Mercurial > repos > guerler > springsuite
annotate planemo/lib/python3.7/site-packages/rdflib/plugins/parsers/structureddata.py @ 1:56ad4e20f292 draft
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
author | guerler |
---|---|
date | Fri, 31 Jul 2020 00:32:28 -0400 |
parents | |
children |
rev | line source |
---|---|
1
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
1 #!/usr/bin/env python |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
2 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
3 Extraction parsers for structured data embedded into HTML or XML files. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
4 The former may include RDFa or microdata. The syntax and the extraction |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
5 procedures are based on: |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
6 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
7 * The RDFa specifications: http://www.w3.org/TR/#tr_RDFa |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
8 * The microdata specification: http://www.w3.org/TR/microdata/ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
9 * The specification of the microdata to RDF conversion: |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
10 http://www.w3.org/TR/microdata-rdf/ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
11 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
12 License: W3C Software License, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
13 http://www.w3.org/Consortium/Legal/copyright-software |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
14 Author: Ivan Herman |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
15 Copyright: W3C |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
16 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
17 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
18 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
19 from rdflib.parser import ( |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
20 Parser, StringInputSource, URLInputSource, FileInputSource) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
21 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
22 try: |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
23 import html5lib |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
24 assert html5lib |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
25 html5lib = True |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
26 except ImportError: |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
27 import warnings |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
28 warnings.warn( |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
29 'html5lib not found! RDFa and Microdata ' + |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
30 'parsers will not be available.') |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
31 html5lib = False |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
32 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
33 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
34 def _get_orig_source(source): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
35 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
36 A bit of a hack; the RDFa/microdata parsers need more than what the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
37 upper layers of RDFLib provide... |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
38 This method returns the original source references. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
39 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
40 if isinstance(source, StringInputSource): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
41 orig_source = source.getByteStream() |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
42 elif isinstance(source, URLInputSource): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
43 orig_source = source.url |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
44 elif isinstance(source, FileInputSource): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
45 orig_source = source.file.name |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
46 source.file.close() |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
47 else: |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
48 orig_source = source.getByteStream() |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
49 baseURI = source.getPublicId() |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
50 return (baseURI, orig_source) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
51 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
52 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
53 def _check_error(graph): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
54 from .pyRdfa import RDFA_Error, ns_rdf |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
55 from .pyRdfa.options import ns_dc |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
56 for (s, p, o) in graph.triples((None, ns_rdf["type"], RDFA_Error)): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
57 for (x, y, msg) in graph.triples((s, ns_dc["description"], None)): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
58 raise Exception("RDFa parsing Error! %s" % msg) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
59 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
60 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
61 # This is the parser interface as it would look when called from the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
62 # rest of RDFLib |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
63 class RDFaParser(Parser): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
64 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
65 Wrapper around the RDFa 1.1 parser. For further details on the RDFa 1.1 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
66 processing, see the relevant W3C documents at |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
67 http://www.w3.org/TR/#tr_RDFa. RDFa 1.1 is defined for XHTML, HTML5, SVG |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
68 and, in general, for any XML language. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
69 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
70 Note that the parser can also handle RDFa 1.0 if the extra parameter is |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
71 used and/or the input source uses RDFa 1.0 specific @version or DTD-s. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
72 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
73 def parse(self, source, graph, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
74 pgraph=None, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
75 media_type="", |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
76 rdfa_version=None, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
77 embedded_rdf=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
78 space_preserve=True, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
79 vocab_expansion=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
80 vocab_cache=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
81 refresh_vocab_cache=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
82 vocab_cache_report=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
83 check_lite=False): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
84 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
85 @param source: one of the input sources that the RDFLib package defined |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
86 @type source: InputSource class instance |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
87 @param graph: target graph for the triples; output graph, in RDFa spec. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
88 parlance |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
89 @type graph: RDFLib Graph |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
90 @keyword pgraph: target for error and warning triples; processor graph, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
91 in RDFa spec. parlance. If set to None, these triples are ignored |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
92 @type pgraph: RDFLib Graph |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
93 @keyword media_type: explicit setting of the preferred media type |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
94 (a.k.a. content type) of the the RDFa source. None means the content |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
95 type of the HTTP result is used, or a guess is made based on the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
96 suffix of a file |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
97 @type media_type: string |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
98 @keyword rdfa_version: 1.0 or 1.1. If the value is "", then, by |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
99 default, 1.1 is used unless the source has explicit signals to use |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
100 1.0 (e.g., using a @version attribute, using a DTD set up for 1.0, etc) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
101 @type rdfa_version: string |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
102 @keyword embedded_rdf: some formats allow embedding RDF in other |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
103 formats: (X)HTML can contain turtle in a special <script> element, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
104 SVG can have RDF/XML embedded in a <metadata> element. This flag |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
105 controls whether those triples should be interpreted and added to |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
106 the output graph. Some languages (e.g., SVG) require this, and the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
107 flag is ignored. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
108 @type embedded_rdf: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
109 @keyword space_preserve: by default, space in the HTML source must be preserved in the generated literal; |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
110 this behavior can be switched off |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
111 @type space_preserve: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
112 @keyword vocab_expansion: whether the RDFa @vocab attribute should |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
113 also mean vocabulary expansion (see the RDFa 1.1 spec for further |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
114 details) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
115 @type vocab_expansion: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
116 @keyword vocab_cache: in case vocab expansion is used, whether the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
117 expansion data (i.e., vocabulary) should be cached locally. This |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
118 requires the ability for the local application to write on the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
119 local file system |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
120 @type vocab_chache: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
121 @keyword vocab_cache_report: whether the details of vocabulary file caching process should be reported |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
122 in the processor graph as information (mainly useful for debug) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
123 @type vocab_cache_report: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
124 @keyword refresh_vocab_cache: whether the caching checks of vocabs should be by-passed, ie, if caches should be re-generated regardless of the stored date (important for vocab development) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
125 @type refresh_vocab_cache: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
126 @keyword check_lite: generate extra warnings in case the input source is not RDFa 1.1 check_lite |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
127 @type check_lite: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
128 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
129 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
130 if html5lib is False: |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
131 raise ImportError( |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
132 'html5lib is not installed, cannot use ' + |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
133 'RDFa and Microdata parsers.') |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
134 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
135 (baseURI, orig_source) = _get_orig_source(source) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
136 self._process(graph, pgraph, baseURI, orig_source, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
137 media_type=media_type, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
138 rdfa_version=rdfa_version, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
139 embedded_rdf=embedded_rdf, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
140 space_preserve=space_preserve, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
141 vocab_expansion=vocab_expansion, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
142 vocab_cache=vocab_cache, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
143 vocab_cache_report=vocab_cache_report, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
144 refresh_vocab_cache=refresh_vocab_cache, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
145 check_lite=check_lite |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
146 ) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
147 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
148 def _process(self, graph, pgraph, baseURI, orig_source, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
149 media_type="", |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
150 rdfa_version=None, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
151 embedded_rdf=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
152 space_preserve=True, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
153 vocab_expansion=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
154 vocab_cache=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
155 vocab_cache_report=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
156 refresh_vocab_cache=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
157 check_lite=False): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
158 from .pyRdfa import pyRdfa, Options |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
159 from rdflib import Graph |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
160 processor_graph = pgraph if pgraph is not None else Graph() |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
161 self.options = Options(output_processor_graph=True, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
162 embedded_rdf=embedded_rdf, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
163 space_preserve=space_preserve, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
164 vocab_expansion=vocab_expansion, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
165 vocab_cache=vocab_cache, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
166 vocab_cache_report=vocab_cache_report, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
167 refresh_vocab_cache=refresh_vocab_cache, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
168 check_lite=check_lite) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
169 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
170 if media_type is None: |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
171 media_type = "" |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
172 processor = pyRdfa(self.options, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
173 base=baseURI, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
174 media_type=media_type, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
175 rdfa_version=rdfa_version) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
176 processor.graph_from_source(orig_source, graph=graph, pgraph=processor_graph, rdfOutput=False) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
177 # This may result in an exception if the graph parsing led to an error |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
178 _check_error(processor_graph) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
179 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
180 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
181 class RDFa10Parser(Parser): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
182 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
183 This is just a convenience class to wrap around the RDFa 1.0 parser. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
184 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
185 def parse(self, source, graph, pgraph=None, media_type=""): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
186 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
187 @param source: one of the input sources that the RDFLib package defined |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
188 @type source: InputSource class instance |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
189 @param graph: target graph for the triples; output graph, in RDFa |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
190 spec. parlance |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
191 @type graph: RDFLib Graph |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
192 @keyword pgraph: target for error and warning triples; processor |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
193 graph, in RDFa spec. parlance. If set to None, these triples are |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
194 ignored |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
195 @type pgraph: RDFLib Graph |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
196 @keyword media_type: explicit setting of the preferred media type |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
197 (a.k.a. content type) of the the RDFa source. None means the content |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
198 type of the HTTP result is used, or a guess is made based on the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
199 suffix of a file |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
200 @type media_type: string |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
201 @keyword rdfOutput: whether Exceptions should be catched and added, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
202 as triples, to the processor graph, or whether they should be raised. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
203 @type rdfOutput: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
204 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
205 RDFaParser().parse(source, graph, pgraph=pgraph, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
206 media_type=media_type, rdfa_version="1.0") |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
207 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
208 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
209 class MicrodataParser(Parser): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
210 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
211 Wrapper around an HTML5 microdata, extracted and converted into RDF. For |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
212 the specification of microdata, see the relevant section of the HTML5 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
213 spec: http://www.w3.org/TR/microdata/; for the algorithm used to extract |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
214 microdata into RDF, see http://www.w3.org/TR/microdata-rdf/. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
215 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
216 def parse(self, source, graph, vocab_expansion=False, vocab_cache=False): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
217 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
218 @param source: one of the input sources that the RDFLib package defined |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
219 @type source: InputSource class instance |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
220 @param graph: target graph for the triples; output graph, in RDFa |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
221 spec. parlance |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
222 @type graph: RDFLib Graph |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
223 @keyword vocab_expansion: whether the RDFa @vocab attribute should |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
224 also mean vocabulary expansion (see the RDFa 1.1 spec for further |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
225 details) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
226 @type vocab_expansion: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
227 @keyword vocab_cache: in case vocab expansion is used, whether the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
228 expansion data (i.e., vocabulary) should be cached locally. This |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
229 requires the ability for the local application to write on the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
230 local file system |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
231 @type vocab_chache: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
232 @keyword rdfOutput: whether Exceptions should be catched and added, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
233 as triples, to the processor graph, or whether they should be raised. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
234 @type rdfOutput: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
235 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
236 if html5lib is False: |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
237 raise ImportError( |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
238 'html5lib is not installed, cannot use RDFa ' + |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
239 'and Microdata parsers.') |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
240 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
241 (baseURI, orig_source) = _get_orig_source(source) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
242 self._process(graph, baseURI, orig_source, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
243 vocab_expansion=vocab_expansion, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
244 vocab_cache=vocab_cache) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
245 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
246 def _process(self, graph, baseURI, orig_source, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
247 vocab_expansion=False, vocab_cache=False): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
248 from .pyMicrodata import pyMicrodata |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
249 processor = pyMicrodata(base=baseURI, vocab_expansion=vocab_expansion, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
250 vocab_cache=vocab_cache) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
251 processor.graph_from_source( |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
252 orig_source, graph=graph, rdfOutput=False) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
253 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
254 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
255 class StructuredDataParser(Parser): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
256 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
257 Convenience parser to extract both RDFa (including embedded Turtle) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
258 and microdata from an HTML file. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
259 It is simply a wrapper around the specific parsers. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
260 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
261 def parse(self, source, graph, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
262 pgraph=None, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
263 rdfa_version="", |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
264 vocab_expansion=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
265 vocab_cache=False, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
266 media_type='text/html' |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
267 ): |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
268 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
269 @param source: one of the input sources that the RDFLib package defined |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
270 @type source: InputSource class instance |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
271 @param graph: target graph for the triples; output graph, in RDFa |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
272 spec. parlance |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
273 @keyword rdfa_version: 1.0 or 1.1. If the value is "", then, by |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
274 default, 1.1 is used unless the source has explicit signals to use 1.0 |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
275 (e.g., using a @version attribute, using a DTD set up for 1.0, etc) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
276 @type rdfa_version: string |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
277 @type graph: RDFLib Graph |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
278 @keyword pgraph: target for error and warning triples; processor |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
279 graph, in RDFa spec. parlance. If set to None, these triples are |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
280 ignored |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
281 @type pgraph: RDFLib Graph |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
282 @keyword vocab_expansion: whether the RDFa @vocab attribute should |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
283 also mean vocabulary expansion (see the RDFa 1.1 spec for further |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
284 details) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
285 @type vocab_expansion: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
286 @keyword vocab_cache: in case vocab expansion is used, whether the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
287 expansion data (i.e., vocabulary) should be cached locally. This |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
288 requires the ability for the local application to write on the |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
289 local file system |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
290 @type vocab_chache: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
291 @keyword rdfOutput: whether Exceptions should be catched and added, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
292 as triples, to the processor graph, or whether they should be raised. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
293 @type rdfOutput: Boolean |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
294 """ |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
295 # Note that the media_type argument is ignored, and is here only to avoid an 'unexpected argument' error. |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
296 # This parser works for text/html only anyway... |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
297 (baseURI, orig_source) = _get_orig_source(source) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
298 if rdfa_version == "" : rdfa_version = "1.1" |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
299 RDFaParser()._process(graph, pgraph, baseURI, orig_source, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
300 media_type='text/html', |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
301 rdfa_version=rdfa_version, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
302 vocab_expansion=vocab_expansion, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
303 vocab_cache=vocab_cache) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
304 MicrodataParser()._process(graph, baseURI, orig_source, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
305 vocab_expansion=vocab_expansion, |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
306 vocab_cache=vocab_cache) |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
307 from .hturtle import HTurtleParser |
56ad4e20f292
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
guerler
parents:
diff
changeset
|
308 HTurtleParser()._process(graph, baseURI, orig_source, media_type='text/html') |