Upload pptx_builder.py with huggingface_hub

pptx_builder.py

@@ -54,52 +54,62 @@ THEMES = {
 
 # ── Image fetching ─────────────────────────────────────────────────────────
 
-def _fetch_image(keyword: str, width: int = 540, height: int = 400) -> str | None:
-    """Download a relevant image from Unsplash. Returns temp file path or None."""
+IMAGE_SOURCES = [
+    # loremflickr — keyword-based, reliable, no API key
+    lambda kw, w, h: f"https://loremflickr.com/{w}/{h}/{urllib.parse.quote(kw)}",
+    # picsum — random beautiful photo (fallback)
+    lambda kw, w, h: f"https://picsum.photos/seed/{urllib.parse.quote(kw)}/{w}/{h}",
+]
+
+def _fetch_image(keyword: str, width: int = 560, height: int = 420) -> str | None:
     if not keyword:
         return None
-    try:
-        encoded = urllib.parse.quote(keyword.strip())
-        url = f"https://source.unsplash.com/{width}x{height}/?{encoded}"
-        tmp = tempfile.NamedTemporaryFile(suffix=".jpg", delete=False)
-        req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
-        with urllib.request.urlopen(req, timeout=8) as resp:
-            tmp.write(resp.read())
-        tmp.close()
-        return tmp.name if os.path.getsize(tmp.name) > 1000 else None
-    except Exception:
-        return None
+    for source_fn in IMAGE_SOURCES:
+        try:
+            url = source_fn(keyword.strip(), width, height)
+            tmp = tempfile.NamedTemporaryFile(suffix=".jpg", delete=False)
+            req = urllib.request.Request(
+                url, headers={"User-Agent": "Mozilla/5.0 SlideAI/1.0"}
+            )
+            with urllib.request.urlopen(req, timeout=10) as resp:
+                data = resp.read()
+            if len(data) > 2000:          # real image, not an error page
+                tmp.write(data)
+                tmp.close()
+                return tmp.name
+            tmp.close()
+        except Exception:
+            continue
+    return None
 
 # ── Drawing helpers ────────────────────────────────────────────────────────
 
 def _blank_slide(prs):
     return prs.slides.add_slide(prs.slide_layouts[6])
 
-def _fill(shape, color: RGBColor):
+def _fill(shape, color):
     shape.fill.solid()
     shape.fill.fore_color.rgb = color
     shape.line.fill.background()
 
-def _set_bg(slide, color: RGBColor):
-    bg = slide.background
-    bg.fill.solid()
-    bg.fill.fore_color.rgb = color
+def _set_bg(slide, color):
+    slide.background.fill.solid()
+    slide.background.fill.fore_color.rgb = color
 
-def _rect(slide, l, t, w, h, color: RGBColor):
+def _rect(slide, l, t, w, h, color):
     s = slide.shapes.add_shape(1, l, t, w, h)
     _fill(s, color)
     return s
 
-def _oval(slide, l, t, w, h, color: RGBColor, alpha=0.15):
+def _oval(slide, l, t, w, h, color, alpha=0.15):
     s = slide.shapes.add_shape(9, l, t, w, h)
     _fill(s, color)
     try:
-        spPr = s.element.spPr
-        solidFill = spPr.find(qn("a:solidFill"))
-        if solidFill is not None:
-            srgb = solidFill.find(qn("a:srgbClr"))
-            if srgb is not None:
-                ae = etree.SubElement(srgb, qn("a:alpha"))
+        sf = s.element.spPr.find(qn("a:solidFill"))
+        if sf is not None:
+            sc = sf.find(qn("a:srgbClr"))
+            if sc is not None:
+                ae = etree.SubElement(sc, qn("a:alpha"))
                 ae.set("val", str(int(alpha * 100000)))
     except Exception:
         pass
@@ -121,13 +131,14 @@ def _txbox(slide, text, l, t, w, h, size, bold, color,
     r.font.color.rgb = color
     return tb
 
-def _add_picture_safe(slide, img_path, l, t, w, h):
-    """Add picture with a rounded-corner placeholder if loading fails."""
-    try:
-        slide.shapes.add_picture(img_path, l, t, w, h)
-        return True
-    except Exception:
-        return False
+def _spc(para, before_pt=0, after_pt=0):
+    pPr = para._p.get_or_add_pPr()
+    if before_pt:
+        sb = etree.SubElement(pPr, qn("a:spcBef"))
+        etree.SubElement(sb, qn("a:spcPts")).set("val", str(before_pt * 100))
+    if after_pt:
+        sa = etree.SubElement(pPr, qn("a:spcAft"))
+        etree.SubElement(sa, qn("a:spcPts")).set("val", str(after_pt * 100))
 
 # ── Slide builders ─────────────────────────────────────────────────────────
 
@@ -135,68 +146,49 @@ def _build_title_slide(prs, data, t, total):
     slide = _blank_slide(prs)
     _set_bg(slide, t["dark"])
 
-    # Try background image (blended behind content)
-    keyword = data.get("image_keyword", "")
-    img_path = _fetch_image(keyword, width=1000, height=600)
+    # Background image with dark overlay
+    img_path = _fetch_image(data.get("image_keyword", ""), 1200, 700)
     if img_path:
         try:
             pic = slide.shapes.add_picture(img_path, 0, 0, SLIDE_W, SLIDE_H)
-            # Send image to back
             slide.shapes._spTree.remove(pic._element)
             slide.shapes._spTree.insert(2, pic._element)
-            # Dark overlay rectangle
-            overlay = _rect(slide, 0, 0, SLIDE_W, SLIDE_H, t["dark"])
-            # Set overlay alpha ~80%
-            spPr = overlay.element.spPr
-            solidFill = spPr.find(qn("a:solidFill"))
-            if solidFill is not None:
-                srgb = solidFill.find(qn("a:srgbClr"))
-                if srgb is not None:
-                    ae = etree.SubElement(srgb, qn("a:alpha"))
-                    ae.set("val", "80000")
+            ov = _rect(slide, 0, 0, SLIDE_W, SLIDE_H, t["dark"])
+            sf = ov.element.spPr.find(qn("a:solidFill"))
+            if sf is not None:
+                sc = sf.find(qn("a:srgbClr"))
+                if sc is not None:
+                    ae = etree.SubElement(sc, qn("a:alpha"))
+                    ae.set("val", "78000")
         except Exception:
             pass
 
-    # Decorative circles
-    _oval(slide, Inches(9.5), Inches(-1.5), Inches(6), Inches(6), t["primary"], 0.20)
+    _oval(slide, Inches(9.5), Inches(-1.5), Inches(6), Inches(6), t["primary"], 0.22)
     _oval(slide, Inches(-0.8), Inches(4.8), Inches(3.5), Inches(3.5), t["secondary"], 0.18)
-
-    # Bottom accent bar
     _rect(slide, 0, SLIDE_H - Inches(0.08), SLIDE_W, Inches(0.08), t["secondary"])
-    # Left accent
     _rect(slide, 0, Inches(1.5), Inches(0.06), Inches(4.5), t["primary"])
 
-    # Title
     tb = slide.shapes.add_textbox(Inches(0.5), Inches(1.8), Inches(12), Inches(2.5))
-    tf = tb.text_frame
-    tf.word_wrap = True
-    p = tf.paragraphs[0]
-    p.alignment = PP_ALIGN.LEFT
+    tf = tb.text_frame; tf.word_wrap = True
+    p = tf.paragraphs[0]; p.alignment = PP_ALIGN.LEFT
     r = p.add_run()
     r.text = data.get("title", "Presentation")
-    r.font.name = "Calibri"
-    r.font.size = Pt(52)
-    r.font.bold = True
-    r.font.color.rgb = t["white"]
+    r.font.name = "Calibri"; r.font.size = Pt(52)
+    r.font.bold = True; r.font.color.rgb = t["white"]
 
-    # Subtitle
     subtitle = data.get("subtitle", "")
     if subtitle:
         tb2 = slide.shapes.add_textbox(Inches(0.5), Inches(4.5), Inches(10), Inches(1))
-        tf2 = tb2.text_frame
-        tf2.word_wrap = True
-        p2 = tf2.paragraphs[0]
-        p2.alignment = PP_ALIGN.LEFT
+        tf2 = tb2.text_frame; tf2.word_wrap = True
+        p2 = tf2.paragraphs[0]; p2.alignment = PP_ALIGN.LEFT
         r2 = p2.add_run()
         r2.text = subtitle
-        r2.font.name = "Calibri Light"
-        r2.font.size = Pt(24)
+        r2.font.name = "Calibri Light"; r2.font.size = Pt(24)
         r2.font.color.rgb = t["secondary"]
 
-    # Slide count badge
     _rect(slide, Inches(0.5), Inches(6.2), Inches(2.4), Inches(0.5), t["primary"])
-    _txbox(slide, f"{total} slides", Inches(0.5), Inches(6.2), Inches(2.4), Inches(0.5),
-           size=11, bold=False, color=t["white"], align=PP_ALIGN.CENTER)
+    _txbox(slide, f"{total} slides", Inches(0.5), Inches(6.2),
+           Inches(2.4), Inches(0.5), 11, False, t["white"], PP_ALIGN.CENTER)
 
     slide.notes_slide.notes_text_frame.text = data.get("speaker_notes", "")
 
@@ -205,137 +197,115 @@ def _build_content_slide(prs, data, slide_num, total, t):
     slide = _blank_slide(prs)
     _set_bg(slide, t["light"])
 
-    keyword = data.get("image_keyword", "")
-    img_path = _fetch_image(keyword, width=540, height=400)
+    img_path = _fetch_image(data.get("image_keyword", ""), 560, 430)
     has_image = img_path is not None
 
     # ── Header band ──────────────────────────────────────────────────────
-    _rect(slide, 0, 0, SLIDE_W, Inches(1.5), t["primary"])
-    _rect(slide, SLIDE_W - Inches(1.5), 0, Inches(1.5), Inches(1.5), t["secondary"])
-
-    # Slide number in corner block
+    _rect(slide, 0, 0, SLIDE_W, Inches(1.4), t["primary"])
+    _rect(slide, SLIDE_W - Inches(1.4), 0, Inches(1.4), Inches(1.4), t["secondary"])
     _txbox(slide, f"{slide_num:02d}",
-           SLIDE_W - Inches(1.5), Inches(0.1), Inches(1.5), Inches(1.2),
-           size=36, bold=True, color=t["white"], align=PP_ALIGN.CENTER)
-
-    # Title in header
+           SLIDE_W - Inches(1.4), Inches(0.05), Inches(1.4), Inches(1.3),
+           36, True, t["white"], PP_ALIGN.CENTER)
     _txbox(slide, data.get("title", ""),
-           Inches(0.5), Inches(0.15), SLIDE_W - Inches(2.5), Inches(1.2),
-           size=28, bold=True, color=t["white"])
+           Inches(0.5), Inches(0.12), SLIDE_W - Inches(2.2), Inches(1.15),
+           26, True, t["white"])
 
-    # ── Left accent stripe ────────────────────────────────────────────────
-    _rect(slide, 0, Inches(1.5), Inches(0.06), Inches(5.6), t["secondary"])
+    # ── Left accent ───────────────────────────���───────────────────────────
+    _rect(slide, 0, Inches(1.4), Inches(0.06), SLIDE_H - Inches(1.75), t["secondary"])
 
-    # ── Bullet content area ───────────────────────────────────────────────
+    # ── Bullets ───────────────────────────────────────────────────────────
     bullets = data.get("bullets", [])
-    text_width = Inches(7.8) if has_image else Inches(12.5)
+    text_w = Inches(7.6) if has_image else Inches(12.8)
 
-    content_box = slide.shapes.add_textbox(
-        Inches(0.35), Inches(1.8), text_width, Inches(5.0)
-    )
-    tf = content_box.text_frame
+    cb = slide.shapes.add_textbox(Inches(0.3), Inches(1.6), text_w, Inches(5.55))
+    tf = cb.text_frame
     tf.word_wrap = True
 
-    ICONS = ["①", "②", "③", "④", "⑤", "⑥"]
-    for idx, bullet_text in enumerate(bullets[:6]):
+    ICONS = ["①", "②", "③", "④", "⑤", "⑥", "⑦"]
+    for idx, bullet in enumerate(bullets[:7]):
         p = tf.paragraphs[0] if idx == 0 else tf.add_paragraph()
         p.alignment = PP_ALIGN.LEFT
-
-        # Spacing between bullets
-        pPr = p._p.get_or_add_pPr()
         if idx > 0:
-            spcBef = etree.SubElement(pPr, qn("a:spcBef"))
-            spcPts = etree.SubElement(spcBef, qn("a:spcPts"))
-            spcPts.set("val", "700")
-
-        # Numbered icon
-        r_num = p.add_run()
-        r_num.text = f"{ICONS[idx]}  "
-        r_num.font.name = "Segoe UI"
-        r_num.font.size = Pt(21)
-        r_num.font.bold = True
-        r_num.font.color.rgb = t["primary"]
-
-        # Bullet text
-        r_txt = p.add_run()
-        r_txt.text = bullet_text
-        r_txt.font.name = "Calibri"
-        r_txt.font.size = Pt(20)
-        r_txt.font.bold = False
-        r_txt.font.color.rgb = t["text_dark"]
-
-    # ── Image panel (right side) ──────────────────────────────────────────
+            _spc(p, before_pt=5)
+
+        rn = p.add_run()
+        rn.text = f"{ICONS[idx]}  "
+        rn.font.name = "Segoe UI"
+        rn.font.size = Pt(17)
+        rn.font.bold = True
+        rn.font.color.rgb = t["primary"]
+
+        rt = p.add_run()
+        rt.text = bullet
+        rt.font.name = "Calibri"
+        rt.font.size = Pt(17)
+        rt.font.bold = False
+        rt.font.color.rgb = t["text_dark"]
+
+    # ── Image panel ───────────────────────────────────────────────────────
     if has_image:
-        img_l = Inches(8.3)
-        img_t = Inches(1.7)
-        img_w = Inches(4.7)
-        img_h = Inches(4.8)
-
-        # Colored image border / frame
-        _rect(slide, img_l - Inches(0.06), img_t - Inches(0.06),
-              img_w + Inches(0.12), img_h + Inches(0.12), t["secondary"])
-
-        success = _add_picture_safe(slide, img_path, img_l, img_t, img_w, img_h)
-        if not success:
-            # Fallback: colored placeholder with keyword label
-            _rect(slide, img_l, img_t, img_w, img_h, t["primary"])
-            _txbox(slide, keyword.title(), img_l + Inches(0.2),
-                   img_t + Inches(1.8), img_w - Inches(0.4), Inches(1.2),
-                   size=16, bold=False, color=t["white"], align=PP_ALIGN.CENTER)
+        il = Inches(8.2); it = Inches(1.55)
+        iw = Inches(4.85); ih = Inches(5.55)
+
+        # Border frame
+        _rect(slide, il - Inches(0.07), it - Inches(0.07),
+              iw + Inches(0.14), ih + Inches(0.14), t["secondary"])
+
+        try:
+            slide.shapes.add_picture(img_path, il, it, iw, ih)
+        except Exception:
+            _rect(slide, il, it, iw, ih, t["primary"])
+            _txbox(slide, data.get("image_keyword", "").title(),
+                   il + Inches(0.3), it + Inches(2.2), iw - Inches(0.6), Inches(1),
+                   14, False, t["white"], PP_ALIGN.CENTER)
 
     # ── Footer ────────────────────────────────────────────────────────────
-    _rect(slide, 0, SLIDE_H - Inches(0.35), SLIDE_W, Inches(0.35), t["primary"])
+    _rect(slide, 0, SLIDE_H - Inches(0.32), SLIDE_W, Inches(0.32), t["primary"])
     _txbox(slide, f"Slide {slide_num} of {total}",
-           SLIDE_W - Inches(2.5), SLIDE_H - Inches(0.35),
-           Inches(2.4), Inches(0.35),
-           size=10, bold=False, color=t["white"], align=PP_ALIGN.RIGHT)
+           SLIDE_W - Inches(2.5), SLIDE_H - Inches(0.32),
+           Inches(2.4), Inches(0.32), 9, False, t["white"], PP_ALIGN.RIGHT)
 
     slide.notes_slide.notes_text_frame.text = data.get("speaker_notes", "")
 
 
-def _build_closing_slide(prs, presentation_title, t):
+def _build_closing_slide(prs, title, t):
     slide = _blank_slide(prs)
     _set_bg(slide, t["dark"])
-
     _oval(slide, Inches(-1), Inches(-1), Inches(5), Inches(5), t["secondary"], 0.15)
     _oval(slide, Inches(10), Inches(3.5), Inches(5), Inches(5), t["primary"], 0.18)
     _rect(slide, 0, SLIDE_H - Inches(0.08), SLIDE_W, Inches(0.08), t["secondary"])
     _rect(slide, 0, Inches(1.5), Inches(0.06), Inches(4.5), t["secondary"])
-
     _txbox(slide, "Thank You", Inches(0.5), Inches(1.8), Inches(12), Inches(2),
-           size=60, bold=True, color=t["white"])
-    _txbox(slide, presentation_title, Inches(0.5), Inches(4.1), Inches(10), Inches(0.8),
-           size=22, bold=False, color=t["secondary"])
+           60, True, t["white"])
+    _txbox(slide, title, Inches(0.5), Inches(4.1), Inches(10), Inches(0.8),
+           22, False, t["secondary"])
     _rect(slide, Inches(0.5), Inches(5.1), Inches(3), Inches(0.05), t["primary"])
     _txbox(slide, "Questions & Discussion",
            Inches(0.5), Inches(5.3), Inches(8), Inches(0.6),
-           size=16, bold=False, color=t["white"])
+           16, False, t["white"])
 
 
 # ── Entry point ────────────────────────────────────────────────────────────
 
 def build_pptx(presentation_data: dict, style: str) -> bytes:
     t = THEMES.get(style, THEMES["Professional"])
-
     prs = Presentation()
     prs.slide_width = SLIDE_W
     prs.slide_height = SLIDE_H
 
     slides = presentation_data.get("slides", [])
     pres_title = presentation_data.get("title", "Presentation")
+    total = len(slides) + 1  # +1 for closing slide
+
+    title_slides  = [s for s in slides if s.get("type") == "title"]
     content_slides = [s for s in slides if s.get("type") != "title"]
-    title_slides = [s for s in slides if s.get("type") == "title"]
-    total_display = len(slides) + 1   # +1 for closing
 
-    # 1 — Title slide
     for s in title_slides:
-        _build_title_slide(prs, s, t, total_display)
+        _build_title_slide(prs, s, t, total)
 
-    # 2 — Content slides
     for i, s in enumerate(content_slides):
-        _build_content_slide(prs, s, i + 2, total_display, t)
+        _build_content_slide(prs, s, i + 2, total, t)
 
-    # 3 — Closing slide
     _build_closing_slide(prs, pres_title, t)
 
     buf = io.BytesIO()