mirror of
https://github.com/pewdiepie-archdaemon/odysseus.git
synced 2026-10-29 17:22:25 +01:00
Merge current dev and refresh PR #6738 from the 7011 integration
This commit is contained in:
pewdiepie-archdaemon
+113
-6
@@ -2,7 +2,8 @@
|
||||
import json
|
||||
import time
|
||||
import copy
|
||||
from dataclasses import dataclass
|
||||
import re
|
||||
from dataclasses import dataclass, replace
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
@@ -19,7 +20,7 @@ _DEPENDENCIES = {
|
||||
'web': {'web_search', 'web_fetch', 'private_browser'},
|
||||
'email': {'list_email_accounts', 'list_emails', 'search_emails', 'read_email',
|
||||
'download_attachment'},
|
||||
'contacts': {'resolve_contact'},
|
||||
'contacts': {'resolve_contact', 'manage_contact'},
|
||||
'documents': {'search_documents', 'read_document'},
|
||||
}
|
||||
_DRAFT_TOOLS = {'ask_user', 'update_plan', 'draft_email', 'draft_email_reply',
|
||||
@@ -116,8 +117,9 @@ def parse_email_task_intent(value):
|
||||
|
||||
def scope_email_tools(schemas, intent, *, active_editor=False):
|
||||
if intent.operation == 'read' and intent.dependencies:
|
||||
allowed = set().union(*(_DEPENDENCIES[d] for d in intent.dependencies))
|
||||
if active_editor:
|
||||
allowed = {tool for dependency in intent.dependencies
|
||||
for tool in _DEPENDENCIES[dependency]}
|
||||
if active_editor and intent.dependencies:
|
||||
allowed.update(_OPEN_EDITOR_TOOLS)
|
||||
if intent.needs_clarification:
|
||||
allowed.add('ask_user')
|
||||
@@ -126,6 +128,9 @@ def scope_email_tools(schemas, intent, *, active_editor=False):
|
||||
if intent.operation not in {'draft', 'revise'}:
|
||||
return list(schemas)
|
||||
allowed = _DRAFT_TOOLS.union(*(_DEPENDENCIES[d] for d in intent.dependencies))
|
||||
# manage_contact is a read-only runtime source for lookups, not a
|
||||
# composition tool. Drafts still resolve recipients through resolve_contact.
|
||||
allowed.discard('manage_contact')
|
||||
if active_editor:
|
||||
allowed.update(_OPEN_EDITOR_TOOLS)
|
||||
if not intent.needs_clarification:
|
||||
@@ -145,6 +150,62 @@ def scope_email_tools(schemas, intent, *, active_editor=False):
|
||||
# as though it were the whole task. This byte budget is deliberately conservative.
|
||||
CLASSIFIER_CONTEXT_BYTES = 24000
|
||||
|
||||
# The model can produce a structurally valid classification while forgetting
|
||||
# the source or mistaking a list of message headers for a request for bodies.
|
||||
_OWN_MAILBOX = re.compile(
|
||||
r'\b(?:my|our)\s+(?:(?:last|latest|recent|new|unread|read|first|most|\d+)\s+){0,5}'
|
||||
r'(?:e-?mails?|mail|inbox|mailbox)\b', re.IGNORECASE,
|
||||
)
|
||||
_MAILBOX_HEADER_LIST = re.compile(
|
||||
r'^\s*(?:list|show|display|what(?:\'s| are))\s+(?:me\s+)?(?:my|our)\s+'
|
||||
r'(?:(?:last|latest|recent|new|unread|read|first|most|\d+)\s+){0,5}'
|
||||
r'(?:e-?mails|messages)\s*'
|
||||
r'(?:with\s+(?:(?:sender|subject|date|time|recipient|status)\s*'
|
||||
r'(?:(?:,|and)\s*)?)+)?[?.!]*\s*$', re.IGNORECASE,
|
||||
)
|
||||
_EMAIL_HEADER_FIELDS = re.compile(r'\b(?:subjects?|senders?|dates?)\b', re.IGNORECASE)
|
||||
_EMAIL_HEADER_FOLLOWUP = re.compile(r'\b(?:those|these|them|their|same)\b', re.IGNORECASE)
|
||||
_READ_REQUEST = re.compile(r'^\s*(?:please\s+)?(?:show|list|display|find|search|read|what|how many|are there|do i have)\b', re.IGNORECASE)
|
||||
_EMAIL_COUNT = re.compile(r'^\s*how many\s+(?:(?:unread|read|new)\s+)?(?:e-?mails|messages)\b',
|
||||
re.IGNORECASE)
|
||||
_EMAIL_BODY_QUERY = re.compile(
|
||||
r'\b(?:mention(?:s|ed|ing)?|body|contents?|amount|owe|says?|said|when|where|why|'
|
||||
r'starts?|deadline|attachments?)\b|\bhow much\b|\bwhat time\b',
|
||||
re.IGNORECASE)
|
||||
_CONTACT_EMAIL_FIELD = re.compile(
|
||||
r'\bcontacts?\b[^.!?\n]{0,100}\bemail\b', re.IGNORECASE,
|
||||
)
|
||||
_MAILBOX_SOURCE = re.compile(
|
||||
r'\b(?:my|our)\s+(?:(?:last|latest|recent|new|unread|read|first|most|\d+)\s+){0,5}'
|
||||
r'(?:e-?mails?|inbox|mailbox)\b|\b(?:inbox|mailbox|email\s+history)\b|'
|
||||
r'\b(?:e-?mails?|messages?)\s+(?:from|to|sent|received)\b|'
|
||||
r'\bemailed?\s+(?:me|us)\b', re.IGNORECASE,
|
||||
)
|
||||
|
||||
EMAIL_TASK_RESPONSE_FORMAT = {
|
||||
'type': 'json_schema',
|
||||
'json_schema': {
|
||||
'name': 'email_task_intent',
|
||||
'strict': True,
|
||||
'schema': {
|
||||
'type': 'object',
|
||||
'additionalProperties': False,
|
||||
'properties': {
|
||||
'operation': {'type': 'string', 'enum': ['draft', 'revise', 'read', 'send', 'other']},
|
||||
'requires_content': {'type': 'boolean'},
|
||||
'dependencies': {'type': 'array', 'items': {
|
||||
'type': 'string', 'enum': list(_DEPENDENCIES),
|
||||
}},
|
||||
'destination': {'type': 'string', 'enum': ['chat', 'mailbox']},
|
||||
'needs_clarification': {'type': 'boolean'},
|
||||
'summary': {'type': 'string'},
|
||||
},
|
||||
'required': ['operation', 'requires_content', 'dependencies', 'destination',
|
||||
'needs_clarification', 'summary'],
|
||||
},
|
||||
},
|
||||
}
|
||||
|
||||
|
||||
async def classify_email_task(client, *, endpoint_url, headers, model, history,
|
||||
supplied_context=None, accounting=None):
|
||||
@@ -152,6 +213,8 @@ async def classify_email_task(client, *, endpoint_url, headers, model, history,
|
||||
# text from multimodal messages, so an attached image cannot hide the latest
|
||||
# instruction and leave us classifying an earlier task instead.
|
||||
dialogue = []
|
||||
latest_user_text = ''
|
||||
prior_user_text = ''
|
||||
for row in history:
|
||||
if row.get('role') not in {'user', 'assistant'} or row.get('_harness_control'):
|
||||
continue
|
||||
@@ -166,6 +229,9 @@ async def classify_email_task(client, *, endpoint_url, headers, model, history,
|
||||
and isinstance(block.get('text'), str))
|
||||
if isinstance(content, str):
|
||||
dialogue.append({'role': row['role'], 'content': content})
|
||||
if row['role'] == 'user':
|
||||
prior_user_text = latest_user_text
|
||||
latest_user_text = content
|
||||
payload = json.dumps({'dialogue': dialogue, 'supplied_context': supplied_context},
|
||||
ensure_ascii=False)
|
||||
if len(payload.encode('utf-8')) > CLASSIFIER_CONTEXT_BYTES:
|
||||
@@ -174,7 +240,7 @@ async def classify_email_task(client, *, endpoint_url, headers, model, history,
|
||||
response = await client.post(endpoint_url, headers=headers, timeout=20, json={
|
||||
'model': model, 'stream': False, 'temperature': 0, 'max_tokens': 500,
|
||||
'chat_template_kwargs': {'enable_thinking': False},
|
||||
'response_format': {'type': 'json_object'},
|
||||
'response_format': EMAIL_TASK_RESPONSE_FORMAT,
|
||||
'messages': [{'role': 'system', 'content': (
|
||||
'Classify the current conversational task. Return JSON only with operation '
|
||||
'(draft, revise, read, send, other), requires_content (boolean), dependencies (array containing only web, '
|
||||
@@ -250,6 +316,47 @@ async def classify_email_task(client, *, endpoint_url, headers, model, history,
|
||||
'response_time': round(time.monotonic() - started, 3),
|
||||
})
|
||||
try:
|
||||
return parse_email_task_intent(json.loads(body['choices'][0]['message']['content']))
|
||||
intent = parse_email_task_intent(json.loads(body['choices'][0]['message']['content']))
|
||||
if intent.operation == 'read' and not intent.needs_clarification:
|
||||
# In a contact-record question, "email" can name an address field
|
||||
# rather than the mailbox as a source. Keep both dependencies when
|
||||
# the request also names a mailbox/history source.
|
||||
contact_email_field = bool(
|
||||
_CONTACT_EMAIL_FIELD.search(latest_user_text)
|
||||
and not _MAILBOX_SOURCE.search(latest_user_text)
|
||||
)
|
||||
if contact_email_field:
|
||||
intent = replace(intent, dependencies=tuple(dict.fromkeys(
|
||||
(*[d for d in intent.dependencies if d != 'email'], 'contacts')
|
||||
)))
|
||||
# Reuse the product's existing source classifier rather than a
|
||||
# second vocabulary of mailbox phrases. The classifier has a known
|
||||
# gap for possessive single-message questions, so retain that
|
||||
# narrow fallback only when it found no competing source family.
|
||||
from src.turn_contract import requested_capabilities
|
||||
capabilities = requested_capabilities(latest_user_text)
|
||||
# A header-only reference can inherit the immediately preceding
|
||||
# mailbox read, but an intervening request or a competing source
|
||||
# breaks that link. Never infer a read from an earlier write.
|
||||
header_followup = (
|
||||
not capabilities
|
||||
and bool(_EMAIL_HEADER_FIELDS.search(latest_user_text))
|
||||
and bool(_EMAIL_HEADER_FOLLOWUP.search(latest_user_text))
|
||||
and bool(_READ_REQUEST.match(prior_user_text))
|
||||
and requested_capabilities(prior_user_text) == {'email'}
|
||||
)
|
||||
mailbox_read = not contact_email_field and (capabilities == {'email'} or (
|
||||
not capabilities and bool(_OWN_MAILBOX.search(latest_user_text))
|
||||
) or header_followup)
|
||||
else:
|
||||
mailbox_read = False
|
||||
if mailbox_read:
|
||||
intent = replace(intent, dependencies=tuple(dict.fromkeys((*intent.dependencies, 'email'))))
|
||||
if (not _EMAIL_BODY_QUERY.search(latest_user_text)
|
||||
and (_MAILBOX_HEADER_LIST.fullmatch(latest_user_text)
|
||||
or _EMAIL_HEADER_FIELDS.search(latest_user_text)
|
||||
or _EMAIL_COUNT.match(latest_user_text))):
|
||||
intent = replace(intent, requires_content=False)
|
||||
return intent
|
||||
except (KeyError, IndexError, TypeError) as exc:
|
||||
raise ValueError('Invalid classifier response') from exc
|
||||
|
||||
Reference in New Issue
Block a user