Test and fix communication with Tsakorpus

common/config.py

@@ -19,11 +19,12 @@ class ResourceConfig:
     rxExt = re.compile('\\.[^.]*$')
 
     def __init__(self, fnameConfig=None):
-        self.platform = CorpPlatform.annis
+        self.platform = 'tsakorpus'
         self.transport_protocol = 'https'
         self.host = '127.0.0.1'
         self.port = '5000'
         self.url_path = '127.0.0.1'
+        self.resource_base_url = 'http://127.0.0.1'
         self.titles = []
         self.descriptions = []
         self.authors = []
@@ -38,6 +39,7 @@ class ResourceConfig:
         self.adv_supported = False
         self.supported_layers = []
         self.resources = []
+        self.search_lang_id = ''
 
         self.query_timeout = 60
 

common/tsakorpus_query_parser.py

 from urllib.parse import quote
 import re
+import json
+import urllib.request
 from .query_parser import QueryParser
 from .config import ResourceConfig
 from .diagnostics import Diagnostic, DiagnosticTypes
@@ -16,15 +18,19 @@ class TsakorpusQueryParser(QueryParser):
         Build a GET string (everything after the ?) from a description
         of the GET parameters in the getParams list.
         """
-        nWords = len(self.term_indexes(getParams))
+        termIndexes = self.term_indexes(getParams)
+        nWords = len(termIndexes)
         s = 'n_words=' + str(nWords)
         for param in getParams:
-            if type(param[1]) is list:
-                index = '_'.join(str(v) for v in param[1])
+            if param[0] in ('word_rel_', 'word_dist_from_', 'word_dist_to_'):
+                sfx = '_0'
             else:
-                index = str(param[1])
-            s += '&' + param[0] + index + '=' + quote(param[2])
+                sfx = ''
+            s += '&' + param[0] + str(param[1]) + sfx + '=' + quote(str(param[2]))
+        for i in termIndexes:
+            s += '&lang' + str(i) + '=' + config.search_lang_id
         s += '&page_size=' + str(config.max_hits)
+        s += '&precise=on&sort=random&response_format=json&distance_strict=on'
         return s
 
     def term_query(self, query, config):
@@ -42,9 +48,10 @@ class TsakorpusQueryParser(QueryParser):
                 iTerm += 1
                 getParams.append(['wf', iTerm, term])
                 if iTerm >= 2:
-                    getParams.append(['word_rel_', [iTerm-1, iTerm], '1'])
-                    getParams.append(['word_dist_from_', [iTerm-1, iTerm], '1'])
-                    getParams.append(['word_dist_to_', [iTerm-1, iTerm], '1'])
+                    # A maximum of one distance constraint per term
+                    getParams.append(['word_rel_', iTerm, iTerm-1])
+                    getParams.append(['word_dist_from_', iTerm, '1'])
+                    getParams.append(['word_dist_to_', iTerm, '1'])
         return getParams
 
 
@@ -70,10 +77,10 @@ class TsakorpusQueryParser(QueryParser):
         """
         getParamsShifted = []
         for param in getParams:
-            if type(param[1]) is int:
-                newParam = (param[0], param[1] + shift, param[2])
+            if type(param[2]) is int:
+                newParam = (param[0], param[1] + shift, param[2] + shift)
             else:
-                newParam = (param[0], [i + shift for i in param[1]], param[2])
+                newParam = (param[0], param[1] + shift, param[2])
             getParamsShifted.append(newParam)
         return getParamsShifted
 
@@ -152,6 +159,18 @@ class TsakorpusQueryParser(QueryParser):
             return self.not_bool(resultRight)
         return {}
 
+    def send_query(self, strGetParams: str, config: ResourceConfig):
+        """
+        Send the translated query to the Tsakorpus instance. Return JSON results
+        returned by the corpus.
+        """
+        url = config.resource_base_url.strip('/') + '/search_sent?' + strGetParams
+        response = urllib.request.urlopen(url)
+        data = response.read()
+        encoding = response.info().get_content_charset('utf-8')
+        responseJSON = json.loads(data.decode(encoding))
+        return responseJSON
+
 
 if __name__ == '__main__':
     pass

config/test.json

 {
 	"host": "0.0.0.0",
 	"port": "80",
-	"max_hits": 20,
-	"platform": "tsakorpus"
+	"max_hits": 15,
+	"platform": "tsakorpus",
+	"resource_base_url": "http://127.0.0.1:7342",
+	"search_lang_id": "beserman"
 }
\ No newline at end of file

main.py

@@ -67,7 +67,9 @@ def endpoint(
     if operation == Operation.searchRetrieve:
         if config.platform == CorpPlatform.tsakorpus:
             try:
-                res = app.qp_tsakorpus.translate_fcsql(query, config)
+                strGetParams = app.qp_tsakorpus.translate_fcsql(query, config)
+                print(strGetParams)
+                res = app.qp_tsakorpus.send_query(strGetParams, config)
             except Diagnostic as diag:
                 print('diag', str(diag))
                 return Response(content=str(diag), media_type='application/xml')