1		% (c) 2018-2019 Lehrstuhl fuer Softwaretechnik und Programmiersprachen,
2		% Heinrich Heine Universitaet Duesseldorf
3		% This software is licenced under EPL 1.0 (http://www.eclipse.org/org/documents/epl-v10.html)
4
5		:- module(regexp, [regexp_init/0,
6		regexp_match/2,
7		is_regexp/1,
8		regexp_replace/4,
9		regexp_search_first/3,
10		regexp_search_first_detailed/4,
11		regexp_search_all/3]).
12
13
14		/* the interface to a heap global variable stored and maintained in C */
15
16
17		foreign(cpp_regex_match,cpp_regex_match(+string,+string,+integer,[-integer])).
18		foreign(cpp_regex_replace,cpp_regex_replace(+string,+string,+string,[-atom])).
19		foreign(cpp_regex_search_all_str,cpp_regex_search_all_str(+string,+string,[-term])).
20		foreign(cpp_regex_search_first_str,cpp_regex_search_first_str(+string,+string,[-atom])).
21		foreign(cpp_regex_search_first,cpp_regex_search_first(+string,+integer,+string,[-term])).
22		foreign_resource(regexp,
23		[
24		cpp_regex_match,
25		cpp_regex_replace,
26		cpp_regex_search_first_str,
27		cpp_regex_search_first,
28		cpp_regex_search_all_str
29		]).
30
31		:- use_module(probsrc(error_manager)).
32
33		% match an atom agains a regular expression in ECMA syntax also represented as an atom
34		regexp_match(String,Pattern) :-
35		atom(String), atom(Pattern),
36		regexp_init,
37		!,
38		cpp_regex_match(String,Pattern,1,Res),
39		%format('String ~w for Pattern ~w: Result=~w~n',[String,Pattern,Res]),
40		(Res=1 -> true
41		; Res>1
42		-> add_error(regexp,'Illegal Regular Expression Pattern (use ECMAScript syntax): ',Pattern),
43		fail).
44		regexp_match(String,Pattern) :-
45		format(user_error,'Illegal call: ',regepx_match(String,Pattern)),
46		fail.
47
48		% check if an atom is a valid regular expression pattern
49		is_regexp(Pattern) :- atom(Pattern),
50		regexp_init,
51		!,
52		cpp_regex_match('',Pattern,0,Res), % TO DO: will still print error messages
53		Res<2. % 0 or 1 means pattern was ok
54		is_regexp(Pattern) :-
55		format(user_error,'Illegal call: ',is_regexp(Pattern)),
56		fail.
57
58
59		% match an atom agains a regular expression in ECMA syntax also represented as an atom
60		regexp_replace(String,Pattern,ReplStr,Result) :-
61		atom(String), atom(Pattern), atom(ReplStr), regexp_init,
62		!,
63		cpp_regex_replace(String,Pattern,ReplStr,Result).
64		%,format('Replace ~w for Pattern ~w with ~w: Result=~w~n',[String,Pattern,ReplStr,Result]).
65		regexp_replace(String,Pattern,ReplStr,Result) :-
66		format(user_error,'Illegal call: ',regexp_replace(String,Pattern,ReplStr,Result)),
67		fail.
68
69		% search for a first match of a pattern; returns empty string if no match found
70		regexp_search_first(String,Pattern,Result) :-
71		atom(String), atom(Pattern), regexp_init,
72		!,
73		cpp_regex_search_first_str(String,Pattern,Result).
74		%,format('Search ~w for Pattern ~w: Result=~w~n',[String,Pattern,Result]).
75		regexp_search_first(String,Pattern,Result) :-
76		format(user_error,'Illegal call: ',regexp_search_first(String,Pattern,Result)),
77		fail.
78
79		% search for a first match of a pattern;
80		% returns either match(Pos,Len,[MatchAtom|SubMatches]) or no-match
81		regexp_search_first_detailed(String,FromPosition,Pattern,Result) :-
82		atom(String), atom(Pattern), FromPosition>0,
83		regexp_init,
84		!,
85		Offset is FromPosition-1,
86		cpp_regex_search_first(String,Offset,Pattern,Result).
87		%, format('Search ~w for Pattern ~w: Result=~w~n',[String,Pattern,Result]).
88		regexp_search_first_detailed(String,From,Pattern,Result) :-
89		format(user_error,'Illegal call: ',regexp_search_first_detailed(String,From,Pattern,Result)),
90		fail.
91
92		:- use_module(library(lists),[reverse/2]).
93		% search for a first match of a pattern; returns empty string if no match found
94		regexp_search_all(String,Pattern,Result) :-
95		atom(String), atom(Pattern), regexp_init,
96		!,
97		cpp_regex_search_all_str(String,Pattern,CRes),
98		reverse(CRes,Result).
99		%,format('Search ~w for Pattern ~w: Result=~w~n',[String,Pattern,Result]).
100		regexp_search_all(String,Pattern,Result) :-
101		format(user_error,'Illegal call: ',regexp_search_first(String,Pattern,Result)),
102		fail.
103
104		:- dynamic loaded/0.
105
106		regexp_init :-
107		(loaded -> true
108		; (assert(loaded),
109		%assert_dir,
110		%assert(user:library_directory(prob_home('.'))),
111		%print(loading_foreign_resource(myheap)),nl,
112		load_foreign_resource(library(regexp)))
113		).
114
115		assert_dir :- (user:library_directory('.') -> true ; assert(user:library_directory('.'))).
116