1 | #include "abstractlistaction.h"
|
---|
2 | #include "OIDtools.h"
|
---|
3 | #include "recptprototools.h"
|
---|
4 |
|
---|
5 | #include "oaitools.h"
|
---|
6 |
|
---|
7 |
|
---|
8 | bool abstractlistaction::check_classifier(recptproto *protocol, const text_t &collection, const text_t &set_name)
|
---|
9 | {
|
---|
10 | text_tset metadata;
|
---|
11 | FilterResponse_t response;
|
---|
12 | return get_info(set_name, collection, "", metadata, false, protocol, response, *this->logout);
|
---|
13 | }
|
---|
14 |
|
---|
15 |
|
---|
16 | bool abstractlistaction::output_content(ostream &output, recptproto *protocol, oaiargs ¶ms)
|
---|
17 | {
|
---|
18 | // Reset variables
|
---|
19 | this->output_docs = 0;
|
---|
20 |
|
---|
21 | text_t set_name = params["set"];
|
---|
22 | text_t position = "";
|
---|
23 |
|
---|
24 | // Process the resumptionToken if there is one
|
---|
25 | if (params["resumptionToken"] != "")
|
---|
26 | {
|
---|
27 | ResumptionToken resumption_token(params["resumptionToken"]);
|
---|
28 | set_name = resumption_token.getSet();
|
---|
29 | position = resumption_token.getPosition();
|
---|
30 | }
|
---|
31 |
|
---|
32 | // Case for "set" argument present -- output just the records in the specified set
|
---|
33 | if (set_name != "")
|
---|
34 | {
|
---|
35 | // Separate the collection name and Greenstone classifier OID from the set name
|
---|
36 | text_t collection_name = "";
|
---|
37 | text_t gsdl_classifier_OID = set_name;
|
---|
38 | oaiclassifier::toGSDL(collection_name, gsdl_classifier_OID);
|
---|
39 |
|
---|
40 | // If output_content_for_set() returns false a resumption token has been output, so it's time to stop
|
---|
41 | if (output_content_for_set(output, protocol, params, collection_name, gsdl_classifier_OID, set_name) == false)
|
---|
42 | {
|
---|
43 | return true;
|
---|
44 | }
|
---|
45 | }
|
---|
46 |
|
---|
47 | // Case for no "set" argument present -- output all records in all collections
|
---|
48 | else
|
---|
49 | {
|
---|
50 | // Get a list of the collections available
|
---|
51 | text_tarray& collections = this->configuration->getCollectionsList();
|
---|
52 | if (collections.size() == 0)
|
---|
53 | {
|
---|
54 | return false;
|
---|
55 | }
|
---|
56 |
|
---|
57 | // Get the current collection from the position value
|
---|
58 | text_t collection_name = "";
|
---|
59 | oaiclassifier::toGSDL(collection_name, position);
|
---|
60 |
|
---|
61 | // Find the starting collection
|
---|
62 | text_tarray::iterator collection_iterator = collections.begin();
|
---|
63 | while (collection_iterator != collections.end())
|
---|
64 | {
|
---|
65 | if (collection_name == "" || collection_name == *collection_iterator)
|
---|
66 | {
|
---|
67 | break;
|
---|
68 | }
|
---|
69 |
|
---|
70 | collection_iterator++;
|
---|
71 | }
|
---|
72 |
|
---|
73 | // Now loop through the remaining collections
|
---|
74 | while (collection_iterator != collections.end())
|
---|
75 | {
|
---|
76 | // If output_content_for_set() returns false a resumption token has been output, so it's time to stop
|
---|
77 | if (output_content_for_set(output, protocol, params, *collection_iterator, "", "") == false)
|
---|
78 | {
|
---|
79 | return true;
|
---|
80 | }
|
---|
81 |
|
---|
82 | collection_iterator++;
|
---|
83 | }
|
---|
84 | }
|
---|
85 |
|
---|
86 | // If no records were output throw an error
|
---|
87 | if (this->configuration->getOAIVersion() >= 200 && this->output_docs == 0)
|
---|
88 | {
|
---|
89 | errorType = "noRecordsMatch";
|
---|
90 | this->output_error(output, errorType);
|
---|
91 | return false;
|
---|
92 | }
|
---|
93 |
|
---|
94 | return true;
|
---|
95 | }
|
---|
96 |
|
---|
97 |
|
---|
98 | bool abstractlistaction::output_content_for_set(ostream &output, recptproto *protocol, oaiargs ¶ms, text_t collection_name, text_t gsdl_classifier_OID, text_t set_name)
|
---|
99 | {
|
---|
100 | // Check if the set is actually a collection
|
---|
101 | if (gsdl_classifier_OID == "")
|
---|
102 | {
|
---|
103 | gsdl_classifier_OID = "oai";
|
---|
104 | }
|
---|
105 |
|
---|
106 | text_t metadata_prefix = params["metadataPrefix"];
|
---|
107 | text_t from = params["from"];
|
---|
108 | text_t until = params["until"];
|
---|
109 | text_t position = "";
|
---|
110 |
|
---|
111 | // Process the resumptionToken if there is one
|
---|
112 | if (params["resumptionToken"] != "")
|
---|
113 | {
|
---|
114 | ResumptionToken resumption_token(params["resumptionToken"]);
|
---|
115 | metadata_prefix = resumption_token.getMetadataPrefix();
|
---|
116 | from = resumption_token.getFrom();
|
---|
117 | until = resumption_token.getUntil();
|
---|
118 | position = resumption_token.getPosition();
|
---|
119 | }
|
---|
120 |
|
---|
121 | // Get the list of identifiers in this collection
|
---|
122 | // Collections should not contain too many identifiers otherwise this will use a lot of time and memory
|
---|
123 | text_tset metadata; // Must be empty for efficiency
|
---|
124 | FilterResponse_t identifiers_response;
|
---|
125 | get_children(gsdl_classifier_OID, collection_name, "", metadata, false, protocol, identifiers_response, *this->logout);
|
---|
126 |
|
---|
127 | // Find the starting position, if necessary
|
---|
128 | ResultDocInfo_tarray::iterator identifier_iterator = identifiers_response.docInfo.begin();
|
---|
129 | if (output_docs == 0)
|
---|
130 | {
|
---|
131 | while (identifier_iterator != identifiers_response.docInfo.end())
|
---|
132 | {
|
---|
133 | if (position == "" || position == (collection_name + ":" + (*identifier_iterator).OID))
|
---|
134 | {
|
---|
135 | break;
|
---|
136 | }
|
---|
137 |
|
---|
138 | identifier_iterator++;
|
---|
139 | }
|
---|
140 | }
|
---|
141 |
|
---|
142 | // Now loop through displaying the next matching records
|
---|
143 | while (identifier_iterator != identifiers_response.docInfo.end())
|
---|
144 | {
|
---|
145 | position = (*identifier_iterator).OID;
|
---|
146 |
|
---|
147 | text_t document_OID = position;
|
---|
148 | if (starts_with(document_OID, "oai."))
|
---|
149 | {
|
---|
150 | document_OID = oaiclassifier::getGSDL_OID(collection_name, document_OID, protocol, *this->logout);
|
---|
151 | }
|
---|
152 |
|
---|
153 | // Check this OID is in the (optional) date range specified
|
---|
154 | if (this->in_date_range(output, protocol, params, collection_name, document_OID, from, until))
|
---|
155 | {
|
---|
156 | // If we've output the desired number of records return a resumptionToken and we're done
|
---|
157 | if (this->output_docs == this->configuration->resumeAfter())
|
---|
158 | {
|
---|
159 | // If a set has been specified, we can use the collection's buildDate in the resumption token
|
---|
160 | text_t date_stamp = "";
|
---|
161 | if (set_name != "")
|
---|
162 | {
|
---|
163 | ColInfoResponse_t cinfo;
|
---|
164 | comerror_t err;
|
---|
165 | protocol->get_collectinfo(collection_name, cinfo, err, cerr);
|
---|
166 | date_stamp = cinfo.buildDate;
|
---|
167 | }
|
---|
168 |
|
---|
169 | ResumptionToken resumption_token(date_stamp, set_name, metadata_prefix, from, until, collection_name + ":" + position);
|
---|
170 |
|
---|
171 | // Don't add any whitespace around the resumption token as it can confuse harvesters/validators
|
---|
172 | output << " <resumptionToken>" << resumption_token.getResumptionTokenString() << "</resumptionToken>" << endl;
|
---|
173 | return false;
|
---|
174 | }
|
---|
175 |
|
---|
176 | // Otherwise output this record and increment the count
|
---|
177 | this->output_document(output, protocol, collection_name, document_OID, metadata_prefix);
|
---|
178 | this->output_docs++;
|
---|
179 | }
|
---|
180 |
|
---|
181 | identifier_iterator++;
|
---|
182 | }
|
---|
183 |
|
---|
184 | return true;
|
---|
185 | }
|
---|
186 |
|
---|
187 |
|
---|
188 | bool abstractlistaction::in_date_range(ostream &output, recptproto *protocol, oaiargs ¶ms,
|
---|
189 | text_t& collection, text_t oai_OID, text_t from, text_t until)
|
---|
190 | {
|
---|
191 | // If no "from" or "until" value is specified every record matches, so we don't need to go any further
|
---|
192 | if (from == "" && until == "")
|
---|
193 | {
|
---|
194 | return true;
|
---|
195 | }
|
---|
196 |
|
---|
197 | // Get the datestamp from the document as sections do not have this metadata
|
---|
198 | text_t document_OID;
|
---|
199 | get_top(oai_OID, document_OID);
|
---|
200 |
|
---|
201 | // Request the lastmodified value for this document
|
---|
202 | text_tset metadata;
|
---|
203 | metadata.insert("lastmodified");
|
---|
204 | FilterResponse_t response;
|
---|
205 | if (!get_info(document_OID, collection, "", metadata, false, protocol, response, *this->logout))
|
---|
206 | {
|
---|
207 | return false;
|
---|
208 | }
|
---|
209 |
|
---|
210 | text_t last_modified_date;
|
---|
211 | this->getLastModifiedDate(response.docInfo[0], last_modified_date);
|
---|
212 |
|
---|
213 | // Check this record is not before the "from" value, if it exists
|
---|
214 | if (from != "" && last_modified_date < from)
|
---|
215 | {
|
---|
216 | // Too early
|
---|
217 | return false;
|
---|
218 | }
|
---|
219 |
|
---|
220 | // Check this record is not after the "until" value, if it exists
|
---|
221 | if (until != "" && last_modified_date > until)
|
---|
222 | {
|
---|
223 | // Too late
|
---|
224 | return false;
|
---|
225 | }
|
---|
226 |
|
---|
227 | // Just right
|
---|
228 | return true;
|
---|
229 | }
|
---|