author | František Kučera <franta-hg@frantovo.cz> |
Sat, 06 Apr 2019 18:25:14 +0200 | |
branch | v_0 |
changeset 2 | 2390e2949a36 |
parent 1 | 8dfb42e5c088 |
child 3 | 891fe13d7397 |
permissions | -rw-r--r-- |
0 | 1 |
/** |
2 |
* Relational pipes |
|
3 |
* Copyright © 2019 František Kučera (Frantovo.cz, GlobalCode.info) |
|
4 |
* |
|
5 |
* This program is free software: you can redistribute it and/or modify |
|
6 |
* it under the terms of the GNU General Public License as published by |
|
7 |
* the Free Software Foundation, either version 3 of the License, or |
|
8 |
* (at your option) any later version. |
|
9 |
* |
|
10 |
* This program is distributed in the hope that it will be useful, |
|
11 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
12 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
13 |
* GNU General Public License for more details. |
|
14 |
* |
|
15 |
* You should have received a copy of the GNU General Public License |
|
16 |
* along with this program. If not, see <http://www.gnu.org/licenses/>. |
|
17 |
*/ |
|
18 |
#pragma once |
|
19 |
||
20 |
#include <iostream> |
|
21 |
#include <string> |
|
22 |
#include <sstream> |
|
23 |
#include <vector> |
|
24 |
||
25 |
#include <relpipe/writer/typedefs.h> |
|
26 |
||
27 |
namespace relpipe { |
|
28 |
namespace in { |
|
29 |
namespace recfile { |
|
30 |
||
31 |
using namespace relpipe::writer; |
|
32 |
||
33 |
class RecfileCommand { |
|
34 |
private: |
|
35 |
||
36 |
enum class RecfileLineType { |
|
37 |
METADATA, |
|
1
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
38 |
DATA, |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
39 |
SEPARATOR, |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
40 |
COMMENT, |
2
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
41 |
END, |
0 | 42 |
}; |
43 |
||
44 |
class RecfileHandler { |
|
45 |
private: |
|
46 |
RelationalWriter* writer; |
|
2
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
47 |
|
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
48 |
void metadata(const string_t& name, const string_t& value) { |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
49 |
std::wcerr << L"metadata(" << name << L", " << value << L");" << std::endl; // TODO: remove debug |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
50 |
} |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
51 |
|
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
52 |
void data(const string_t& name, const string_t& value) { |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
53 |
std::wcerr << L"data(" << name << L", " << value << L");" << std::endl; // TODO: remove debug |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
54 |
} |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
55 |
|
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
56 |
void comment(const string_t& value) { |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
57 |
std::wcerr << L"comment(" << value << L");" << std::endl; // TODO: remove debug |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
58 |
} |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
59 |
|
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
60 |
void separator() { |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
61 |
std::wcerr << L"separator()" << std::endl; // TODO: remove debug |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
62 |
} |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
63 |
|
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
64 |
void end() { |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
65 |
std::wcerr << L"end()" << std::endl; // TODO: remove debug |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
66 |
} |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
67 |
|
0 | 68 |
public: |
69 |
||
70 |
RecfileHandler(RelationalWriter* writer) : writer(writer) { |
|
71 |
} |
|
72 |
||
73 |
virtual ~RecfileHandler() { |
|
74 |
} |
|
75 |
||
1
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
76 |
void logicalLine(RecfileLineType type, const string_t& name = L"", const string_t& value = L"") { |
0 | 77 |
// TODO: writer->startRelation() |
78 |
// TODO: writer->writeAttribute() |
|
2
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
79 |
switch (type) { |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
80 |
case RecfileLineType::METADATA: return metadata(name, value); |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
81 |
case RecfileLineType::DATA: return data(name, value); |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
82 |
case RecfileLineType::COMMENT: return comment(value); |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
83 |
case RecfileLineType::SEPARATOR: return separator(); |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
84 |
case RecfileLineType::END: return end(); |
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
85 |
} |
0 | 86 |
} |
87 |
||
88 |
}; |
|
89 |
||
1
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
90 |
enum class ParserState { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
91 |
START, |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
92 |
NAME, |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
93 |
VALUE, |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
94 |
VALUE_CONTINUATION, |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
95 |
COMMENT, |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
96 |
END, |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
97 |
}; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
98 |
|
0 | 99 |
class RecfileParser { |
100 |
private: |
|
1
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
101 |
wstring_convert<codecvt_utf8<wchar_t>> convertor; // TODO: support also other encodings or are recfiles always in UTF-8? |
0 | 102 |
RecfileHandler& handler; |
1
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
103 |
|
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
104 |
void emitLogicalLine(RecfileLineType& type, std::stringstream& name, std::stringstream& value) { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
105 |
handler.logicalLine(type, convertor.from_bytes(name.str()), convertor.from_bytes(value.str())); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
106 |
|
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
107 |
name.str(""); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
108 |
name.clear(); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
109 |
value.str(""); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
110 |
value.clear(); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
111 |
type = RecfileLineType::DATA; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
112 |
} |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
113 |
|
0 | 114 |
public: |
115 |
||
116 |
RecfileParser(RecfileHandler& handler) : handler(handler) { |
|
117 |
} |
|
118 |
||
119 |
virtual ~RecfileParser() { |
|
120 |
} |
|
121 |
||
122 |
void parse(std::istream& input) { |
|
1
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
123 |
|
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
124 |
ParserState state = ParserState::START; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
125 |
RecfileLineType type = RecfileLineType::DATA; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
126 |
std::stringstream name; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
127 |
std::stringstream value; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
128 |
char ch; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
129 |
|
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
130 |
while (state != ParserState::END && input.good()) { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
131 |
ch = input.get(); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
132 |
if (input.eof()) continue; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
133 |
|
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
134 |
switch (state) { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
135 |
case ParserState::START: |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
136 |
if (ch == '%') { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
137 |
type = RecfileLineType::METADATA; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
138 |
break; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
139 |
} else if (ch == ' ') { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
140 |
break; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
141 |
} else if (ch == '\n') { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
142 |
handler.logicalLine(RecfileLineType::SEPARATOR); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
143 |
break; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
144 |
} else if (ch == '#') { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
145 |
type = RecfileLineType::COMMENT; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
146 |
state = ParserState::COMMENT; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
147 |
if (input.get() != ' ') input.unget(); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
148 |
break; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
149 |
} // else → name |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
150 |
case ParserState::NAME: |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
151 |
if (ch == ':') { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
152 |
state = ParserState::VALUE; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
153 |
if (input.get() != ' ') input.unget(); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
154 |
} else { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
155 |
name << ch; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
156 |
} |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
157 |
break; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
158 |
case ParserState::VALUE: |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
159 |
if (ch == '\n') state = ParserState::VALUE_CONTINUATION; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
160 |
else value << ch; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
161 |
break; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
162 |
case ParserState::VALUE_CONTINUATION: |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
163 |
if (ch == '+') { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
164 |
state = ParserState::VALUE; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
165 |
if (value.tellp()) value << '\n'; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
166 |
if (input.get() != ' ') input.unget(); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
167 |
} else { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
168 |
input.unget(); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
169 |
state = ParserState::START; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
170 |
emitLogicalLine(type, name, value); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
171 |
} |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
172 |
break; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
173 |
case ParserState::COMMENT: |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
174 |
if (ch == '\n') { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
175 |
state = ParserState::START; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
176 |
emitLogicalLine(type, name, value); |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
177 |
} else { |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
178 |
value << ch; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
179 |
} |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
180 |
break; |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
181 |
default: |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
182 |
throw RelpipeWriterException(L"Unknown ParserState: " + std::to_wstring((int) state) + L" in RecfileParser."); // TODO: better exception |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
183 |
} |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
184 |
} |
8dfb42e5c088
parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
185 |
emitLogicalLine(type, name, value); |
2
2390e2949a36
parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents:
1
diff
changeset
|
186 |
handler.logicalLine(RecfileLineType::END); |
0 | 187 |
} |
188 |
}; |
|
189 |
||
190 |
public: |
|
191 |
||
192 |
void process(std::istream& input, std::ostream& output) { |
|
193 |
unique_ptr<RelationalWriter> writer(Factory::create(output)); |
|
194 |
RecfileHandler handler(writer.get()); |
|
195 |
RecfileParser parser(handler); |
|
196 |
parser.parse(input); |
|
197 |
} |
|
198 |
}; |
|
199 |
||
200 |
} |
|
201 |
} |
|
202 |
} |