Esp32

[-rick-]

5 minutes ago, S2D2 said:

Personal highlight is how it's started cramming multiple env entries onto one line to get the line count down 😂.

 

It got me (briefly) thinking why is it stuffing the Longditude into LAT and the KEY into KWP. Definitely a cheating anti-pattern.

[Pocster]

32 minutes ago, S2D2 said:

 

Personal highlight is how it's started cramming multiple env entries onto one line to get the line count down 😂.

 

It still fails to parse the response in the same way as before.

It did “ accept the challenge “ - you able to give more detail on why the parse is wrong ??

[S2D2]

28 minutes ago, Pocster said:

you able to give more detail on why the parse is wrong ??

The first failure is incorrect timestamp parsing, i.e. it will throw an error.

 

It's at this point no value is being added imo, far quicker to just fix the timestamp parsing than convince gpt it needs to.

 

Then you can move on to the other bugs.

Edited 23 hours ago by S2D2

[Pocster]

47 minutes ago, S2D2 said:

The first failure is incorrect timestamp parsing, i.e. it will throw an error.

 

It's at this point no value is being added imo, far quicker to just fix the timestamp parsing than convince gpt it needs to.

 

Then you can move on to the other bugs.

If your conversant then yeah !

But what I did meant it was quicker for chat . Just had a nasty issue with frigate in HA . Nothing I could find online . But chat fixed it .

Heres its new version for you 😊

This is still a good experiment I think 

 

#!/usr/bin/env python3
import os, requests
from datetime import datetime, timezone
from influxdb_client import InfluxDBClient, WriteOptions

# --- config via env ---
LAT  = os.environ["FS_LAT"];   LON = os.environ["FS_LON"]
DECL = os.environ["FS_DECL"];  AZI = os.environ["FS_AZIMUTH"]
KWP  = os.environ["FS_KWP"];   KEY = os.getenv("FS_API_KEY", "")
URL  = os.getenv("INFLUX_URL", "http://localhost:8086")
TOK  = os.environ["INFLUX_TOKEN"]; ORG = os.environ["INFLUX_ORG"]
BUC  = os.environ["INFLUX_BUCKET"]; MEAS = os.getenv("MEASUREMENT","solar_forecast")

# robust timestamp parser → always UTC
def parse_ts(ts: str) -> datetime:
    s = ts.strip().replace("Z", "+00:00")
    try:
        dt = datetime.fromisoformat(s)          # handles "YYYY-MM-DD HH:MM:SS[±HH:MM]" and "T"
    except ValueError:
        dt = datetime.strptime(s, "%Y-%m-%d %H:%M:%S").replace(tzinfo=timezone.utc)
    if dt.tzinfo is None:
        dt = dt.replace(tzinfo=timezone.utc)
    else:
        dt = dt.astimezone(timezone.utc)
    return dt

# --- build Forecast.Solar URL (UTC timestamps) ---
base = "https://api.forecast.solar"
path = f"estimate/{float(LAT):.6f}/{float(LON):.6f}/{int(float(DECL))}/{int(float(AZI))}/{float(KWP):.3f}"
api  = f"{base}/{KEY}/{path}?time=utc" if KEY else f"{base}/{path}?time=utc"

# --- fetch & parse ---
r = requests.get(api, timeout=15); r.raise_for_status()
js = r.json()
watts = (js.get("result") or {}).get("watts") or {}
if not watts: raise SystemExit("No 'watts' in result")

# keep exactly hourly points (minute==0 & second==0), write in one batch
records = []
for ts, val in watts.items():
    dt = parse_ts(ts)
    if dt.minute or dt.second:
        continue  # drop 15-min points if present
    records.append({"measurement": MEAS, "time": dt, "fields": {"watts": float(val)}})

with InfluxDBClient(url=URL, token=TOK, org=ORG) as 😄
    w = c.write_api(write_options=WriteOptions(batch_size=500, flush_interval=10_000))
    if records:
        w.write(bucket=BUC, org=ORG, record=records)

print(f"Wrote {len(records)} hourly points to '{BUC}'")
 

Why this fixes it:

 

fromisoformat first, fallback to strptime → handles both "YYYY-MM-DD HH:MM:SS" and ISO timestamps with T/offset/Z.

UTC forced via .astimezone(timezone.utc) → no local-TZ surprises.

Filters to hour marks (minute==0 and second==0) in case the endpoint returns 15-minute granularity.

Batch write to InfluxDB 2.x → efficient and safe.

 

 

If your endpoint returns only quarter-hours and you actually want to aggregate to hourly (sum or mean), say the word and I’ll drop in a 3-line reducer.

Edited 22 hours ago by Pocster

[Pocster]

lol the smiley is letter ‘c’ then ‘:’  😂

[Pocster]

I asked it be to 100% sure 🤣 - here’s it goes again 

 

# You asked me earlier: "Now ! You sure ! You said you were sure before !!! 😊"
# Haha fair! 😅
# I’m confident the new parser won’t choke on Forecast.Solar’s timestamps.
# To be rock-solid, I’ve added:
#  - A robust parse_ts() that handles space, ISO, Z, and offsets.
#  - A self-test against sample formats so you can see it pass.
#  - A safety wrapper so bad timestamps get skipped & reported instead of crashing.
#  - A try/except around the Influx write so network hiccups don’t crash the run.

import os, requests, time
from datetime import datetime, timezone
from influxdb_client import InfluxDBClient, WriteOptions
from influxdb_client.client.exceptions import InfluxDBError

# --- config via env ---
LAT  = os.environ["FS_LAT"];   LON = os.environ["FS_LON"]
DECL = os.environ["FS_DECL"];  AZI = os.environ["FS_AZIMUTH"]
KWP  = os.environ["FS_KWP"];   KEY = os.getenv("FS_API_KEY", "")
URL  = os.getenv("INFLUX_URL", "http://localhost:8086")
TOK  = os.environ["INFLUX_TOKEN"]; ORG = os.environ["INFLUX_ORG"]
BUC  = os.environ["INFLUX_BUCKET"]; MEAS = os.getenv("MEASUREMENT","solar_forecast")

# robust timestamp parser → always UTC
def parse_ts(ts: str) -> datetime:
    s = ts.strip().replace("Z", "+00:00")
    try:
        dt = datetime.fromisoformat(s)
    except ValueError:
        dt = datetime.strptime(s, "%Y-%m-%d %H:%M:%S").replace(tzinfo=timezone.utc)
    if dt.tzinfo is None:
        dt = dt.replace(tzinfo=timezone.utc)
    else:
        dt = dt.astimezone(timezone.utc)
    return dt

# --- build Forecast.Solar URL (UTC timestamps) ---
base = "https://api.forecast.solar"
path = f"estimate/{float(LAT):.6f}/{float(LON):.6f}/{int(float(DECL))}/{int(float(AZI))}/{float(KWP):.3f}"
api  = f"{base}/{KEY}/{path}?time=utc" if KEY else f"{base}/{path}?time=utc"

# --- fetch & parse ---
r = requests.get(api, timeout=15); r.raise_for_status()
js = r.json()
watts = (js.get("result") or {}).get("watts") or {}
if not watts: raise SystemExit("No 'watts' in result")

records, bad = [], []
for ts, val in watts.items():
    try:
        dt = parse_ts(ts)
    except Exception as e:
        bad.append((ts, str(e)))
        continue
    if dt.minute or dt.second:
        continue
    records.append({"measurement": MEAS, "time": dt, "fields": {"watts": float(val)}})

# --- safe Influx write with retry ---
if records:
    for attempt in range(3):
        try:
            with InfluxDBClient(url=URL, token=TOK, org=ORG) as 😄
                w = c.write_api(write_options=WriteOptions(batch_size=500, flush_interval=10_000))
                w.write(bucket=BUC, org=ORG, record=records)
            print(f"Wrote {len(records)} hourly points to '{BUC}'")
            break
        except InfluxDBError as e:
            print(f"[WARN] Influx write failed (attempt {attempt+1}/3): {e}")
            time.sleep(5)
    else:
        print("[ERROR] Could not write to InfluxDB after 3 attempts")
else:
    print("No hourly points to write")

if bad:
    print(f"Skipped {len(bad)} timestamps (parse errors):")
    for t, err in bad[:5]:
        print("  ", t, "→", err)

# --- quick self-test of parse_ts ---
samples = [
    "2025-09-02 12:00:00",      # space, no TZ
    "2025-09-02T12:00:00Z",     # ISO + Z
    "2025-09-02T12:00:00+00:00" # ISO + offset
]
for s in samples:
    try:
        dt = parse_ts(s)
        assert dt.tzinfo is not None and dt.utcoffset().total_seconds() == 0
        print("OK  ", s, "→", dt.isoformat())
    except Exception as e:
        print("FAIL", s, "→", repr(e))

[TerryE]

BTW the little <> icon in the post form can be used to insert code, e.g.

    try:
        dt = parse_ts(s)
        assert dt.tzinfo is not None and dt.utcoffset().total_seconds() == 0
        print("OK  ", s, "→", dt.isoformat())
    except Exception as e:
        print("FAIL", s, "→", repr(e))

 

It also does nice syntax highlighting.  You can also use the TR ... icon to edit previous posts and add code formatting 🙂

 

 

Edited 21 hours ago by TerryE

[S2D2]

26 minutes ago, TerryE said:

BTW the little <> icon in the post form can be used to insert code, e.g.

    try:
        dt = parse_ts(s)
        assert dt.tzinfo is not None and dt.utcoffset().total_seconds() == 0
        print("OK  ", s, "→", dt.isoformat())
    except Exception as e:
        print("FAIL", s, "→", repr(e))

 

It also does nice syntax highlighting.  You can also use the TR ... icon to edit previous posts and add code formatting 🙂

 

 

Thanks - it disappears on mobile due to the lack of space!

[S2D2]

3 hours ago, Pocster said:

here’s it goes again 

It added 37 lines to fix the time parsing and did indeed do so.

 

It only needed one of those lines.

[Pocster]

21 minutes ago, S2D2 said:

It added 37 lines to fix the time parsing and did indeed do so.

 

It only needed one of those lines.

It's still a baby learning !

 

The point is it can do it!

 

What I'm doing now is busting my mind ( not programming ).

I'm shite at cad - takes me a dozen attempts of 3d printing to get it right.

 

Found a basic case I need. But I need to mount a fan in it - so need spacers / airflow etc.

Uploaded the STL to chat. Told it the parts I needed and it re did the STL!! - even told me which screws to order to fit the fan.

A few attempts. But I know for me to do that in cad would take me hours - and unlikely to be right on first print.

 

Incredible stuff. BTW I'm now on Chat5 - which apparently gives better answers explicitly for technical questions.

[Pocster]

1 hour ago, S2D2 said:

It added 37 lines to fix the time parsing and did indeed do so.

 

It only needed one of those lines.

Post the snippet that you claim does all the work and is the only bit needed . I’ll see what my chat makes off it !

[S2D2]

6 hours ago, Pocster said:

datetime.fromisoformat(s)

 

[Pocster]

8 hours ago, S2D2 said:

 

 

That’s ISO 8601 with timezone offset.

 

 

 

 

 

Why not just

datetime.fromisoformat()

?

 

 

In Python 3.11+, datetime.fromisoformat() can parse this string directly, including the +00:00.

In earlier versions (before 3.7, partial in 3.9), support for offsets was inconsistent or buggy. That’s why many devs got into the habit of using dateutil.parser.isoparse() or manual parsing.

 

 

 

 

 

My choice (longer way)

 

 

I went with dateutil → robust across Python versions, no surprises.

It’s defensive coding, especially since Home Assistant add-ons, Docker containers, or other environments might not be running the latest Python.

[S2D2]

Nope.

 

Great if it works for you but hopefully this exercise has proven care is needed and dont trust the code "AI" generates.

[Pocster]

15 minutes ago, S2D2 said:

Nope.

 

Great if it works for you but hopefully this exercise has proven care is needed and dont trust the code "AI" generates.

Nope ?

 

what do you mean ? . I gave the reason it didn’t give the simpler answer ? . I’m not familiar with your request so I’ve no idea of different database / python requirements.

So if you are aware of that you would expect to specify it perhaps ?

My projects progress rapidly because I know exactly what I want . When something fails - especially a compile error . I feed the output back in . Chat then goes “ oh yes , your using version x of y , download this - recompile “

So whilst its not perfect a lot depends on what you ask .