File size: 11,077 Bytes
2605d55
aa4f694
2605d55
 
1d82a0b
2605d55
1d82a0b
 
 
2605d55
 
 
 
 
abd7b16
3e68ccf
 
57daf6a
3e68ccf
9fcc9ee
2605d55
 
 
 
 
 
3e68ccf
2605d55
 
aa4f694
 
f845b93
2605d55
f845b93
2605d55
f845b93
2605d55
 
f845b93
 
2605d55
 
 
 
 
 
 
 
f845b93
 
abd7b16
 
 
 
 
 
 
 
 
 
 
f845b93
 
2605d55
f845b93
2605d55
 
 
 
f845b93
2605d55
 
f845b93
2605d55
 
 
 
 
 
 
 
f845b93
 
2605d55
3e68ccf
2605d55
3e68ccf
 
 
 
2605d55
 
 
3e68ccf
e55d16a
2605d55
 
 
 
e55d16a
2605d55
3e68ccf
6b5a020
2605d55
 
 
 
3e68ccf
2605d55
e55d16a
2605d55
 
e55d16a
2605d55
 
3e68ccf
 
2605d55
3e68ccf
2605d55
3e68ccf
 
2605d55
 
 
 
 
 
3e68ccf
2605d55
e55d16a
2605d55
 
 
 
 
3e68ccf
2605d55
3e68ccf
2605d55
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
57daf6a
 
 
 
 
 
2605d55
 
6e4a820
2605d55
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
abd7b16
 
 
 
 
 
 
 
2605d55
 
 
 
 
 
 
57daf6a
 
 
 
 
 
 
2605d55
 
6e4a820
2605d55
 
 
abd7b16
 
 
 
 
2605d55
 
 
 
3e68ccf
2605d55
 
3e68ccf
2605d55
 
3e68ccf
2605d55
 
 
3e68ccf
1189403
 
 
2605d55
3e68ccf
2605d55
 
 
 
 
 
 
 
 
abd7b16
 
 
 
 
 
 
 
 
2605d55
 
 
 
 
3e68ccf
2605d55
 
 
 
 
 
 
c0b5a2b
2605d55
 
3e68ccf
2605d55
 
 
 
3e68ccf
2605d55
3e68ccf
 
2605d55
e55d16a
2605d55
3e68ccf
2605d55
e55d16a
33d58d5
2605d55
 
 
e55d16a
94d93d2
2605d55
 
 
e55d16a
2605d55
 
 
 
33d58d5
c6643c7
2605d55
 
 
 
 
 
33d58d5
2605d55
 
c6643c7
2605d55
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3e68ccf
 
 
ae312d0
 
 
 
3e68ccf
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
import datetime
import logging
import pathlib
import random
import tempfile
from typing import List

import json5
import streamlit as st
from langchain_community.chat_message_histories import (
    StreamlitChatMessageHistory
)
from langchain_core.messages import HumanMessage
from langchain_core.prompts import ChatPromptTemplate
# from transformers import AutoTokenizer

from global_config import GlobalConfig
from helpers import llm_helper, pptx_helper, text_helper


@st.cache_data
def _load_strings() -> dict:
    """
    Load various strings to be displayed in the app.
    :return: The dictionary of strings.
    """

    with open(GlobalConfig.APP_STRINGS_FILE, 'r', encoding='utf-8') as in_file:
        return json5.loads(in_file.read())


@st.cache_data
def _get_prompt_template(is_refinement: bool) -> str:
    """
    Return a prompt template.

    :param is_refinement: Whether this is the initial or refinement prompt.
    :return: The prompt template as f-string.
    """

    if is_refinement:
        with open(GlobalConfig.REFINEMENT_PROMPT_TEMPLATE, 'r', encoding='utf-8') as in_file:
            template = in_file.read()
    else:
        with open(GlobalConfig.INITIAL_PROMPT_TEMPLATE, 'r', encoding='utf-8') as in_file:
            template = in_file.read()

    return template


# @st.cache_resource
# def _get_tokenizer() -> AutoTokenizer:
#     """
#     Get Mistral tokenizer for counting tokens.
#
#     :return: The tokenizer.
#     """
#
#     return AutoTokenizer.from_pretrained(
#         pretrained_model_name_or_path=GlobalConfig.HF_LLM_MODEL_NAME
#     )


APP_TEXT = _load_strings()

# Session variables
CHAT_MESSAGES = 'chat_messages'
DOWNLOAD_FILE_KEY = 'download_file_name'
IS_IT_REFINEMENT = 'is_it_refinement'

logger = logging.getLogger(__name__)
progress_bar = st.progress(0, text='Setting up SlideDeck AI...')

texts = list(GlobalConfig.PPTX_TEMPLATE_FILES.keys())
captions = [GlobalConfig.PPTX_TEMPLATE_FILES[x]['caption'] for x in texts]
pptx_template = st.sidebar.radio(
    'Select a presentation template:',
    texts,
    captions=captions,
    horizontal=True
)


def display_page_header_content():
    """
    Display content in the page header.
    """

    st.title(APP_TEXT['app_name'])
    st.subheader(APP_TEXT['caption'])
    # st.markdown(
    #     '![Visitors](https://api.visitorbadge.io/api/visitors?path=https%3A%2F%2Fhuggingface.co%2Fspaces%2Fbarunsaha%2Fslide-deck-ai&countColor=%23263759)'  # noqa: E501
    # )


def display_page_footer_content():
    """
    Display content in the page footer.
    """

    st.text(APP_TEXT['tos'] + '\n\n' + APP_TEXT['tos2'])


def build_ui():
    """
    Display the input elements for content generation.
    """

    display_page_header_content()

    with st.expander('Usage Policies and Limitations'):
        display_page_footer_content()

    progress_bar.progress(50, text='Setting up chat interface...')
    set_up_chat_ui()


def set_up_chat_ui():
    """
    Prepare the chat interface and related functionality.
    """

    with st.expander('Usage Instructions'):
        st.write(GlobalConfig.CHAT_USAGE_INSTRUCTIONS)
        st.markdown(
            'SlideDeck AI is powered by'
            ' [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2)'
        )

    # view_messages = st.expander('View the messages in the session state')

    st.chat_message('ai').write(
        random.choice(APP_TEXT['ai_greetings'])
    )
    progress_bar.progress(100, text='Done!')
    progress_bar.empty()

    history = StreamlitChatMessageHistory(key=CHAT_MESSAGES)

    if _is_it_refinement():
        template = _get_prompt_template(is_refinement=True)
    else:
        template = _get_prompt_template(is_refinement=False)

    prompt_template = ChatPromptTemplate.from_template(template)

    # Since Streamlit app reloads at every interaction, display the chat history
    # from the save session state
    for msg in history.messages:
        msg_type = msg.type
        if msg_type == 'user':
            st.chat_message(msg_type).write(msg.content)
        else:
            st.chat_message(msg_type).code(msg.content, language='json')

    if prompt := st.chat_input(
        placeholder=APP_TEXT['chat_placeholder'],
        max_chars=GlobalConfig.LLM_MODEL_MAX_INPUT_LENGTH
    ):

        progress_bar_pptx = st.progress(0, 'Preparing to run...')
        if not text_helper.is_valid_prompt(prompt):
            st.error(
                'Not enough information provided!'
                ' Please be a little more descriptive and type a few words'
                ' with a few characters :)'
            )
            return

        logger.info('User input: %s | #characters: %d', prompt, len(prompt))
        st.chat_message('user').write(prompt)

        user_messages = _get_user_messages()
        user_messages.append(prompt)
        list_of_msgs = [
            f'{idx + 1}. {msg}' for idx, msg in enumerate(user_messages)
        ]
        list_of_msgs = '\n'.join(list_of_msgs)

        if _is_it_refinement():
            formatted_template = prompt_template.format(
                **{
                    'instructions': list_of_msgs,
                    'previous_content': _get_last_response()
                }
            )
        else:
            formatted_template = prompt_template.format(
                **{
                    'question': prompt,
                }
            )

        progress_bar_pptx.progress(5, 'Calling LLM...will retry if connection times out...')
        response: dict = llm_helper.hf_api_query({
            'inputs': formatted_template,
            'parameters': {
                'temperature': GlobalConfig.LLM_MODEL_TEMPERATURE,
                'min_length': GlobalConfig.LLM_MODEL_MIN_OUTPUT_LENGTH,
                'max_length': GlobalConfig.LLM_MODEL_MAX_OUTPUT_LENGTH,
                'max_new_tokens': GlobalConfig.LLM_MODEL_MAX_OUTPUT_LENGTH,
                'num_return_sequences': 1,
                'return_full_text': False,
                # "repetition_penalty": 0.0001
            },
            'options': {
                'wait_for_model': True,

                'use_cache': True
            }
        })

        if len(response) > 0 and 'generated_text' in response[0]:
            response: str = response[0]['generated_text'].strip()

        st.chat_message('ai').code(response, language='json')

        history.add_user_message(prompt)
        history.add_ai_message(response)

        # if GlobalConfig.COUNT_TOKENS:
        #     tokenizer = _get_tokenizer()
        #     tokens_count_in = len(tokenizer.tokenize(formatted_template))
        #     tokens_count_out = len(tokenizer.tokenize(response))
        #     logger.debug(
        #         'Tokens count:: input: %d, output: %d',
        #         tokens_count_in, tokens_count_out
        #     )

        # _display_messages_history(view_messages)

        # The content has been generated as JSON
        # There maybe trailing ``` at the end of the response -- remove them
        # To be careful: ``` may be part of the content as well when code is generated
        progress_bar_pptx.progress(50, 'Analyzing response...')
        response_cleaned = text_helper.get_clean_json(response)

        logger.info(
            'Cleaned JSON response:: original length: %d | cleaned length: %d',
            len(response), len(response_cleaned)
        )
        logger.debug('Cleaned JSON: %s', response_cleaned)

        # Now create the PPT file
        progress_bar_pptx.progress(75, 'Creating the slide deck...give it a moment...')
        generate_slide_deck(response_cleaned)
        progress_bar_pptx.progress(100, text='Done!')

        logger.info(
            '#messages in history / 2: %d',
            len(st.session_state[CHAT_MESSAGES]) / 2
        )


def generate_slide_deck(json_str: str):
    """
    Create a slide deck.

    :param json_str: The content in *valid* JSON format.
    """

    if DOWNLOAD_FILE_KEY in st.session_state:
        path = pathlib.Path(st.session_state[DOWNLOAD_FILE_KEY])
    else:
        temp = tempfile.NamedTemporaryFile(delete=False, suffix='.pptx')
        path = pathlib.Path(temp.name)
        st.session_state[DOWNLOAD_FILE_KEY] = str(path)

        if temp:
            temp.close()

    logger.debug('Creating PPTX file: %s...', st.session_state[DOWNLOAD_FILE_KEY])

    try:
        pptx_helper.generate_powerpoint_presentation(
            json_str,
            slides_template=pptx_template,
            output_file_path=path
        )

        _display_download_button(path)
    except ValueError as ve:
        st.error(
            f"{APP_TEXT['json_parsing_error']}"
            f"\n\nAdditional error info: {ve}"
            f"\n\nHere are some sample instructions that you could try to possibly fix this error;"
            f"if these don't work, try rephrasing or refreshing:"
            f"\n\n"
            "- Regenerate content and fix the JSON error."
            "\n- Regenerate content and fix the JSON error. Quotes inside quotes should be escaped."
        )
        logger.error('%s', APP_TEXT['json_parsing_error'])
        logger.error('Additional error info: %s', str(ve))
    except Exception as ex:
        st.error(APP_TEXT['content_generation_error'])
        logger.error('Caught a generic exception: %s', str(ex))


def _is_it_refinement() -> bool:
    """
    Whether it is the initial prompt or a refinement.

    :return: True if it is the initial prompt; False otherwise.
    """

    if IS_IT_REFINEMENT in st.session_state:
        return True

    if len(st.session_state[CHAT_MESSAGES]) >= 2:
        # Prepare for the next call
        st.session_state[IS_IT_REFINEMENT] = True
        return True

    return False


def _get_user_messages() -> List[str]:
    """
    Get a list of user messages submitted until now from the session state.

    :return: The list of user messages.
    """

    return [
        msg.content for msg in st.session_state[CHAT_MESSAGES] if isinstance(msg, HumanMessage)
    ]


def _get_last_response() -> str:
    """
    Get the last response generated by AI.

    :return: The response text.
    """

    return st.session_state[CHAT_MESSAGES][-1].content


def _display_messages_history(view_messages: st.expander):
    """
    Display the history of messages.

    :param view_messages: The list of AI and Human messages.
    """

    with view_messages:
        view_messages.json(st.session_state[CHAT_MESSAGES])


def _display_download_button(file_path: pathlib.Path):
    """
    Display a download button to download a slide deck.

    :param file_path: The path of the .pptx file.
    """

    with open(file_path, 'rb') as download_file:
        st.download_button(
            'Download PPTX file ⬇️',
            data=download_file,
            file_name='Presentation.pptx',
            key=datetime.datetime.now()
        )


def main():
    """
    Trigger application run.
    """

    build_ui()


if __name__ == '__main__':
    main()