Auto-update: Thu Jun 27 13:16:34 PDT 2024

This commit is contained in:
sanj 2024-06-27 13:16:34 -07:00
parent 8f095e5e71
commit 12b4b53705
9 changed files with 225 additions and 103 deletions

View file

@ -12,7 +12,7 @@ from typing import List, Optional
import traceback
import logging
from .logs import Logger
from .classes import AutoResponder, IMAPConfig, SMTPConfig, EmailAccount, EmailContact, IncomingEmail, TimezoneTracker, Database
from .classes import AutoResponder, IMAPConfig, SMTPConfig, EmailAccount, EmailContact, IncomingEmail, TimezoneTracker, Database, PyGeolocator
# from sijapi.config.config import load_config
# cfg = load_config()
@ -56,7 +56,6 @@ os.makedirs(REQUESTS_DIR, exist_ok=True)
REQUESTS_LOG_PATH = LOGS_DIR / "requests.log"
### LOCATE AND WEATHER LOCALIZATIONS
USER_FULLNAME = os.getenv('USER_FULLNAME')
USER_BIO = os.getenv('USER_BIO')
@ -68,7 +67,8 @@ VISUALCROSSING_API_KEY = os.getenv("VISUALCROSSING_API_KEY")
GEONAMES_TXT = DATA_DIR / "geonames.txt"
LOCATIONS_CSV = DATA_DIR / "US.csv"
TZ = tz.gettz(os.getenv("TZ", "America/Los_Angeles"))
DynamicTZ = TimezoneTracker(DB)
TZ_CACHE = DATA_DIR / "tzcache.json"
DynamicTZ = TimezoneTracker(TZ_CACHE)
### Obsidian & notes
ALLOWED_FILENAME_CHARS = r'[^\w \.-]'
@ -90,13 +90,14 @@ YEAR_FMT = os.getenv("YEAR_FMT")
MONTH_FMT = os.getenv("MONTH_FMT")
DAY_FMT = os.getenv("DAY_FMT")
DAY_SHORT_FMT = os.getenv("DAY_SHORT_FMT")
GEOLOCATOR = PyGeolocator
### Large language model
LLM_URL = os.getenv("LLM_URL", "http://localhost:11434")
LLM_SYS_MSG = os.getenv("SYSTEM_MSG", "You are a helpful AI assistant.")
SUMMARY_INSTRUCT = os.getenv('SUMMARY_INSTRUCT', "You are an AI assistant that provides accurate summaries of text -- nothing more and nothing less. You must not include ANY extraneous text other than the sumary. Do not include comments apart from the summary, do not preface the summary, and do not provide any form of postscript. Do not add paragraph breaks. Do not add any kind of formatting. Your response should begin with, consist of, and end with an accurate plaintext summary.")
SUMMARY_INSTRUCT_TTS = os.getenv('SUMMARY_INSTRUCT_TTS', "You are an AI assistant that provides email summaries for Sanjay. Your response will undergo Text-To-Speech conversion and added to Sanjay's private podcast. Providing adequate context (Sanjay did not send this question to you, he will only hear your response) but aiming for conciseness and precision, and bearing in mind the Text-To-Speech conversion (avoiding acronyms and formalities), summarize the following email.")
DEFAULT_LLM = os.getenv("DEFAULT_LLM", "dolphin-mistral")
DEFAULT_LLM = os.getenv("DEFAULT_LLM", "llama3")
DEFAULT_VISION = os.getenv("DEFAULT_VISION", "llava")
DEFAULT_VOICE = os.getenv("DEFAULT_VOICE", "Luna")
DEFAULT_11L_VOICE = os.getenv("DEFAULT_11L_VOICE", "Victoria")

View file

@ -8,6 +8,11 @@ from pydantic import BaseModel, Field
from typing import Optional
import asyncpg
import os
from typing import Optional, Tuple, Union
from datetime import datetime, timedelta
import json
from timezonefinder import TimezoneFinder
from pathlib import Path
from pydantic import BaseModel, Field
from typing import Optional
@ -21,6 +26,38 @@ from typing import Optional
import asyncpg
from contextlib import asynccontextmanager
import reverse_geocoder as rg
from timezonefinder import TimezoneFinder
from srtm import get_data
class PyGeolocator:
def __init__(self):
self.tf = TimezoneFinder()
self.srtm_data = get_data()
def get_location(self, lat, lon):
result = rg.search((lat, lon))
return result[0]['name'], result[0]['admin1'], result[0]['cc']
def get_elevation(self, lat, lon):
return self.srtm_data.get_elevation(lat, lon)
def get_timezone(self, lat, lon):
return self.tf.timezone_at(lat=lat, lng=lon)
def lookup(self, lat, lon):
city, state, country = self.get_location(lat, lon)
elevation = self.get_elevation(lat, lon)
timezone = self.get_timezone(lat, lon)
return {
"city": city,
"state": state,
"country": country,
"elevation": elevation,
"timezone": timezone
}
class Database(BaseModel):
host: str = Field(..., description="Database host")
port: int = Field(5432, description="Database port")
@ -108,7 +145,6 @@ class IncomingEmail(BaseModel):
attachments: List[dict] = []
class Location(BaseModel):
latitude: float
longitude: float
@ -141,24 +177,18 @@ class Location(BaseModel):
}
class TimezoneTracker:
def __init__(self, db_config: Database, cache_file: str = 'timezone_cache.json'):
self.db_config = db_config
self.cache_file = cache_file
def __init__(self, cache_file: Union[str, Path] = 'timezone_cache.json'):
self.cache_file = Path(cache_file)
self.last_timezone: str = "America/Los_Angeles"
self.last_update: Optional[datetime] = None
self.last_location: Optional[Tuple[float, float]] = None
self.tf = TimezoneFinder()
async def find(self, lat: float, lon: float) -> str:
query = """
SELECT tzid
FROM timezones
WHERE ST_Contains(geom, ST_SetSRID(ST_MakePoint($1, $2), 4326))
LIMIT 1;
"""
async with await self.db_config.get_connection() as conn:
result = await conn.fetchrow(query, lon, lat)
return result['tzid'] if result else 'Unknown'
def find(self, lat: float, lon: float) -> str:
timezone = self.tf.timezone_at(lat=lat, lng=lon)
return timezone if timezone else 'Unknown'
async def refresh(self, location: Union[Location, Tuple[float, float]], force: bool = False) -> str:
if isinstance(location, Location):
@ -167,22 +197,15 @@ class TimezoneTracker:
lat, lon = location
current_time = datetime.now()
if (force or
not self.last_update or
current_time - self.last_update > timedelta(hours=1) or
self.last_location != (lat, lon)):
new_timezone = await self.find(lat, lon)
new_timezone = self.find(lat, lon)
self.last_timezone = new_timezone
self.last_update = current_time
self.last_location = (lat, lon)
await self.save_to_cache()
return new_timezone
return self.last_timezone
async def save_to_cache(self):
@ -191,12 +214,12 @@ class TimezoneTracker:
'last_update': self.last_update.isoformat() if self.last_update else None,
'last_location': self.last_location
}
with open(self.cache_file, 'w') as f:
with self.cache_file.open('w') as f:
json.dump(cache_data, f)
async def load_from_cache(self):
try:
with open(self.cache_file, 'r') as f:
with self.cache_file.open('r') as f:
cache_data = json.load(f)
self.last_timezone = cache_data.get('last_timezone')
self.last_update = datetime.fromisoformat(cache_data['last_update']) if cache_data.get('last_update') else None
@ -208,7 +231,7 @@ class TimezoneTracker:
async def get_current(self, location: Union[Location, Tuple[float, float]]) -> str:
await self.load_from_cache()
return await self.refresh(location)
async def get_last(self) -> Optional[str]:
await self.load_from_cache()
return self.last_timezone

51
sijapi/data/osm.sh Executable file
View file

@ -0,0 +1,51 @@
#!/bin/bash
set -e # Exit immediately if a command exits with a non-zero status.
# Set variables
DB_NAME="sij"
DB_USER="sij"
OSM_FILE="north-america-latest.osm.pbf"
FLAT_NODES="/Users/sij/workshop/sijapi/sijapi/data/db/flat-nodes.bin"
# Ensure the directory for flat-nodes exists
mkdir -p "$(dirname "$FLAT_NODES")"
# Determine total system memory in MB
TOTAL_MEM=$(sysctl hw.memsize | awk '{print $2 / 1024 / 1024}')
# Calculate cache size (50% of total memory, max 32GB)
CACHE_SIZE=$(echo "scale=0; $TOTAL_MEM * 0.5 / 1" | bc)
CACHE_SIZE=$(( CACHE_SIZE > 32768 ? 32768 : CACHE_SIZE ))
# Calculate number of processes (number of CPU cores minus 1, min 1)
NUM_PROCESSES=$(sysctl -n hw.ncpu)
NUM_PROCESSES=$(( NUM_PROCESSES > 1 ? NUM_PROCESSES - 1 : 1 ))
echo "Starting OSM data import..."
# Run osm2pgsql
osm2pgsql -d $DB_NAME \
--create \
--slim \
-G \
--hstore \
--tag-transform-script /opt/homebrew/Cellar/osm2pgsql/1.11.0_1/share/osm2pgsql/openstreetmap-carto.lua \
-C $CACHE_SIZE \
--number-processes $NUM_PROCESSES \
-S /opt/homebrew/Cellar/osm2pgsql/1.11.0_1/share/osm2pgsql/default.style \
--prefix osm \
-H localhost \
-P 5432 \
-U $DB_USER \
--flat-nodes $FLAT_NODES \
$OSM_FILE
echo "OSM data import completed. Creating indexes..."
# Create indexes (adjust table names if necessary)
psql -d $DB_NAME -U $DB_USER -c "CREATE INDEX IF NOT EXISTS idx_osm_point_way ON osm_point USING GIST (way);"
psql -d $DB_NAME -U $DB_USER -c "CREATE INDEX IF NOT EXISTS idx_osm_line_way ON osm_line USING GIST (way);"
psql -d $DB_NAME -U $DB_USER -c "CREATE INDEX IF NOT EXISTS idx_osm_polygon_way ON osm_polygon USING GIST (way);"
echo "Import completed and indexes created."

View file

@ -51,7 +51,7 @@
"inputs": {
"batch_size": 1,
"width": 1023,
"height": 1024,
"height": 1025,
"resampling": "bicubic",
"X": 0,
"Y": 0,

View file

@ -42,7 +42,7 @@ if MS365_TOGGLE is True:
@calendar.get("/o365/oauth_redirect")
async def oauth_redirect(code: str = None, error: str = None):
L.INFO(f"Received request to /o365/oauth_redirect")
L.DEBUG(f"Received request to /o365/oauth_redirect")
if error:
L.ERR(f"OAuth2 Error: {error}")
raise HTTPException(
@ -75,7 +75,7 @@ if MS365_TOGGLE is True:
@calendar.get("/o365/me")
async def read_items():
L.INFO(f"Received request to /o365/me")
L.DEBUG(f"Received request to /o365/me")
token = await load_token()
if not token:
raise HTTPException(
@ -204,7 +204,7 @@ def get_calendar_ids() -> Dict[str, str]:
calendar_identifiers = {
calendar.title() : calendar.calendarIdentifier() for calendar in all_calendars
}
L.INFO(f"{calendar_identifiers}")
L.DEBUG(f"{calendar_identifiers}")
return calendar_identifiers
# Helper to convert datetime to NSDate

View file

@ -204,7 +204,7 @@ tags:
with open(md_path, 'w', encoding='utf-8') as md_file:
md_file.write(markdown_content)
L.INFO(f"Saved markdown to {md_path}")
L.DEBUG(f"Saved markdown to {md_path}")
return True
@ -221,7 +221,12 @@ async def autorespond(this_email: IncomingEmail, account: EmailAccount):
auto_response_subject = f"Auto-Response Re: {this_email.subject}"
auto_response_body = await generate_auto_response_body(this_email, profile, account)
L.DEBUG(f"Auto-response: {auto_response_body}")
await send_auto_response(this_email.sender, auto_response_subject, auto_response_body, profile, account)
success = await send_auto_response(this_email.sender, auto_response_subject, auto_response_body, profile, account)
if success == True:
return True
L.WARN(f"We were not able to successfully auto-respond to {this_email.subject}")
return False
async def send_auto_response(to_email, subject, body, profile, account):
try:
@ -264,13 +269,16 @@ async def save_processed_uid(filename: Path, account_name: str, uid: str):
f.write(f"{account_name}:{uid}\n")
async def process_account_summarization(account: EmailAccount):
summarized_log = EMAIL_LOGS / "summarized.txt"
summarized_log = EMAIL_LOGS / account.name / "summarized.txt"
os.makedirs(summarized_log.parent, exist_ok = True)
while True:
try:
processed_uids = await load_processed_uids(summarized_log)
L.DEBUG(f"{len(processed_uids)} emails marked as already summarized are being ignored.")
with get_imap_connection(account) as inbox:
unread_messages = inbox.messages(unread=True)
L.DEBUG(f"There are {len(unread_messages)} unread messages.")
for uid, message in unread_messages:
uid_str = uid.decode() if isinstance(uid, bytes) else str(uid)
if uid_str not in processed_uids:
@ -282,27 +290,35 @@ async def process_account_summarization(account: EmailAccount):
recipients=recipients,
subject=message.subject,
body=clean_email_content(message.body['html'][0]) if message.body['html'] else clean_email_content(message.body['plain'][0]) or "",
attachments=message.attachments
attachments=message.attachments
)
if account.summarize:
save_success = await save_email(this_email, account)
if save_success:
await save_processed_uid(summarized_log, account.name, uid_str)
L.INFO(f"Summarized email: {uid_str}")
else:
L.WARN(f"Failed to summarize {this_email.subject}")
else:
L.INFO(f"account.summarize shows as false.")
else:
L.DEBUG(f"Skipping {uid_str} because it was already processed.")
except Exception as e:
L.ERR(f"An error occurred during summarization for account {account.name}: {e}")
await asyncio.sleep(account.refresh)
async def process_account_autoresponding(account: EmailAccount):
autoresponded_log = EMAIL_LOGS / "autoresponded.txt"
autoresponded_log = EMAIL_LOGS / account.name / "autoresponded.txt"
os.makedirs(autoresponded_log.parent, exist_ok = True)
while True:
try:
processed_uids = await load_processed_uids(autoresponded_log)
L.DEBUG(f"{len(processed_uids)} already processed emails are being ignored.")
L.DEBUG(f"{len(processed_uids)} emails marked as already responded to are being ignored.")
with get_imap_connection(account) as inbox:
unread_messages = inbox.messages(unread=True)
L.DEBUG(f"There are {len(unread_messages)} unread messages.")
for uid, message in unread_messages:
uid_str = uid.decode() if isinstance(uid, bytes) else str(uid)
if uid_str not in processed_uids:
@ -320,7 +336,10 @@ async def process_account_autoresponding(account: EmailAccount):
respond_success = await autorespond(this_email, account)
if respond_success:
await save_processed_uid(autoresponded_log, account.name, uid_str)
L.WARN(f"Auto-responded to email: {uid_str}")
L.WARN(f"Auto-responded to email: {this_email.subject}")
else:
L.WARN(f"Failed auto-response to {this_email.subject}")
L.DEBUG(f"Skipping {uid_str} because it was already processed.")
except Exception as e:
L.ERR(f"An error occurred during auto-responding for account {account.name}: {e}")

View file

@ -16,12 +16,15 @@ import aiohttp
import folium
import time as timer
from dateutil.parser import parse as dateutil_parse
from concurrent.futures import ThreadPoolExecutor
from functools import partial
from srtm import get_data
from pathlib import Path
from pydantic import BaseModel
from typing import Optional, Any, Dict, List, Union
from datetime import datetime, timedelta, time
from sijapi import L, DB, TZ, NAMED_LOCATIONS, DynamicTZ
from sijapi.classes import Location
from sijapi import L, DB, TZ, NAMED_LOCATIONS, DynamicTZ, GEOLOCATOR
from sijapi.classes import Location, PyGeolocator
from sijapi.utilities import haversine
# from osgeo import gdal
# import elevation
@ -29,9 +32,23 @@ from sijapi.utilities import haversine
locate = APIRouter()
async def reverse_geocode_local_1(lat, lon, date_time: datetime = datetime.now()) -> Optional[Location]:
date_time = await localize_datetime(date_time)
loc = GEOLOCATOR.lookup(lat, lon)
location = Location(
latitude = lat,
longitude = lon,
elevation = loc['elevation'],
datetime = date_time,
city = loc['city'],
state = loc['state'],
country = loc['country']
)
return location
async def reverse_geocode(latitude: float, longitude: float, elevation: float = None) -> Optional[Location]:
url = f"https://nominatim.openstreetmap.org/reverse?format=json&lat={latitude}&lon={longitude}"
L.INFO(f"Calling Nominatim API at {url}")
L.DEBUG(f"Calling Nominatim API at {url}")
headers = {
'User-Agent': 'sij.law/1.0 (sij@sij.law)', # replace with your app name and email
}
@ -67,7 +84,7 @@ async def reverse_geocode(latitude: float, longitude: float, elevation: float =
county=address.get("county"),
country_code=address.get("country_code")
)
L.INFO(f"Created Location object: {location}")
L.DEBUG(f"Created Location object: {location}")
return location
except aiohttp.ClientError as e:
L.ERR(f"Error: {e}")
@ -140,35 +157,55 @@ async def geocode(zip_code: Optional[str] = None, latitude: Optional[float] = No
raise Exception("An error occurred while processing your request")
async def localize_datetime(dt, fetch_loc: bool = False):
initial_dt = dt
if fetch_loc:
loc = await get_last_location()
tz = await DynamicTZ.get_current(loc)
else:
tz = await DynamicTZ.get_last()
async def localize_datetime(dt: Union[str, datetime], fetch_loc: bool = False) -> datetime:
"""
Localize a datetime object or string to the appropriate timezone.
Args:
dt (Union[str, datetime]): The datetime to localize.
fetch_loc (bool): Whether to fetch the current location for timezone.
Returns:
datetime: A timezone-aware datetime object.
Raises:
ValueError: If the input cannot be parsed as a datetime.
"""
try:
# Convert string to datetime if necessary
if isinstance(dt, str):
dt = dateutil_parse(dt)
L.DEBUG(f"{initial_dt} was a string so we attempted converting to datetime. Result: {dt}")
L.DEBUG(f"Converted string '{dt}' to datetime object.")
if isinstance(dt, datetime):
L.DEBUG(f"{dt} is a datetime object, so we will ensure it is tz-aware.")
if dt.tzinfo is None:
dt = dt.replace(tzinfo=TZ)
# L.DEBUG(f"{dt} should now be tz-aware. Returning it now.")
return dt
else:
# L.DEBUG(f"{dt} already was tz-aware. Returning it now.")
return dt
if not isinstance(dt, datetime):
raise ValueError("Input must be a string or datetime object.")
# Fetch timezone
if fetch_loc:
loc = await get_last_location()
tz = await DynamicTZ.get_current(loc)
L.DEBUG(f"Fetched current timezone: {tz}")
else:
L.ERR(f"Conversion failed")
raise TypeError("Conversion failed")
except Exception as e:
tz = await DynamicTZ.get_last()
L.DEBUG(f"Using last known timezone: {tz}")
# Localize datetime
if dt.tzinfo is None:
dt = dt.replace(tzinfo=tz)
L.DEBUG(f"Localized naive datetime to {tz}")
elif dt.tzinfo != tz:
dt = dt.astimezone(tz)
L.DEBUG(f"Converted datetime from {dt.tzinfo} to {tz}")
return dt
except ValueError as e:
L.ERR(f"Error parsing datetime: {e}")
raise TypeError("Input must be a string or datetime object")
raise
except Exception as e:
L.ERR(f"Unexpected error in localize_datetime: {e}")
raise ValueError(f"Failed to localize datetime: {e}")
@ -199,37 +236,27 @@ async def find_override_locations(lat: float, lon: float) -> Optional[str]:
return closest_location
async def get_elevation(latitude, longitude):
url = "https://api.open-elevation.com/api/v1/lookup"
async def get_elevation(latitude: float, longitude: float, unit: str = "m") -> float:
loop = asyncio.get_running_loop()
payload = {
"locations": [
{
"latitude": latitude,
"longitude": longitude
}
]
}
async with aiohttp.ClientSession() as session:
try:
async with session.post(url, json=payload) as response:
response.raise_for_status() # Raise an exception for unsuccessful requests
data = await response.json()
if "results" in data:
elevation = data["results"][0]["elevation"]
return elevation
else:
return None
# Create a thread pool executor
with ThreadPoolExecutor() as pool:
# Run get_data() in a separate thread
srtm_data = await loop.run_in_executor(pool, get_data)
except aiohttp.ClientError as e:
L.ERR(f"Error: {e}")
return None
# Run get_elevation() in a separate thread
elevation = await loop.run_in_executor(
pool,
partial(srtm_data.get_elevation, latitude, longitude)
)
if unit == "m":
return elevation
elif unit == "km":
return elevation / 1000
elif unit == "ft" or unit == "'":
return elevation * 3.280839895
else:
raise ValueError(f"Unsupported unit: {unit}")
async def fetch_locations(start: datetime, end: datetime = None) -> List[Location]:
start_datetime = await localize_datetime(start)
@ -502,8 +529,6 @@ async def post_locate_endpoint(locations: Union[Location, List[Location]]):
return {"message": "Locations and weather updated", "results": responses}
# Assuming post_location and get_elevation are async functions. If not, they should be modified to be async as well.
async def get_last_location() -> Optional[Location]:

View file

@ -211,7 +211,7 @@ async def process_for_daily_note(file: Optional[UploadFile] = File(None), text:
text_entry = text if text else ""
L.INFO(f"transcription: {transcription}\nfile_entry: {file_entry}\ntext_entry: {text_entry}")
L.DEBUG(f"transcription: {transcription}\nfile_entry: {file_entry}\ntext_entry: {text_entry}")
return await add_to_daily_note(transcription, file_entry, text_entry, now)
@ -520,19 +520,22 @@ async def process_archive(
markdown_content = f"---\n"
markdown_content += f"title: {readable_title}\n"
markdown_content += f"added: {timestamp}\n"
markdown_content += f"url: {url}"
markdown_content += f"date: {datetime.now().strftime('%Y-%m-%d')}"
markdown_content += f"---\n\n"
markdown_content += f"# {readable_title}\n\n"
markdown_content += f"Clipped from [{url}]({url}) on {timestamp}"
markdown_content += content
try:
markdown_path.parent.mkdir(parents=True, exist_ok=True)
with open(markdown_path, 'w', encoding=encoding) as md_file:
md_file.write(markdown_content)
L.INFO(f"Successfully saved to {markdown_path}")
L.DEBUG(f"Successfully saved to {markdown_path}")
return markdown_path
except Exception as e:
L.ERR(f"Failed to write markdown file: {str(e)}")
raise HTTPException(status_code=500, detail=f"Failed to write markdown file: {str(e)}")
L.WARN(f"Failed to write markdown file: {str(e)}")
return None
def download_file(url, folder):
os.makedirs(folder, exist_ok=True)
@ -800,7 +803,7 @@ async def update_dn_weather(date_time: datetime):
city = city if city else loc.city
city = city if city else loc.house_number + ' ' + loc.road
L.INFO(f"City geocoded: {city}")
L.DEBUG(f"City geocoded: {city}")
# Assemble journal path
absolute_path, relative_path = assemble_journal_path(date_time, filename="Weather", extension=".md", no_timestamp = True)

View file

@ -295,7 +295,7 @@ async def extract_text_from_pdf(file_path: str) -> str:
L.ERR(f"Error extracting text with pdfminer.six: {e}")
# If both methods fail or are deemed insufficient, use OCR as the last resort
L.INFO("Falling back to OCR for text extraction...")
L.DEBUG("Falling back to OCR for text extraction...")
return await ocr_pdf(file_path)
async def is_valid_pdf(file_path: str) -> bool:
@ -331,7 +331,7 @@ async def extract_text_from_pdf(file_path: str) -> str:
L.ERR(f"Error extracting text with pdfminer.six: {str(e)}")
# Fall back to OCR
L.INFO("Falling back to OCR for text extraction...")
L.DEBUG("Falling back to OCR for text extraction...")
try:
images = convert_from_path(file_path)
ocr_texts = await asyncio.gather(*(asyncio.to_thread(pytesseract.image_to_string, img) for img in images))