Merge pull request #6040 from tomachalek/fcs_refacts_and_fixes

Fcs refacts and fixes

lib/action/model/fcs.py

@@ -40,6 +40,10 @@
 
 @dataclass
 class FCSResourceInfo:
+    pid: str
+    """
+    pid should be a universaly accepted persistent identifier (Handle.Net, DOI etc.)
+    """
     title: str
     landing_page_uri: Optional[str] = None
     language: Optional[str] = None
@@ -100,6 +104,7 @@ async def corpora_info(self, value: str, max_items: int) -> List[FCSResourceInfo
                     lang_code = get_lang_code(a2=cinfo.collator_locale.split('_')[0])
                 resources.append(
                     FCSResourceInfo(
+                        pid=corpus_id, # TODO we need something like an ID from Handle.Net or DOI
                         title=corpus_id,
                         description=cinfo.localized_desc('en'),
                         landing_page_uri=cinfo.web,
@@ -148,7 +153,13 @@ async def fcs_scan(self, corpname: str, scan_query: str, max_ter: int, start: in
         wl = await wordlist(corp, args, max_ter)
         return [(d['str'], d['freq']) for d in wl][start:][:max_ter]
 
-    async def fcs_search(self, corp: AbstractKCorpus, corpname: str, fcs_query: str, max_rec: int, start: int) -> Tuple[FCSSearchResult, str]:
+    async def fcs_search(
+            self,
+            corp: AbstractKCorpus,
+            fcs_query: str,
+            max_rec: int,
+            start: int
+    ) -> Tuple[FCSSearchResult, str]:
         """
         aux function for federated content search: operation=searchRetrieve
         """
@@ -209,7 +220,7 @@ async def fcs_search(self, corp: AbstractKCorpus, corpname: str, fcs_query: str,
         try:
             with plugins.runtime.AUTH as auth:
                 anon_id = auth.anonymous_user(self.plugin_ctx)['id']
-            q = ['q' + rq]
+            q = ('q' + rq,)
 
             # try to locate concordance in cache
             lock = asyncio.Lock()

lib/plugin_types/corparch/corpus.py

@@ -197,6 +197,7 @@ class   CorpusInfo:
     """
 
     id: Optional[str] = None
+    pid: Optional[str] = None # TODO see https://github.com/czcorpus/kontext/issues/6039
     name: Optional[str] = None
     description: Optional[str] = None  # filled in during localization
     _description_cs: Optional[str] = None

lib/views/fcs.py

@@ -17,8 +17,8 @@
 from collections import defaultdict
 from dataclasses import asdict, dataclass, field
 from typing import Any, Dict, List, Optional, Tuple
+import random
 
-import aiofiles
 import plugins
 import settings
 from action.argmapping.conc import QueryFormArgs
@@ -78,6 +78,7 @@ async def op_explain(amodel: FCSActionModel, req: KRequest, resp_common: FCSResp
                     lang_code = get_lang_code(a2=cinfo.collator_locale.split('_')[0])
                 resp['resources'].append(
                     FCSResourceInfo(
+                        pid=corp, # TODO we need something like an ID from Handle.Net or DOI
                         title=corp,
                         description=cinfo.localized_desc('en'),
                         landing_page_uri=cinfo.web,
@@ -151,7 +152,6 @@ async def op_search_retrieve(amodel: FCSActionModel, req: KRequest, resp_common:
     tasks = [
         amodel.fcs_search(
             await amodel.cf.get_corpus(corp),
-            amodel.args.corpname,
             query,
             resp_common.maximumRecords,
             resp_common.startRecord,
@@ -160,9 +160,12 @@ async def op_search_retrieve(amodel: FCSActionModel, req: KRequest, resp_common:
     ]
     results: List[List[FCSSearchResult], str] = await asyncio.gather(*tasks)
     # merging results
+    merged_rows = [row for result, _ in results for row in result.rows]
+    if len(merged_rows) > resp_common.maximumRecords:
+        merged_rows = random.sample(merged_rows, k=resp_common.maximumRecords)
     merged_results = FCSSearchResult(
-        rows=[row for result, _ in results for row in result.rows],
-        size=sum(r.size for r, _ in results),
+        rows=merged_rows,
+        size=len(merged_rows),
     )
     cql_query = results[0][1]
 

templates/fcs/v1_complete.html

@@ -64,7 +64,7 @@
             </ed:SupportedDataViews>
             <ed:Resources>
                 {% for resource in resources %}
-                <ed:Resource>
+                <ed:Resource pid="{{ resource.pid }}">
                     <ed:Title xml:lang="en">{{ resource.title }}</ed:Title>
                     {% if resource.description %}
                     <ed:Description xml:lang="en">{{ resource.description }}</ed:Description>
@@ -134,7 +134,8 @@
             <sru:recordSchema>http://clarin.eu/fcs/resource</sru:recordSchema>
             <sru:recordPacking>{{ recordPacking }}</sru:recordPacking>
             <sru:recordData>
-                <fcs:Resource xmlns:fcs="http://clarin.eu/fcs/resource" pid="{{ corppid }}" ref="{{ conc_view_url_tpl }}{{ corppid }}">
+                <fcs:Resource xmlns:fcs="http://clarin.eu/fcs/resource"
+                              {% if corppid %} pid="{{ corppid }}" ref="{{ conc_view_url_tpl }}{{ corppid }}" {% endif %}>
                     <fcs:DataView type="application/x-clarin-fcs-hits+xml">
                         <hits:Result xmlns:hits="http://clarin.eu/fcs/dataview/hits">
                             {{ left|xmle }}<hits:Hit>{{ word|xmle }}</hits:Hit> {{ right|xmle }}