From: Sarah Hoffmann <lonvia@denofr.de>
Date: Wed, 21 Jul 2021 09:37:14 +0000 (+0200)
Subject: convert word info column to json before copying
X-Git-Tag: v4.0.0~45^2~6
X-Git-Url: https://git.openstreetmap.org./nominatim.git/commitdiff_plain/eb6814d74e509fb49986989bb4c60539a2871d76

convert word info column to json before copying
---

diff --git a/nominatim/db/utils.py b/nominatim/db/utils.py
index 9a4a41a5..bb7faa25 100644
--- a/nominatim/db/utils.py
+++ b/nominatim/db/utils.py
@@ -65,6 +65,7 @@ _SQL_TRANSLATION = {ord(u'\\'): u'\\\\',
                     ord(u'\t'): u'\\t',
                     ord(u'\n'): u'\\n'}
 
+
 class CopyBuffer:
     """ Data collector for the copy_from command.
     """
diff --git a/nominatim/tokenizer/legacy_icu_tokenizer.py b/nominatim/tokenizer/legacy_icu_tokenizer.py
index 14fa5b60..e019ef67 100644
--- a/nominatim/tokenizer/legacy_icu_tokenizer.py
+++ b/nominatim/tokenizer/legacy_icu_tokenizer.py
@@ -4,6 +4,7 @@ libICU instead of the PostgreSQL module.
 """
 from collections import Counter
 import itertools
+import json
 import logging
 import re
 from textwrap import dedent
@@ -173,7 +174,7 @@ class LegacyICUTokenizer:
             # copy them back into the word table
             with CopyBuffer() as copystr:
                 for k, v in words.items():
-                    copystr.add('w', k, {'count': v})
+                    copystr.add('w', k, json.dumps({'count': v}))
 
                 with conn.cursor() as cur:
                     copystr.copy_out(cur, 'word',
@@ -287,7 +288,7 @@ class LegacyICUNameAnalyzer:
                         to_delete.append(word)
                     else:
                         copystr.add(self.name_processor.get_search_normalized(postcode),
-                                    'P', {'postcode': postcode})
+                                    'P', json.dumps({'postcode': postcode}))
 
                 if to_delete:
                     cur.execute("""DELETE FROM WORD
@@ -337,8 +338,8 @@ class LegacyICUNameAnalyzer:
                 term = self.name_processor.get_search_normalized(word)
                 if term:
                     copystr.add(term, 'S',
-                                {'word': word, 'class': cls, 'type': typ,
-                                 'op': oper if oper in ('in', 'near') else None})
+                                json.dumps({'word': word, 'class': cls, 'type': typ,
+                                            'op': oper if oper in ('in', 'near') else None}))
                     added += 1
 
             copystr.copy_out(cursor, 'word',
diff --git a/test/python/test_db_utils.py b/test/python/test_db_utils.py
index 545cc58f..9eea7ed1 100644
--- a/test/python/test_db_utils.py
+++ b/test/python/test_db_utils.py
@@ -1,6 +1,8 @@
 """
 Tests for DB utility functions in db.utils
 """
+import json
+
 import pytest
 
 import nominatim.db.utils as db_utils
@@ -115,3 +117,38 @@ class TestCopyBuffer:
 
 
 
+class TestCopyBufferJson:
+    TABLE_NAME = 'copytable'
+
+    @pytest.fixture(autouse=True)
+    def setup_test_table(self, table_factory):
+        table_factory(self.TABLE_NAME, 'colA INT, colB JSONB')
+
+
+    def table_rows(self, cursor):
+        cursor.execute('SELECT * FROM ' + self.TABLE_NAME)
+        results = {k: v for k,v in cursor}
+
+        assert len(results) == cursor.rowcount
+
+        return results
+
+
+    def test_json_object(self, temp_db_cursor):
+        with db_utils.CopyBuffer() as buf:
+            buf.add(1, json.dumps({'test': 'value', 'number': 1}))
+
+            buf.copy_out(temp_db_cursor, self.TABLE_NAME)
+
+        assert self.table_rows(temp_db_cursor) == \
+                   {1: {'test': 'value', 'number': 1}}
+
+
+    def test_json_object_special_chras(self, temp_db_cursor):
+        with db_utils.CopyBuffer() as buf:
+            buf.add(1, json.dumps({'te\tst': 'va\nlue', 'nu"mber': None}))
+
+            buf.copy_out(temp_db_cursor, self.TABLE_NAME)
+
+        assert self.table_rows(temp_db_cursor) == \
+                   {1: {'te\tst': 'va\nlue', 'nu"mber': None}}