(match.Find_end(), Scrib_kv_utl_.base1_list_(tmp_list));
|
public boolean Gmatch_init(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
||||||
}
|
// String text = Scrib_kv_utl_.Val_to_str(values, 0);
|
||||||
private int To_java_by_lua(int bgn_as_codes_base1, int len_in_codes) {
|
String regx = args.Pull_str(1);
|
||||||
// convert bgn from base_1 to base_0
|
Scrib_regx_converter regx_converter = new Scrib_regx_converter();
|
||||||
int bgn_as_codes = bgn_as_codes_base1;
|
if (Scrib_pattern_matcher.Mode_is_xowa())
|
||||||
if (bgn_as_codes > 0)
|
regx_converter.patternToRegex(regx, Scrib_regx_converter.Anchor_null, true);
|
||||||
bgn_as_codes -= Scrib_lib_ustring.Base1;
|
else
|
||||||
// TOMBSTONE: do not adjust negative numbers for base1; fails tests
|
regx = regx_converter.patternToRegex(regx, Scrib_regx_converter.Anchor_null, true);
|
||||||
// else if (bgn_as_codes < 0) bgn_as_codes += Scrib_lib_ustring.Base1;
|
return rslt.Init_many_objs(regx, regx_converter.Capt_ary());
|
||||||
|
}
|
||||||
// adjust bgn for negative-numbers and large positive-numbers
|
public boolean Gmatch_callback(Scrib_proc_args args, Scrib_proc_rslt rslt) {
|
||||||
// NOTE: MW uses mb_strlen which returns len of mb chars as 1; REF.PHP: http://php.net/manual/en/function.mb-strlen.php
|
String text = args.Xstr_str_or_null(0); // NOTE: UstringLibrary.php!ustringGmatchCallback calls preg_match directly; $s can be any type, and php casts automatically;
|
||||||
// NOTE: MW does additional +1 for PHP.base_1. This is not needed for JAVA; noted below as IGNORE_BASE_1_ADJ
|
String regx = args.Pull_str(1);
|
||||||
if (bgn_as_codes < 0) // negative number means search from rear of String
|
Keyval[] capt = args.Cast_kv_ary_or_null(2);
|
||||||
bgn_as_codes += len_in_codes; // NOTE:IGNORE_BASE_1_ADJ
|
int pos = args.Pull_int(3);
|
||||||
else if (bgn_as_codes > len_in_codes) // bgn_as_codes > text_len; confine to text_len; NOTE:IGNORE_BASE_1_ADJ
|
|
||||||
bgn_as_codes = len_in_codes; // NOTE:IGNORE_BASE_1_ADJ
|
Ustring text_ucs = Ustring_.New_codepoints(text);
|
||||||
|
// int pos_as_codes = To_java_by_lua(pos, text_ucs.Len_in_data());
|
||||||
// will be negative if Abs(bgn_as_codes) > text.length; ISSUE#:366; DATE:2019-02-23
|
Regx_match match = Scrib_pattern_matcher.New(core.Page_url()).Match_one(text_ucs, regx, pos, false);
|
||||||
if (bgn_as_codes < 0)
|
if (match.Rslt_none()) return rslt.Init_many_objs(pos, Keyval_.Ary_empty);
|
||||||
bgn_as_codes = 0;
|
List_adp tmp_list = List_adp_.New();
|
||||||
return bgn_as_codes;
|
AddCapturesFromMatch(tmp_list, match, text, capt, true); // NOTE: was incorrectly set as false; DATE:2014-04-23
|
||||||
}
|
return rslt.Init_many_objs(match.Find_end(), Scrib_kv_utl_.base1_list_(tmp_list));
|
||||||
private void AddCapturesFromMatch(List_adp tmp_list, Regx_match rslt, String text, Keyval[] capts, boolean op_is_match) {// NOTE: this matches behavior in UstringLibrary.php!addCapturesFromMatch
|
}
|
||||||
int capts_len = capts == null ? 0 : capts.length;
|
private int To_java_by_lua(int bgn_as_codes_base1, int len_in_codes) {
|
||||||
if (capts_len > 0) { // NOTE: changed from "grps_len > 0"; PAGE:en.w:Portal:Constructed_languages/Intro DATE:2018-07-02
|
// convert bgn from base_1 to base_0
|
||||||
Regx_group[] grps = rslt.Groups();
|
int bgn_as_codes = bgn_as_codes_base1;
|
||||||
int grps_len = grps.length;
|
if (bgn_as_codes > 0)
|
||||||
for (int j = 0; j < grps_len; j++) {
|
bgn_as_codes -= Scrib_lib_ustring.Base1;
|
||||||
Regx_group grp = grps[j];
|
// TOMBSTONE: do not adjust negative numbers for base1; fails tests
|
||||||
if ( j < capts_len // bounds check b/c null can be passed
|
// else if (bgn_as_codes < 0) bgn_as_codes += Scrib_lib_ustring.Base1;
|
||||||
&& Bool_.Cast(capts[j].Val()) // check if true; indicates that group is "()" or "anypos" see regex converter; DATE:2014-04-23
|
|
||||||
)
|
// adjust bgn for negative-numbers and large positive-numbers
|
||||||
tmp_list.Add(grp.Bgn() + Scrib_lib_ustring.Base1); // return index only for "()"; NOTE: do not return as String; callers expect int and will fail typed comparisons; DATE:2016-01-21
|
// NOTE: MW uses mb_strlen which returns len of mb chars as 1; REF.PHP: http://php.net/manual/en/function.mb-strlen.php
|
||||||
else
|
// NOTE: MW does additional +1 for PHP.base_1. This is not needed for JAVA; noted below as IGNORE_BASE_1_ADJ
|
||||||
tmp_list.Add(grp.Val()); // return match
|
if (bgn_as_codes < 0) // negative number means search from rear of String
|
||||||
}
|
bgn_as_codes += len_in_codes; // NOTE:IGNORE_BASE_1_ADJ
|
||||||
}
|
else if (bgn_as_codes > len_in_codes) // bgn_as_codes > text_len; confine to text_len; NOTE:IGNORE_BASE_1_ADJ
|
||||||
else if ( op_is_match // if op_is_match, and no captures, extract find_txt; note that UstringLibrary.php says "$arr[] = $m[0][0];" which means get the 1st match;
|
bgn_as_codes = len_in_codes; // NOTE:IGNORE_BASE_1_ADJ
|
||||||
&& tmp_list.Count() == 0) // only add match once; EX: "aaaa", "a" will have four matches; get 1st; DATE:2014-04-02
|
|
||||||
tmp_list.Add(String_.Mid(text, rslt.Find_bgn(), rslt.Find_end()));
|
// will be negative if Abs(bgn_as_codes) > text.length; ISSUE#:366; DATE:2019-02-23
|
||||||
}
|
if (bgn_as_codes < 0)
|
||||||
public static Regx_adp RegxAdp_new_(byte[] page_url, String regx) {
|
bgn_as_codes = 0;
|
||||||
Regx_adp rv = Regx_adp_.new_(regx);
|
return bgn_as_codes;
|
||||||
if (rv.Pattern_is_invalid()) {
|
}
|
||||||
// try to identify [z-a] errors; PAGE:https://en.wiktionary.org/wiki/Module:scripts/data; DATE:2017-04-23
|
private void AddCapturesFromMatch(List_adp tmp_list, Regx_match rslt, String text, Keyval[] capts, boolean op_is_match) {// NOTE: this matches behavior in UstringLibrary.php!addCapturesFromMatch
|
||||||
Exception exc = rv.Pattern_is_invalid_exception();
|
int capts_len = capts == null ? 0 : capts.length;
|
||||||
Gfo_usr_dlg_.Instance.Log_many("", "", "regx is invalid: regx=~{0} page=~{1} exc=~{2}", regx, page_url, Err_.Message_gplx_log(exc));
|
if (capts_len > 0) { // NOTE: changed from "grps_len > 0"; PAGE:en.w:Portal:Constructed_languages/Intro DATE:2018-07-02
|
||||||
}
|
Regx_group[] grps = rslt.Groups();
|
||||||
return rv;
|
int grps_len = grps.length;
|
||||||
}
|
for (int j = 0; j < grps_len; j++) {
|
||||||
private static final int
|
Regx_group grp = grps[j];
|
||||||
Base1 = 1
|
if ( j < capts_len // bounds check b/c null can be passed
|
||||||
, End_adj = 1; // lua / php uses "end" as <= not <; EX: "abc" and bgn=0, end= 1; for XOWA, this is "a"; for MW / PHP it is "ab"
|
&& Bool_.Cast(capts[j].Val()) // check if true; indicates that group is "()" or "anypos" see regex converter; DATE:2014-04-23
|
||||||
}
|
)
|
||||||
|
tmp_list.Add(grp.Bgn() + Scrib_lib_ustring.Base1); // return index only for "()"; NOTE: do not return as String; callers expect int and will fail typed comparisons; DATE:2016-01-21
|
||||||
|
else
|
||||||
|
tmp_list.Add(grp.Val()); // return match
|
||||||
|
}
|
||||||
|
}
|
||||||
|
else if ( op_is_match // if op_is_match, and no captures, extract find_txt; note that UstringLibrary.php says "$arr[] = $m[0][0];" which means get the 1st match;
|
||||||
|
&& tmp_list.Count() == 0) // only add match once; EX: "aaaa", "a" will have four matches; get 1st; DATE:2014-04-02
|
||||||
|
tmp_list.Add(String_.Mid(text, rslt.Find_bgn(), rslt.Find_end()));
|
||||||
|
}
|
||||||
|
public static Regx_adp RegxAdp_new_(byte[] page_url, String regx) {
|
||||||
|
Regx_adp rv = Regx_adp_.new_(regx);
|
||||||
|
if (rv.Pattern_is_invalid()) {
|
||||||
|
// try to identify [z-a] errors; PAGE:https://en.wiktionary.org/wiki/Module:scripts/data; DATE:2017-04-23
|
||||||
|
Exception exc = rv.Pattern_is_invalid_exception();
|
||||||
|
Gfo_usr_dlg_.Instance.Log_many("", "", "regx is invalid: regx=~{0} page=~{1} exc=~{2}", regx, page_url, Err_.Message_gplx_log(exc));
|
||||||
|
}
|
||||||
|
return rv;
|
||||||
|
}
|
||||||
|
private static final int
|
||||||
|
Base1 = 1
|
||||||
|
, End_adj = 1; // lua / php uses "end" as <= not <; EX: "abc" and bgn=0, end= 1; for XOWA, this is "a"; for MW / PHP it is "ab"
|
||||||
|
}
|
||||||
|
|||||||
@ -1,27 +1,35 @@
|
|||||||
/*
|
/*
|
||||||
XOWA: the XOWA Offline Wiki Application
|
XOWA: the XOWA Offline Wiki Application
|
||||||
Copyright (C) 2012-2017 gnosygnu@gmail.com
|
Copyright (C) 2012-2020 gnosygnu@gmail.com
|
||||||
|
|
||||||
XOWA is licensed under the terms of the General Public License (GPL) Version 3,
|
XOWA is licensed under the terms of the General Public License (GPL) Version 3,
|
||||||
or alternatively under the terms of the Apache License Version 2.0.
|
or alternatively under the terms of the Apache License Version 2.0.
|
||||||
|
|
||||||
You may use XOWA according to either of these licenses as is most appropriate
|
You may use XOWA according to either of these licenses as is most appropriate
|
||||||
for your project on a case-by-case basis.
|
for your project on a case-by-case basis.
|
||||||
|
|
||||||
The terms of each license can be found in the source code repository:
|
The terms of each license can be found in the source code repository:
|
||||||
|
|
||||||
GPLv3 License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-GPLv3.txt
|
GPLv3 License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-GPLv3.txt
|
||||||
Apache License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-APACHE2.txt
|
Apache License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-APACHE2.txt
|
||||||
*/
|
*/
|
||||||
package gplx.xowa.xtns.scribunto.libs; import gplx.*; import gplx.xowa.*; import gplx.xowa.xtns.*; import gplx.xowa.xtns.scribunto.*;
|
package gplx.xowa.xtns.scribunto.libs;
|
||||||
import org.junit.*; import gplx.xowa.xtns.scribunto.engines.mocks.*;
|
|
||||||
|
import gplx.Keyval_;
|
||||||
|
import gplx.String_;
|
||||||
|
import gplx.xowa.xtns.scribunto.Scrib_kv_utl_;
|
||||||
|
import gplx.xowa.xtns.scribunto.Scrib_lib;
|
||||||
|
import gplx.xowa.xtns.scribunto.engines.mocks.Mock_scrib_fxt;
|
||||||
|
import org.junit.Before;
|
||||||
|
import org.junit.Test;
|
||||||
|
|
||||||
public class Scrib_lib_ustring__match__tst {
|
public class Scrib_lib_ustring__match__tst {
|
||||||
private final Mock_scrib_fxt fxt = new Mock_scrib_fxt(); private Scrib_lib lib;
|
private final Mock_scrib_fxt fxt = new Mock_scrib_fxt(); private Scrib_lib lib;
|
||||||
@Before public void init() {
|
@Before public void init() {
|
||||||
fxt.Clear();
|
fxt.Clear();
|
||||||
lib = fxt.Core().Lib_ustring().Init();
|
lib = fxt.Core().Lib_ustring().Init();
|
||||||
}
|
}
|
||||||
@Test public void Basic() {
|
@Test public void Basic() {
|
||||||
Exec_match("abcd" , "bc" , 1, "bc"); // basic
|
Exec_match("abcd" , "bc" , 1, "bc"); // basic
|
||||||
Exec_match("abcd" , "x" , 1, String_.Null_mark); // empty
|
Exec_match("abcd" , "x" , 1, String_.Null_mark); // empty
|
||||||
Exec_match("abcd" , "a" , 2, String_.Null_mark); // bgn
|
Exec_match("abcd" , "a" , 2, String_.Null_mark); // bgn
|
||||||
@ -35,24 +43,26 @@ public class Scrib_lib_ustring__match__tst {
|
|||||||
Exec_match(1 , "a" , 1, String_.Null_mark); // Module can pass raw ints; PAGE:en.w:Budget_of_the_European_Union; DATE:2015-01-22
|
Exec_match(1 , "a" , 1, String_.Null_mark); // Module can pass raw ints; PAGE:en.w:Budget_of_the_European_Union; DATE:2015-01-22
|
||||||
Exec_match("" , "a?" , 1, ""); // no results with ? should return "" not nil; PAGE:en.d:民; DATE:2015-01-30
|
Exec_match("" , "a?" , 1, ""); // no results with ? should return "" not nil; PAGE:en.d:民; DATE:2015-01-30
|
||||||
}
|
}
|
||||||
@Test public void Args_out_of_order() {
|
@Test public void Args_out_of_order() {
|
||||||
fxt.Test__proc__kvps__empty(lib, Scrib_lib_ustring.Invk_match, Keyval_.Ary(Keyval_.int_(2, "[a]")));
|
fxt.Test__proc__kvps__empty(lib, Scrib_lib_ustring.Invk_match, Keyval_.Ary(Keyval_.int_(2, "[a]")));
|
||||||
}
|
}
|
||||||
@Test public void Balanced__trailing_whitespace() { // PURPOSE: match trailing whitespace; PAGE:en.w:Portal:Constructed_languages/Intro DATE:2018-07-02
|
@Test public void Balanced__trailing_whitespace() { // PURPOSE: match trailing whitespace; PAGE:en.w:Portal:Constructed_languages/Intro DATE:2018-07-02
|
||||||
Exec_match("[[a]] b", "%b[]%s*", 1, "[[a]] ");
|
Exec_match("[[a]] b", "%b[]%s*", 1, "[[a]] ");
|
||||||
}
|
}
|
||||||
@Test public void Balanced__numbered_1() { // PURPOSE: handle mix of balanced and regular capture; PAGE:en.w:Bahamas
|
@Test public void Balanced__numbered_1() { // PURPOSE: handle mix of balanced and regular capture; PAGE:en.w:Bahamas
|
||||||
Exec_match("[[5]]X99Y", "%b[]X(%d)%1Y", 1, "9");
|
Exec_match("[[5]]X99Y", "%b[]X(%d)%1Y", 1, "9");
|
||||||
}
|
}
|
||||||
@Test public void Balanced__numbered_2() {
|
@Test public void Balanced__numbered_2() {
|
||||||
Exec_match("A88B[[5]]X99Y", "A(%d)%1B%b[]X(%d)%2Y", 1, "8;9");
|
Exec_match("A88B[[5]]X99Y", "A(%d)%1B%b[]X(%d)%2Y", 1, "8;9");
|
||||||
}
|
}
|
||||||
@Test public void Unicode_alpha() {// ISSUE#:502; DATE:2019-07-01
|
@Test public void Unicode_alpha() {// ISSUE#:502; DATE:2019-07-01
|
||||||
Exec_match("ä" , "%a", 1, "ä");
|
Exec_match("ä" , "%a", 1, "ä");
|
||||||
}
|
}
|
||||||
|
@Test public void Number() {// 2019-20-01|ISSUE#:802|passing integer should return NULL, not throw error
|
||||||
|
Exec_match_obj("A" , 0, 0, String_.Null_mark);
|
||||||
|
}
|
||||||
|
|
||||||
|
// @Test public void Match_viwiktionary() {
|
||||||
// @Test public void Match_viwiktionary() {
|
|
||||||
// fxt.Init_cbk(Scrib_core.Key_mw_interface, fxt.Core().Lib_ustring(), Scrib_lib_ustring.Invk_match);
|
// fxt.Init_cbk(Scrib_core.Key_mw_interface, fxt.Core().Lib_ustring(), Scrib_lib_ustring.Invk_match);
|
||||||
// Exec_match("tr" , "()(r)", 1, ";"); // should return all matches
|
// Exec_match("tr" , "()(r)", 1, ";"); // should return all matches
|
||||||
// Exec_match("tr" , "^([b]*).-([c]*)$", 1, ";"); // should return all matches
|
// Exec_match("tr" , "^([b]*).-([c]*)$", 1, ";"); // should return all matches
|
||||||
@ -60,4 +70,7 @@ public class Scrib_lib_ustring__match__tst {
|
|||||||
private void Exec_match(Object text, String regx, int bgn, String expd) {
|
private void Exec_match(Object text, String regx, int bgn, String expd) {
|
||||||
fxt.Test__proc__kvps__flat(lib, Scrib_lib_ustring.Invk_match, Scrib_kv_utl_.base1_many_(text, regx, bgn), expd);
|
fxt.Test__proc__kvps__flat(lib, Scrib_lib_ustring.Invk_match, Scrib_kv_utl_.base1_many_(text, regx, bgn), expd);
|
||||||
}
|
}
|
||||||
|
private void Exec_match_obj(Object text, Object regx, int bgn, String expd) {
|
||||||
|
fxt.Test__proc__kvps__flat(lib, Scrib_lib_ustring.Invk_match, Scrib_kv_utl_.base1_many_(text, regx, bgn), expd);
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user