Merge pull request #233 from Inist-CNRS/add-importer-corpus

[RFR] support .corpus files

package.json

@@ -43,7 +43,8 @@
     "csv-parse": "1.1.10",
     "expect": "1.20.2",
     "ezs": "1.1.3",
-    "ezs-basics": "2.2.5",
+    "ezs-basics": "2.3.0",
+    "ezs-istex": "2.0.1",
     "file-loader": "0.9.0",
     "file-saver": "1.3.3",
     "html-webpack-plugin": "2.26.0",

src/api/loaders/corpus.js

@@ -0,0 +1,22 @@
+import ezs from 'ezs';
+import ezsBasics from 'ezs-basics';
+import ezsISTEX from 'ezs-istex';
+
+ezs.use(ezsBasics);
+ezs.use(ezsISTEX);
+
+const output = 'corpusName,author,language,abstract,title,pmid,'
+             + 'genre,host,publicationDate,copyrightDate,id,'
+             + 'score,serie,fulltext';
+
+
+export default config => stream =>
+    stream
+        .pipe(ezs('stringify'))
+        .pipe(ezs('concat'))
+        .pipe(ezs('ISTEXCorpus'))
+        .pipe(ezs('ISTEXQuery', { params: { output, ...config } }))
+        .pipe(ezs('ISTEXHarvest'))
+        .pipe(ezs('ISTEXRequest'))
+        .pipe(ezs('ISTEXHits'))
+        .pipe(ezs('flatten'));

src/api/loaders/index.js

@@ -4,6 +4,7 @@ import tsv from './tsv'; // eslint-disable-line
 import skos from './skos'; // eslint-disable-line
 import json from './json'; // eslint-disable-line
 import xml from './xml'; // eslint-disable-line
+import corpus from './corpus'; // eslint-disable-line
 
 export default {
     CSV,
@@ -16,4 +17,5 @@ export default {
     atom: xml,
     mods: xml,
     tei: xml,
+    corpus,
 };