spsays-training/scripts/ochkm_extract.py

#!/usr/bin/env python3
"""Extract On Chanting the Hare Kṛṣṇa Mantra (OCHKM) from other.html.

The book has 1 intro segment (front-pages) and 4 chapters/sections.
There are NO p13 bookmarks — chapters are detected via p27 headings.

Output:
    lib/jsons/books/ochkm/intro.json          — Book metadata
    lib/jsons/books/ochkm/intro/*.json         — Intro segments
    lib/jsons/books/ochkm/chapters/*.json      — Chapters

Usage:
    python3 scripts/ochkm_extract.py
"""

import json
import os
import re
import sys

from bs4 import BeautifulSoup, Tag

# ---------------------------------------------------------------------------
# ScaGoudy → IAST Unicode mapping
# ---------------------------------------------------------------------------
SCAGOUDY_TO_IAST = {
    "ä": "ā",
    "Ä": "Ā",
    "å": "ṛ",
    "Å": "Ṛ",
    "ñ": "ṣ",
    "Ñ": "Ṣ",
    "ö": "ṭ",
    "Ö": "Ṭ",
    "ë": "ṇ",
    "Ë": "Ṇ",
    "ò": "ḍ",
    "Ò": "Ḍ",
    "ç": "ś",
    "Ç": "Ś",
    "ì": "ṅ",
    "ï": "ñ",
    "ü": "ū",
    "Ü": "Ū",
    "é": "ī",
    "É": "Ī",
    "à": "ṁ",
    "ù": "ḥ",
    "è": "ṝ",
    "û": "ṁl",
    "ÿ": "ḷ",
}


def decode_scagoudy(text: str) -> str:
    """Replace ScaGoudy-encoded characters with proper IAST Unicode."""
    for src, dst in SCAGOUDY_TO_IAST.items():
        text = text.replace(src, dst)
    return text


def decode_html_content(tag: Tag) -> str:
    """Extract inner HTML from a tag, decode ScaGoudy, preserve formatting tags."""
    inner = "".join(str(child) for child in tag.children)
    inner = decode_scagoudy(inner)
    # Remove span tags with CSS classes (RTF conversion artifacts)
    inner = re.sub(
        r'<span[^>]*class="[^"]*"[^>]*>(.*?)</span>', r"\1", inner, flags=re.DOTALL
    )
    # Clean up Apple-tab-span leftovers (tab characters)
    inner = inner.replace("\t", " ")
    # Normalize whitespace (but preserve <br> tags)
    inner = re.sub(r"\s+", " ", inner).strip()
    return inner


def decode_text(tag: Tag) -> str:
    """Get plain text from a tag with ScaGoudy decoding."""
    return decode_scagoudy(tag.get_text().strip())


def get_css_class(tag: Tag) -> str:
    """Get the first CSS class of a tag, or empty string."""
    classes = tag.get("class", [])
    return classes[0] if classes else ""


def is_empty(tag: Tag) -> bool:
    """Check if a tag is empty or whitespace-only."""
    text = tag.get_text().strip()
    return not text or text == "\xa0"


def is_dash_line(text: str) -> bool:
    """Return True if text is only dashes, spaces, and/or whitespace."""
    cleaned = text.strip()
    if not cleaned:
        return False
    return all(c in "-—–\u2014\u2013 \t" for c in cleaned)


# ---------------------------------------------------------------------------
# CSS class categorization for OCHKM
# ---------------------------------------------------------------------------

# Book title header class
BOOK_HEADER_CLASS = "p12"

# Chapter heading class (used as chapter boundary since no p13 bookmarks)
CHAPTER_HEADING_CLASS = "p27"

# Body paragraph classes (all treated as paragraphs)
BODY_CLASSES = frozenset(["p6", "p19", "p76", "p25", "p21", "p100"])

# Classes to skip entirely
SKIP_CLASSES = frozenset(["p9", "p15", "p70"])

# Chapter titles mapping (proper IAST decoded)
CHAPTER_TITLES = {
    1: "Brief Biography of the Spiritual Master",
    2: "On Chanting the Hare Kṛṣṇa Mantra",
    3: "What is ISKCON?",
    4: "Śrī Śrī Śikṣāṣṭakam",
}


# ---------------------------------------------------------------------------
# Extraction helpers
# ---------------------------------------------------------------------------


def find_ochkm_boundaries(all_paragraphs: list) -> tuple:
    """Find the start and end paragraph indices of the OCHKM content.

    Start: <p class="p12"> containing "ON CHANTING THE HARE"
    End: next <p class="p12"> after start
    """
    start_idx = None
    end_idx = len(all_paragraphs)

    for i, p in enumerate(all_paragraphs):
        cls = get_css_class(p)
        if cls != BOOK_HEADER_CLASS:
            continue
        text = p.get_text().strip().upper()
        if "ON CHANTING THE HARE" in text:
            start_idx = i
        elif start_idx is not None:
            # Next p12 after OCHKM start = end of OCHKM
            end_idx = i
            break

    if start_idx is None:
        print(
            "ERROR: Could not find 'ON CHANTING THE HARE KṚṢṆA MANTRA' header in other.html!"
        )
        sys.exit(1)

    return start_idx, end_idx


def find_chapter_headings(all_paragraphs: list, start_idx: int, end_idx: int) -> list:
    """Find all chapter headings (p27) in the OCHKM range.

    Returns list of (paragraph_index, sequential_chapter_number, chapter_title).
    """
    headings = []
    seq = 0

    for i in range(start_idx, end_idx):
        p = all_paragraphs[i]
        cls = get_css_class(p)
        if cls != CHAPTER_HEADING_CLASS:
            continue

        text = decode_text(p)
        # Remove bold/italic tags from text content
        text = re.sub(r"</?[bi]>", "", text).strip()

        if not text:
            continue

        seq += 1
        # Use pre-defined proper IAST titles
        ch_title = CHAPTER_TITLES.get(seq, text)
        headings.append((i, seq, ch_title))

    return headings


def extract_paragraphs(
    all_paragraphs: list,
    start_idx: int,
    end_idx: int,
    is_front_pages: bool = False,
) -> list:
    """Extract paragraphs for a section (chapter or intro segment).

    Handles all CSS classes within the OCHKM content.
    Detects italic+br verses in any body class → centered divs.
    """
    result = []
    verse_accumulator = []

    def flush_verse():
        if verse_accumulator:
            joined = "<br>".join(verse_accumulator)
            result.append(f'<div style="text-align:center"><i>{joined}</i></div>')
            verse_accumulator.clear()

    for i in range(start_idx, end_idx):
        p = all_paragraphs[i]
        cls = get_css_class(p)

        # Skip empty paragraphs
        if is_empty(p):
            continue

        # Skip noise classes
        if cls in SKIP_CLASSES:
            continue

        # Skip book title header (p12)
        if cls == BOOK_HEADER_CLASS:
            continue

        # Skip chapter headings (p27) — they are boundaries, not content
        if cls == CHAPTER_HEADING_CLASS:
            continue

        # Check for dash lines — skip them
        text = decode_text(p)
        if is_dash_line(text):
            continue

        # Non-verse paragraph — flush any accumulated verse first
        flush_verse()

        # Body paragraphs (p6, p19, p76, p25, p21, p100)
        if cls in BODY_CLASSES:
            html = decode_html_content(p)
            if html:
                # Detect Sanskrit verses disguised as body paragraphs:
                # they start with <i> and contain <br> (multi-line italic = verse)
                stripped_html = html.strip()
                if stripped_html.startswith("<i>") and "<br" in stripped_html:
                    # This is a verse — wrap in centered italic div
                    inner = re.sub(r"</?i>", "", stripped_html)
                    result.append(
                        f'<div style="text-align:center"><i>{inner}</i></div>'
                    )
                else:
                    result.append(html)
            continue

        # Any other class — try to extract content (be permissive)
        html = decode_html_content(p)
        if html and not is_dash_line(text):
            result.append(html)

    # Flush any trailing verse
    flush_verse()

    return result


# ---------------------------------------------------------------------------
# Main extraction logic
# ---------------------------------------------------------------------------


def extract_ochkm(soup: BeautifulSoup) -> dict:
    """Extract the entire OCHKM book from the parsed HTML.

    Returns a dict with:
        "intro_segments": list of intro segment dicts
        "chapters": list of chapter dicts
        "chapter_listing": list of {number, title} for intro.json
    """
    all_paragraphs = list(soup.find_all("p"))

    # Step 1: Find OCHKM boundaries
    ochkm_start, ochkm_end = find_ochkm_boundaries(all_paragraphs)
    print(f"  OCHKM content: paragraphs {ochkm_start} to {ochkm_end}")
    print(f"  Total paragraphs in range: {ochkm_end - ochkm_start}")

    # Step 2: Find all chapter headings (p27)
    chapter_headings = find_chapter_headings(all_paragraphs, ochkm_start, ochkm_end)
    print(f"  Found {len(chapter_headings)} chapter headings")
    for pos, ch_num, ch_title in chapter_headings:
        print(f"    [chapter {ch_num}] paragraph {pos}: {ch_title}")

    # Step 3: Extract front pages
    # From book start (ochkm_start) to first chapter heading
    intro_segments = []

    if chapter_headings:
        front_pages_end = chapter_headings[0][0]
    else:
        front_pages_end = ochkm_end

    front_pages_paragraphs = extract_paragraphs(
        all_paragraphs, ochkm_start, front_pages_end, is_front_pages=True
    )
    intro_segments.append(
        {
            "slug": "front-pages",
            "title": "Front Pages",
            "type": "intro-segment",
            "book": "ochkm",
            "order": 1,
            "paragraphs": front_pages_paragraphs,
        }
    )

    # Step 4: Extract chapters
    chapters = []
    for idx, (ch_pos, ch_num, ch_title) in enumerate(chapter_headings):
        # Content starts after the heading
        content_start = ch_pos + 1

        # Content ends at next chapter heading or ochkm_end
        if idx + 1 < len(chapter_headings):
            content_end = chapter_headings[idx + 1][0]
        else:
            content_end = ochkm_end

        paragraphs = extract_paragraphs(all_paragraphs, content_start, content_end)

        ch = {
            "book": "ochkm",
            "chapter": ch_num,
            "title": ch_title,
            "type": "chapter",
            "paragraphs": paragraphs,
        }
        chapters.append(ch)

    # Step 5: Build chapter listing for intro.json
    chapter_listing = [
        {"number": ch["chapter"], "title": ch["title"]} for ch in chapters
    ]

    return {
        "intro_segments": intro_segments,
        "chapters": chapters,
        "chapter_listing": chapter_listing,
    }


# ---------------------------------------------------------------------------
# Output
# ---------------------------------------------------------------------------


def write_json(data, filepath):
    """Write data as formatted JSON to filepath, creating directories as needed."""
    os.makedirs(os.path.dirname(filepath), exist_ok=True)
    with open(filepath, "w", encoding="utf-8") as f:
        json.dump(data, f, ensure_ascii=False, indent=2)


def main():
    base_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
    html_path = os.path.join(base_dir, "public", "html", "other.html")
    output_base = os.path.join(base_dir, "lib", "jsons", "books", "ochkm")

    if not os.path.exists(html_path):
        print(f"ERROR: {html_path} not found!")
        sys.exit(1)

    print(f"Reading {html_path}...")
    with open(html_path, "r", encoding="utf-8") as f:
        soup = BeautifulSoup(f, "html.parser")

    print("Extracting On Chanting the Hare Kṛṣṇa Mantra...\n")
    result = extract_ochkm(soup)

    intro_segments = result["intro_segments"]
    chapters = result["chapters"]
    chapter_listing = result["chapter_listing"]

    # --- Write intro.json ---
    intro_meta = {
        "book": "ochkm",
        "title": "On Chanting the Hare Kṛṣṇa Mantra",
        "author": "His Divine Grace A.C. Bhaktivedanta Swami Prabhupāda",
        "year": 1967,
        "description": "The sublime method for reviving our transcendental consciousness through the chanting of the holy names.",
        "totalChapters": len(chapters),
        "chapters": chapter_listing,
    }
    intro_path = os.path.join(output_base, "intro.json")
    write_json(intro_meta, intro_path)
    print(f"\n  Written: {intro_path}")

    # --- Write intro segments ---
    print(f"\n--- Intro Segments ---")
    for seg in intro_segments:
        filepath = os.path.join(output_base, "intro", f"{seg['slug']}.json")
        write_json(seg, filepath)
        print(
            f"  {seg['order']}. {seg['title']}: {len(seg['paragraphs'])} paragraphs → {filepath}"
        )

    # --- Write chapters ---
    print(f"\n--- Chapters ---")
    for ch in chapters:
        ch_num_str = str(ch["chapter"]).zfill(2)
        filepath = os.path.join(output_base, "chapters", f"{ch_num_str}.json")
        write_json(ch, filepath)
        print(
            f"  Ch {ch['chapter']:>2}: {len(ch['paragraphs']):>3} paragraphs — "
            f"{ch['title']}"
        )

    # --- Summary ---
    total_intro_paras = sum(len(s["paragraphs"]) for s in intro_segments)
    total_chapter_paras = sum(len(ch["paragraphs"]) for ch in chapters)
    total_paras = total_intro_paras + total_chapter_paras

    print(f"\n{'=' * 60}")
    print(f"  EXTRACTION SUMMARY")
    print(f"{'=' * 60}")
    print(f"  Intro segments:     {len(intro_segments)}")
    print(f"  Chapters:           {len(chapters)}")
    print(f"  Total paragraphs:   {total_paras}")
    print(f"    Intro:            {total_intro_paras}")
    print(f"    Chapters:         {total_chapter_paras}")
    if chapters:
        print(f"  Avg paras/chapter:  {total_chapter_paras / len(chapters):.1f}")
    print(f"  Output directory:   {output_base}/")

    # --- Samples ---
    print(f"\n--- Sample: Front Pages ---")
    if intro_segments:
        front = intro_segments[0]
        for i, p in enumerate(front["paragraphs"][:5]):
            preview = re.sub(r"<[^>]+>", "", p)[:100]
            print(f"  [{i}] {preview}...")

    if chapters:
        print(f"\n--- Sample: Chapter 1 ({chapters[0]['title']}) ---")
        ch1 = chapters[0]
        for i, p in enumerate(ch1["paragraphs"][:5]):
            preview = re.sub(r"<[^>]+>", "", p)[:100]
            print(f"  [{i}] {preview}...")

    if chapters and len(chapters) > 1:
        print(
            f"\n--- Sample: Chapter {chapters[-1]['chapter']} ({chapters[-1]['title']}) ---"
        )
        last = chapters[-1]
        for i, p in enumerate(last["paragraphs"][:5]):
            preview = re.sub(r"<[^>]+>", "", p)[:100]
            print(f"  [{i}] {preview}...")

    # --- Quality Checks ---
    print(f"\n--- Quality Checks ---")

    # Check all 4 chapters present
    expected_chapters = set(range(1, 5))
    actual_chapters = set(ch["chapter"] for ch in chapters)
    missing = expected_chapters - actual_chapters
    extra = actual_chapters - expected_chapters
    if missing:
        print(f"  ⚠ Missing chapters: {sorted(missing)}")
    if extra:
        print(f"  ⚠ Extra chapters: {sorted(extra)}")
    if not missing and not extra:
        print(f"  ✅ All 4 chapters present")

    # Check all 1 intro segment present
    expected_intros = {"front-pages"}
    actual_intros = {s["slug"] for s in intro_segments}
    missing_intros = expected_intros - actual_intros
    extra_intros = actual_intros - expected_intros
    if missing_intros:
        print(f"  ⚠ Missing intro segments: {missing_intros}")
    if extra_intros:
        print(f"  ⚠ Extra intro segments: {extra_intros}")
    if not missing_intros and not extra_intros:
        print(f"  ✅ All 1 intro segments present")

    # Check empty chapters
    empty_chapters = [ch["chapter"] for ch in chapters if len(ch["paragraphs"]) == 0]
    if empty_chapters:
        print(f"  ⚠ Empty chapters (0 paragraphs): {empty_chapters}")
    else:
        print(f"  ✅ No empty chapters")

    # Check empty intro segments
    empty_intros = [s["slug"] for s in intro_segments if len(s["paragraphs"]) == 0]
    if empty_intros:
        print(f"  ⚠ Empty intro segments: {empty_intros}")
    else:
        print(f"  ✅ No empty intro segments")

    # Check ScaGoudy decoding
    print(f"\n--- ScaGoudy Decoding Check ---")
    all_text = ""
    for ch in chapters:
        all_text += " ".join(ch["paragraphs"][:20])
    for seg in intro_segments:
        all_text += " ".join(seg["paragraphs"][:10])

    if "Kṛṣṇa" in all_text:
        print(f"  ✅ 'Kṛṣṇa' found (ScaGoudy properly decoded)")
    else:
        print(f"  ⚠ 'Kṛṣṇa' not found — ScaGoudy decoding may have failed")

    if "Prabhupāda" in all_text:
        print(f"  ✅ 'Prabhupāda' found (ScaGoudy properly decoded)")

    if "saṅkīrtana" in all_text or "saṅkértana" in all_text or "saṁkīrtana" in all_text:
        print(f"  ✅ 'saṅkīrtana' variant found (ScaGoudy properly decoded)")

    # Check for un-decoded ScaGoudy remnants
    scagoudy_remnants = 0
    for ch in chapters:
        for p in ch["paragraphs"]:
            if "Kåñëa" in p or "Prabhupäda" in p:
                scagoudy_remnants += 1
    for seg in intro_segments:
        for p in seg["paragraphs"]:
            if "Kåñëa" in p or "Prabhupäda" in p:
                scagoudy_remnants += 1
    if scagoudy_remnants == 0:
        print(f"  ✅ No ScaGoudy remnants detected")
    else:
        print(f"  ⚠ {scagoudy_remnants} paragraphs with ScaGoudy remnants")

    # Check for centered verse blocks
    verse_block_count = 0
    for ch in chapters:
        for p in ch["paragraphs"]:
            if '<div style="text-align:center"><i>' in p:
                verse_block_count += 1
    for seg in intro_segments:
        for p in seg["paragraphs"]:
            if '<div style="text-align:center"><i>' in p:
                verse_block_count += 1
    if verse_block_count > 0:
        print(f"  ✅ {verse_block_count} centered Sanskrit verse blocks found")
    else:
        print(f"  ℹ No centered verse blocks found (may be expected for this book)")

    # Check for residual span tags
    span_count = 0
    for ch in chapters:
        for p in ch["paragraphs"]:
            if 'class="' in p and "<span" in p:
                span_count += 1
    for seg in intro_segments:
        for p in seg["paragraphs"]:
            if 'class="' in p and "<span" in p:
                span_count += 1
    if span_count == 0:
        print(f"  ✅ No residual span class artifacts found")
    else:
        print(f"  ⚠ {span_count} paragraphs still contain span class artifacts")

    # Check for dash-line paragraphs that shouldn't be there
    dash_count = 0
    for ch in chapters:
        for p in ch["paragraphs"]:
            plain = re.sub(r"<[^>]+>", "", p)
            if is_dash_line(plain):
                dash_count += 1
    for seg in intro_segments:
        for p in seg["paragraphs"]:
            plain = re.sub(r"<[^>]+>", "", p)
            if is_dash_line(plain):
                dash_count += 1
    if dash_count == 0:
        print(f"  ✅ No dashed-line paragraphs in output")
    else:
        print(f"  ⚠ {dash_count} dashed-line paragraphs found in output")

    # Chapter title decoding check
    print(f"\n--- Chapter Title Decoding Check ---")
    for ch in chapters:
        title = ch["title"]
        if "å" in title or "ë" in title or "ñ" in title or "ä" in title:
            print(f"  ⚠ Chapter {ch['chapter']} title has un-decoded ScaGoudy: {title}")
        else:
            print(f"  ✅ Ch {ch['chapter']}: {title}")

    print(f"\nDone! ✨")


if __name__ == "__main__":
    main()