Create app.py

app.py

@@ -0,0 +1,126 @@
+import os, io, re, json
+import streamlit as st
+from PIL import Image
+
+import pdfplumber           # text layer for PDFs
+from pdf2image import convert_from_bytes  # render image-only PDFs
+import pytesseract          # OCR
+
+APP_TITLE = "Flyer → Image-to-Text (Where & When)"
+st.set_page_config(page_title=APP_TITLE, page_icon="📰", layout="centered")
+st.title("📰 Flyer → Where & When")
+st.caption("Upload a flyer (PDF/JPG/PNG). I’ll read it and pull out the date, time, and location.")
+
+# ----------- helpers -----------
+
+DATE_PAT = r"(\b\w{3,9}\b\s+\d{1,2}(?:,\s*\d{4})?)|\b\d{1,2}/\d{1,2}/\d{2,4}\b"
+TIME_PAT = r"\b(\d{1,2}:\d{2}\s*[AP]M|\d{1,2}\s*[AP]M)\b"
+LOC_HINTS = r"(Location|Where|Address|Venue|At)[:\-\s]+(.+)"
+
+def pdf_bytes_to_text(pdf_bytes: bytes) -> str:
+    # Try selectable text first
+    parts = []
+    try:
+        with pdfplumber.open(io.BytesIO(pdf_bytes)) as pdf:
+            for page in pdf.pages:
+                t = page.extract_text() or ""
+                if t.strip():
+                    parts.append(t)
+    except Exception:
+        pass
+    if parts:
+        return "\n\n".join(parts)
+
+    # Fallback to OCR (image-only PDFs)
+    txts = []
+    images = convert_from_bytes(pdf_bytes, fmt="png", dpi=200)
+    for img in images:
+        txts.append(pytesseract.image_to_string(img))
+    return "\n".join(txts)
+
+def image_bytes_to_text(img_bytes: bytes) -> str:
+    img = Image.open(io.BytesIO(img_bytes)).convert("RGB")
+    return pytesseract.image_to_string(img)
+
+def extract_where_when(text: str):
+    # Date(s)
+    dates = re.findall(DATE_PAT, text, flags=re.I)
+    # re.findall returns tuples from the alternation; flatten/clean:
+    clean_dates = []
+    for tup in dates:
+        s = [t for t in tup if t]
+        if s: clean_dates.append(s[0])
+
+    # Time(s)
+    times = re.findall(TIME_PAT, text, flags=re.I)
+
+    # Location (first good hint line)
+    loc_match = re.search(LOC_HINTS, text, flags=re.I)
+    location = loc_match.group(2).strip() if loc_match else ""
+
+    return {
+        "dates_found": clean_dates,
+        "times_found": times,
+        "location": location
+    }
+
+# ----------- UI -----------
+
+uploaded = st.file_uploader("Upload flyer", type=["pdf", "png", "jpg", "jpeg"])
+
+manual_text = st.text_area(
+    "Or paste text manually (optional)",
+    height=160,
+    placeholder="If OCR struggles, paste the flyer words here…"
+)
+
+if st.button("Read flyer"):
+    if not uploaded and not manual_text.strip():
+        st.warning("Please upload a file or paste some text.")
+        st.stop()
+
+    raw_text = ""
+    if uploaded:
+        if uploaded.type == "application/pdf":
+            raw_text = pdf_bytes_to_text(uploaded.read())
+        else:
+            raw_text = image_bytes_to_text(uploaded.read())
+
+    # If user pasted text, prefer that (it may be cleaner than OCR)
+    combined_text = manual_text.strip() or raw_text
+
+    if not combined_text.strip():
+        st.error("I couldn’t read any text. Try exporting your Canva flyer as a higher-resolution PNG or a text-based PDF.")
+        st.stop()
+
+    results = extract_where_when(combined_text)
+
+    st.success("Done! Here’s what I found:")
+    col1, col2 = st.columns(2)
+    with col1:
+        st.markdown("**Date(s) detected:**")
+        if results["dates_found"]:
+            st.write(",  ".join(results["dates_found"]))
+        else:
+            st.write("—")
+
+        st.markdown("**Time(s) detected:**")
+        if results["times_found"]:
+            st.write(",  ".join(results["times_found"]))
+        else:
+            st.write("—")
+
+    with col2:
+        st.markdown("**Location detected:**")
+        st.write(results["location"] or "—")
+
+    st.markdown("---")
+    st.markdown("**All extracted text (for reference):**")
+    st.code(combined_text[:3000])  # avoid huge pastes
+
+    st.download_button(
+        "Download JSON",
+        data=json.dumps(results, indent=2),
+        file_name="where_when.json",
+        mime="application/json"
+    )