Merge pull request #90 from michalc/feat/raise-httpx-exceptions

feat: raise any httpx exceptions, e.g. from failed auth

Author: michalc

README.md

@@ -238,7 +238,13 @@ This means that sqlite-s3-query is not for all use cases of querying SQLite data
 This is not necessarily a permanent decision - it is possible that in future sqlite-s3-query will support unversioned buckets.
 
 
-## Exception hierarchy
+## Exceptions
+
+Under the hood [HTTPX](https://www.python-httpx.org/) is used to communicate with S3, but any [exceptions raised by HTTPX](https://www.python-httpx.org/exceptions/) are passed through to client code unchanged. This includes `httpx.HTTPStatusError` when S3 returns a non-200 response. Most commonly this will be when S3 returns a 403 in the case of insufficient permissions on the database object being queried.
+
+All other exceptions raised inherit from `sqlite_s3_query.SQLiteS3QueryError` as described in the following hierarchy.
+
+### Exception hierarchy
 
 - `SQLiteS3QueryError`
 

sqlite_s3_query.py

@@ -1,5 +1,6 @@
 import hmac
 import os
+import threading
 from contextlib import contextmanager
 from ctypes import CFUNCTYPE, POINTER, Structure, create_string_buffer, pointer, cast, memmove, memset, sizeof, addressof, cdll, byref, string_at, c_char_p, c_int, c_double, c_int64, c_void_p, c_char
 from ctypes.util import find_library
@@ -69,13 +70,33 @@ def sqlite_s3_query_multi(url, get_credentials=lambda now: (
     body_hash = sha256(b'').hexdigest()
     scheme, netloc, path, _, _ = urlsplit(url)
 
+    # We could use contextvars, but they aren't introduced until Python 3.7
+    pending_exceptions = {}
+    pending_exception_lock = threading.Lock()
+
+    def set_pending_exception(exception):
+        thread_id = threading.get_ident()
+        with pending_exception_lock:
+            pending_exceptions[thread_id] = exception
+
+    def raise_any_pending_exception():
+        thread_id = threading.get_ident()
+        with pending_exception_lock:
+            try:
+                raise pending_exceptions.pop(thread_id)
+            except KeyError:
+                pass
+
     def run(func, *args):
         res = func(*args)
+        raise_any_pending_exception()
         if res != 0:
             raise SQLiteError(libsqlite3.sqlite3_errstr(res).decode())
 
     def run_with_db(db, func, *args):
-        if func(*args) != 0:
+        res = func(*args)
+        raise_any_pending_exception()
+        if res != 0:
             raise SQLiteError(libsqlite3.sqlite3_errmsg(db).decode())
 
     @contextmanager
@@ -183,7 +204,8 @@ def x_read(p_file, p_out, i_amt, i_ofst):
                         offset += len(chunk)
                         if offset > i_amt:
                             break
-            except Exception:
+            except Exception as exception:
+                set_pending_exception(exception)
                 return SQLITE_IOERR
 
             if offset != i_amt:
@@ -328,6 +350,7 @@ def rows(get_pp_stmt, columns):
             if res == SQLITE_DONE:
                 break
             if res != SQLITE_ROW:
+                raise_any_pending_exception()
                 raise SQLiteError(libsqlite3.sqlite3_errstr(res).decode())
 
             yield tuple(

test.py

@@ -15,6 +15,7 @@
 import uuid
 
 import httpx
+from httpx import HTTPStatusError
 
 from sqlite_s3_query import (
     VersioningNotEnabledError,
@@ -411,9 +412,56 @@ def test_empty_object(self):
             'wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY',
             None,
         ), get_libsqlite3=get_libsqlite3) as query:
-            with self.assertRaisesRegex(SQLiteError, 'disk I/O error'):
+            with self.assertRaisesRegex(HTTPStatusError, r"\b416\b"):
                 query('SELECT 1').__enter__()
 
+    def test_incorrect_permission_on_context_enter(self):
+        with get_db([("CREATE TABLE my_table (my_col_a text, my_col_b text);",())]) as db:
+            put_object_with_versioning('my-bucket', 'my.db', db)
+
+        with self.assertRaisesRegex(HTTPStatusError, r"\b403\b"):
+            sqlite_s3_query('http://localhost:9000/my-bucket/my.db', get_credentials=lambda now: (
+                'us-east-1',
+                'AKIAIOSFODNN7EXAMPLE',
+                'not-the-right-key',
+                None,
+            ), get_libsqlite3=get_libsqlite3).__enter__()
+
+    def test_incorrect_permission_on_run_query(self):
+        with get_db([("CREATE TABLE my_table (my_col_a text, my_col_b text);",())]) as db:
+            put_object_with_versioning('my-bucket', 'my.db', db)
+
+        creds = (
+            (
+                'us-east-1',
+                'AKIAIOSFODNN7EXAMPLE',
+                'wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY',
+                None,
+            ), (
+                'us-east-1',
+                'AKIAIOSFODNN7EXAMPLE',
+                'not-the-right-key',
+                None,
+            )
+        )
+        creds_it = iter(creds)
+
+        with sqlite_s3_query('http://localhost:9000/my-bucket/my.db', get_credentials=lambda now: next(creds_it), get_libsqlite3=get_libsqlite3) as query:
+            with self.assertRaisesRegex(HTTPStatusError, r"\b403\b"):
+                query('SELECT 1').__enter__()
+
+    def test_short_db_header(self):
+        put_object_with_versioning('my-bucket', 'my.db', lambda: (b'*' * 99,))
+
+        with sqlite_s3_query('http://localhost:9000/my-bucket/my.db', get_credentials=lambda now: (
+            'us-east-1',
+            'AKIAIOSFODNN7EXAMPLE',
+            'wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY',
+            None,
+        ), get_libsqlite3=get_libsqlite3) as query:
+            with self.assertRaisesRegex(SQLiteError, 'disk I/O error'):
+                query("SELECT * FROM non_table").__enter__()
+
     def test_bad_db_header(self):
         put_object_with_versioning('my-bucket', 'my.db', lambda: (b'*' * 100,))
 
@@ -426,6 +474,18 @@ def test_bad_db_header(self):
             with self.assertRaisesRegex(SQLiteError, 'disk I/O error'):
                 query("SELECT * FROM non_table").__enter__()
 
+    def test_bad_db_first_page(self):
+        put_object_with_versioning('my-bucket', 'my.db', lambda: (b'*' * 4096,))
+
+        with sqlite_s3_query('http://localhost:9000/my-bucket/my.db', get_credentials=lambda now: (
+            'us-east-1',
+            'AKIAIOSFODNN7EXAMPLE',
+            'wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY',
+            None,
+        ), get_libsqlite3=get_libsqlite3) as query:
+            with self.assertRaisesRegex(SQLiteError, 'not a database'):
+                query("SELECT * FROM non_table").__enter__()
+
     def test_bad_db_second_half(self):
         with get_db([("CREATE TABLE my_table (my_col_a text, my_col_b text);",())] + [
             ("INSERT INTO my_table VALUES " + ','.join(["('some-text-a', 'some-text-b')"] * 500),()),