Add logging; replace all print statements

app.py

@@ -1,6 +1,7 @@
 import base64
 import os
 import json5
+import logging
 import shutil
 import time
 import streamlit as st
@@ -16,6 +17,12 @@ APP_TEXT = json5.loads(open(GlobalConfig.APP_STRINGS_FILE, 'r').read())
 GB_CONVERTER = 2 ** 30
 
 
+logging.basicConfig(
+    level=GlobalConfig.LOG_LEVEL,
+    format='%(asctime)s - %(message)s',
+)
+
+
 @st.cache_data
 def get_contents_wrapper(text: str) -> str:
     """
@@ -82,11 +89,12 @@ def get_disk_used_percentage() -> float:
     used = used // GB_CONVERTER
     free = free // GB_CONVERTER
     used_perc = 100.0 * used / total
-    print(f'Total: {total} GB\n'
-          f'Used: {used} GB\n'
-          f'Free: {free} GB')
 
-    print('\n'.join(os.listdir()))
+    logging.debug(f'Total: {total} GB\n'
+                  f'Used: {used} GB\n'
+                  f'Free: {free} GB')
+
+    logging.debug('\n'.join(os.listdir()))
 
     return used_perc
 
@@ -147,24 +155,25 @@ def process_topic_inputs(topic: str, progress_bar):
     """
 
     topic_length = len(topic)
-    print(f'Input length:: topic: {topic_length}')
+    logging.debug(f'Input length:: topic: {topic_length}')
 
     if topic_length >= 10:
-        print(
+        logging.debug(
             f'Topic: {topic}\n'
         )
-        print('=' * 20)
+        logging.debug('=' * 20)
 
         target_length = min(topic_length, GlobalConfig.LLM_MODEL_MAX_INPUT_LENGTH)
 
         try:
+            logging.info('Calling LLM for content generation...')
             slides_content = get_contents_wrapper(topic[:target_length])
             content_length = len(slides_content)
 
-            print('=' * 20)
-            print(f'Slides content:\n{slides_content}')
-            print(f'Content length: {content_length}')
-            print('=' * 20)
+            logging.debug('=' * 20)
+            logging.debug(f'Slides content:\n{slides_content}')
+            logging.debug(f'Content length: {content_length}')
+            logging.debug('=' * 20)
             st.write(f'''Slides content:\n{slides_content}''')
             progress_bar.progress(100, text='Done!')
 
@@ -213,10 +222,11 @@ def process_slides_contents(text: str, progress_bar: st.progress):
     :param progress_bar: Progress bar for this step
     """
 
-    print('JSON button clicked')
+    logging.debug('JSON button clicked')
     json_str = ''
 
     try:
+        logging.info('Calling LLM for conversion...')
         json_str = get_json_wrapper(text)
     except Exception as ex:
         st.error(f'An exception occurred while trying to convert to JSON.'
@@ -226,9 +236,9 @@ def process_slides_contents(text: str, progress_bar: st.progress):
         # st.stop()
 
     # yaml_str = llm_helper.text_to_yaml(text)
-    print('=' * 20)
-    print(f'JSON:\n{json_str}')
-    print('=' * 20)
+    logging.debug('=' * 20)
+    logging.debug(f'JSON:\n{json_str}')
+    logging.debug('=' * 20)
     st.code(json_str, language='json')
 
     if len(json_str) > 0:
@@ -275,6 +285,7 @@ def process_slides_contents(text: str, progress_bar: st.progress):
         timestamp = time.time()
         output_file_name = f'{session_id}_{timestamp}.pptx'
 
+        logging.info('Creating PPTX file...')
         all_headers = pptx_helper.generate_powerpoint_presentation(
             json_str,
             as_yaml=False,
@@ -347,6 +358,7 @@ def show_bonus_stuff(
     urls_text.write(APP_TEXT['urls_info'])
 
     # Use the presentation title and the slides headers to find relevant info online
+    logging.info('Using Metaphor search...')
     ppt_text = ' '.join(ppt_headers)
     search_results = get_web_search_results_wrapper(ppt_text)
     md_text_items = []
@@ -356,6 +368,7 @@ def show_bonus_stuff(
 
     urls_list.markdown('\n\n'.join(md_text_items))
 
+    logging.info('Calling SDXL for image generation...')
     img_empty.write('')
     img_text.write(APP_TEXT['image_info'])
     image = get_ai_image_wrapper(ppt_text)

global_config.py

@@ -29,6 +29,7 @@ class GlobalConfig:
 
     METAPHOR_API_KEY = os.environ.get('METAPHOR_API_KEY', '')
 
+    LOG_LEVEL = 'INFO'
     APP_STRINGS_FILE = 'strings.json'
     PRELOAD_DATA_FILE = 'examples/example_02.json'
     SLIDES_TEMPLATE_FILE = 'langchain_templates/template_07.txt'

llm_helper.py

@@ -1,4 +1,5 @@
 import json
+import logging
 import time
 import metaphor_python as metaphor
 import requests
@@ -8,6 +9,11 @@ from langchain.llms import Clarifai
 from global_config import GlobalConfig
 
 
+logging.basicConfig(
+    level=GlobalConfig.LOG_LEVEL,
+    format='%(asctime)s - %(message)s',
+)
+
 prompt = None
 llm_contents = None
 llm_json = None
@@ -39,7 +45,7 @@ def get_llm(use_gpt: bool) -> Clarifai:
             verbose=True,
             # temperature=0.1,
         )
-    print(llm)
+    # print(llm)
 
     return llm
 
@@ -62,7 +68,7 @@ def generate_slides_content(topic: str) -> str:
         prompt = PromptTemplate.from_template(template_txt)
 
     formatted_prompt = prompt.format(topic=topic)
-    print(f'formatted_prompt:\n{formatted_prompt}')
+    # print(f'formatted_prompt:\n{formatted_prompt}')
 
     if llm_contents is None:
         llm_contents = get_llm(use_gpt=False)
@@ -91,7 +97,7 @@ def text_to_json(content: str) -> str:
         text = text.replace('<REPLACE_PLACEHOLDER>', content)
 
     text = text.strip()
-    print(text)
+    # print(text)
 
     if llm_json is None:
         llm_json = get_llm(use_gpt=True)
@@ -152,7 +158,7 @@ Output:
 '''
 
     text = text.strip()
-    print(text)
+    # print(text)
 
     if llm_json is None:
         llm_json = get_llm(use_gpt=True)
@@ -208,8 +214,8 @@ def get_ai_image(text: str) -> str:
         ]
     }
 
-    print('*** AI image generator...')
-    print(url)
+    # print('*** AI image generator...')
+    # print(url)
 
     start = time.time()
     response = requests.post(
@@ -219,16 +225,16 @@ def get_ai_image(text: str) -> str:
     )
     stop = time.time()
 
-    print('Response:', response, response.status_code)
-    print('Image generation took', stop - start, 'seconds')
+    # print('Response:', response, response.status_code)
+    logging.debug('Image generation took', stop - start, 'seconds')
     img_data = ''
 
     if response.ok:
-        print('*** Clarifai SDXL request: Response OK')
+        # print('*** Clarifai SDXL request: Response OK')
         json_data = json.loads(response.text)
         img_data = json_data['outputs'][0]['data']['image']['base64']
     else:
-        print('Image generation failed:', response.text)
+        logging.error('*** Image generation failed:', response.text)
 
     return img_data
 

pptx_helper.py

@@ -1,5 +1,6 @@
 from typing import List, Tuple
 import json5
+import logging
 import pptx
 import re
 import yaml
@@ -9,6 +10,11 @@ from global_config import GlobalConfig
 
 PATTERN = re.compile(r"^slide[ ]+\d+:", re.IGNORECASE)
 
+logging.basicConfig(
+    level=GlobalConfig.LOG_LEVEL,
+    format='%(asctime)s - %(message)s',
+)
+
 
 def remove_slide_number_from_heading(header: str) -> str:
     """
@@ -45,13 +51,13 @@ def generate_powerpoint_presentation(
         try:
             parsed_data = yaml.safe_load(structured_data)
         except yaml.parser.ParserError as ype:
-            print(f'*** YAML parse error: {ype}')
+            logging.error(f'*** YAML parse error: {ype}')
             parsed_data = {'title': '', 'slides': []}
     else:
         # The structured "JSON" might contain trailing commas, so using json5
         parsed_data = json5.loads(structured_data)
 
-    print(f"*** Using PPTX template: {GlobalConfig.PPTX_TEMPLATE_FILES[slides_template]['file']}")
+    logging.debug(f"*** Using PPTX template: {GlobalConfig.PPTX_TEMPLATE_FILES[slides_template]['file']}")
     presentation = pptx.Presentation(GlobalConfig.PPTX_TEMPLATE_FILES[slides_template]['file'])
 
     # The title slide
@@ -60,7 +66,7 @@ def generate_powerpoint_presentation(
     title = slide.shapes.title
     subtitle = slide.placeholders[1]
     title.text = parsed_data['title']
-    print(f'Title is: {title.text}')
+    logging.debug(f'Title is: {title.text}')
     subtitle.text = 'by Myself and SlideDeck AI :)'
     all_headers = [title.text, ]