Implement set logic for documents

Ref #103

src/zbmath_rest2oai/getAsXml.py

@@ -58,6 +58,19 @@ def apply_zbmath_api_fixes(result, prefix):
     result['states'] = states
 
 
+def extract_tags(result):
+    mscs = result.get('msc',[])
+    tags = []
+    for msc in mscs:
+        msc0 = msc['code'][:2]
+        if msc0 not in tags:
+            tags.append(msc0)
+    tags.sort()
+    if result.get('database') == 'JFM':
+        tags.append('JFM')
+    return tags
+
+
 def final_xml2(api_source, prefix):
     headers = {'Accept': 'application/json'}
     r = requests.get(api_source, headers=headers)
@@ -69,12 +82,15 @@ def final_xml2(api_source, prefix):
         raise Exception(f"Unexpected response with status code {r.status_code}: {r.text}")
     json = r.json()
     dict_math_entities = {}
+    tags = {}
     for result in json["result"]:
         apply_zbmath_api_fixes(result, prefix)
-        dict_math_entities[result["id"]] = _illegal_xml_chars_RE.sub("", Converter(wrap="root").build(
+        identifier = result["id"]
+        dict_math_entities[identifier] = _illegal_xml_chars_RE.sub("", Converter(wrap="root").build(
             result,
             closed_tags_for=[[], '', [None], None]))
-    return [dict_math_entities, r.elapsed.total_seconds()]
+        tags[identifier] = extract_tags(result)
+    return [dict_math_entities, r.elapsed.total_seconds(), tags]
 
 
 if __name__ == "__main__":

src/zbmath_rest2oai/writeOai.py

@@ -48,9 +48,9 @@ async def async_write_oai(xml_contents, ingest_format, tags=None):
 
 
 def write_oai(api_source, prefix, ingest_format):
-    test_xml, time_rest = getAsXml.final_xml2(api_source, prefix)
+    xml_contents, time_rest, tags = getAsXml.final_xml2(api_source, prefix)
     start = timer()
-    records, last_id = asyncio.run(async_write_oai(test_xml, ingest_format))
+    records, last_id = asyncio.run(async_write_oai(xml_contents, ingest_format, tags))
 
     last_id = int(last_id.removeprefix(prefix))
     return {

test/test_extract_tags.py

@@ -0,0 +1,22 @@
+import unittest
+
+import requests
+
+from zbmath_rest2oai.getAsXml import extract_tags
+
+from zbmath_rest2oai import getAsXml
+
+API_SOURCE = 'https://api.zbmath.org/v1/document/_structured_search?page=0&results_per_page=10&zbmath%20id=2500495'
+
+
+class PlainXmlTest(unittest.TestCase):
+    def test_similarity(self):
+        headers = {'Accept': 'application/json'}
+        r = requests.get(API_SOURCE, headers=headers)
+        real_tags = extract_tags(r.json()['result'][0])
+        assert real_tags == ['11','JFM']
+
+
+
+if __name__ == '__main__':
+    unittest.main()