ima-claude 2.9.0 → 2.13.0

package/plugins/ima-claude/skills/ima-doc2pdf/scripts/generate_pdf.py

@@ -0,0 +1,663 @@
+"""
+Generate a styled IMA branded document PDF directly from a Word doc.
+
+Usage:
+    python3 generate_pdf.py <path_to_docx> [--out <output.pdf>]
+
+Required: pip install reportlab python-docx
+"""
+
+import sys
+import io
+import re
+import base64
+import tempfile
+import urllib.request
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).parent))
+from extract_docx import extract_document
+
+try:
+    from reportlab.lib.pagesizes import LETTER
+    from reportlab.lib.units import inch
+    from reportlab.lib.colors import HexColor, white, black
+    from reportlab.lib.styles import ParagraphStyle
+    from reportlab.lib.enums import TA_LEFT, TA_CENTER, TA_JUSTIFY
+    from reportlab.platypus import (
+        BaseDocTemplate, PageTemplate, Frame, Paragraph, Spacer,
+        ListFlowable, ListItem, HRFlowable, PageBreak, NextPageTemplate,
+        KeepTogether, Image
+    )
+    from reportlab.platypus.flowables import Flowable
+    from reportlab.pdfgen import canvas as pdfcanvas
+    from reportlab.pdfbase import pdfmetrics
+    from reportlab.pdfbase.ttfonts import TTFont
+except ImportError:
+    print("ERROR: reportlab not installed. Run:")
+    print("  pip install reportlab")
+    sys.exit(1)
+
+
+# ── Register Lato fonts ──────────────────────────────────────────────────────
+FONT_DIR = Path(__file__).parent.parent / "fonts"
+LATO_FONTS = ("Lato-Regular.ttf", "Lato-Bold.ttf", "Lato-Italic.ttf", "Lato-BoldItalic.ttf")
+LATO_BASE_URL = "https://github.com/google/fonts/raw/main/ofl/lato/"
+
+
+def ensure_fonts():
+    """Download Lato fonts from Google Fonts GitHub repo if any are missing."""
+    if FONT_DIR.exists() and all((FONT_DIR / f).exists() for f in LATO_FONTS):
+        return
+
+    print("Downloading Lato fonts...")
+    FONT_DIR.mkdir(parents=True, exist_ok=True)
+
+    for font_file in LATO_FONTS:
+        dest = FONT_DIR / font_file
+        if dest.exists():
+            continue
+        url = LATO_BASE_URL + font_file
+        urllib.request.urlretrieve(url, dest)
+        print(f"  {font_file}")
+
+
+def register_fonts():
+    """Register Lato TTF fonts with reportlab."""
+    ensure_fonts()
+    fonts = {
+        "Lato":            "Lato-Regular.ttf",
+        "Lato-Bold":       "Lato-Bold.ttf",
+        "Lato-Italic":     "Lato-Italic.ttf",
+        "Lato-BoldItalic": "Lato-BoldItalic.ttf",
+    }
+    for name, filename in fonts.items():
+        path = FONT_DIR / filename
+        if path.exists():
+            pdfmetrics.registerFont(TTFont(name, str(path)))
+        else:
+            print(f"WARNING: Font not found: {path}")
+
+    # Register font family so <b> and <i> markup works in Paragraphs
+    from reportlab.pdfbase.pdfmetrics import registerFontFamily
+    registerFontFamily(
+        "Lato",
+        normal="Lato",
+        bold="Lato-Bold",
+        italic="Lato-Italic",
+        boldItalic="Lato-BoldItalic",
+    )
+
+
+# ── Brand colours (IMA Brand Book v4.0) ──────────────────────────────────────
+NAVY        = HexColor("#00066F")   # Trustworthy Indigo
+GOLD        = HexColor("#FFCC00")   # Vital Gold
+BODY_TEXT   = HexColor("#000000")
+GREY_LIGHT  = HexColor("#CCCCCC")
+GREY_TEXT   = HexColor("#666666")
+
+PAGE_W, PAGE_H = LETTER   # 612 × 792 pt
+MARGIN = 0.5 * inch
+
+
+# ── Styles (confirmed from Canva) ────────────────────────────────────────────
+def build_styles():
+    S = {}
+
+    def s(name, **kw):
+        S[name] = ParagraphStyle(name=name, **kw)
+
+    # ── Cover ────────────────────────────────────────────────────────────
+    s("cover_title_1",
+      fontName="Lato-Bold", fontSize=90.5, leading=95,
+      textColor=white, alignment=TA_CENTER, spaceAfter=8)
+
+    s("cover_title_2",
+      fontName="Lato-Bold", fontSize=67, leading=72,
+      textColor=white, alignment=TA_CENTER, spaceAfter=8)
+
+    s("cover_subtitle",
+      fontName="Lato-Bold", fontSize=24, leading=30,
+      textColor=white, alignment=TA_CENTER, spaceAfter=6)
+
+    s("cover_authors",
+      fontName="Lato", fontSize=20, leading=26,
+      textColor=white, alignment=TA_CENTER, spaceAfter=4)
+
+    s("cover_disclaimer",
+      fontName="Lato", fontSize=12, leading=16,
+      textColor=white, alignment=TA_CENTER, spaceAfter=3)
+
+    s("cover_date",
+      fontName="Lato-Italic", fontSize=12, leading=16,
+      textColor=white, alignment=TA_CENTER)
+
+    # ── Content headings ─────────────────────────────────────────────────
+    s("intro_heading",
+      fontName="Lato-Bold", fontSize=15, leading=20,
+      textColor=NAVY, alignment=TA_CENTER,
+      spaceBefore=6, spaceAfter=10)
+
+    s("section_heading",
+      fontName="Lato-Bold", fontSize=15, leading=20,
+      textColor=NAVY, spaceBefore=16, spaceAfter=5)
+
+    s("sub_heading",
+      fontName="Lato-Bold", fontSize=13, leading=17,
+      textColor=NAVY, spaceBefore=10, spaceAfter=3)
+
+    # ── Body text ────────────────────────────────────────────────────────
+    s("body",
+      fontName="Lato", fontSize=12, leading=14.5,
+      textColor=BODY_TEXT, spaceBefore=0, spaceAfter=6,
+      alignment=TA_JUSTIFY)
+
+    s("bullet",
+      fontName="Lato", fontSize=12, leading=14.5,
+      textColor=BODY_TEXT, spaceBefore=1, spaceAfter=1)
+
+    # ── Warning box — navy bg, white + Vital Gold text ───────────────────
+    s("warning",
+      fontName="Lato-Bold", fontSize=12, leading=16,
+      textColor=white, alignment=TA_CENTER,
+      backColor=NAVY, borderColor=NAVY,
+      borderWidth=0, borderPad=10,
+      spaceBefore=10, spaceAfter=10)
+
+    # ── Q&A ──────────────────────────────────────────────────────────────
+    s("qa_question",
+      fontName="Lato-Bold", fontSize=12, leading=16,
+      textColor=NAVY, spaceBefore=10, spaceAfter=2)
+
+    s("qa_answer",
+      fontName="Lato", fontSize=12, leading=14.5,
+      textColor=BODY_TEXT, spaceBefore=2, spaceAfter=4,
+      alignment=TA_JUSTIFY)
+
+    # ── References ───────────────────────────────────────────────────────
+    s("ref_heading",
+      fontName="Lato-Bold", fontSize=13, leading=17,
+      textColor=NAVY, spaceBefore=10, spaceAfter=8)
+
+    s("reference",
+      fontName="Lato", fontSize=8, leading=11,
+      textColor=HexColor("#333333"), spaceBefore=1, spaceAfter=1,
+      leftIndent=14, firstLineIndent=-14)
+
+    # ── Captions ─────────────────────────────────────────────────────────
+    s("caption",
+      fontName="Lato", fontSize=12, leading=14.5,
+      textColor=BODY_TEXT, spaceBefore=4, spaceAfter=4)
+
+    s("disclaimer",
+      fontName="Lato", fontSize=12, leading=16,
+      textColor=white, alignment=TA_CENTER,
+      spaceBefore=4, spaceAfter=4)
+
+    s("content_disclaimer",
+      fontName="Lato", fontSize=12, leading=16,
+      textColor=BODY_TEXT, alignment=TA_JUSTIFY,
+      spaceBefore=4, spaceAfter=6)
+
+    # ── Footer ───────────────────────────────────────────────────────────
+    s("footer",
+      fontName="Lato", fontSize=10, leading=12,
+      textColor=GREY_TEXT, alignment=TA_CENTER)
+
+    return S
+
+
+# ── Page canvas callbacks ─────────────────────────────────────────────────────
+def make_cover_canvas(footer_h):
+    def on_cover(canv, doc):
+        canv.saveState()
+        # Full navy background
+        canv.setFillColor(NAVY)
+        canv.rect(0, 0, PAGE_W, PAGE_H, fill=1, stroke=0)
+        canv.restoreState()
+    return on_cover
+
+
+def make_content_canvas(title_short, date_str, footer_h):
+    if date_str:
+        clean_date = date_str.replace("Updated ", "").strip()
+        footer_text = f"{title_short} ({clean_date})"
+    else:
+        footer_text = title_short
+
+    def on_content(canv, doc):
+        canv.saveState()
+        # Bottom footer rule
+        canv.setStrokeColor(GREY_LIGHT)
+        canv.setLineWidth(0.5)
+        canv.line(MARGIN, footer_h + 8, PAGE_W - MARGIN, footer_h + 8)
+        # Single centered footer line — Lato 10pt
+        canv.setFont("Lato", 10)
+        canv.setFillColor(GREY_TEXT)
+        canv.drawCentredString(PAGE_W / 2, footer_h - 2, footer_text)
+        canv.restoreState()
+    return on_content
+
+
+# ── Markup helpers ────────────────────────────────────────────────────────────
+def safe(text):
+    return (text.replace("&", "&amp;")
+               .replace("<", "&lt;")
+               .replace(">", "&gt;"))
+
+
+def runs_to_markup(runs):
+    if not runs:
+        return ""
+    out = []
+    for r in runs:
+        t = safe(r.get("text", ""))
+        if r.get("bold") and r.get("italic"):
+            out.append(f"<b><i>{t}</i></b>")
+        elif r.get("bold"):
+            out.append(f"<b>{t}</b>")
+        elif r.get("italic"):
+            out.append(f"<i>{t}</i>")
+        else:
+            out.append(t)
+    return "".join(out)
+
+
+def para_markup(entry):
+    markup = runs_to_markup(entry.get("runs", []))
+    return markup if markup else safe(entry["text"])
+
+
+def warning_markup(text):
+    """Format warning text with Vital Gold for emphasis portions."""
+    # The warning box has white text with gold emphasis
+    # For now, render all as white; gold portions need manual Word markup
+    return safe(text)
+
+
+# ── Extract cover metadata ────────────────────────────────────────────────────
+def extract_cover_meta(sections):
+    """
+    Pull title, authors, disclaimer, date from the document preamble.
+    Only the FIRST heading is the guide title — stop taking headings after that
+    so section headings like 'Introduction' don't bleed onto the cover.
+    """
+    title = disclaimer = date_str = ""
+    author_list = []
+    got_title = False
+
+    for entry in sections[:30]:
+        t = entry["type"]
+        text = entry["text"].strip()
+
+        if t in ("h1", "heading_bold"):
+            if not got_title:
+                title = text
+                got_title = True
+            else:
+                break
+
+        elif t == "author":
+            author_list.append(text)
+
+        elif t == "disclaimer":
+            disclaimer = text
+
+        elif t == "date":
+            date_str = text
+
+        elif t == "body" and got_title:
+            break
+
+    authors = "<br/>".join(safe(a) for a in author_list) if author_list else ""
+
+    return title, authors, disclaimer, date_str
+
+
+# ── Image extraction ─────────────────────────────────────────────────────────
+def extract_images_from_docx(docx_path):
+    """Extract embedded images from DOCX, return dict of paragraph_index -> temp file path."""
+    from docx import Document
+    from docx.oxml.ns import qn
+    
+    doc = Document(docx_path)
+    images = {}  # rId -> (blob, content_type)
+    for rel in doc.part.rels.values():
+        if 'image' in rel.reltype:
+            images[rel.rId] = (rel.target_part.blob, rel.target_part.content_type)
+    
+    # Find which paragraphs have images
+    positions = {}  # para_index -> [(rId, blob, content_type)]
+    for i, para in enumerate(doc.paragraphs):
+        for run in para.runs:
+            drawings = run._element.findall(qn('w:drawing'))
+            for d in drawings:
+                blips = d.findall('.//' + qn('a:blip'))
+                for blip in blips:
+                    embed = blip.get(qn('r:embed'))
+                    if embed and embed in images:
+                        blob, ct = images[embed]
+                        positions.setdefault(i, []).append((embed, blob, ct))
+    
+    # Write to temp files and return para_index -> [filepath]
+    result = {}
+    for para_idx, items in positions.items():
+        paths = []
+        for rId, blob, ct in items:
+            ext = '.png' if 'png' in ct else '.jpg'
+            tf = tempfile.NamedTemporaryFile(suffix=ext, delete=False)
+            tf.write(blob)
+            tf.close()
+            paths.append(tf.name)
+        result[para_idx] = paths
+    
+    return result
+
+
+# ── Build story helpers ───────────────────────────────────────────────────────
+def group_consecutive_bullets(sections):
+    """Merge consecutive bullet entries into grouped blocks.
+
+    Returns a new list where runs of {'type': 'bullet'} entries are replaced
+    by a single {'type': 'bullet_group', 'items': [...]} dict.
+    """
+    result = []
+    i = 0
+    while i < len(sections):
+        if sections[i]["type"] == "bullet":
+            group = []
+            while i < len(sections) and sections[i]["type"] == "bullet":
+                group.append(sections[i])
+                i += 1
+            result.append({"type": "bullet_group", "items": group})
+        else:
+            result.append(sections[i])
+            i += 1
+    return result
+
+
+def block_to_flowables(block, styles, intro_added, heading_count):
+    """Map a single block dict to a list of ReportLab flowables.
+
+    Returns (flowables, intro_added, heading_count) — the bool state is
+    threaded through so this stays a pure transformation.
+    """
+    t = block["type"]
+
+    if t in ("h1", "date", "author"):
+        return [], intro_added, heading_count
+
+    if t == "heading_bold":
+        heading_count += 1
+        if heading_count == 1:
+            return [], intro_added, heading_count
+        if not intro_added and block["text"].strip().lower() == "introduction":
+            return (
+                [Spacer(1, 0.15 * inch),
+                 Paragraph("Introduction", styles["intro_heading"]),
+                 Spacer(1, 0.05 * inch)],
+                True,
+                heading_count,
+            )
+        return [Paragraph(para_markup(block), styles["section_heading"])], intro_added, heading_count
+
+    if t == "h2":
+        if not intro_added and block["text"].strip().lower() == "introduction":
+            return (
+                [Spacer(1, 0.15 * inch),
+                 Paragraph("Introduction", styles["intro_heading"]),
+                 Spacer(1, 0.05 * inch)],
+                True,
+                heading_count,
+            )
+        return [Paragraph(para_markup(block), styles["section_heading"])], intro_added, heading_count
+
+    if t == "h3":
+        return [Paragraph(para_markup(block), styles["sub_heading"])], intro_added, heading_count
+
+    if t == "ref_heading":
+        return (
+            [PageBreak(),
+             Paragraph("References", styles["ref_heading"]),
+             HRFlowable(width="100%", color=NAVY, thickness=1.5, spaceAfter=8)],
+            intro_added,
+            heading_count,
+        )
+
+    if t == "warning":
+        return [Paragraph(warning_markup(block["text"]), styles["warning"])], intro_added, heading_count
+
+    if t == "disclaimer":
+        return [Paragraph(safe(block["text"]), styles["content_disclaimer"])], intro_added, heading_count
+
+    if t == "body":
+        return [Paragraph(para_markup(block), styles["body"])], intro_added, heading_count
+
+    if t == "bullet_group":
+        items = [
+            ListItem(
+                Paragraph(para_markup(entry), styles["bullet"]),
+                bulletColor=NAVY, bulletFontSize=10, leftIndent=18,
+            )
+            for entry in block["items"]
+        ]
+        return (
+            [ListFlowable(
+                items, bulletType="bullet",
+                bulletFontName="Lato", bulletFontSize=10,
+                leftIndent=18, bulletOffsetY=-1,
+                spaceBefore=4, spaceAfter=6,
+            )],
+            intro_added,
+            heading_count,
+        )
+
+    if t in ("figure_caption", "table_caption"):
+        return [Paragraph(safe(block["text"]), styles["caption"])], intro_added, heading_count
+
+    # Default
+    return [Paragraph(para_markup(block), styles["body"])], intro_added, heading_count
+
+
+# ── Build story ───────────────────────────────────────────────────────────────
+def build_story(data, styles, title_short, date_str, image_positions=None):
+    story = []
+    footer_h = 0.4 * inch
+    if image_positions is None:
+        image_positions = {}
+
+    # ── Cover page (all on navy background) ──────────────────────────────
+    title, authors, disclaimer, cover_date = extract_cover_meta(
+        data["sections"]
+    )
+
+    story.append(Spacer(1, 1.2 * inch))
+    if title:
+        # Split title visually: first half as hero text, second half as subtitle.
+        # This is a placeholder cover — ima-cover-creator replaces it for final output.
+        words = title.split()
+        mid = max(1, len(words) // 2)
+        line1 = " ".join(words[:mid]).upper()
+        line2 = " ".join(words[mid:]).upper() if len(words) > mid else ""
+
+        story.append(Paragraph(safe(line1), styles["cover_title_1"]))
+        if line2:
+            story.append(Spacer(1, 0.2 * inch))
+            story.append(Paragraph(safe(line2), styles["cover_title_2"]))
+
+    story.append(Spacer(1, 0.5 * inch))
+    if authors:
+        story.append(Paragraph(authors, styles["cover_authors"]))
+    if disclaimer:
+        story.append(Spacer(1, 0.4 * inch))
+        story.append(Paragraph(safe(disclaimer), styles["cover_disclaimer"]))
+    if cover_date:
+        story.append(Spacer(1, 0.15 * inch))
+        story.append(Paragraph(safe(cover_date), styles["cover_date"]))
+
+    story.append(NextPageTemplate("content"))
+    story.append(PageBreak())
+
+    # ── Content — pipeline: group bullets → map blocks → flatten ─────────
+    grouped = group_consecutive_bullets(data["sections"])
+    intro_added = False
+    heading_count = 0
+    rendered_images = set()
+    
+    # Build a lookup of section index -> paragraph index for image placement
+    for block in grouped:
+        flowables, intro_added, heading_count = block_to_flowables(
+            block, styles, intro_added, heading_count
+        )
+        story.extend(flowables)
+        
+        # Insert images at this paragraph position
+        para_idx = block.get("index", -1)
+        if para_idx in image_positions and para_idx not in rendered_images:
+            max_w = PAGE_W - 2 * MARGIN - 0.5 * inch
+            for img_path in image_positions[para_idx]:
+                try:
+                    img = Image(img_path, width=max_w, height=None)
+                    # Let reportlab scale proportionally
+                    img._restrictSize(max_w, PAGE_H - 3 * inch)
+                    story.append(Spacer(1, 6))
+                    story.append(img)
+                    story.append(Spacer(1, 6))
+                except Exception as e:
+                    print(f"  Warning: Could not embed image at para {para_idx}: {e}")
+            rendered_images.add(para_idx)
+        
+        # Also check for images between consecutive section indices
+        if block.get("type") == "bullet_group":
+            for item in block.get("items", []):
+                item_idx = item.get("index", -1)
+                if item_idx in image_positions and item_idx not in rendered_images:
+                    max_w = PAGE_W - 2 * MARGIN - 0.5 * inch
+                    for img_path in image_positions[item_idx]:
+                        try:
+                            img = Image(img_path, width=max_w, height=None)
+                            img._restrictSize(max_w, PAGE_H - 3 * inch)
+                            story.append(Spacer(1, 6))
+                            story.append(img)
+                            story.append(Spacer(1, 6))
+                        except Exception as e:
+                            print(f"  Warning: Could not embed image: {e}")
+                    rendered_images.add(item_idx)
+    
+    # Insert any remaining images not yet rendered (e.g. image-only paragraphs)
+    for para_idx, paths in image_positions.items():
+        if para_idx not in rendered_images:
+            max_w = PAGE_W - 2 * MARGIN - 0.5 * inch
+            for img_path in paths:
+                try:
+                    img = Image(img_path, width=max_w, height=None)
+                    img._restrictSize(max_w, PAGE_H - 3 * inch)
+                    story.append(Spacer(1, 6))
+                    story.append(img)
+                    story.append(Spacer(1, 6))
+                except Exception as e:
+                    print(f"  Warning: Could not embed orphan image: {e}")
+            rendered_images.add(para_idx)
+
+    # ── Q&A ───────────────────────────────────────────────────────────────
+    if data.get("qa_pairs"):
+        story.append(Spacer(1, 0.1 * inch))
+        story.append(HRFlowable(width="100%", color=GREY_LIGHT,
+                                 thickness=0.5, spaceAfter=8))
+        for qa in data["qa_pairs"]:
+            block = [Paragraph(para_markup(qa["question"]), styles["qa_question"])]
+            for ans in qa["answer_parts"]:
+                block.append(Paragraph(para_markup(ans), styles["qa_answer"]))
+            story.append(KeepTogether(block))
+
+    # ── References ────────────────────────────────────────────────────────
+    if data.get("references"):
+        story.append(PageBreak())
+        story.append(Paragraph("References", styles["ref_heading"]))
+        story.append(HRFlowable(width="100%", color=NAVY, thickness=1.5,
+                                 spaceAfter=8))
+        for ref in data["references"]:
+            story.append(Paragraph(safe(ref["text"]), styles["reference"]))
+
+    return story
+
+
+# ── Main ──────────────────────────────────────────────────────────────────────
+def generate_pdf(docx_path, out_path):
+    print(f"Extracting: {docx_path}")
+    data = extract_document(docx_path)
+    
+    # Extract images from DOCX
+    print("Extracting images...")
+    image_positions = extract_images_from_docx(docx_path)
+    print(f"  Found images at {len(image_positions)} paragraph positions")
+
+    register_fonts()
+    styles = build_styles()
+
+    title, authors, disclaimer, date_str = extract_cover_meta(
+        data["sections"]
+    )
+    title_short = (title[:60] + "...") if len(title) > 60 else title
+
+    footer_h = 0.4 * inch
+
+    doc = BaseDocTemplate(
+        str(out_path),
+        pagesize=LETTER,
+        leftMargin=MARGIN, rightMargin=MARGIN,
+        topMargin=MARGIN, bottomMargin=MARGIN,
+    )
+
+    cover_frame = Frame(
+        MARGIN, footer_h + 0.3 * inch,
+        PAGE_W - 2 * MARGIN,
+        PAGE_H - 2 * MARGIN - footer_h,
+        id="cover_frame"
+    )
+    content_frame = Frame(
+        MARGIN, footer_h + 0.4 * inch,
+        PAGE_W - 2 * MARGIN,
+        PAGE_H - 2 * MARGIN - footer_h,
+        id="content_frame"
+    )
+
+    doc.addPageTemplates([
+        PageTemplate(
+            id="cover",
+            frames=[cover_frame],
+            onPage=make_cover_canvas(footer_h)
+        ),
+        PageTemplate(
+            id="content",
+            frames=[content_frame],
+            onPage=make_content_canvas(title_short, date_str, footer_h)
+        ),
+    ])
+
+    story = build_story(data, styles, title_short, date_str, image_positions)
+
+    print(f"Building PDF...")
+    doc.build(story)
+    print(f"Done → {out_path}")
+
+
+# ── Entry point ───────────────────────────────────────────────────────────────
+if __name__ == "__main__":
+    sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding="utf-8", errors="replace")
+
+    if len(sys.argv) < 2:
+        print("Usage: python generate_pdf.py <path_to_docx> [--out output.pdf]")
+        sys.exit(1)
+
+    docx_path = Path(sys.argv[1])
+    if not docx_path.exists():
+        print(f"Error: File not found: {docx_path}")
+        sys.exit(1)
+
+    if "--out" in sys.argv:
+        out_path = Path(sys.argv[sys.argv.index("--out") + 1])
+    else:
+        out_path = docx_path.with_suffix(".pdf")
+
+    generate_pdf(docx_path, out_path)