fix(openai-sdk-python): restore complete async middleware update

This commit is contained in:
abhinav7x94 2026-08-16 09:07:55 +05:30
parent 879a4707cd
commit 25ebd8ca81
2 changed files with 545 additions and 154 deletions

View file

@ -439,164 +439,405 @@ class SupermemoryOpenAIWrapper:
def _create_client_variant_facade(self, name: str) -> _ResourceFacade:
"""Wrap completions reached through a client response variant."""
client_resource = getattr(self._client, name)
chat_resource = client_resource.chaëN½‰žËkºwµçXÛÛ\][ÛœË\˜]È‹˜]רܙX]WÙ˜XÝÜžBˆ
BˆÛY[<EFBFBD> ˜Ú] <EFBFBD>Ú]ܘ]×Ü™\ÜÛœÙHHÚ[\S˜[Y\ÜXÙJˆÛÛ\][ÛœÏXÛÛ\][Û—Ü™\ÛÝ\˜ÙJ˜Ú] \˜]È‹˜]רܙX]WÙ˜XÝÜžJBˆ
BˆÛY[<EFBFBD> <EFBFBD>Ú]ܘ]×Ü™\ÜÛœÙHHÚ[\S˜[Y\ÜXÙJˆÚ]TÚ[\S˜[Y\ÜXÙJˆÛÛ\][ÛœÏXÛÛ\][Û—Ü™\ÛÝ\˜ÙJ˜ÛY[<EFBFBD> \˜]È‹˜]רܙX]WÙ˜XÝÜžJBˆ
Bˆ
BˆÛY[<EFBFBD> ˜Ú] ˜ÛÛ\][ÛœË<EFBFBD>Ú]ÜÝ™X[Z[™×Ü™\ÜÛœÙHHÛÛ\][Û—Ü™\ÛÝ\˜ÙJˆ˜ÛÛ\][ÛœË\Ý™X[H‹Ý™X[Z[™×ØÜ™X]WÙ˜XÝÜžBˆ
BˆÛY[<EFBFBD> ˜Ú] <EFBFBD>Ú]ÜÝ™X[Z[™×Ü™\ÜÛœÙHHÚ[\S˜[Y\ÜXÙJˆÛÛ\][ÛœÏXÛÛ\][Û—Ü™\ÛÝ\˜ÙJ˜Ú] \Ý™X[H‹Ý™X[Z[™×ØÜ™X]WÙ˜XÝÜžJBˆ
BˆÛY[<EFBFBD> <EFBFBD>Ú]ÜÝ™X[Z[™×Ü™\ÜÛœÙHHÚ[\S˜[Y\ÜXÙJˆÚ]TÚ[\S˜[Y\ÜXÙJˆÛÛ\][ÛœÏXÛÛ\][Û—Ü™\ÛÝ\˜ÙJ˜ÛY[<EFBFBD> \Ý™X[H‹Ý™X[Z[™×ØÜ™X]WÙ˜XÝÜžJBˆ
Bˆ
Bˆ™]\›ˆØ[‚‚™Yˆ™\ÜÛœÙWݘ\šX[<EFBFBD>ØÜ™X]\ÊÛY[<EFBFBD>ˆ[žK˜[YNˆÝŠH Oˆ\ÝÐ[žWN‚ˆ™]\›ˆˆÙ]]ŠÛY[<EFBFBD> ˜Ú] ˜ÛÛ\][ۜ˘[YJK˜Ü™X]KˆÙ]]ŠÛY[<EFBFBD> ˜Ú] ˜[YJK˜ÛÛ\][ۜ˘ܙX]KˆÙ]]ŠÛY[<EFBFBD> ˜[YJK˜Ú] ˜ÛÛ\][ۜ˘ܙX]KˆB‚‚”‘PSÐTÖS<EFBFBD>×ÐÓÓTUSÓ—ÔUÈH
ˆ››Ü›X[‹ˆ˜ÛÛ\][ۜ˜˜]È‹ˆ˜Ú] œ˜]È‹ˆ˜ÛY[<EFBFBD> œ˜]È‹ˆ˜ÛÛ\][ۜ˜Ý™X[Z[™È‹ˆ˜Ú] œÝ™X[Z[™È‹ˆ˜ÛY[<EFBFBD> œÝ™X[Z[™È‹ŠB‚‚™Yˆ™X[Ø\Þ[˜×ØÛÛ\][Û—ØÜ™X]JÛY[<EFBFBD>ˆ[žK]ˆÝŠH Oˆ[žN‚ˆYˆ]OH››Ü›X[Ž‚ˆ™]\›ˆÛY[<EFBFBD> ˜Ú] ˜ÛÛ\][ۜ˘ܙX]BˆYˆ]OH˜ÛÛ\][ۜ˜˜]ÈŽ‚ˆ™]\›ˆÛY[<EFBFBD> ˜Ú] ˜ÛÛ\][ÛœË<EFBFBD>Ú]ܘ]×Ü™\ÜÛœÙK˜Ü™X]BˆYˆ]OH˜Ú] œ˜]ÈŽ‚ˆ™]\›ˆÛY[<EFBFBD> ˜Ú] <EFBFBD>Ú]ܘ]×Ü™\ÜÛœÙK˜ÛÛ\][ۜ˘ܙX]BˆYˆ]OH˜ÛY[<EFBFBD> œ˜]ÈŽ‚ˆ™]\›ˆÛY[<EFBFBD> <EFBFBD>Ú]ܘ]×Ü™\ÜÛœÙK˜Ú] ˜ÛÛ\][ۜ˘ܙX]BˆYˆ]OH˜ÛÛ\][ۜ˜Ý™X[Z[™ÈŽ‚ˆ™]\›ˆÛY[<EFBFBD> ˜Ú] ˜ÛÛ\][ÛœË<EFBFBD>Ú]ÜÝ™X[Z[™×Ü™\ÜÛœÙK˜Ü™X]BˆYˆ]OH˜Ú] œÝ™X[Z[™ÈŽ‚ˆ™]\›ˆÛY[<EFBFBD> ˜Ú] <EFBFBD>Ú]ÜÝ™X[Z[™×Ü™\ÜÛœÙK˜ÛÛ\][ۜ˘ܙX]BˆYˆ]OH˜ÛY[<EFBFBD> œÝ™X[Z[™ÈŽ‚ˆ™]\›ˆÛY[<EFBFBD> <EFBFBD>Ú]ÜÝ™X[Z[™×Ü™\ÜÛœÙK˜Ú] ˜ÛÛ\][ۜ˘ܙX]Bˆ˜Z\ÙH\ÜÙ\<EFBFBD>[Û‘\œ›ÜŠˆ•[šÛ›ÝÛˆÛÛ\][Ûˆ]ˆÜ]HŠB‚‚<EFBFBD>]\Ý ™š^\™J]]Ý\ÙOU<EFBFBD>YJHÈ\NˆYۛܙVÝ[<EFBFBD>\Y YXÛܘ]Ü—B™YˆÝ\\›Y[[ÜžWØ\WÚÙ^J
H OˆÙ[™\˜]Ü–Ó›Û™K›Û™K›Û™WN‚ˆÚ]]Ú ™XÝ
ÜË™[<EFBFBD>š\›Û‹È”ÕTT“QSSÔ–WÐTWÒÑVHŽˆ<EFBFBD>\Ý ZÙ^HŸJN‚ˆZY[‚‚™Yˆ\ÝÜÚ\™YÜÞ[˜×ØÛY[<EFBFBD>ÙÙ\×Û›ÝÜÝXÚ×Ý[˜[<EFBFBD>ÛZY]Ø\™J
H Oˆ›Û™N‚ˆ˜\ÙWØÛY[<EFBFBD> ÜšYÚ[˜[ØÜ™X]HHÜ™X]WÜÞ[˜×ØÛY[<EFBFBD>
chat_resource = client_resource.chat
completions_resource = chat_resource.completions
return _ResourceFacade(
client_resource,
chat=_ResourceFacade(
chat_resource,
completions=_ResourceFacade(
completions_resource,
create=self._create_completion_method(
completions_resource.create,
streaming_response=name == "with_streaming_response",
),
),
),
)
def _create_completion_method(
self,
original_create: Any,
*,
streaming_response: bool = False,
) -> Any:
"""Wrap one completion create implementation with memory injection."""
if self._is_async_client and streaming_response:
BˆÛÚÝ\Έ\ÝÜÝ—HH×B‚ˆ\Þ[˜ÈYˆ˜ZÙWÜ›Û\
ˆY\ÜØYÙ\Έ\ÝÐ[žWKˆÛÛ<EFBFBD>Z[™\—ÝYΈÝ‹ˆÙÙÙ\Žˆ[žKˆ[ÙNˆ[žKˆ\WÚÙ^NˆÝ‹ˆ
H Oˆ\ÝÐ[žWN‚ˆÛÚÝ\˘\[™
ÛÛ<EFBFBD>Z[™\—ÝYÊBˆ™]\›ˆˆÈœ›ÛHŽˆœÞ\Ý[H‹˜ÛÛ<EFBFBD>[<EFBFBD>ŽˆˆœÙXÜ™] ^ØÛÛ<EFBFBD>Z[™\—ÝYßHŸKˆ
›Y\ÜØYÙ\ˈB‚ˆÚ]]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™KœÝ\\›Y[[ÜžK”Ý\\›Y[[ÜžH‹ˆ™]\›—ݘ[YOS[ØÚÊ
Kˆ
K]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™K˜YÜÞ\Ý[WÜ›Û\‹ˆÚYWÙY™™XÝY˜ZÙWÜ›Û\ ˆ
N‚ˆ[˜[<EFBFBD>ØNˆ[žHHÚ]ÜÝ\\›Y[[ÜžJˆ˜\ÙWØÛY[<EFBFBD> ˆZY]Ø\™WÛÜ[ÛœÊ<EFBFBD>[˜[<EFBFBD> XHŠKˆ
Bˆ[˜[<EFBFBD>ØŽˆ[žHHÚ]ÜÝ\\›Y[[ÜžJˆ˜\ÙWØÛY[<EFBFBD> ˆZY]Ø\™WÛÜ[ÛœÊ<EFBFBD>[˜[<EFBFBD> XˆŠKˆ
B‚ˆ\ÜÙ\<EFBFBD>˜\ÙWØÛY[<EFBFBD> ˜Ú] ˜ÛÛ\][ۜ˘ܙX]H\ÈÜšYÚ[˜[ØÜ™X]Bˆ\ÜÙ\<EFBFBD>[˜[<EFBFBD>ØK˜Ú]\țݘ\ÙWØÛY[<EFBFBD> ˜Ú]ˆ\ÜÙ\<EFBFBD>[˜[<EFBFBD>Ø‹˜Ú] ›X\šÙ\ˆOH˜Ú] [X\šÙ\ˆ‚ˆ\ÜÙ\<EFBFBD>[˜[<EFBFBD>Ø‹˜Ú] ˜ÛÛ\][ۜ˛X\šÙ\ˆOH˜ÛÛ\][ÛœË[X\šÙ\ˆ‚‚ˆ[˜[<EFBFBD>Ø‹˜Ú] ˜ÛÛ\][ۜ˘ܙX]Jˆ[Ù[H™Ü ]\Ý‹ˆY\ÜØYÙ\ÏVÞÈœ›ÛHŽˆ<EFBFBD>\Ù\ˆ‹˜ÛÛ<EFBFBD>[<EFBFBD>Žˆœš]˜]HŸWKˆ
B‚ˆ\ÜÙ\<EFBFBD>ÛÚÝ\ÈOHÈ<EFBFBD>[˜[<EFBFBD> Xˆ—BˆY\ÜØYÙ\ÈHÜšYÚ[˜[ØÜ™X]K˜Ø[Ø\™ÜËšÝØ\™ÜÖÈ›Y\ÜØYÙ\È—Bˆ\ÜÙ\<EFBFBD>Y\ÜØYÙ\ÖÌVȘÛÛ<EFBFBD>[<EFBFBD>—HOHœÙXÜ™] ][˜[<EFBFBD> Xˆ‚ˆ\ÜÙ\<EFBFBD>[
<EFBFBD>[˜[<EFBFBD> XHˆ›Ý[ˆÝŠY\ÜØYÙJH›ÜˆY\ÜØYÙH[ˆY\ÜØYÙ\ÊB‚ˆ˜\ÙWØÛY[<EFBFBD> ˜Ú] ˜ÛÛ\][ۜ˘ܙX]Jˆ[Ù[H™Ü ]\Ý‹ˆY\ÜØYÙ\ÏVÞÈœ›ÛHŽˆ<EFBFBD>\Ù\ˆ‹˜ÛÛ<EFBFBD>[<EFBFBD>Žˆ<EFBFBD>[<EFBFBD>ܘ\YŸWKˆ
Bˆ\ÜÙ\<EFBFBD>ÛÚÝ\ÈOHÈ<EFBFBD>[˜[<EFBFBD> Xˆ—B‚‚™Yˆ\ÝÜ™]ܘ\[™×ØWÙ˜XØYWÜ™XÛÝ™\œ×ÝWÜš\Ý[™WØÛY[<EFBFBD>
def create_streaming_with_memory(
**kwargs: Any,
) -> _AsyncMemoryResponseContextManager:
return _AsyncMemoryResponseContextManager(
self,
original_create,
kwargs,
)
H Oˆ›Û™N‚ˆ˜\ÙWØÛY[<EFBFBD> ÜšYÚ[˜[ØÜ™X]HHÜ™X]WÜÞ[˜×ØÛY[<EFBFBD>
return create_streaming_with_memory
BˆÛÚÝ\Έ\ÝÜÝ—HH×B‚ˆ\Þ[˜ÈYˆ˜ZÙWÜ›Û\
ˆY\ÜØYÙ\Έ\ÝÐ[žWKˆÛÛ<EFBFBD>Z[™\—ÝYΈÝ‹ˆÙÙÙ\Žˆ[žKˆ[ÙNˆ[žKˆ\WÚÙ^NˆÝ‹ˆ
H Oˆ\ÝÐ[žWN‚ˆÛÚÝ\˘\[™
ÛÛ<EFBFBD>Z[™\—ÝYÊBˆ™]\›ˆˆÈœ›ÛHŽˆœÞ\Ý[H‹˜ÛÛ<EFBFBD>[<EFBFBD>ŽˆˆœÙXÜ™] ^ØÛÛ<EFBFBD>Z[™\—ÝYßHŸKˆ
›Y\ÜØYÙ\ˈB‚ˆÚ]]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™KœÝ\\›Y[[ÜžK”Ý\\›Y[[ÜžH‹ˆ™]\›—ݘ[YOS[ØÚÊ
Kˆ
K]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™K˜YÜÞ\Ý[WÜ›Û\‹ˆÚYWÙY™™XÝY˜ZÙWÜ›Û\ ˆ
N‚ˆ[˜[<EFBFBD>ØNˆ[žHHÚ]ÜÝ\\›Y[[ÜžJˆ˜\ÙWØÛY[<EFBFBD> ˆZY]Ø\™WÛÜ[ÛœÊ<EFBFBD>[˜[<EFBFBD> XHŠKˆ
Bˆ[˜[<EFBFBD>ØŽˆ[žHHÚ]ÜÝ\\›Y[[ÜžJˆ[˜[<EFBFBD>ØKˆZY]Ø\™WÛÜ[ÛœÊ<EFBFBD>[˜[<EFBFBD> XˆŠKˆ
B‚ˆ[˜[<EFBFBD>Ø‹˜Ú] ˜ÛÛ\][ۜ˘ܙX]Jˆ[Ù[H™Ü ]\Ý‹ˆY\ÜØYÙ\ÏVÞÈœ›ÛHŽˆ<EFBFBD>\Ù\ˆ‹˜ÛÛ<EFBFBD>[<EFBFBD>Žˆœš]˜]HŸWKˆ
B‚ˆ\ÜÙ\<EFBFBD>ÛÚÝ\ÈOHÈ<EFBFBD>[˜[<EFBFBD> Xˆ—Bˆ\ÜÙ\<EFBFBD>ÜšYÚ[˜[ØÜ™X]K˜Ø[ØÛÝ[<EFBFBD>OH Bˆ\ÜÙ\<EFBFBD>˜\ÙWØÛY[<EFBFBD> ˜Ú] ˜ÛÛ\][ۜ˘ܙX]H\ÈÜšYÚ[˜[ØÜ™X]B‚‚™Yˆ\Ýܘ]ר[™ÜÝ™X[Z[™×Ü™\ÜÛœÙWÙ˜XØY\×Ü™[XZ[—ÛY[[ÜžWØ]Ø\™J
H Oˆ›Û™N‚ˆ˜\ÙWØÛY[<EFBFBD> ÈHÜ™X]WÜÞ[˜×ØÛY[<EFBFBD>
if self._is_async_client:
BˆØ[ÈH]XÚÜ™\ÜÛœÙWݘ\šX[<EFBFBD>ʈ˜\ÙWØÛY[<EFBFBD> ˆ[X™HX™[ˆ[ØÚÊ™]\›—ݘ[YOT˜]Ô™\ÜÛœÙJX™[
JKˆ[X™HX™[ˆ[ØÚÊ™]\›—ݘ[YOTÞ[˜ÔÝ™X[PÛÛ<EFBFBD>^
X™[
JKˆ
BˆÛÚÝ\Έ\ÝÜÝ—HH×B‚ˆ\Þ[˜ÈYˆ˜ZÙWÜ›Û\
ˆY\ÜØYÙ\Έ\ÝÐ[žWKˆÛÛ<EFBFBD>Z[™\—ÝYΈÝ‹ˆÙÙÙ\Žˆ[žKˆ[ÙNˆ[žKˆ\WÚÙ^NˆÝ‹ˆ
H Oˆ\ÝÐ[žWN‚ˆÛÚÝ\˘\[™
ÛÛ<EFBFBD>Z[™\—ÝYÊBˆ™]\›ˆˆÈœ›ÛHŽˆœÞ\Ý[H‹˜ÛÛ<EFBFBD>[<EFBFBD>ŽˆˆœÙXÜ™] ^ØÛÛ<EFBFBD>Z[™\—ÝYßHŸKˆ
›Y\ÜØYÙ\ˈB‚ˆÚ]]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™KœÝ\\›Y[[ÜžK”Ý\\›Y[[ÜžH‹ˆ™]\›—ݘ[YOS[ØÚÊ
Kˆ
K]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™K˜YÜÞ\Ý[WÜ›Û\‹ˆÚYWÙY™™XÝY˜ZÙWÜ›Û\ ˆ
N‚ˆ[˜[<EFBFBD>ØŽˆ[žHHÚ]ÜÝ\\›Y[[ÜžJˆ˜\ÙWØÛY[<EFBFBD> ˆZY]Ø\™WÛÜ[ÛœÊ<EFBFBD>[˜[<EFBFBD> XˆŠKˆ
B‚ˆ›ÜˆÜ™X]H[ˆ™\ÜÛœÙWݘ\šX[<EFBFBD>ØÜ™X]\Ê[˜[<EFBFBD>Ø‹<EFBFBD>Ú]ܘ]×Ü™\ÜÛœÙHŠN‚ˆ™\ÜÛœÙHHÜ™X]Jˆ[Ù[H™Ü ]\Ý‹ˆY\ÜØYÙ\ÏVÞÈœ›ÛHŽˆ<EFBFBD>\Ù\ˆ‹˜ÛÛ<EFBFBD>[<EFBFBD>Žˆœ˜]ÈŸWKˆ
Bˆ\ÜÙ\<EFBFBD>™\ÜÛœÙKœ\œÙJ
KœÝ\<EFBFBD>ÝÚ]
œ\œÙY HŠB‚ˆ›ÜˆÜ™X]H[ˆ™\ÜÛœÙWݘ\šX[<EFBFBD>ØÜ™X]\Ê[˜[<EFBFBD>Ø‹<EFBFBD>Ú]ÜÝ™X[Z[™×Ü™\ÜÛœÙHŠN‚ˆÚ]Ü™X]Jˆ[Ù[H™Ü ]\Ý‹ˆY\ÜØYÙ\ÏVÞÈœ›ÛHŽˆ<EFBFBD>\Ù\ˆ‹˜ÛÛ<EFBFBD>[<EFBFBD>ŽˆœÝ™X[Z[™ÈŸWKˆ
H\ÈÝ™X[N‚ˆ\ÜÙ\<EFBFBD>Ý™X[K›X™[ ™[™ÝÚ]
œÝ™X[HŠB‚ˆ\ÜÙ\<EFBFBD>ÛÚÝ\ÈOHÈ<EFBFBD>[˜[<EFBFBD> Xˆ—H
ˆ ‚ˆ›ÜˆÜ™X]H[ˆØ[Ë<EFBFBD>˜[Y\Ê
N‚ˆ\ÜÙ\<EFBFBD>œÙXÜ™] ][˜[<EFBFBD> Xˆˆ[ˆÝŠÜ™X]K˜Ø[Ø\™ÜËšÝØ\™ÜÖÈ›Y\ÜØYÙ\È—JB‚‚™Yˆ\ÝØ\Þ[˜×ܘ]ר[™ÜÝ™X[Z[™×Ü™\ÜÛœÙWÜ™Yš^\×Ü™[XZ[—ÛY[[ÜžWØ]Ø\™J
H Oˆ›Û™N‚ˆ˜\ÙWØÛY[<EFBFBD> ÈHÜ™X]WØ\Þ[˜×ØÛY[<EFBFBD>
async def create_async_with_memory(
**kwargs: Any,
) -> Any:
return await self._create_with_memory_async(original_create, **kwargs)
BˆØ[ÈH]XÚÜ™\ÜÛœÙWݘ\šX[<EFBFBD>ʈ˜\ÙWØÛY[<EFBFBD> ˆ[X™HX™[ˆ\Þ[˜Ó[ØÚÊ™]\›—ݘ[YOT˜]Ô™\ÜÛœÙJX™[
JKˆ[X™HX™[ˆ[ØÚÊ™]\›—ݘ[YOP\Þ[˜ÔÝ™X[PÛÛ<EFBFBD>^
X™[
JKˆ
BˆÛÚÝ\Έ\ÝÜÝ—HH×B‚ˆ\Þ[˜ÈYˆ˜ZÙWÜ›Û\
ˆY\ÜØYÙ\Έ\ÝÐ[žWKˆÛÛ<EFBFBD>Z[™\—ÝYΈÝ‹ˆÙÙÙ\Žˆ[žKˆ[ÙNˆ[žKˆ\WÚÙ^NˆÝ‹ˆ
H Oˆ\ÝÐ[žWN‚ˆÛÚÝ\˘\[™
ÛÛ<EFBFBD>Z[™\—ÝYÊBˆ™]\›ˆˆÈœ›ÛHŽˆœÞ\Ý[H‹˜ÛÛ<EFBFBD>[<EFBFBD>ŽˆˆœÙXÜ™] ^ØÛÛ<EFBFBD>Z[™\—ÝYßHŸKˆ
›Y\ÜØYÙ\ˈB‚ˆ\Þ[˜ÈYˆØ[ܘ]ÊÜ™X]Nˆ[žJH Oˆ›Û™N‚ˆ™\ÜÛœÙHH]ØZ]Ü™X]Jˆ[Ù[H™Ü ]\Ý‹ˆY\ÜØYÙ\ÏVÞÈœ›ÛHŽˆ<EFBFBD>\Ù\ˆ‹˜ÛÛ<EFBFBD>[<EFBFBD>Žˆœ˜]ÈŸWKˆ
Bˆ\ÜÙ\<EFBFBD>™\ÜÛœÙKœ\œÙJ
KœÝ\<EFBFBD>ÝÚ]
œ\œÙY HŠB‚ˆ\Þ[˜ÈYˆÛÛœÝ[YWÜÝ™X[JÝ™X[WØÛÛ<EFBFBD>^ˆ[žJH Oˆ›Û™N‚ˆ\Þ[˜ÈÚ]Ý™X[WØÛÛ<EFBFBD>^\ÈÝ™X[N‚ˆ\ÜÙ\<EFBFBD>Ý™X[K›X™[ ™[™ÝÚ]
œÝ™X[HŠB‚ˆÚ]]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™KœÝ\\›Y[[ÜžK”Ý\\›Y[[ÜžH‹ˆ™]\›—ݘ[YOS[ØÚÊ
Kˆ
K]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™K˜YÜÞ\Ý[WÜ›Û\‹ˆÚYWÙY™™XÝY˜ZÙWÜ›Û\ ˆ
N‚ˆ[˜[<EFBFBD>ØŽˆ[žHHÚ]ÜÝ\\›Y[[ÜžJˆ˜\ÙWØÛY[<EFBFBD> ˆZY]Ø\™WÛÜ[ÛœÊ<EFBFBD>[˜[<EFBFBD> XˆŠKˆ
B‚ˆ›ÜˆÜ™X]H[ˆ™\ÜÛœÙWݘ\šX[<EFBFBD>ØÜ™X]\Ê[˜[<EFBFBD>Ø‹<EFBFBD>Ú]ܘ]×Ü™\ÜÛœÙHŠN‚ˆ\Þ[˜Ú[Ëœ<EFBFBD>[ŠØ[ܘ]ÊÜ™X]JJB‚ˆ›ÜˆÜ™X]H[ˆ™\ÜÛœÙWݘ\šX[<EFBFBD>ØÜ™X]\Ê[˜[<EFBFBD>Ø‹<EFBFBD>Ú]ÜÝ™X[Z[™×Ü™\ÜÛœÙHŠN‚ˆÝ™X[WØÛÛ<EFBFBD>^HÜ™X]Jˆ[Ù[H™Ü ]\Ý‹ˆY\ÜØYÙ\ÏVÞÈœ›ÛHŽˆ<EFBFBD>\Ù\ˆ‹˜ÛÛ<EFBFBD>[<EFBFBD>ŽˆœÝ™X[Z[™ÈŸWKˆ
Bˆ\Þ[˜Ú[Ëœ<EFBFBD>[ŠÛÛœÝ[YWÜÝ™X[JÝ™X[WØÛÛ<EFBFBD>^
JB‚ˆ\ÜÙ\<EFBFBD>ÛÚÝ\ÈOHÈ<EFBFBD>[˜[<EFBFBD> Xˆ—H
ˆ ‚ˆ›ÜˆÜ™X]H[ˆØ[Ë<EFBFBD>˜[Y\Ê
N‚ˆ\ÜÙ\<EFBFBD>œÙXÜ™] ][˜[<EFBFBD> Xˆˆ[ˆÝŠÜ™X]K˜Ø[Ø\™ÜËšÝØ\™ÜÖÈ›Y\ÜØYÙ\È—JB‚‚<EFBFBD>]\Ý ›X\šËœ\˜[Y]š^™Jœ]‹‘PSÐTÖS<EFBFBD>×ÐÓÓTUSÓ—ÔUÊB<EFBFBD>]\Ý ›X\šË˜\Þ[˜Ú[˜\Þ[˜ÈYˆ\ÝÜ™X[Ø\Þ[˜×ÛÜ[˜ZWÜ]×Ý\ÙWØ\Þ[˜×ÛZY]Ø\™J]ˆÝŠH Oˆ›Û™N‚ˆÛÚÝ\Έ\ÝÜÝ—HH×BˆÜš]\Έ\ÝÝ\VÜÝ‹Ü[Û˜[ÜÝ—KÝ—WHH×BˆÙ[<EFBFBD>ÛY\ÜØYÙ\Έ\ÝÛ\ÝÐ[žWWHH×B‚ˆ\Þ[˜ÈYˆ˜ZÙWÜ›Û\
ˆY\ÜØYÙ\Έ\ÝÐ[žWKˆÛÛ<EFBFBD>Z[™\—ÝYΈÝ‹ˆÙÙÙ\Žˆ[žKˆ[ÙNˆ[žKˆ\WÚÙ^NˆÝ‹ˆ
H Oˆ\ÝÐ[žWN‚ˆÛÚÝ\˘\[™
ÛÛ<EFBFBD>Z[™\—ÝYÊBˆ™]\›ˆˆÈœ›ÛHŽˆœÞ\Ý[H‹˜ÛÛ<EFBFBD>[<EFBFBD>ŽˆˆœÙXÜ™] ^ØÛÛ<EFBFBD>Z[™\—ÝYßHŸKˆ
›Y\ÜØYÙ\ˈB‚ˆ\Þ[˜ÈYˆ˜ZÙWØYÛY[[ÜžJˆÛY[<EFBFBD>ˆ[žKˆÛÛ<EFBFBD>Z[™\—ÝYΈÝ‹ˆÛÛ<EFBFBD>[<EFBFBD>ˆÝ‹ˆÝ\ÝÛWÚYˆÜ[Û˜[ÜÝ—KˆÙÙÙ\Žˆ[žKˆ
H Oˆ›Û™N‚ˆÜš]\˘\[™
return create_async_with_memory
ÛÛ<EFBFBD>Z[™\—ÝYËÝ\ÝÛWÚY ÛÛ<EFBFBD>[<EFBFBD>
JB‚ˆYˆ[™WÜ™\]Y\Ý
™\]Y\݈ ”™\]Y\Ý
H Oˆ ”™\ÜÛœÙN‚ˆ›ÙHHœÛÛ‹›ØYÊ™\]Y\Ý ˜ÛÛ<EFBFBD>[<EFBFBD>
BˆÙ[<EFBFBD>ÛY\ÜØYÙ\˘\[™
›ÙVÈ›Y\ÜØYÙ\È—JBˆ™]\›ˆ ”™\ÜÛœÙJˆ Œ ˆ™\]Y\Ý\™\]Y\Ý ˆXY\œÏ^ȘÛÛ<EFBFBD>[<EFBFBD> ]\HŽˆ˜\XØ][Û‹ÚœÛÛˆŸKˆœÛÛ<EFBFBD>^ˆšYŽˆ˜Ú]Û\ ]\Ý‹ˆ›Øš™XÝŽˆ˜Ú] ˜ÛÛ\][Ûˆ‹ˆ˜Ü™X]YŽˆ ˆ›[Ù[Žˆ™Ü ]\Ý‹ˆ˜ÚÚXÙ\ÈŽˆˆˆš[™^Žˆ ˆ›Y\ÜØYÙHŽˆÈœ›ÛHŽˆ˜\ÜÚ\Ý[<EFBFBD>‹˜ÛÛ<EFBFBD>[<EFBFBD>Žˆ›ÚÈŸKˆ™š[š\ÚÜ™X\ÛÛˆŽˆœÝÜ‹ˆBˆKˆKˆ
B‚ˆØÛY[<EFBFBD>H <EFBFBD>\Þ[˜ÐÛY[<EFBFBD>
˜[œÜÜ<EFBFBD>Z “[ØÚÕ˜[œÜÜ<EFBFBD>
[™WÜ™\]Y\Ý
JBˆ˜\ÙWØÛY[<EFBFBD>H\Þ[˜ÓÜ[<EFBFBD>RJ\WÚÙ^OH›Ü[˜ZK]\Ý‹ØÛY[<EFBFBD>ZØÛY[<EFBFBD>
B‚ˆžN‚ˆÚ]]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™KœÝ\\›Y[[ÜžK”Ý\\›Y[[ÜžH‹ˆ™]\›—ݘ[YOS[ØÚÊ
Kˆ
K]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™K˜YÜÞ\Ý[WÜ›Û\‹ˆ™]ÏY˜ZÙWÜ›Û\ ˆ
K]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™K˜YÛY[[ÜžWÝÛÛ‹ˆ™]ÏY˜ZÙWØYÛY[[ÜžKˆ
KØ\›š[™Ü˘Ø]ÚÝØ\›š[™Üʈ™XÛÜ™U<EFBFBD>YBˆ
H\ÈØ]YÚ‚ˆØ\›š[™ÜËœÚ[\Yš[\Š˜[Ø^\È‹<EFBFBD>[<EFBFBD>[YUØ\›š[™ÊBˆܘ\Yˆ[žHHÚ]ÜÝ\\›Y[[ÜžJˆ˜\ÙWØÛY[<EFBFBD> ˆZY]Ø\™WÛÜ[ÛœÊ<EFBFBD>[˜[<EFBFBD> \™X[‹YÛY[[ÜžOH˜[Ø^\ÈŠKˆ
BˆÜ™X]HH™X[Ø\Þ[˜×ØÛÛ\][Û—ØÜ™X]Jܘ\Y ]
BˆÝØ\™ÜÈHˆ›[Ù[Žˆ™Ü ]\Ý‹ˆ›Y\ÜØYÙ\ÈŽˆÞÈœ›ÛHŽˆ<EFBFBD>\Ù\ˆ‹˜ÛÛ<EFBFBD>[<EFBFBD>Žˆœš]˜]HY\ÜØYÙHŸWKˆB‚ˆYˆ]OH››Ü›X[Ž‚ˆ™\ÜÛœÙHH]ØZ]Ü™X]J
ŠšÝØ\™ÜÊBˆ\ÜÙ\<EFBFBD>™\ÜÛœÙKšYOH˜Ú]Û\ ]\Ý‚ˆ[Yˆ] ™[™ÝÚ]
‹œ˜]ÈŠN‚ˆ˜]×Ü™\ÜÛœÙHH]ØZ]Ü™X]J
ŠšÝØ\™ÜÊBˆ\ÜÙ\<EFBFBD>˜]×Ü™\ÜÛœÙKœ\œÙJ
KšYOH˜Ú]Û\ ]\Ý‚ˆ[ÙN‚ˆ™\ÜÛœÙWØÛÛ<EFBFBD>^HÜ™X]J
ŠšÝØ\™ÜÊBˆ\ÜÙ\<EFBFBD>›Ý[œÜXÝ š\Ø]ØZ]X›J™\ÜÛœÙWØÛÛ<EFBFBD>^
Bˆ\Þ[˜ÈÚ]™\ÜÛœÙWØÛÛ<EFBFBD>^\ÈÝ™X[Z[™×Ü™\ÜÛœÙN‚ˆ\ÜÙ\<EFBFBD>Ý™X[Z[™×Ü™\ÜÛœÙKœÝ]\רÛÙHOH Œ ‚ˆ]ØZ]ܘ\Y <EFBFBD>ØZ]ٛܗؘXÚÙÜ›Ý[™Ý\ÚÜÊ
Bˆ]ØZ]\Þ[˜Ú[ËœÛY\
BˆØË˜ÛÛXÝ
def create_sync_with_memory(
**kwargs: Any,
) -> Any:
return self._create_with_memory_sync(original_create, **kwargs)
B‚ˆ<EFBFBD>[<EFBFBD>[YWÝØ\›š[™ÜÈHˆØ\›š[™Âˆ›ÜˆØ\›š[™È[ˆØ]YÚˆYˆ\ÜÝX˜Û\ÜÊØ\›š[™Ë˜Ø]YÛÜžK<EFBFBD>[<EFBFBD>[YUØ\›š[™ÊBˆB‚ˆ\ÜÙ\<EFBFBD>ÛÚÝ\ÈOHÈ<EFBFBD>[˜[<EFBFBD> \™X[—Bˆ\ÜÙ\<EFBFBD>Üš]\ÈOHˆ
ˆ<EFBFBD>[˜[<EFBFBD> \™X[‹ˆ˜ÛÛ<EFBFBD>™\œØ][ÛŽ<EFBFBD>™XY ][˜[<EFBFBD> \™X[‹ˆ•\Ù\Žˆš]˜]HY\ÜØYÙH‹ˆ
BˆBˆ\ÜÙ\<EFBFBD>[ŠÙ[<EFBFBD>ÛY\ÜØYÙ\ÊHOH Bˆ\ÜÙ\<EFBFBD>Ù[<EFBFBD>ÛY\ÜØYÙ\ÖÌVÌVȘÛÛ<EFBFBD>[<EFBFBD>—HOHœÙXÜ™] ][˜[<EFBFBD> \™X[‚ˆ\ÜÙ\<EFBFBD><EFBFBD>[<EFBFBD>[YWÝØ\›š[™ÜÈOH×Bˆš[˜[N‚ˆ]ØZ]˜\ÙWØÛY[<EFBFBD> ˜ÛÜÙJ
B‚‚<EFBFBD>]\Ý ›X\šË˜\Þ[˜Ú[ÈÈ\NˆYۛܙVÝ[<EFBFBD>\Y YXÛܘ]Ü—B˜\Þ[˜ÈYˆ\ÝÜÚ\™YØ\Þ[˜×ØÛY[<EFBFBD>ÜØ]™\×ÛÛ›WٛܗÜÙ[XÝYÝ[˜[<EFBFBD>
H Oˆ›Û™N‚ˆ˜\ÙWØÛY[<EFBFBD> ÜšYÚ[˜[ØÜ™X]HHÜ™X]WØ\Þ[˜×ØÛY[<EFBFBD>
BˆÛÚÝ\Έ\ÝÜÝ—HH×BˆÜš]\Έ\ÝÝ\VÜÝ‹Ü[Û˜[ÜÝ—KÝ—WHH×B‚ˆ\Þ[˜ÈYˆ˜ZÙWÜ›Û\
ˆY\ÜØYÙ\Έ\ÝÐ[žWKˆÛÛ<EFBFBD>Z[™\—ÝYΈÝ‹ˆÙÙÙ\Žˆ[žKˆ[ÙNˆ[žKˆ\WÚÙ^NˆÝ‹ˆ
H Oˆ\ÝÐ[žWN‚ˆÛÚÝ\˘\[™
ÛÛ<EFBFBD>Z[™\—ÝYÊBˆ™]\›ˆˆÈœ›ÛHŽˆœÞ\Ý[H‹˜ÛÛ<EFBFBD>[<EFBFBD>ŽˆˆœÙXÜ™] ^ØÛÛ<EFBFBD>Z[™\—ÝYßHŸKˆ
›Y\ÜØYÙ\ˈB‚ˆ\Þ[˜ÈYˆ˜ZÙWØYÛY[[ÜžJˆÛY[<EFBFBD>ˆ[žKˆÛÛ<EFBFBD>Z[™\—ÝYΈÝ‹ˆÛÛ<EFBFBD>[<EFBFBD>ˆÝ‹ˆÝ\ÝÛWÚYˆÜ[Û˜[ÜÝ—KˆÙÙÙ\Žˆ[žKˆ
H Oˆ›Û™N‚ˆÜš]\˘\[™
ÛÛ<EFBFBD>Z[™\—ÝYËÝ\ÝÛWÚY ÛÛ<EFBFBD>[<EFBFBD>
JB‚ˆÚ]]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™KœÝ\\›Y[[ÜžK”Ý\\›Y[[ÜžH‹ˆ™]\›—ݘ[YOS[ØÚÊ
Kˆ
K]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™K˜YÜÞ\Ý[WÜ›Û\‹ˆÚYWÙY™™XÝY˜ZÙWÜ›Û\ ˆ
K]Ú
ˆœÝ\\›Y[[ÜžWÛÜ[˜ZK›ZY]Ø\™K˜YÛY[[ÜžWÝÛÛ‹ˆÚYWÙY™™XÝY˜ZÙWØYÛY[[ÜžKˆ
N‚ˆ[˜[<EFBFBD>ØNˆ[žHHÚ]ÜÝ\\›Y[[ÜžJˆ˜\ÙWØÛY[<EFBFBD> ˆZY]Ø\™WÛÜ[ÛœÊ<EFBFBD>[˜[<EFBFBD> XH‹YÛY[[ÜžOH˜[Ø^\ÈŠKˆ
Bˆ[˜[<EFBFBD>ØŽˆ[žHHÚ]ÜÝ\\›Y[[ÜžJˆ˜\ÙWØÛY[<EFBFBD> ˆZY]Ø\™WÛÜ[ÛœÊ<EFBFBD>[˜[<EFBFBD> Xˆ‹YÛY[[ÜžOH˜[Ø^\ÈŠKˆ
B‚ˆ]ØZ][˜[<EFBFBD>Ø‹˜Ú] ˜ÛÛ\][ۜ˘ܙX]Jˆ[Ù[H™Ü ]\Ý‹ˆY\ÜØYÙ\ÏVÞÈœ›ÛHŽˆ<EFBFBD>\Ù\ˆ‹˜ÛÛ<EFBFBD>[<EFBFBD>Žˆœš]˜]H[˜[<EFBFBD>ˆY\ÜØYÙHŸWKˆ
Bˆ]ØZ][˜[<EFBFBD>ØK<EFBFBD>ØZ]ٛܗؘXÚÙÜ›Ý[™Ý\ÚÜÊ
Bˆ]ØZ][˜[<EFBFBD>Ø‹<EFBFBD>ØZ]ٛܗؘXÚÙÜ›Ý[™Ý\ÚÜÊ
B‚ˆ\ÜÙ\<EFBFBD>˜\ÙWØÛY[<EFBFBD> ˜Ú] ˜ÛÛ\][ۜ˘ܙX]H\ÈÜšYÚ[˜[ØÜ™X]Bˆ\ÜÙ\<EFBFBD>ÛÚÝ\ÈOHÈ<EFBFBD>[˜[<EFBFBD> Xˆ—Bˆ\ÜÙ\<EFBFBD>Üš]\ÈOHˆ
ˆ<EFBFBD>[˜[<EFBFBD> Xˆ‹ˆ˜ÛÛ<EFBFBD>™\œØ][ÛŽ<EFBFBD>™XY ][˜[<EFBFBD> Xˆ‹ˆ•\Ù\Žˆš]˜]H[˜[<EFBFBD>ˆY\ÜØYÙH‹ˆ
BˆBˆ\ÜÙ\<EFBFBD>ÜšYÚ[˜[ØÜ™X]K˜Ø[ØÛÝ[<EFBFBD>OH B
return create_sync_with_memory
async def _create_with_memory_async(
self,
original_create: Any,
**kwargs: Any,
) -> Any:
"""Async version of create with memory injection."""
messages = kwargs.get("messages", [])
if self._options.add_memory == "always":
user_message = get_last_user_message(messages)
if user_message and user_message.strip():
content = (
get_conversation_content(messages)
if self._options.custom_id
else user_message
)
custom_id = (
f"conversation:{self._options.custom_id}"
if self._options.custom_id
else None
)
# Create background task for memory storage
task = asyncio.create_task(
add_memory_tool(
self._supermemory_client,
self._container_tag,
content,
custom_id,
self._logger,
)
)
# Track the task and set up cleanup
self._background_tasks.add(task)
task.add_done_callback(self._background_tasks.discard)
# Log any exceptions but don't fail the main request
def handle_task_exception(task_obj):
try:
if task_obj.exception() is not None:
exception = task_obj.exception()
if isinstance(
exception,
(SupermemoryNetworkError, SupermemoryAPIError),
):
self._logger.warn(
"Background memory storage failed",
{
"error": str(exception),
"type": type(exception).__name__,
},
)
else:
self._logger.error(
"Unexpected error in background memory storage",
{
"error": str(exception),
"type": type(exception).__name__,
},
)
except asyncio.CancelledError:
self._logger.debug("Memory storage task was cancelled")
task.add_done_callback(handle_task_exception)
if self._options.mode != "profile":
user_message = get_last_user_message(messages)
if not user_message:
self._logger.debug("No user message found, skipping memory search")
return await original_create(**kwargs)
self._logger.info(
"Starting memory search",
{
"container_tag": self._container_tag,
"conversation_id": self._options.custom_id,
"mode": self._options.mode,
},
)
enhanced_messages = await add_system_prompt(
messages,
self._container_tag,
self._logger,
self._options.mode,
self._get_api_key(),
)
kwargs["messages"] = enhanced_messages
return await original_create(**kwargs)
def _create_with_memory_sync(
self,
original_create: Any,
**kwargs: Any,
) -> Any:
"""Sync version of create with memory injection."""
# For sync clients, we implement a simplified version without background tasks
messages = kwargs.get("messages", [])
# Handle memory addition synchronously if needed
if self._options.add_memory == "always":
user_message = get_last_user_message(messages)
if user_message and user_message.strip():
content = (
get_conversation_content(messages)
if self._options.custom_id
else user_message
)
custom_id = (
f"conversation:{self._options.custom_id}"
if self._options.custom_id
else None
)
# Use asyncio.run() for the memory addition
try:
asyncio.run(
add_memory_tool(
self._supermemory_client,
self._container_tag,
content,
custom_id,
self._logger,
)
)
except RuntimeError as e:
if "cannot be called from a running event loop" in str(e):
# We're in an async context, log warning and skip memory saving
self._logger.warn(
"Cannot save memory in sync client from async context",
{"error": str(e)},
)
else:
raise
except SupermemoryNetworkError as e:
# Network errors are expected, log as warning
self._logger.warn("Network error saving memory", {"error": str(e)})
except (SupermemoryAPIError, SupermemoryMemoryOperationError) as e:
# API/memory errors are concerning, log as error
self._logger.error("Failed to save memory", {"error": str(e)})
except Exception as e:
# Unexpected errors should be investigated
self._logger.error(
"Unexpected error saving memory",
{"error": str(e), "type": type(e).__name__},
)
# Handle memory search and injection
if self._options.mode != "profile":
user_message = get_last_user_message(messages)
if not user_message:
self._logger.debug("No user message found, skipping memory search")
return original_create(**kwargs)
self._logger.info(
"Starting memory search",
{
"container_tag": self._container_tag,
"conversation_id": self._options.custom_id,
"mode": self._options.mode,
},
)
# Use asyncio.run() for memory search and injection
try:
enhanced_messages = asyncio.run(
add_system_prompt(
messages,
self._container_tag,
self._logger,
self._options.mode,
self._get_api_key(),
)
)
except RuntimeError as e:
if "cannot be called from a running event loop" in str(e):
# We're in an async context, run in a separate thread
import concurrent.futures
with concurrent.futures.ThreadPoolExecutor(max_workers=1) as executor:
future = executor.submit(
asyncio.run,
add_system_prompt(
messages,
self._container_tag,
self._logger,
self._options.mode,
self._get_api_key(),
),
)
enhanced_messages = future.result()
else:
raise
kwargs["messages"] = enhanced_messages
return original_create(**kwargs)
async def wait_for_background_tasks(self, timeout: Optional[float] = 10.0) -> None:
"""
Wait for all background memory storage tasks to complete.
Args:
timeout: Maximum time to wait in seconds. None for no timeout.
Raises:
asyncio.TimeoutError: If tasks don't complete within timeout
"""
if not self._background_tasks:
return
self._logger.debug(
f"Waiting for {len(self._background_tasks)} background tasks to complete"
)
try:
if timeout is not None:
await asyncio.wait_for(
asyncio.gather(*self._background_tasks, return_exceptions=True),
timeout=timeout,
)
else:
await asyncio.gather(*self._background_tasks, return_exceptions=True)
self._logger.debug("All background tasks completed")
except asyncio.TimeoutError:
self._logger.warn(
f"Background tasks did not complete within {timeout}s timeout"
)
# Cancel remaining tasks
tasks_to_cancel = [
task for task in self._background_tasks if not task.done()
]
for task in tasks_to_cancel:
task.cancel()
if tasks_to_cancel:
await asyncio.gather(*tasks_to_cancel, return_exceptions=True)
raise
def cancel_background_tasks(self) -> None:
"""Cancel all pending background tasks."""
cancelled_count = 0
for task in self._background_tasks:
if not task.done():
task.cancel()
cancelled_count += 1
if cancelled_count > 0:
self._logger.debug(f"Cancelled {cancelled_count} pending background tasks")
async def __aenter__(self):
"""Async context manager entry."""
return self
async def __aexit__(self, exc_type, exc_val, exc_tb):
"""Async context manager exit - wait for background tasks."""
try:
await self.wait_for_background_tasks(timeout=5.0)
except asyncio.TimeoutError:
self._logger.warn("Some background memory tasks did not complete on exit")
def __enter__(self):
"""Sync context manager entry."""
return self
def __exit__(self, exc_type, exc_val, exc_tb):
"""Sync context manager exit - attempt to wait for background tasks."""
if self._background_tasks:
try:
# Try to wait for background tasks in sync context
asyncio.run(self.wait_for_background_tasks(timeout=5.0))
except RuntimeError as e:
if "cannot be called from a running event loop" in str(e):
# In async context, just cancel the tasks
self._logger.warn(
"Cannot wait for background tasks in sync context from async environment. "
"Use async context manager or call wait_for_background_tasks() manually."
)
self.cancel_background_tasks()
else:
raise
except asyncio.TimeoutError:
self._logger.warn(
"Some background memory tasks did not complete on exit"
)
self.cancel_background_tasks()
def __getattr__(self, name: str) -> Any:
"""Delegate all other attributes to the wrapped client."""
if name in {"with_raw_response", "with_streaming_response"}:
value = self._create_client_variant_facade(name)
setattr(self, name, value)
return value
return getattr(self._client, name)
def with_supermemory(
openai_client: Union[OpenAI, AsyncOpenAI],
options: OpenAIMiddlewareOptions,
) -> Union[OpenAI, AsyncOpenAI]:
"""
Wraps an OpenAI client with SuperMemory middleware to automatically inject relevant memories
into the system prompt based on the user's message content.
This middleware searches the supermemory API for relevant memories using the container tag
and user message, then either appends memories to an existing system prompt or creates
a new system prompt with the memories.
Args:
openai_client: The OpenAI client to wrap with SuperMemory middleware
options: Configuration options for the middleware (container_tag and custom_id are required)
Returns:
An OpenAI client with SuperMemory middleware injected
Example:
```python
from supermemory_openai import with_supermemory, OpenAIMiddlewareOptions
from openai import OpenAI
# Create OpenAI client with supermemory middleware
openai = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
openai_with_supermemory = with_supermemory(
openai,
OpenAIMiddlewareOptions(
container_tag="user-123",
custom_id="conversation-456",
mode="full",
add_memory="always"
)
)
# Use normally - memories will be automatically injected
response = await openai_with_supermemory.chat.completions.create(
model="gpt-4",
messages=[
{"role": "user", "content": "What's my favorite programming language?"}
]
)
```
Raises:
ValueError: When SUPERMEMORY_API_KEY environment variable is not set
Exception: When supermemory API request fails
"""
wrapper = SupermemoryOpenAIWrapper(openai_client, options)
# Return the wrapper, which delegates all attributes to the original client
return cast(Union[OpenAI, AsyncOpenAI], wrapper)

View file

@ -3,12 +3,18 @@
from __future__ import annotations
import asyncio
import gc
import inspect
import json
import os
import warnings
from types import SimpleNamespace
from typing import Any, Generator, Literal, Optional
from unittest.mock import AsyncMock, Mock, patch
import httpx
import pytest
from openai import AsyncOpenAI
from supermemory_openai import OpenAIMiddlewareOptions, with_supermemory
@ -113,6 +119,35 @@ def response_variant_creates(client: Any, name: str) -> list[Any]:
]
REAL_ASYNC_COMPLETION_PATHS = (
"normal",
"completions.raw",
"chat.raw",
"client.raw",
"completions.streaming",
"chat.streaming",
"client.streaming",
)
def real_async_completion_create(client: Any, path: str) -> Any:
if path == "normal":
return client.chat.completions.create
if path == "completions.raw":
return client.chat.completions.with_raw_response.create
if path == "chat.raw":
return client.chat.with_raw_response.completions.create
if path == "client.raw":
return client.with_raw_response.chat.completions.create
if path == "completions.streaming":
return client.chat.completions.with_streaming_response.create
if path == "chat.streaming":
return client.chat.with_streaming_response.completions.create
if path == "client.streaming":
return client.with_streaming_response.chat.completions.create
raise AssertionError(f"Unknown completion path: {path}")
@pytest.fixture(autouse=True) # type: ignore[untyped-decorator]
def supermemory_api_key() -> Generator[None, None, None]:
with patch.dict(os.environ, {"SUPERMEMORY_API_KEY": "test-key"}):
@ -330,6 +365,121 @@ def test_async_raw_and_streaming_response_prefixes_remain_memory_aware() -> None
assert "secret-tenant-b" in str(create.call_args.kwargs["messages"])
@pytest.mark.parametrize("path", REAL_ASYNC_COMPLETION_PATHS)
@pytest.mark.asyncio
async def test_real_async_openai_paths_use_async_middleware(path: str) -> None:
lookups: list[str] = []
writes: list[tuple[str, Optional[str], str]] = []
sent_messages: list[list[Any]] = []
async def fake_prompt(
messages: list[Any],
container_tag: str,
logger: Any,
mode: Any,
api_key: str,
) -> list[Any]:
lookups.append(container_tag)
return [
{"role": "system", "content": f"secret-{container_tag}"},
*messages,
]
async def fake_add_memory(
client: Any,
container_tag: str,
content: str,
custom_id: Optional[str],
logger: Any,
) -> None:
writes.append((container_tag, custom_id, content))
def handle_request(request: httpx.Request) -> httpx.Response:
body = json.loads(request.content)
sent_messages.append(body["messages"])
return httpx.Response(
200,
request=request,
headers={"content-type": "application/json"},
json={
"id": "chatcmpl-test",
"object": "chat.completion",
"created": 0,
"model": "gpt-test",
"choices": [
{
"index": 0,
"message": {"role": "assistant", "content": "ok"},
"finish_reason": "stop",
}
],
},
)
http_client = httpx.AsyncClient(transport=httpx.MockTransport(handle_request))
base_client = AsyncOpenAI(api_key="openai-test", http_client=http_client)
try:
with patch(
"supermemory_openai.middleware.supermemory.Supermemory",
return_value=Mock(),
), patch(
"supermemory_openai.middleware.add_system_prompt",
new=fake_prompt,
), patch(
"supermemory_openai.middleware.add_memory_tool",
new=fake_add_memory,
), warnings.catch_warnings(
record=True
) as caught:
warnings.simplefilter("always", RuntimeWarning)
wrapped: Any = with_supermemory(
base_client,
middleware_options("tenant-real", add_memory="always"),
)
create = real_async_completion_create(wrapped, path)
kwargs = {
"model": "gpt-test",
"messages": [{"role": "user", "content": "private message"}],
}
if path == "normal":
response = await create(**kwargs)
assert response.id == "chatcmpl-test"
elif path.endswith(".raw"):
raw_response = await create(**kwargs)
assert raw_response.parse().id == "chatcmpl-test"
else:
response_context = create(**kwargs)
assert not inspect.isawaitable(response_context)
async with response_context as streaming_response:
assert streaming_response.status_code == 200
await wrapped.wait_for_background_tasks()
await asyncio.sleep(0)
gc.collect()
runtime_warnings = [
warning
for warning in caught
if issubclass(warning.category, RuntimeWarning)
]
assert lookups == ["tenant-real"]
assert writes == [
(
"tenant-real",
"conversation:thread-tenant-real",
"User: private message",
)
]
assert len(sent_messages) == 1
assert sent_messages[0][0]["content"] == "secret-tenant-real"
assert runtime_warnings == []
finally:
await base_client.close()
@pytest.mark.asyncio # type: ignore[untyped-decorator]
async def test_shared_async_client_saves_only_for_selected_tenant() -> None:
base_client, original_create = create_async_client()