Merge pull request #5 from NetherlandsForensicInstitute/bugfix/remove…

…_limit_in_queries Remove LIMIT from queries
NetherlandsForensicInstitute · Oct 31, 2024 · da57d9e · da57d9e
2 parents de2054a + 14fef7f
commit da57d9e
Showing 1 changed file with 2 additions and 6 deletions.
diff --git a/copietje/__init__.py b/copietje/__init__.py
@@ -95,13 +95,11 @@ def make_index(self, zip_file: str):
         with ZipFile(zip_file) as documents_zip:
             cur = self.database.cursor()
 
-            # LIMIT should be removed in final script
             cur.execute("""
                 SELECT path, uid, stream, minhash
                 FROM documents
                 WHERE tags IS NOT NULL
-                ORDER BY size ASC
-                LIMIT 200
+                ORDER BY uid ASC
             """)
             for row in cur:
                 if minhash := self._get_or_update_minhash(row, documents_zip):
@@ -111,13 +109,11 @@ def query_index(self, zip_file: str) -> Iterable[Tuple[str, Iterable[Tuple[str,
         with ZipFile(zip_file) as documents_zip:
             cur = self.database.cursor()
 
-            # LIMIT should be removed in final script
             cur.execute("""
                 SELECT path, uid, stream, minhash
                 FROM documents
                 WHERE tags IS NULL
-                ORDER BY size ASC
-                LIMIT 200
+                ORDER BY uid ASC
             """)
             for row in cur:
                 if query_hash := self._get_or_update_minhash(row, documents_zip):