diffron 0.1.4__tar.gz → 0.1.6__tar.gz

{diffron-0.1.4/diffron.egg-info → diffron-0.1.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: diffron
-Version: 0.1.4
+Version: 0.1.6
 Summary: Git commit message and PR description generator using Lemonade
 Home-page: https://github.com/diffron/diffron
 Author: Diffron Contributors
@@ -47,7 +47,7 @@ Git commit message and PR description generator using AMD Lemonade via lemonade-
 
 **Diffron is a production-ready reference implementation of the lemonade-python-sdk — submitted to the AMD Lemonade Developer Challenge 2026.**
 
-![Version](https://img.shields.io/badge/version-0.1.3-blue)
+![Version](https://img.shields.io/badge/version-0.1.6-blue)
 ![Python](https://img.shields.io/badge/python-3.9+-blue)
 ![License](https://img.shields.io/badge/license-MIT-green)
 ![Platform](https://img.shields.io/badge/platform-Windows-lightgrey)

{diffron-0.1.4 → diffron-0.1.6}/README.md

@@ -4,7 +4,7 @@ Git commit message and PR description generator using AMD Lemonade via lemonade-
 
 **Diffron is a production-ready reference implementation of the lemonade-python-sdk — submitted to the AMD Lemonade Developer Challenge 2026.**
 
-![Version](https://img.shields.io/badge/version-0.1.3-blue)
+![Version](https://img.shields.io/badge/version-0.1.6-blue)
 ![Python](https://img.shields.io/badge/python-3.9+-blue)
 ![License](https://img.shields.io/badge/license-MIT-green)
 ![Platform](https://img.shields.io/badge/platform-Windows-lightgrey)

{diffron-0.1.4 → diffron-0.1.6}/diffron/lemonade.py

@@ -229,7 +229,10 @@ class LemonadeClient:
                 )
                 if new_port and new_port != self._port:
                     self._port = new_port
-                    # Reinitialize client with new port
+                    self.port = new_port  # Update the port attribute
+                    # Rebuild the base_url with the new port and the same host
+                    self.base_url = f"http://{self.host}:{new_port}"
+                    # Reinitialize client with new base_url
                     self._init_client()
         except Exception:
             # Silently continue on reload failure

{diffron-0.1.4 → diffron-0.1.6/diffron.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: diffron
-Version: 0.1.4
+Version: 0.1.6
 Summary: Git commit message and PR description generator using Lemonade
 Home-page: https://github.com/diffron/diffron
 Author: Diffron Contributors
@@ -47,7 +47,7 @@ Git commit message and PR description generator using AMD Lemonade via lemonade-
 
 **Diffron is a production-ready reference implementation of the lemonade-python-sdk — submitted to the AMD Lemonade Developer Challenge 2026.**
 
-![Version](https://img.shields.io/badge/version-0.1.3-blue)
+![Version](https://img.shields.io/badge/version-0.1.6-blue)
 ![Python](https://img.shields.io/badge/python-3.9+-blue)
 ![License](https://img.shields.io/badge/license-MIT-green)
 ![Platform](https://img.shields.io/badge/platform-Windows-lightgrey)

{diffron-0.1.4 → diffron-0.1.6}/diffron.egg-info/SOURCES.txt

@@ -36,9 +36,6 @@ diffron/hooks/prepare-commit-msg
 diffron/hooks/prepare-commit-msg-global.py
 diffron/hooks/prepare-commit-msg.py
 docs/HOOKS.md
-docs/LEMONADE_PORT_VERIFICATION_FIX.md
-docs/LEMONADE_UNRESPONSIVE_FIX.md
-docs/MODEL_NAME_CASE_FIX.md
 docs/SETUP.md
 docs/USAGE.md
 hooks/aipr.py

{diffron-0.1.4 → diffron-0.1.6}/diffron.egg-info/top_level.txt

@@ -1,6 +1,4 @@
-build
 diffron
 dist
 docs
 hooks
-internal

{diffron-0.1.4 → diffron-0.1.6}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "diffron"
-version = "0.1.4"
+version = "0.1.6"
 description = "Git commit message and PR description generator using Lemonade"
 readme = "README.md"
 license = {text = "MIT"}

diffron-0.1.4/docs/LEMONADE_PORT_VERIFICATION_FIX.md

@@ -1,105 +0,0 @@
-# Lemonade Port Verification Fix
-
-## Problem
-
-Diffron incorrectly connected to **port 8000** (running Axo Biomarker Analyzer) instead of **port 8020** (running Lemonade), causing:
-
-```
-Error code: 404 - {'detail': 'Not Found'}
-```
-
-### Root Cause
-
-The `detect_lemonade_port()` function in `diffron/lemonade.py` only checked if a **port was open** (`is_port_open`), not whether it was actually a **Lemonade server**.
-
-When port 8000 responded (Axo Biomarker web app), Diffron mistakenly identified it as Lemonade and sent API requests to the wrong service.
-
-| Port | Service | Response |
-|------|---------|----------|
-| 8000 | Axo Biomarker Analyzer | HTML web page |
-| 8020 | **Lemonade** (correct) | `{"data": [...]}` API |
-
-## Solution
-
-### Use lemonade-python-sdk's `verify_lemonade_server`
-
-The **lemonade-python-sdk** provides `verify_lemonade_server(port, host)` which:
-
-1. Queries the native Lemonade API endpoint `/api/v1/models`
-2. Validates the response format (`{"data": [...]}` or list)
-3. Distinguishes Lemonade from other services (Ollama, web apps, etc.)
-
-### Changes
-
-**1. Added `lemonade-sdk` dependency**
-
-Updated `requirements.txt`:
-```txt
-lemonade-sdk>=0.1.0
-```
-
-**2. Updated `detect_lemonade_port()`**
-
-`diffron/lemonade.py` now:
-
-- Imports `verify_lemonade_server` and `find_available_lemonade_port` from `lemonade_sdk.port_scanner`
-- **Verifies** each open port is actually Lemonade before returning it
-- Falls back to legacy `is_port_open` behavior if SDK is not installed
-
-```python
-# Import from lemonade-python-sdk for proper Lemonade verification
-try:
-    from lemonade_sdk.port_scanner import verify_lemonade_server, find_available_lemonade_port
-    HAS_LEMONADE_SDK = True
-except ImportError:
-    HAS_LEMONADE_SDK = False
-
-def detect_lemonade_port(...) -> Optional[int]:
-    # ... existing env var checks ...
-
-    # Try SDK's find_available_lemonade_port first
-    if HAS_LEMONADE_SDK:
-        found = find_available_lemonade_port(hostname, scan_ports)
-        if found:
-            return found
-
-    # Fallback: manual scan with SDK verification
-    for port in ports_to_scan:
-        if is_port_open(hostname, port):
-            if HAS_LEMONADE_SDK:
-                if verify_lemonade_server(port, hostname):  # ← Key fix!
-                    return port
-            else:
-                return port  # Legacy: no verification
-```
-
-**3. Added graceful fallback**
-
-If `lemonade-sdk` is not installed, Diffron falls back to the original `is_port_open` behavior (first open port wins), maintaining backward compatibility.
-
-## Testing
-
-```bash
-# Before fix: returned 8000 (Axo Biomarker)
-# After fix: returns 8020 (Lemonade)
-python -c "from diffron.lemonade import detect_lemonade_port; print(detect_lemonade_port())"
-# Output: 8020
-```
-
-## Impact
-
-- **Fixes 404 errors** when other services run on ports 8000/8080
-- **No breaking changes** - existing setups continue to work
-- **Better diagnostics** - clear separation between Lemonade and non-Lemonade services
-
-## Files Changed
-
-| File | Change |
-|------|--------|
-| `diffron/lemonade.py` | Use `verify_lemonade_server` for port detection |
-| `requirements.txt` | Add `lemonade-sdk>=0.1.0` dependency |
-
-## Version
-
-- **Fixed in**: 0.1.3
-- **Date**: 2026-04-13

diffron-0.1.4/docs/LEMONADE_UNRESPONSIVE_FIX.md

@@ -1,100 +0,0 @@
-# Lemonade Model Unresponsive Fix
-
-## Problem
-
-Das Lemonade-Modell wurde manchmal unresponsive, was zu einem Fehler in Diffron führte:
-
-```
-Diffron error: 'NoneType' object is not subscriptable
-```
-
-Dies trat auf, wenn die Antwort des Modells `None` war oder `response.choices` leer war, was beim Zugriff auf `response.choices[0].message.content` zu einem Fehler führte.
-
-## Lösung
-
-In [`diffron/lemonade.py`](../diffron/lemonade.py:170) wurde eine Retry-Logik mit automatischem Entladen und Neuladen des Modells implementiert.
-
-### Änderungen in `LemonadeClient.chat_completion()`
-
-```python
-max_retries = 3
-for attempt in range(1, max_retries + 1):
-    try:
-        response = self.client.chat.completions.create(...)
-        # Handle None response (model unresponsive)
-        if response is None or not response.choices:
-            if attempt < max_retries:
-                self._reload_model()
-                continue
-            raise RuntimeError("Lemonade model returned no response")
-        return response.choices[0].message.content
-    except Exception as e:
-        if attempt < max_retries:
-            self._reload_model()
-            continue
-        raise RuntimeError(
-            f"Lemonade model failed after {max_retries} attempts: {e}"
-        )
-```
-
-### Neue Methoden
-
-#### `_reload_model()`
-
-Entlädt und lädt das Lemonade-Modell neu, wenn es unresponsive ist:
-
-```python
-def _reload_model(self) -> None:
-    """Unload and reload the Lemonade model when unresponsive."""
-    try:
-        if HAS_LEMONADE_SDK:
-            new_port = find_available_lemonade_port(
-                hostname=self.host or DEFAULT_HOST,
-                ports=DEFAULT_PORTS
-            )
-            if new_port and new_port != self._port:
-                self._port = new_port
-                self._init_client()
-    except Exception:
-        pass
-```
-
-#### `_init_client()`
-
-Initialisiert den OpenAI-Client neu mit der aktuellen Konfiguration:
-
-```python
-def _init_client(self) -> None:
-    """Reinitialize the OpenAI client with current configuration."""
-    self.client = OpenAI(
-        base_url=f"{self.base_url}/api/v1",
-        api_key=self.api_key
-    )
-```
-
-## Ablauf bei Unresponsiveness
-
-1. **Erster Versuch**: Modell antwortet nicht → `_reload_model()` aufgerufen
-2. **Zweiter Versuch**: Modell antwortet immer noch nicht → `_reload_model()` erneut aufgerufen
-3. **Dritter Versuch**: Erfolgt mit neu geladenem Modell
-4. **Falls alle Versuche fehlschlagen**: `RuntimeError` mit detaillierter Fehlermeldung
-
-## Voraussetzungen
-
-Die automatische Wiederherstellung erfordert das `lemonade-python-sdk`:
-
-```python
-try:
-    from lemonade_sdk.port_scanner import verify_lemonade_server
-    from lemonade_sdk.port_scanner import find_available_lemonade_port
-    HAS_LEMONADE_SDK = True
-except ImportError:
-    HAS_LEMONADE_SDK = False
-```
-
-Ohne das SDK wird weiterhin versucht, einen offenen Port zu finden, aber die automatische Wiederherstellung ist eingeschränkt.
-
-## Siehe auch
-
-- [LEMONADE_PORT_VERIFICATION_FIX.md](LEMONADE_PORT_VERIFICATION_FIX.md) - Port-Erkennungsfix
-- [MODEL_NAME_CASE_FIX.md](MODEL_NAME_CASE_FIX.md) - Modellnamen-Fallfix

diffron-0.1.4/docs/MODEL_NAME_CASE_FIX.md

@@ -1,88 +0,0 @@
-# Model Name Case Sensitivity Fix
-
-## Problem
-
-Diffron returned a **404 Not Found** error from the Lemonade server when using the `Qwen3.5-0.8B-GGUF` model.
-
-### Root Cause
-
-The Lemonade server is **case-sensitive** when matching model names. Diffron's curated model configuration used the wrong casing:
-
-| Location | Model Name (Before) | Model Name (Expected) |
-|----------|---------------------|------------------------|
-| `diffron/models.py` | `qwen3.5-0.8b-gguf` | `Qwen3.5-0.8B-GGUF` |
-
-This caused the OpenAI SDK to send a request with the wrong model identifier, resulting in:
-
-```
-Error code: 404 - {'detail': 'Not Found'}
-```
-
-## Solution
-
-### 1. Corrected Model Name
-
-Updated the canonical model name in `diffron/models.py`:
-
-```python
-# Before
-ModelConfig(name="qwen3.5-0.8b-gguf", ...)
-
-# After
-ModelConfig(name="Qwen3.5-0.8B-GGUF", ...)
-```
-
-### 2. Case-Insensitive Model Lookup
-
-Enhanced `get_model_config()` to accept case-insensitive input while always returning the **canonical name**:
-
-```python
-def get_model_config(model_name: str) -> Optional[ModelConfig]:
-    # Exact match first
-    for model in AVAILABLE_MODELS:
-        if model.name == model_name:
-            return model
-    # Case-insensitive fallback
-    for model in AVAILABLE_MODEOLS:
-        if model.name.lower() == model_name.lower():
-            return model
-    return None
-```
-
-This allows users to type `qwen3.5-0.8b-gguf`, `QWEN3.5-0.8B-GGUF`, etc., but Diffron always sends the correct name to the server.
-
-### 3. CLI Canonical Name Enforcement
-
-The `setup_model_cli()` now stores the **canonical name** from the config, not the user's raw input:
-
-```python
-# Always uses config.name (canonical) instead of args.model (user input)
-os.environ["DIFFRON_MODEL"] = config.name
-subprocess.run(["setx", "DIFFRON_MODEL", config.name], check=True)
-```
-
-## Testing
-
-```bash
-# All of these now work correctly:
-diffron-setup-model --model Qwen3.5-0.8B-GGUF   # exact
-diffron-setup-model --model qwen3.5-0.8b-gguf   # lowercase (fallback)
-diffron-setup-model --model QWEN3.5-0.8B-GGUF   # uppercase (fallback)
-```
-
-## Impact
-
-- **No breaking changes** - existing `DIFFRON_MODEL` values with wrong casing are now resolved correctly
-- **Improved UX** - users don't need to memorize exact casing
-- **Server compatibility** - correct model name is always sent to Lemonade
-
-## Files Changed
-
-| File | Change |
-|------|--------|
-| `diffron/models.py` | Fixed `Qwen3.5-0.8B-GGUF` name, added case-insensitive lookup |
-
-## Version
-
-- **Fixed in**: 0.1.3
-- **Date**: 2026-04-13