Merge pull request #15 from barun-saha/visual

global_config.py

@@ -25,8 +25,8 @@ class GlobalConfig:
     PRELOAD_DATA_FILE = 'examples/example_02.json'
     SLIDES_TEMPLATE_FILE = 'langchain_templates/template_combined.txt'
     # JSON_TEMPLATE_FILE = 'langchain_templates/text_to_json_template_02.txt'
-    INITIAL_PROMPT_TEMPLATE = 'langchain_templates/chat_prompts/initial_template_v2_steps.txt'
-    REFINEMENT_PROMPT_TEMPLATE = 'langchain_templates/chat_prompts/refinement_template_v2_steps.txt'
+    INITIAL_PROMPT_TEMPLATE = 'langchain_templates/chat_prompts/initial_template_v3_two_cols.txt'
+    REFINEMENT_PROMPT_TEMPLATE = 'langchain_templates/chat_prompts/refinement_template_v3_two_cols.txt'
 
     PPTX_TEMPLATE_FILES = {
         'Basic': {

helpers/pptx_helper.py

@@ -79,10 +79,7 @@ def generate_powerpoint_presentation(
         GlobalConfig.PPTX_TEMPLATE_FILES[slides_template]['file']
     )
     presentation = pptx.Presentation(GlobalConfig.PPTX_TEMPLATE_FILES[slides_template]['file'])
-
-    slide_width_inch = EMU_TO_INCH_SCALING_FACTOR * presentation.slide_width
-    slide_height_inch = EMU_TO_INCH_SCALING_FACTOR * presentation.slide_height
-    logger.debug('Slide width: %f, height: %f', slide_width_inch, slide_height_inch)
+    slide_width_inch, slide_height_inch = _get_slide_width_height_inches(presentation)
 
     # The title slide
     title_slide_layout = presentation.slide_layouts[0]
@@ -97,26 +94,31 @@ def generate_powerpoint_presentation(
     subtitle.text = 'by Myself and SlideDeck AI :)'
     all_headers = [title.text, ]
 
-    # Add contents in a loop
+    # Add content in a loop
     for a_slide in parsed_data['slides']:
-        bullet_slide_layout = presentation.slide_layouts[1]
-        slide = presentation.slides.add_slide(bullet_slide_layout)
-
-        if not _handle_step_by_step_process(
-            slide=slide,
-            slide_json=a_slide,
-            slide_width_inch=slide_width_inch,
-            slide_height_inch=slide_height_inch,
-        ):
-            _handle_default_display(slide, a_slide)
-
-        _handle_key_message(
-            slide=slide,
+        is_processing_done = _handle_double_col_layout(
+            presentation=presentation,
             slide_json=a_slide,
             slide_width_inch=slide_width_inch,
-            slide_height_inch=slide_height_inch,
+            slide_height_inch=slide_height_inch
         )
 
+        if not is_processing_done:
+            is_processing_done = _handle_step_by_step_process(
+                presentation=presentation,
+                slide_json=a_slide,
+                slide_width_inch=slide_width_inch,
+                slide_height_inch=slide_height_inch
+            )
+
+        if not is_processing_done:
+            _handle_default_display(
+                presentation=presentation,
+                slide_json=a_slide,
+                slide_width_inch=slide_width_inch,
+                slide_height_inch=slide_height_inch
+            )
+
     # The thank-you slide
     last_slide_layout = presentation.slide_layouts[0]
     slide = presentation.slides.add_slide(last_slide_layout)
@@ -150,16 +152,23 @@ def get_flat_list_of_contents(items: list, level: int) -> List[Tuple]:
 
 
 def _handle_default_display(
-        slide: pptx.slide.Slide,
+        presentation: pptx.Presentation,
         slide_json: dict,
+        slide_width_inch: float,
+        slide_height_inch: float
 ):
     """
     Display a list of text in a slide.
 
-    :param slide: The slide to be processed.
+    :param presentation: The presentation object.
     :param slide_json: The content of the slide as JSON data.
+    :param slide_width_inch: The width of the slide in inches.
+    :param slide_height_inch: The height of the slide in inches.
     """
 
+    bullet_slide_layout = presentation.slide_layouts[1]
+    slide = presentation.slides.add_slide(bullet_slide_layout)
+
     shapes = slide.shapes
     title_shape = shapes.title
     body_shape = shapes.placeholders[1]
@@ -172,14 +181,99 @@ def _handle_default_display(
 
     flat_items_list = get_flat_list_of_contents(slide_json['bullet_points'], level=0)
 
-    for an_item in flat_items_list:
-        paragraph = text_frame.add_paragraph()
-        paragraph.text = an_item[0].removeprefix(STEP_BY_STEP_PROCESS_MARKER)
-        paragraph.level = an_item[1]
+    for idx, an_item in enumerate(flat_items_list):
+        if idx == 0:
+            text_frame.text = an_item[0].removeprefix(STEP_BY_STEP_PROCESS_MARKER)
+        else:
+            paragraph = text_frame.add_paragraph()
+            paragraph.text = an_item[0].removeprefix(STEP_BY_STEP_PROCESS_MARKER)
+            paragraph.level = an_item[1]
+
+    _handle_key_message(
+        the_slide=slide,
+        slide_json=slide_json,
+        slide_height_inch=slide_height_inch,
+        slide_width_inch=slide_width_inch
+    )
+
+
+def _handle_double_col_layout(
+        presentation: pptx.Presentation(),
+        slide_json: dict,
+        slide_width_inch: float,
+        slide_height_inch: float
+) -> bool:
+    """
+    Add a slide with a double column layout for comparison.
+
+    :param presentation: The presentation object.
+    :param slide_json: The content of the slide as JSON data.
+    :param slide_width_inch: The width of the slide in inches.
+    :param slide_height_inch: The height of the slide in inches.
+    :return: True if double col layout has been added; False otherwise.
+    """
+
+    if 'bullet_points' in slide_json and slide_json['bullet_points']:
+        double_col_content = slide_json['bullet_points']
+
+        if double_col_content and (
+                len(double_col_content) == 2
+        ) and isinstance(double_col_content[0], dict) and isinstance(double_col_content[1], dict):
+            slide = presentation.slide_layouts[4]
+            slide = presentation.slides.add_slide(slide)
+
+            shapes = slide.shapes
+            title_placeholder = shapes.title
+            title_placeholder.text = remove_slide_number_from_heading(slide_json['heading'])
+
+            left_heading, right_heading = shapes.placeholders[1], shapes.placeholders[3]
+            left_col, right_col = shapes.placeholders[2], shapes.placeholders[4]
+            left_col_frame, right_col_frame = left_col.text_frame, right_col.text_frame
+
+            if 'heading' in double_col_content[0]:
+                left_heading.text = double_col_content[0]['heading']
+            if 'bullet_points' in double_col_content[0]:
+                flat_items_list = get_flat_list_of_contents(
+                    double_col_content[0]['bullet_points'], level=0
+                )
+
+                for idx, an_item in enumerate(flat_items_list):
+                    if idx == 0:
+                        left_col_frame.text = an_item[0].removeprefix(STEP_BY_STEP_PROCESS_MARKER)
+                    else:
+                        paragraph = left_col_frame.add_paragraph()
+                        paragraph.text = an_item[0].removeprefix(STEP_BY_STEP_PROCESS_MARKER)
+                        paragraph.level = an_item[1]
+
+            if 'heading' in double_col_content[1]:
+                right_heading.text = double_col_content[1]['heading']
+            if 'bullet_points' in double_col_content[1]:
+                flat_items_list = get_flat_list_of_contents(
+                    double_col_content[1]['bullet_points'], level=0
+                )
+
+                for idx, an_item in enumerate(flat_items_list):
+                    if idx == 0:
+                        right_col_frame.text = an_item[0].removeprefix(STEP_BY_STEP_PROCESS_MARKER)
+                    else:
+                        paragraph = right_col_frame.add_paragraph()
+                        paragraph.text = an_item[0].removeprefix(STEP_BY_STEP_PROCESS_MARKER)
+                        paragraph.level = an_item[1]
+
+            _handle_key_message(
+                the_slide=slide,
+                slide_json=slide_json,
+                slide_height_inch=slide_height_inch,
+                slide_width_inch=slide_width_inch
+            )
+
+            return True
+
+    return False
 
 
 def _handle_step_by_step_process(
-        slide: pptx.slide.Slide,
+        presentation: pptx.Presentation,
         slide_json: dict,
         slide_width_inch: float,
         slide_height_inch: float
@@ -187,11 +281,11 @@ def _handle_step_by_step_process(
     """
     Add shapes to display a step-by-step process in the slide, if available.
 
-    :param slide: The slide to be processed.
+    :param presentation: The presentation object.
     :param slide_json: The content of the slide as JSON data.
     :param slide_width_inch: The width of the slide in inches.
     :param slide_height_inch: The height of the slide in inches.
-    :return True is this slide has a step-by-step process depiction; False otherwise.
+    :return True if this slide has a step-by-step process depiction added; False otherwise.
     """
 
     if 'bullet_points' in slide_json and slide_json['bullet_points']:
@@ -228,6 +322,8 @@ def _handle_step_by_step_process(
         ):
             return False
 
+        bullet_slide_layout = presentation.slide_layouts[1]
+        slide = presentation.slides.add_slide(bullet_slide_layout)
         shapes = slide.shapes
         shapes.title.text = remove_slide_number_from_heading(slide_json['heading'])
 
@@ -277,7 +373,7 @@ def _handle_step_by_step_process(
 
 
 def _handle_key_message(
-        slide: pptx.slide.Slide,
+        the_slide: pptx.slide.Slide,
         slide_json: dict,
         slide_width_inch: float,
         slide_height_inch: float
@@ -285,7 +381,7 @@ def _handle_key_message(
     """
     Add a shape to display the key message in the slide, if available.
 
-    :param slide: The slide to be processed.
+    :param the_slide: The slide to be processed.
     :param slide_json: The content of the slide as JSON data.
     :param slide_width_inch: The width of the slide in inches.
     :param slide_height_inch: The height of the slide in inches.
@@ -296,7 +392,7 @@ def _handle_key_message(
         width = pptx.util.Inches(slide_width_inch / 2.3)
         top = pptx.util.Inches(slide_height_inch - height.inches - 0.1)
         left = pptx.util.Inches((slide_width_inch - width.inches) / 2)
-        shape = slide.shapes.add_shape(
+        shape = the_slide.shapes.add_shape(
             MSO_AUTO_SHAPE_TYPE.ROUNDED_RECTANGLE,
             left=left,
             top=top,
@@ -306,6 +402,21 @@ def _handle_key_message(
         shape.text = slide_json['key_message']
 
 
+def _get_slide_width_height_inches(presentation: pptx.Presentation) -> Tuple[float, float]:
+    """
+    Get the dimensions of a slide in inches.
+
+    :param presentation: The presentation object.
+    :return: The width and the height.
+    """
+
+    slide_width_inch = EMU_TO_INCH_SCALING_FACTOR * presentation.slide_width
+    slide_height_inch = EMU_TO_INCH_SCALING_FACTOR * presentation.slide_height
+    # logger.debug('Slide width: %f, height: %f', slide_width_inch, slide_height_inch)
+
+    return slide_width_inch, slide_height_inch
+
+
 if __name__ == '__main__':
     _JSON_DATA = '''
     {
@@ -363,6 +474,28 @@ if __name__ == '__main__':
             ],
             "key_message": ""
         },
+        {
+            "heading": "Pros and Cons: Deep Learning vs. Classical Machine Learning",
+            "bullet_points": [
+                {
+                    "heading": "Classical Machine Learning",
+                    "bullet_points": [
+                        "Interpretability: Easy to understand the model",
+                        "Faster Training: Quicker to train models",
+                        "Scalability: Can handle large datasets"
+                    ]
+                },
+                {
+                    "heading": "Deep Learning",
+                    "bullet_points": [
+                        "Handling Complex Data: Can learn from raw data",
+                        "Feature Extraction: Automatically learns features",
+                        "Improved Accuracy: Achieves higher accuracy"
+                    ]
+                }
+            ],
+            "key_message": ""
+        },
         {
             "heading": "Pros of AI",
             "bullet_points": [

langchain_templates/chat_prompts/initial_template_v3_two_cols.txt

@@ -0,0 +1,83 @@
+You are a helpful, intelligent chatbot. Create the slides for a presentation on the given topic.
+Include main headings for each slide, detailed bullet points for each slide.
+Add relevant content to each slide.
+The content of each slide should be verbose, descriptive, and very detailed.
+If relevant, add one or two examples to illustrate the concept.
+Unless explicitly specified with the topic, create about 10 slides.
+
+
+### Topic:
+{question}
+
+
+The output must be only a valid and syntactically correct JSON adhering to the following schema:
+{{
+    "title": "Presentation Title",
+    "slides": [
+        {{
+            "heading": "Heading for the First Slide",
+            "bullet_points": [
+                "First bullet point",
+                [
+                    "Sub-bullet point 1",
+                    "Sub-bullet point 2"
+                ],
+                "Second bullet point"
+            ],
+            "key_message": ""
+        }},
+        {{
+            "heading": "Heading for the Second Slide",
+            "bullet_points": [
+                "First bullet point",
+                "Second bullet item",
+                "Third bullet point"
+            ],
+            "key_message": "The key message conveyed in this slide"
+        }},
+        {{
+            "heading": "A slide that describes a step-by-step/sequential process",
+            "bullet_points": [
+                ">> The first step of the process (begins with special marker >>)",
+                ">> A second step (begins with >>)",
+                ">> Third step",
+            ],
+            "key_message": ""
+        }},
+        {{
+            "heading": "A slide with a double column layout (useful for side-by-side comparison/contrasting of related concepts, e.g., pros & cons, advantages & risks, old approach vs. modern approach, and so on)",
+            "bullet_points": [
+                {{
+                    "heading": "Heading of the left column",
+                    "bullet_points": [
+                        "First bullet point",
+                        "Second bullet item",
+                        "Third bullet point"
+                    ]
+                }},
+                {{
+                    "heading": "Heading of the right column",
+                    "bullet_points": [
+                        "First bullet point",
+                        "Second bullet item",
+                        "Third bullet point"
+                    ]
+                }}
+            ],
+            "key_message": ""
+        }}
+    ]
+}}
+
+
+### Some more hints on the slide content and JSON output format:
+- For two or three important slides, generate the key message that those slides convey and assign
+them to the `key_message` elements of JSON output.
+- Identify if a slide describes a step-by-step/sequential process, then begin the bullet points
+with a special marker >>. Limit this to max two or three slides.
+- Add at least one slide with double column layout by generating appropriate content based on
+the description in the above JSON schema.
+
+
+### Output:
+```json

langchain_templates/chat_prompts/refinement_template_v3_two_cols.txt

@@ -0,0 +1,94 @@
+You are a helpful, intelligent chatbot. You follow instructions to refine an existing slide deck.
+A list of user instructions is provided below in sequential order -- from the oldest to the latest.
+The previously generated content of the slide deck in JSON format is also provided.
+Follow the instructions to revise the content of the previously generated slides of the presentation on the given topic.
+Include main headings for each slide, detailed bullet points for each slide.
+Add relevant content to each slide.
+The content of the slides should be descriptive, verbose, and detailed.
+If relevant, add one or two examples to illustrate the concept.
+Unless explicitly specified with the topic, create about 10 slides.
+You also fix any syntax error that may be present in the JSON-formatted content.
+
+A slide that describes a step-by-step/sequential process begins the bullet points
+with a special marker >>
+
+
+### List of instructions:
+{instructions}
+
+
+### Previously generated slide deck content as JSON:
+{previous_content}
+
+
+The output must be only a valid and syntactically correct JSON adhering to the following schema:
+{{
+    "title": "Presentation Title",
+    "slides": [
+        {{
+            "heading": "Heading for the First Slide",
+            "bullet_points": [
+                "First bullet point",
+                [
+                    "Sub-bullet point 1",
+                    "Sub-bullet point 2"
+                ],
+                "Second bullet point"
+            ],
+            "key_message": ""
+        }},
+        {{
+            "heading": "Heading for the Second Slide",
+            "bullet_points": [
+                "First bullet point",
+                "Second bullet item",
+                "Third bullet point"
+            ],
+            "key_message": "The key message conveyed in this slide"
+        }},
+        {{
+            "heading": "A slide that describes a step-by-step/sequential process",
+            "bullet_points": [
+                ">> The first step of the process (begins with special marker >>)",
+                ">> A second step (begins with >>)",
+                ">> Third step",
+            ],
+            "key_message": ""
+        }},
+        {{
+            "heading": "A slide with a double column layout (useful for side-by-side comparison/contrasting of related concepts, e.g., pros & cons, advantages & risks, old approach vs. modern approach, and so on)",
+            "bullet_points": [
+                {{
+                    "heading": "Heading of the left column",
+                    "bullet_points": [
+                        "First bullet point",
+                        "Second bullet item",
+                        "Third bullet point"
+                    ]
+                }},
+                {{
+                    "heading": "Heading of the right column",
+                    "bullet_points": [
+                        "First bullet point",
+                        "Second bullet item",
+                        "Third bullet point"
+                    ]
+                }}
+            ],
+            "key_message": ""
+        }}
+    ]
+}}
+
+
+### Some more hints on the slide content and JSON output format:
+- For two or three important slides, generate the key message that those slides convey and assign
+them to the `key_message` elements of JSON output.
+- Identify if a slide describes a step-by-step/sequential process, then begin the bullet points
+with a special marker >>. Limit this to max two or three slides.
+- Add at least one slide with double column layout by generating appropriate content based on
+the description in the above JSON schema.
+
+
+### Output:
+```json