Update app.py
Browse files
app.py
CHANGED
|
@@ -15,12 +15,13 @@ summarizer = pipeline(
|
|
| 15 |
device=-1 # CPU only
|
| 16 |
)
|
| 17 |
|
| 18 |
-
CHUNK_SIZE = 900
|
| 19 |
|
| 20 |
# =========================
|
| 21 |
-
# Utilities
|
| 22 |
# =========================
|
| 23 |
def clean_text(text: str) -> str:
|
|
|
|
| 24 |
text = text.replace("‘", "'").replace("’", "'")
|
| 25 |
text = text.replace("“", '"').replace("”", '"')
|
| 26 |
text = re.sub(r"[.]{2,}", ".", text)
|
|
@@ -36,7 +37,9 @@ def clean_text(text: str) -> str:
|
|
| 36 |
result.append(s.strip())
|
| 37 |
return " ".join(result)
|
| 38 |
|
|
|
|
| 39 |
def chunk_text(text: str):
|
|
|
|
| 40 |
tokens = tokenizer.encode(text, add_special_tokens=False)
|
| 41 |
chunks = []
|
| 42 |
for i in range(0, len(tokens), CHUNK_SIZE):
|
|
@@ -45,11 +48,15 @@ def chunk_text(text: str):
|
|
| 45 |
chunks.append(chunk_text)
|
| 46 |
return chunks
|
| 47 |
|
|
|
|
| 48 |
def summarize_long_text(text: str) -> str:
|
|
|
|
| 49 |
if not text or len(text.strip()) == 0:
|
| 50 |
return "No text provided."
|
|
|
|
| 51 |
chunks = chunk_text(text)
|
| 52 |
summaries = []
|
|
|
|
| 53 |
for chunk in chunks:
|
| 54 |
summary = summarizer(
|
| 55 |
chunk,
|
|
@@ -58,10 +65,36 @@ def summarize_long_text(text: str) -> str:
|
|
| 58 |
do_sample=False
|
| 59 |
)[0]["summary_text"]
|
| 60 |
summaries.append(summary)
|
|
|
|
| 61 |
merged = " ".join(summaries)
|
| 62 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
|
| 64 |
def read_pdf(file) -> str:
|
|
|
|
| 65 |
try:
|
| 66 |
reader = PdfReader(file)
|
| 67 |
pages = [page.extract_text() or "" for page in reader.pages]
|
|
@@ -69,6 +102,7 @@ def read_pdf(file) -> str:
|
|
| 69 |
except Exception as e:
|
| 70 |
return f"PDF read error: {e}"
|
| 71 |
|
|
|
|
| 72 |
# =========================
|
| 73 |
# Main handler
|
| 74 |
# =========================
|
|
@@ -77,111 +111,38 @@ def process_input(text, file):
|
|
| 77 |
text = read_pdf(file)
|
| 78 |
return summarize_long_text(text)
|
| 79 |
|
| 80 |
-
# =========================
|
| 81 |
-
# Custom theme + stronger CSS for black text
|
| 82 |
-
# =========================
|
| 83 |
-
custom_theme = gr.themes.Default(
|
| 84 |
-
primary_hue="blue",
|
| 85 |
-
secondary_hue="gray",
|
| 86 |
-
neutral_hue="gray",
|
| 87 |
-
font=[gr.themes.GoogleFont('Inter'), 'ui-sans-serif', 'sans-serif'],
|
| 88 |
-
).set(
|
| 89 |
-
body_background_fill="#ffffff",
|
| 90 |
-
body_background_fill_dark="#ffffff",
|
| 91 |
-
block_background_fill="#ffffff",
|
| 92 |
-
block_background_fill_dark="#ffffff",
|
| 93 |
-
button_primary_background_fill="#2563eb",
|
| 94 |
-
button_primary_background_fill_hover="#1d4ed8",
|
| 95 |
-
button_primary_text_color="#ffffff",
|
| 96 |
-
button_primary_border_color="#2563eb",
|
| 97 |
-
body_text_color="#111111", # ← dark almost-black
|
| 98 |
-
body_text_color_dark="#111111",
|
| 99 |
-
)
|
| 100 |
-
|
| 101 |
-
custom_css = """
|
| 102 |
-
.gradio-container, body, .main, .app, .wrap {
|
| 103 |
-
background-color: #ffffff !important;
|
| 104 |
-
color: #000000 !important;
|
| 105 |
-
}
|
| 106 |
-
|
| 107 |
-
* {
|
| 108 |
-
color: #000000 !important; /* Force almost everything black */
|
| 109 |
-
}
|
| 110 |
-
|
| 111 |
-
label, .label, span, p, div, h1, h2, h3, h4, h5, h6 {
|
| 112 |
-
color: #000000 !important;
|
| 113 |
-
}
|
| 114 |
-
|
| 115 |
-
.gr-markdown, .prose, .prose p, .prose li, .prose strong {
|
| 116 |
-
color: #111111 !important;
|
| 117 |
-
}
|
| 118 |
-
|
| 119 |
-
input, textarea, .textbox, .gr-text-input, .gr-textarea {
|
| 120 |
-
color: #000000 !important;
|
| 121 |
-
background-color: #ffffff !important;
|
| 122 |
-
border: 1px solid #d1d5db !important;
|
| 123 |
-
}
|
| 124 |
-
|
| 125 |
-
::placeholder {
|
| 126 |
-
color: #6b7280 !important; /* Gray placeholder – easier to read */
|
| 127 |
-
}
|
| 128 |
-
|
| 129 |
-
.gr-button-primary {
|
| 130 |
-
color: #ffffff !important; /* White text on blue button */
|
| 131 |
-
border-radius: 8px !important;
|
| 132 |
-
font-weight: 600 !important;
|
| 133 |
-
}
|
| 134 |
-
|
| 135 |
-
header, .gr-top, .gr-header {
|
| 136 |
-
background-color: #f8f9fa !important;
|
| 137 |
-
border-bottom: 1px solid #e5e7eb !important;
|
| 138 |
-
padding: 12px 24px !important;
|
| 139 |
-
color: #000000 !important;
|
| 140 |
-
}
|
| 141 |
-
|
| 142 |
-
a {
|
| 143 |
-
color: #2563eb !important; /* Blue links for contrast */
|
| 144 |
-
}
|
| 145 |
-
|
| 146 |
-
a:hover {
|
| 147 |
-
color: #1d4ed8 !important;
|
| 148 |
-
}
|
| 149 |
-
"""
|
| 150 |
|
| 151 |
# =========================
|
| 152 |
# Gradio UI
|
| 153 |
# =========================
|
| 154 |
-
with gr.Blocks(
|
|
|
|
| 155 |
gr.Markdown(
|
| 156 |
-
"
|
|
|
|
|
|
|
|
|
|
| 157 |
)
|
| 158 |
-
|
| 159 |
-
|
| 160 |
-
|
| 161 |
-
"
|
|
|
|
| 162 |
)
|
| 163 |
-
|
| 164 |
-
|
| 165 |
-
|
| 166 |
-
|
| 167 |
-
|
| 168 |
-
|
| 169 |
-
placeholder="Paste your long article / text here...",
|
| 170 |
-
)
|
| 171 |
-
with gr.Column(scale=1, min_width=240):
|
| 172 |
-
file_input = gr.File(
|
| 173 |
-
label="Upload PDF (optional)",
|
| 174 |
-
file_types=[".pdf"],
|
| 175 |
-
)
|
| 176 |
-
|
| 177 |
-
summarize_btn = gr.Button("Summarize", variant="primary")
|
| 178 |
-
|
| 179 |
output = gr.Textbox(
|
| 180 |
-
lines=
|
| 181 |
-
label="Summary",
|
| 182 |
-
placeholder="
|
| 183 |
)
|
| 184 |
-
|
|
|
|
|
|
|
| 185 |
summarize_btn.click(
|
| 186 |
fn=process_input,
|
| 187 |
inputs=[text_input, file_input],
|
|
|
|
| 15 |
device=-1 # CPU only
|
| 16 |
)
|
| 17 |
|
| 18 |
+
CHUNK_SIZE = 900 # safe margin
|
| 19 |
|
| 20 |
# =========================
|
| 21 |
+
# Utilities
|
| 22 |
# =========================
|
| 23 |
def clean_text(text: str) -> str:
|
| 24 |
+
"""Fix quotes, spacing, repetition, and broken punctuation."""
|
| 25 |
text = text.replace("‘", "'").replace("’", "'")
|
| 26 |
text = text.replace("“", '"').replace("”", '"')
|
| 27 |
text = re.sub(r"[.]{2,}", ".", text)
|
|
|
|
| 37 |
result.append(s.strip())
|
| 38 |
return " ".join(result)
|
| 39 |
|
| 40 |
+
|
| 41 |
def chunk_text(text: str):
|
| 42 |
+
"""Token-aware chunking to avoid model overflow."""
|
| 43 |
tokens = tokenizer.encode(text, add_special_tokens=False)
|
| 44 |
chunks = []
|
| 45 |
for i in range(0, len(tokens), CHUNK_SIZE):
|
|
|
|
| 48 |
chunks.append(chunk_text)
|
| 49 |
return chunks
|
| 50 |
|
| 51 |
+
|
| 52 |
def summarize_long_text(text: str) -> str:
|
| 53 |
+
"""Summarize arbitrarily long text safely."""
|
| 54 |
if not text or len(text.strip()) == 0:
|
| 55 |
return "No text provided."
|
| 56 |
+
|
| 57 |
chunks = chunk_text(text)
|
| 58 |
summaries = []
|
| 59 |
+
|
| 60 |
for chunk in chunks:
|
| 61 |
summary = summarizer(
|
| 62 |
chunk,
|
|
|
|
| 65 |
do_sample=False
|
| 66 |
)[0]["summary_text"]
|
| 67 |
summaries.append(summary)
|
| 68 |
+
|
| 69 |
merged = " ".join(summaries)
|
| 70 |
+
cleaned_summary = clean_text(merged)
|
| 71 |
+
|
| 72 |
+
# Add study advice section
|
| 73 |
+
study_advice = """
|
| 74 |
+
|
| 75 |
+
---
|
| 76 |
+
|
| 77 |
+
### 📚 How to Study This Summary Effectively
|
| 78 |
+
|
| 79 |
+
Here are some proven techniques to help you learn and remember the material better:
|
| 80 |
+
|
| 81 |
+
- **Active Recall** — Cover the summary (or close your eyes) and try to explain each main point in your own words. This is one of the most powerful ways to strengthen memory.
|
| 82 |
+
- **Spaced Repetition** — Review this summary today, again in 2–3 days, then in one week. Use free apps like Anki or Quizlet to turn key points into flashcards.
|
| 83 |
+
- **Feynman Technique** — Pretend you're teaching this topic to a friend (or a 12-year-old). Explaining it simply reveals what you truly understand.
|
| 84 |
+
- **Self-Testing** — Create 3–5 questions from the summary (e.g. “What is…?”, “Why does…?”, “Give an example of…”). Answer them without looking.
|
| 85 |
+
- **Make Connections** — Draw a quick mind map or diagram linking the main ideas together. This helps see the big picture.
|
| 86 |
+
- **Apply It** — If possible, solve related problems, write a short paragraph, or discuss the topic with someone.
|
| 87 |
+
|
| 88 |
+
Re-reading alone is weak — **active engagement** is what makes information stick!
|
| 89 |
+
|
| 90 |
+
Good luck with your studies! 🚀
|
| 91 |
+
"""
|
| 92 |
+
|
| 93 |
+
return cleaned_summary + study_advice
|
| 94 |
+
|
| 95 |
|
| 96 |
def read_pdf(file) -> str:
|
| 97 |
+
"""Safely extract text from PDF."""
|
| 98 |
try:
|
| 99 |
reader = PdfReader(file)
|
| 100 |
pages = [page.extract_text() or "" for page in reader.pages]
|
|
|
|
| 102 |
except Exception as e:
|
| 103 |
return f"PDF read error: {e}"
|
| 104 |
|
| 105 |
+
|
| 106 |
# =========================
|
| 107 |
# Main handler
|
| 108 |
# =========================
|
|
|
|
| 111 |
text = read_pdf(file)
|
| 112 |
return summarize_long_text(text)
|
| 113 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 114 |
|
| 115 |
# =========================
|
| 116 |
# Gradio UI
|
| 117 |
# =========================
|
| 118 |
+
with gr.Blocks() as demo:
|
| 119 |
+
gr.Markdown("# 📄 Long Text Summarizer (Free-Tier Safe)")
|
| 120 |
gr.Markdown(
|
| 121 |
+
"• Handles **thousands of words**\n"
|
| 122 |
+
"• Supports **PDF upload**\n"
|
| 123 |
+
"• Optimized for **CPU / free tier**\n"
|
| 124 |
+
"• Includes **study tips** to help you learn better"
|
| 125 |
)
|
| 126 |
+
|
| 127 |
+
text_input = gr.Textbox(
|
| 128 |
+
lines=15,
|
| 129 |
+
label="Paste text (optional)",
|
| 130 |
+
placeholder="Paste your lecture notes, article, or book chapter here..."
|
| 131 |
)
|
| 132 |
+
|
| 133 |
+
file_input = gr.File(
|
| 134 |
+
label="Upload PDF (optional)",
|
| 135 |
+
file_types=[".pdf"]
|
| 136 |
+
)
|
| 137 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
output = gr.Textbox(
|
| 139 |
+
lines=14,
|
| 140 |
+
label="Summary + Study Advice",
|
| 141 |
+
placeholder="Your summary and learning tips will appear here..."
|
| 142 |
)
|
| 143 |
+
|
| 144 |
+
summarize_btn = gr.Button("Summarize & Get Study Tips", variant="primary")
|
| 145 |
+
|
| 146 |
summarize_btn.click(
|
| 147 |
fn=process_input,
|
| 148 |
inputs=[text_input, file_input],
|