author | František Kučera <franta-hg@frantovo.cz> |
Sat, 04 Dec 2021 21:14:49 +0100 | |
branch | v_0 |
changeset 24 | 9403bbbd04f3 |
parent 22 | 3f6488171e34 |
permissions | -rw-r--r-- |
0 | 1 |
/** |
2 |
* Relational pipes |
|
3 |
* Copyright © 2018 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 |
|
10
1ae185cac1f3
fix license version: GNU GPLv3
František Kučera <franta-hg@frantovo.cz>
parents:
3
diff
changeset
|
7 |
* the Free Software Foundation, version 3 of the License. |
0 | 8 |
* |
9 |
* This program is distributed in the hope that it will be useful, |
|
10 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
11 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
12 |
* GNU General Public License for more details. |
|
13 |
* |
|
14 |
* You should have received a copy of the GNU General Public License |
|
15 |
* along with this program. If not, see <http://www.gnu.org/licenses/>. |
|
16 |
*/ |
|
17 |
#include <cstdlib> |
|
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
18 |
#include <vector> |
0 | 19 |
#include <memory> |
20
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
20 |
#include <locale> |
0 | 21 |
#include <regex> |
22 |
#include <algorithm> |
|
23 |
#include <unistd.h> |
|
24 |
||
25 |
#include <relpipe/writer/RelationalWriter.h> |
|
26 |
#include <relpipe/writer/RelpipeWriterException.h> |
|
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
27 |
#include <relpipe/writer/AttributeMetadata.h> |
0 | 28 |
#include <relpipe/writer/Factory.h> |
29 |
#include <relpipe/writer/TypeId.h> |
|
30 |
||
31 |
#include <relpipe/cli/CLI.h> |
|
32 |
||
14
012d491e219a
separate code into CSVCommand.h and CSVCommand.h
František Kučera <franta-hg@frantovo.cz>
parents:
10
diff
changeset
|
33 |
#include "CSVCommand.h" |
012d491e219a
separate code into CSVCommand.h and CSVCommand.h
František Kučera <franta-hg@frantovo.cz>
parents:
10
diff
changeset
|
34 |
|
0 | 35 |
using namespace std; |
36 |
using namespace relpipe::cli; |
|
37 |
using namespace relpipe::writer; |
|
38 |
||
14
012d491e219a
separate code into CSVCommand.h and CSVCommand.h
František Kučera <franta-hg@frantovo.cz>
parents:
10
diff
changeset
|
39 |
namespace relpipe { |
012d491e219a
separate code into CSVCommand.h and CSVCommand.h
František Kučera <franta-hg@frantovo.cz>
parents:
10
diff
changeset
|
40 |
namespace in { |
012d491e219a
separate code into CSVCommand.h and CSVCommand.h
František Kučera <franta-hg@frantovo.cz>
parents:
10
diff
changeset
|
41 |
namespace csv { |
012d491e219a
separate code into CSVCommand.h and CSVCommand.h
František Kučera <franta-hg@frantovo.cz>
parents:
10
diff
changeset
|
42 |
|
012d491e219a
separate code into CSVCommand.h and CSVCommand.h
František Kučera <franta-hg@frantovo.cz>
parents:
10
diff
changeset
|
43 |
bool CSVCommand::readValue(std::istream& input, std::stringstream& currentValue, bool& lastInRecord) { |
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
44 |
lastInRecord = false; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
45 |
char ch; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
46 |
input.get(ch); |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
47 |
if (ch == '"') { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
48 |
while (input.get(ch)) { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
49 |
if (ch == '"') { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
50 |
input.get(ch); |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
51 |
if (ch == '"') { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
52 |
currentValue << ch; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
53 |
} else { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
54 |
if (ch == '\r') input.get(ch); |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
55 |
if (ch == '\n') lastInRecord = true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
56 |
else if (ch != ',') throw RelpipeWriterException(L"Unexpected character (should be „\\n“ or „,“)"); |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
57 |
return true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
58 |
} |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
59 |
} else { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
60 |
currentValue << ch; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
61 |
} |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
62 |
} |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
63 |
} else if (ch == ',') { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
64 |
return true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
65 |
} else if (ch == '\n') { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
66 |
lastInRecord = true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
67 |
return true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
68 |
} else if (ch == '\r') { |
3
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
69 |
input.get(ch); |
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
70 |
if (ch == '\n') { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
71 |
lastInRecord = true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
72 |
return true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
73 |
} else { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
74 |
throw RelpipeWriterException(L"Crazy carriage stuck during journey"); |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
75 |
} |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
76 |
} else { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
77 |
for (currentValue << ch; input.get(ch);) { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
78 |
switch (ch) { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
79 |
case ',': return true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
80 |
case '\r': break; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
81 |
case '\n': |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
82 |
lastInRecord = true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
83 |
return true; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
84 |
default: currentValue << ch; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
85 |
} |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
86 |
} |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
87 |
} |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
88 |
return false; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
89 |
} |
0 | 90 |
|
20
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
91 |
/** |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
92 |
* Data types might be encoded in the attribute names: name::type e.g. some_attribute::integer |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
93 |
* |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
94 |
* TODO: share this code through relpipe-lib-infertypes (when available) |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
95 |
*/ |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
96 |
void tryParseTypes(vector<AttributeMetadata>& metadata, std::shared_ptr<writer::RelationalWriter> writer, Configuration& configuration) { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
97 |
std::wregex pattern(L"(.*)::(.*)"); |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
98 |
std::wsmatch match; |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
99 |
|
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
100 |
if (configuration.readTypes == Configuration::ReadTypes::AUTO || configuration.readTypes == Configuration::ReadTypes::TRUE) { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
101 |
bool hasTypes = true; |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
102 |
std::vector<TypeId> types; |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
103 |
std::vector<string_t> names; |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
104 |
for (AttributeMetadata& am : metadata) { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
105 |
if (std::regex_match(am.attributeName, match, pattern)) { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
106 |
names.push_back(match[1]); |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
107 |
if (configuration.readTypes == Configuration::ReadTypes::TRUE) { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
108 |
types.push_back(writer->toTypeId(match[2])); // must be valid type name otherwise exception is thrown |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
109 |
} else { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
110 |
try { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
111 |
types.push_back(writer->toTypeId(match[2])); |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
112 |
} catch (...) { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
113 |
hasTypes = false; // ignore exception and keep original names and default type (string) |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
114 |
} |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
115 |
} |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
116 |
} else { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
117 |
hasTypes = false; |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
118 |
} |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
119 |
} |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
120 |
|
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
121 |
if (hasTypes) { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
122 |
for (int i = 0, count = metadata.size(); i < count; i++) { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
123 |
metadata[i].attributeName = names[i]; |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
124 |
metadata[i].typeId = types[i]; |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
125 |
} |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
126 |
} else if (configuration.readTypes == Configuration::ReadTypes::TRUE) { |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
127 |
throw RelpipeWriterException(L"Types were expected in the CSV header, but not found."); |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
128 |
} |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
129 |
} |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
130 |
} |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
131 |
|
16
15ee963675af
change CLI interface: options: --relation --attribute
František Kučera <franta-hg@frantovo.cz>
parents:
15
diff
changeset
|
132 |
void CSVCommand::process(std::istream& input, std::shared_ptr<writer::RelationalWriter> writer, Configuration& configuration) { |
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
133 |
wstring_convert < codecvt_utf8<wchar_t>> convertor; // UTF-8 is required for CSV |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
134 |
vector<AttributeMetadata> metadata; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
135 |
bool headerDone = false; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
136 |
bool lastInRecord = false; |
21
22eb4838e8d0
check total number of values and throw exception if it does not match the number of columns
František Kučera <franta-hg@frantovo.cz>
parents:
20
diff
changeset
|
137 |
integer_t valueCount = 0; |
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
138 |
stringstream currentValue; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
139 |
|
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
140 |
|
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
141 |
while (readValue(input, currentValue, lastInRecord) && input.good()) { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
142 |
if (headerDone) { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
143 |
writer->writeAttribute(convertor.from_bytes(currentValue.str())); |
22
3f6488171e34
check number of values on each row
František Kučera <franta-hg@frantovo.cz>
parents:
21
diff
changeset
|
144 |
|
21
22eb4838e8d0
check total number of values and throw exception if it does not match the number of columns
František Kučera <franta-hg@frantovo.cz>
parents:
20
diff
changeset
|
145 |
valueCount++; |
22
3f6488171e34
check number of values on each row
František Kučera <franta-hg@frantovo.cz>
parents:
21
diff
changeset
|
146 |
if (valueCount > metadata.size()) throw RelpipeWriterException(L"The number of values " + std::to_wstring(valueCount) + L" exceeds declared column count: " + std::to_wstring(metadata.size())); |
3f6488171e34
check number of values on each row
František Kučera <franta-hg@frantovo.cz>
parents:
21
diff
changeset
|
147 |
if (lastInRecord && valueCount != metadata.size()) throw RelpipeWriterException(L"The number of values " + std::to_wstring(valueCount) + L" is lower than declared column count: " + std::to_wstring(metadata.size())); |
3f6488171e34
check number of values on each row
František Kučera <franta-hg@frantovo.cz>
parents:
21
diff
changeset
|
148 |
if (lastInRecord) valueCount = 0; |
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
149 |
} else { |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
150 |
AttributeMetadata am; |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
151 |
am.attributeName = convertor.from_bytes(currentValue.str()); |
3
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
152 |
am.typeId = TypeId::STRING; |
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
153 |
metadata.push_back(am); |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
154 |
if (lastInRecord) { |
3
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
155 |
|
16
15ee963675af
change CLI interface: options: --relation --attribute
František Kučera <franta-hg@frantovo.cz>
parents:
15
diff
changeset
|
156 |
vector<string_t> firstLine; |
3
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
157 |
|
16
15ee963675af
change CLI interface: options: --relation --attribute
František Kučera <franta-hg@frantovo.cz>
parents:
15
diff
changeset
|
158 |
if (metadata.size() == configuration.attributes.size()) { |
3
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
159 |
for (int i = 0; i < metadata.size(); i++) { |
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
160 |
firstLine.push_back(metadata[i].attributeName); |
16
15ee963675af
change CLI interface: options: --relation --attribute
František Kučera <franta-hg@frantovo.cz>
parents:
15
diff
changeset
|
161 |
metadata[i].attributeName = configuration.attributes[i].name; |
15ee963675af
change CLI interface: options: --relation --attribute
František Kučera <franta-hg@frantovo.cz>
parents:
15
diff
changeset
|
162 |
metadata[i].typeId = configuration.attributes[i].type; |
3
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
163 |
} |
16
15ee963675af
change CLI interface: options: --relation --attribute
František Kučera <franta-hg@frantovo.cz>
parents:
15
diff
changeset
|
164 |
} else if (configuration.attributes.size() == 0) { |
20
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
165 |
// first line contains attribute names and maybe also types |
90ae67de2f68
optionally read data types from the CSV header: --read-types (complements relpipe-out-csv --write-types true)
František Kučera <franta-hg@frantovo.cz>
parents:
16
diff
changeset
|
166 |
tryParseTypes(metadata, writer, configuration); |
16
15ee963675af
change CLI interface: options: --relation --attribute
František Kučera <franta-hg@frantovo.cz>
parents:
15
diff
changeset
|
167 |
} else { |
15ee963675af
change CLI interface: options: --relation --attribute
František Kučera <franta-hg@frantovo.cz>
parents:
15
diff
changeset
|
168 |
throw RelpipeWriterException(L"Declared attribute count (" + std::to_wstring(configuration.attributes.size()) + L") does not match with number of columns of the first line (" + std::to_wstring(metadata.size()) + L")"); |
3
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
169 |
} |
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
170 |
|
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
171 |
headerDone = true; |
16
15ee963675af
change CLI interface: options: --relation --attribute
František Kučera <franta-hg@frantovo.cz>
parents:
15
diff
changeset
|
172 |
writer->startRelation(configuration.relation, metadata, true); |
3
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
173 |
if (firstLine.size()) { |
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
174 |
for (string_t value : firstLine) writer->writeAttribute(value); |
d7907be4cc40
allow also custom relation name, attribute names and types (optional)
František Kučera <franta-hg@frantovo.cz>
parents:
2
diff
changeset
|
175 |
} |
1
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
176 |
} |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
177 |
} |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
178 |
|
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
179 |
currentValue.str(""); |
5eb4d149c6e2
first working version
František Kučera <franta-hg@frantovo.cz>
parents:
0
diff
changeset
|
180 |
currentValue.clear(); |
0 | 181 |
} |
21
22eb4838e8d0
check total number of values and throw exception if it does not match the number of columns
František Kučera <franta-hg@frantovo.cz>
parents:
20
diff
changeset
|
182 |
|
22eb4838e8d0
check total number of values and throw exception if it does not match the number of columns
František Kučera <franta-hg@frantovo.cz>
parents:
20
diff
changeset
|
183 |
/** |
22eb4838e8d0
check total number of values and throw exception if it does not match the number of columns
František Kučera <franta-hg@frantovo.cz>
parents:
20
diff
changeset
|
184 |
* RFC 4180: |
22eb4838e8d0
check total number of values and throw exception if it does not match the number of columns
František Kučera <franta-hg@frantovo.cz>
parents:
20
diff
changeset
|
185 |
* - Each line should contain the same number of fields throughout the file. |
22eb4838e8d0
check total number of values and throw exception if it does not match the number of columns
František Kučera <franta-hg@frantovo.cz>
parents:
20
diff
changeset
|
186 |
* - The last field in the record must not be followed by a comma. |
22eb4838e8d0
check total number of values and throw exception if it does not match the number of columns
František Kučera <franta-hg@frantovo.cz>
parents:
20
diff
changeset
|
187 |
*/ |
22
3f6488171e34
check number of values on each row
František Kučera <franta-hg@frantovo.cz>
parents:
21
diff
changeset
|
188 |
if (valueCount && valueCount != metadata.size()) throw RelpipeWriterException(L"Unexpected EOF: The number of values " + std::to_wstring(valueCount) + L" is lower than declared column count: " + std::to_wstring(metadata.size())); |
0 | 189 |
} |
190 |
||
14
012d491e219a
separate code into CSVCommand.h and CSVCommand.h
František Kučera <franta-hg@frantovo.cz>
parents:
10
diff
changeset
|
191 |
} |
0 | 192 |
} |
14
012d491e219a
separate code into CSVCommand.h and CSVCommand.h
František Kučera <franta-hg@frantovo.cz>
parents:
10
diff
changeset
|
193 |
} |