author | František Kučera <franta-hg@frantovo.cz> |
Thu, 30 Jan 2020 14:19:14 +0100 | |
branch | v_0 |
changeset 75 | ecbf6504915c |
parent 74 | a2aa84f310a5 |
child 77 | a680bcd946cd |
permissions | -rw-r--r-- |
65
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
1 |
/** |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
2 |
* Relational pipes |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
3 |
* Copyright © 2020 František Kučera (Frantovo.cz, GlobalCode.info) |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
4 |
* |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
5 |
* This program is free software: you can redistribute it and/or modify |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
6 |
* it under the terms of the GNU General Public License as published by |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
7 |
* the Free Software Foundation, version 3 of the License. |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
8 |
* |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
9 |
* This program is distributed in the hope that it will be useful, |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
10 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
11 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
12 |
* GNU General Public License for more details. |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
13 |
* |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
14 |
* You should have received a copy of the GNU General Public License |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
15 |
* along with this program. If not, see <http://www.gnu.org/licenses/>. |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
16 |
*/ |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
17 |
|
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
18 |
#include "streamlet-common.h" |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
19 |
|
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
20 |
#include <unistd.h> |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
21 |
#include <regex> |
65
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
22 |
#include <libxml++-2.6/libxml++/libxml++.h> |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
23 |
|
74
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
24 |
/** |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
25 |
* This streamlet provides values from XML files. |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
26 |
* It uses the XPath language to define, what portion of XML should be returned. |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
27 |
* |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
28 |
* With no options it does not provide any attributes. |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
29 |
* |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
30 |
* XPath expressions are passed as 'attribute' options. |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
31 |
* e.g. --option 'attribute' 'name()' will return single attribute with the name of the root node. |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
32 |
* |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
33 |
* Attributes can be renamed using aliases: --option 'attribute' 'name()' --as 'name'. Otherwise the full XPath expression is used as a name. |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
34 |
* Number of aliases should match the number of attributes (otherwise only first attributes are renamed, because aliases are global, not relative to the --option). |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
35 |
* |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
36 |
* Like relpipe-in-xmltable, this streamlet supports several modes: |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
37 |
* - string |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
38 |
* - boolean |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
39 |
* - raw-xml |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
40 |
* - line-number |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
41 |
* - xpath |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
42 |
* |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
43 |
* TODO: more OOP, move to separate repository, proper CMake project, clean-up, stabilize API |
a2aa84f310a5
streamlet examples: documentation
František Kučera <franta-hg@frantovo.cz>
parents:
70
diff
changeset
|
44 |
*/ |
65
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
45 |
class XPathStreamlet : public Streamlet { |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
46 |
private: |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
47 |
xmlpp::Node::PrefixNsMap ns; |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
48 |
|
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
49 |
void findXmlnsInEnvironment() { |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
50 |
std::regex xmlnsEnvPattern("xmlns_(.*)=(.*)"); |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
51 |
std::cmatch match; |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
52 |
for (char **env = environ; *env; env++) if (std::regex_match(*env, match, xmlnsEnvPattern)) ns[std::string(match[1])] = match[2]; |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
53 |
} |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
54 |
|
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
55 |
void findXmlnsInOptions() { |
75
ecbf6504915c
streamlet examples: encapsulate and hide abstract class fields
František Kučera <franta-hg@frantovo.cz>
parents:
74
diff
changeset
|
56 |
for (Option o : getOptions(std::wregex(L"xmlns[:_](.*)"))) ns[toBytes(o.nameMatch[1])] = toBytes(o.value); |
ecbf6504915c
streamlet examples: encapsulate and hide abstract class fields
František Kučera <franta-hg@frantovo.cz>
parents:
74
diff
changeset
|
57 |
for (Option o : getOptions(std::wregex(L"xmlns"), std::wregex(L"([^:]+):(.*)"))) ns[toBytes(o.valueMatch[1])] = toBytes(o.valueMatch[2]); |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
58 |
} |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
59 |
|
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
60 |
// Modes should share the logic of relpipe-in-xmltable |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
61 |
|
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
62 |
enum class Mode { |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
63 |
STRING, |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
64 |
BOOLEAN, |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
65 |
// TODO: support also XML number, when we have a rational or decimal numbers in Relational pipes |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
66 |
RAW_XML, |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
67 |
LINE_NUMBER, |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
68 |
XPATH |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
69 |
}; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
70 |
|
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
71 |
Mode toMode(std::wstring modeName) { |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
72 |
if (modeName == L"string") return Mode::STRING; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
73 |
else if (modeName == L"boolean") return Mode::BOOLEAN; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
74 |
else if (modeName == L"raw-xml") return Mode::RAW_XML; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
75 |
else if (modeName == L"line-number") return Mode::LINE_NUMBER; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
76 |
else if (modeName == L"xpath") return Mode::XPATH; |
75
ecbf6504915c
streamlet examples: encapsulate and hide abstract class fields
František Kučera <franta-hg@frantovo.cz>
parents:
74
diff
changeset
|
77 |
else throw std::invalid_argument("Unsupported mode: " + toBytes(modeName)); |
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
78 |
} |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
79 |
|
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
80 |
std::wstring toType(Mode mode) { |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
81 |
if (mode == Mode::BOOLEAN) return BOOLEAN; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
82 |
else if (mode == Mode::LINE_NUMBER) return INTEGER; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
83 |
else return STRING; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
84 |
} |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
85 |
|
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
86 |
class XPathAttribute { |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
87 |
public: |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
88 |
|
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
89 |
std::wstring name; |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
90 |
std::wstring xpath; |
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
91 |
Mode mode = Mode::STRING; |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
92 |
}; |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
93 |
|
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
94 |
std::vector<XPathAttribute> xpathAttributes; |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
95 |
|
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
96 |
protected: |
65
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
97 |
|
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
98 |
std::vector<AttributeMetadata> getOutputAttributesMetadata() override { |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
99 |
findXmlnsInEnvironment(); |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
100 |
findXmlnsInOptions(); |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
101 |
|
65
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
102 |
std::vector<AttributeMetadata> oam; |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
103 |
|
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
104 |
std::vector<Option> modeOptions = getOptions(L"mode"); |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
105 |
std::vector<Option> attributeOptions = getOptions(L"attribute"); |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
106 |
for (int i = 0, limit = attributeOptions.size(); i < limit; i++) { |
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
107 |
Mode mode = i < modeOptions.size() ? toMode(modeOptions[i].value) : Mode::STRING; |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
108 |
std::wstring alias = getAlias(i, attributeOptions[i].value); |
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
109 |
xpathAttributes.push_back({alias, attributeOptions[i].value, mode}); |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
110 |
oam.push_back({alias, toType(mode)}); |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
111 |
} |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
112 |
|
65
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
113 |
return oam; |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
114 |
} |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
115 |
|
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
116 |
std::vector<OutputAttribute> getOutputAttributes() override { |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
117 |
std::vector<OutputAttribute> oa; |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
118 |
|
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
119 |
try { |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
120 |
xmlpp::DomParser parser; |
75
ecbf6504915c
streamlet examples: encapsulate and hide abstract class fields
František Kučera <franta-hg@frantovo.cz>
parents:
74
diff
changeset
|
121 |
parser.parse_file(toBytes(getCurrentFile())); |
65
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
122 |
xmlpp::Element* root = parser.get_document()->get_root_node(); |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
123 |
|
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
124 |
for (XPathAttribute xpathAttribute : xpathAttributes) { |
75
ecbf6504915c
streamlet examples: encapsulate and hide abstract class fields
František Kučera <franta-hg@frantovo.cz>
parents:
74
diff
changeset
|
125 |
std::string xpath = toBytes(xpathAttribute.xpath); |
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
126 |
std::wstring result; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
127 |
bool isNull = false; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
128 |
|
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
129 |
if (xpathAttribute.mode == Mode::STRING) { |
75
ecbf6504915c
streamlet examples: encapsulate and hide abstract class fields
František Kučera <franta-hg@frantovo.cz>
parents:
74
diff
changeset
|
130 |
result = fromBytes(root->eval_to_string(xpath, ns)); |
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
131 |
} else if (xpathAttribute.mode == Mode::BOOLEAN) { |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
132 |
result = root->eval_to_boolean(xpath, ns) ? L"true" : L"false"; |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
133 |
} else if (xpathAttribute.mode == Mode::LINE_NUMBER) { |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
134 |
xmlpp::NodeSet attributeNodes = root->find(xpath, ns); |
70
018e2609f5bb
streamlets: move NULL handling from particular streamlets to StreamletAttributeFinder
František Kučera <franta-hg@frantovo.cz>
parents:
68
diff
changeset
|
135 |
if (attributeNodes.size()) result = std::to_wstring(attributeNodes[0]->get_line()); |
018e2609f5bb
streamlets: move NULL handling from particular streamlets to StreamletAttributeFinder
František Kučera <franta-hg@frantovo.cz>
parents:
68
diff
changeset
|
136 |
else isNull = true; |
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
137 |
} else if (xpathAttribute.mode == Mode::XPATH) { |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
138 |
xmlpp::NodeSet attributeNodes = root->find(xpath, ns); |
75
ecbf6504915c
streamlet examples: encapsulate and hide abstract class fields
František Kučera <franta-hg@frantovo.cz>
parents:
74
diff
changeset
|
139 |
if (attributeNodes.size()) result = fromBytes(attributeNodes[0]->get_path()); |
70
018e2609f5bb
streamlets: move NULL handling from particular streamlets to StreamletAttributeFinder
František Kučera <franta-hg@frantovo.cz>
parents:
68
diff
changeset
|
140 |
else isNull = true; |
68
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
141 |
} else if (xpathAttribute.mode == Mode::RAW_XML) { |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
142 |
throw std::logic_error("Raw XML mode is not yet implemented."); // TODO: implement also RAW_XML |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
143 |
} else { |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
144 |
throw std::logic_error("Unsupported mode."); // should never happer |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
145 |
} |
5d3d57d9c323
streamlet examples: xpath: support multiple modes (string, boolean, line-number, xpath), TODO: raw-xml
František Kučera <franta-hg@frantovo.cz>
parents:
67
diff
changeset
|
146 |
|
70
018e2609f5bb
streamlets: move NULL handling from particular streamlets to StreamletAttributeFinder
František Kučera <franta-hg@frantovo.cz>
parents:
68
diff
changeset
|
147 |
oa.push_back({result, isNull}); |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
148 |
} |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
149 |
} catch (xmlpp::parse_error& e) { |
70
018e2609f5bb
streamlets: move NULL handling from particular streamlets to StreamletAttributeFinder
František Kučera <franta-hg@frantovo.cz>
parents:
68
diff
changeset
|
150 |
for (XPathAttribute xpathAttribute : xpathAttributes) oa.push_back({L"", true}); |
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
151 |
// invalid XML → xmlpp::parse_error → just skip this file |
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
152 |
// invalid XPath → xmlpp::exception → failure |
65
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
153 |
} |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
154 |
|
67
0766d298eb1c
streamlet examples: xpath
František Kučera <franta-hg@frantovo.cz>
parents:
65
diff
changeset
|
155 |
|
65
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
156 |
return oa; |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
157 |
} |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
158 |
}; |
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
159 |
|
6944a03fb883
streamlet examples: xpath – parse and validate XML document
František Kučera <franta-hg@frantovo.cz>
parents:
diff
changeset
|
160 |
STREAMLET_RUN(XPathStreamlet) |