.

caco3 · Jan 18, 2025 · c917a06 · c917a06
1 parent c634977
commit c917a06
Show file tree

Hide file tree

Showing 8 changed files with 116 additions and 75 deletions.
diff --git a/discussion-bot/readme.md b/discussion-bot/readme.md
@@ -1,12 +1,39 @@
 # Discussion Bot
-To parse a discussion and generate a response, acall it with
+The discussion bot can be used to create auto-responses on discussions.
+The `responses.md` file is used a sa database and contains responses and trigger words.
+When `response_generator.py` gets fed with a text, it parses the text, finds matching trigger words in the database and creates a list of suggestions.
+The suggestions, wrapped between the content of `response_intro.md` and `response_outro.md`, then get put together to a response.
+
+## Usage
+To parse a discussion and generate a response, call it with
 ```bash
 python response_generator.py --actor <ACTOR> --title <TITLE> --body <BODY>
 ```
 
 Example:
 ```bash
-python response_generator.py --actor "CaCO3" --title "Test" --body "--body "Hi all. I have always wrong value and reflections. Also home assistant does not get any data."
+python response_generator.py --actor "CaCO3" --title "Test" --body "Hi all. I have always wrong value and reflections. Also home assistant does not get any data."
+```
+
+## Data Format
+It is important to make sure that the `responses.md` file has the right format!
+Altrough it is a markdown file and can be rendered by a markdown viewer, it actually is parsed by the script.
+Thus the format must be as following:
+- A response starts with a title line and must have a leading `#` followed by a whitespace.
+- The title can be empty. It also can be used to note some comments.
+- The title line is followed by a one-line response. All markdown keywords are allowed, but it must stay on one line.
+- Then one or more trigger words get listed as a list. Each line has one trigger word ar phrase.
+- It is suggested to separate the responses with empty lines for easier visual separation, although this is optional.
+
+Example:
+```
+# Wrong Transitions
+Check parameter numberanalogtodigittransitionstart
+- lagging
+- late transition
+- early transition
 ```
-# Testing
+
+
+## Testing
 See [test/readme.md](test/readme.md)
diff --git a/discussion-bot/response_data.txt b/discussion-bot/response_data.txt
diff --git a/discussion-bot/response_generator.py b/discussion-bot/response_generator.py
@@ -6,39 +6,47 @@
 class Data_Record_Parser():
     data_records = []
 
-    def __init__(self, data_file):
-        records_count = 0
-        record_started = False
-        with open(data_file, "r") as f:
+    def __init__(self, response_file):
+        self.parse_responses(response_file)
+
+    def parse_responses(self, response_file):
+        with open(response_file, "r") as f:
             for line in f:
                 line = line.strip()
-                if line.startswith(";"):  # Comment line
+                if line == "":  # Empty line
                     continue
 
-                if line != "": # Line contains data
-                    if not record_started:
-                        record_started = True
-                        self.data_records.append({"response": None, "trigger_patterns": []})
-                        records_count += 1
-
-                    if self.data_records[records_count - 1]["response"] == None:  # Record has no response yet -> Add line as response
-                        self.data_records[records_count - 1]["response"] = line
-                    else: # Record has a response -> Add line as trigger pattern
-                        self.data_records[records_count - 1]["trigger_patterns"].append(line)
-                else:  # Empty line -> end of record
-                    record_started = False
-                    continue
+                if line.startswith("#"):  # Title line -> response start
+                    if len(self.data_records) > 1:  # This is not the first response
+                        # Make sure the previous response record has at least one trigger pattern
+                        if len(self.data_records[-2]["trigger_patterns"]) == 0:  # Number of trigger patterns of the 2nd last record
+                            raise Exception("The previous response record has no trigger patterns!")
+
+                    self.data_records.append({"title": line.replace("# ", ""), "response": None, "trigger_patterns": []})
+
+                elif line.startswith("-"):  # Trigger pattern line
+                    if self.data_records[-1]["response"] == None:  # Record has no response yet
+                        raise Exception("The response record has no response line!")
+                    self.data_records[-1]["trigger_patterns"].append(line.replace("- ", ""))  # Add to trigger patterns
+
+                else:  # Response line
+                    self.data_records[-1]["response"] = line  # Add as response line
+
+        log.info(f"Found {len(self.data_records)} data records")
+        # import pprint
+        # pprint.pprint(self.data_records, width=300)
 
-        log.info(f"Found {records_count} data records")
 
     def get_records(self):
         return self.data_records
 
 
 class Response_Generator():
-    def __init__(self, data_file):
-        parser = Data_Record_Parser(data_file)
+    def __init__(self, response_file, intro_file, outro_file):
+        parser = Data_Record_Parser(response_file)
         self.data_records = parser.get_records()
+        self.intro = open(intro_file, "r").read()
+        self.outro = open(outro_file, "r").read()
 
     def process_discussion(self, actor, title, body):
         """Analyses the given title and body and creates a response based on the found trigger words
@@ -55,22 +63,13 @@ def process_discussion(self, actor, title, body):
                     break
 
         if len(responses) > 0:  # At least one trigger pattern matched
-            response = f"Hi @{actor}"
-            response += """
-    
-I am the (experimental) AIOTED-Bot 🤖
-
-Have you already checked our [documentation](https://jomjol.github.io/AI-on-the-edge-device-docs)?
-I analyzed your question and I might be able to help you. Here are some useful links based on your input:
-
-"""
+            response = self.intro + "\n"
+            response = response.replace("{actor}", "{" + actor + "}")
 
             for finding in responses:
                 response += f" - **{finding[0]}:** {finding[1]}\n"
 
-            response += """
-If this all does not help and you need support of an experienced user or developer to look into it (after you really studied the documentation), you can write a reply with the text `help-needed`.
-Please be aware that we are a small team and run this project in our private, free time, so our time to give support is really limitted!"""
+            response += "\n" +self.outro
         else:
             response = ""
 
@@ -92,7 +91,9 @@ def process_discussion(self, actor, title, body):
     title = args.title
     body = args.body
 
-    rg = Response_Generator(os.path.dirname(__file__) + "/" + "response_data.txt")
+    rg = Response_Generator(os.path.dirname(__file__) + "/" + "responses.md",
+                            os.path.dirname(__file__) + "/" + "response_intro.md",
+                            os.path.dirname(__file__) + "/" + "response_outro.md")
 
     response = rg.process_discussion(actor, title, body)
 

diff --git a/discussion-bot/response_intro.md b/discussion-bot/response_intro.md
@@ -0,0 +1,6 @@
+Hi @{actor}
+
+I am the (experimental) [AIOTED](https://jomjol.github.io/AI-on-the-edge-device-docs/FAQs/#what-does-aioted-mean)-Bot 🤖
+
+Have you already checked our [documentation](https://jomjol.github.io/AI-on-the-edge-device-docs)?
+I analyzed your question and I might be able to help you. Here are some useful links based on your input:
diff --git a/discussion-bot/response_outro.md b/discussion-bot/response_outro.md
@@ -0,0 +1,2 @@
+If this all does not help and you need support of an experienced user or developer to look into it (after you really studied the documentation), you can write a reply with the text `help-needed`.
+Please be aware that we are a small team and run this project in our private and free time, so our time to give support is really limited!
diff --git a/discussion-bot/responses.md b/discussion-bot/responses.md
@@ -0,0 +1,33 @@
+# Rate too high
+See the [Rate too high in our FAQ](https://jomjol.github.io/AI-on-the-edge-device-docs/FAQs/#rate-too-high-read)
+- Rate too high 
+
+# Homeassistant
+Check HomeAssistant
+- homeassistant
+- Home Assistant
+- Home-assistant
+
+# Wrong Values
+Improve ROI, ...
+- wrong values
+
+# Wrong Transitions
+Check parameter numberanalogtodigittransitionstart
+- lagging
+- late transition
+- early transition
+- transition
+
+# Reflections
+Improve LED, diffusor, see xxx
+- reflection
+
+# LCD, Dot Display, Matrix
+See LCD/Matrix xxx
+LCD, matrix
+
+# shifted
+Make sure to set []() correctly.
+- shift
+- decimal place
diff --git a/discussion-bot/test/readme.md b/discussion-bot/test/readme.md
@@ -7,13 +7,15 @@ pip install -r requirements.txt
 ```
 
 ## Test Data
-The `discussions.json` file contains a list of discussion threads (first comment only).
+The `discussions.json.tar.gz` file contains a list of discussion threads (first comment only).
 The list got generated with the following cmd:
 ```bash
 for i in {1..10}; do wget "https://api.github.com/repos/jomjol/AI-on-the-edge-device/discussions?per_page=100&page=$i" -O discussions_$i.json; done
 ```
 
 Afterward I removed the first and last line of each file (remove the `[` and `]`) merged the files and wrap the new one again with `[` and `]`.
 
+The `discussions.json.tar.gz` file gets extracted to `discussions.json` if it does not exist yet.
+
 ## Run Tests
 If you run the `test.py` script it will generate a HTML page which shows all discussions and the matching responses.
diff --git a/discussion-bot/test/test.py b/discussion-bot/test/test.py
@@ -64,7 +64,9 @@
     output_file.write("<th class=vertical-separator>Response</th>")
     output_file.write("</tr>")
 
-    rg = Response_Generator(os.path.dirname(__file__) + "/../" + "response_data.txt")
+    rg = Response_Generator(os.path.dirname(__file__) + "/../" + "responses.md",
+                            os.path.dirname(__file__) + "/../" + "response_intro.md",
+                            os.path.dirname(__file__) + "/../" + "response_outro.md")
 
     for i in range(len(discussions) - 1, -1, -1):
         log.info(f"Processing discussion {len(discussions) - i}/{len(discussions)}...")
@@ -94,8 +96,11 @@
 
         output_file.write(f"<tr><td><h3>{title_html}</h3>{body_html}</td><td class=vertical-separator>{response_html}</td></tr>")
 
+        # break  # testing
+
+    output_file.write("</table>")
     output_file.write("</body>")
     output_file.write("</html>")
     output_file.close()
 
-    log.info("Done. Check the result.html file.")
+    log.info("Done. Please open the result.html file in a webbrowser.")
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,2 @@
		If this all does not help and you need support of an experienced user or developer to look into it (after you really studied the documentation), you can write a reply with the text `help-needed`.
		Please be aware that we are a small team and run this project in our private and free time, so our time to give support is really limited!