", "
", inner)
# Normalize spaces around text (not across
)
parts = inner.split("
")
parts = [re.sub(r"\s+", " ", p).strip() for p in parts]
inner = "
".join(parts)
return inner.strip()
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"
# ---------------------------------------------------------------------------
# CSS class categorization for GG
# ---------------------------------------------------------------------------
# Book title header class
BOOK_HEADER_CLASS = "p12"
# Subtitle class
SUBTITLE_CLASS = "p31"
# Bookmark class
BOOKMARK_CLASS = "p13"
# Introduction heading / Chapter heading
HEADING_CLASS = "p27"
# Introduction body text
BODY_CLASS = "p6"
# Signature block
SIGNATURE_CLASS = "p222"
# Chapter title heading (contains Chapter N + Title)
CHAPTER_TITLE_CLASS = "p223"
# Bengali verse text (italic, with
line breaks — THE MAIN content)
VERSE_CLASS = "p24"
# Text number headings (bold)
TEXT_HEADING_CLASS = "p224"
# Empty/break classes to skip
SKIP_CLASSES = frozenset(["p225", "p226", "p15"])
# ---------------------------------------------------------------------------
# Chapter titles mapping (from bookmarks)
# ---------------------------------------------------------------------------
CHAPTER_TITLES = {
1: "Observing the Armies on the Battlefield of Kurukṣetra",
2: "Contents of the Gītā Summarized",
3: "Karma-yoga",
4: "Transcendental Knowledge",
5: "Karma-yoga—Action in Kṛṣṇa Consciousness",
6: "Dhyāna-yoga",
7: "Knowledge of the Absolute",
8: "Attaining the Supreme",
9: "The Most Confidential Knowledge",
10: "The Opulence of the Absolute",
11: "The Universal Form",
12: "Devotional Service",
13: "Nature, the Enjoyer, and Consciousness",
14: "The Three Modes of Material Nature",
15: "The Yoga of the Supreme Person",
16: "The Divine and Demoniac Natures",
17: "The Divisions of Faith",
18: "Conclusion—The Perfection of Renunciation",
}
# ---------------------------------------------------------------------------
# Extraction helpers
# ---------------------------------------------------------------------------
def find_gg_boundaries(all_paragraphs: list) -> tuple:
"""Find the start and end paragraph indices of the GG content.
Start: Gétär Gän
(or decoded: Gītār Gān)
End: Next after GG content (Songs of the Vaiṣṇava Ācāryas)
"""
start_idx = None
end_idx = len(all_paragraphs)
for i, p in enumerate(all_paragraphs):
cls = get_css_class(p)
text = p.get_text().strip()
decoded = decode_scagoudy(text)
if cls == BOOK_HEADER_CLASS:
if start_idx is None:
# Look for "Gétär Gän" or decoded "Gītār Gān"
if (
"Gétär Gän" in text
or "Gītār Gān" in decoded
or "G\u00e9t\u00e4r G\u00e4n" in text
):
start_idx = i
continue
else:
# Next p12 after start marks the end
end_idx = i
break
if start_idx is None:
print("ERROR: Could not find 'Gītār Gān' header in other.html!")
sys.exit(1)
return start_idx, end_idx
def find_all_bookmarks(all_paragraphs: list, start_idx: int, end_idx: int) -> list:
"""Find all GG bookmark tags in the GG range.
Bookmarks are p13 class with text starting with "GG:".
Returns list of (paragraph_index, bookmark_text).
"""
bookmarks = []
for i in range(start_idx, end_idx):
p = all_paragraphs[i]
cls = get_css_class(p)
if cls != BOOKMARK_CLASS:
continue
text = decode_text(p).strip()
if text.startswith("GG:"):
bookmarks.append((i, text))
return bookmarks
def parse_chapter_bookmark(text: str):
"""Parse a chapter bookmark like 'GG: Chapter 1'.
Returns chapter_number (int) or None if not a chapter bookmark.
"""
m = re.match(r"GG:\s+Chapter\s+(\d+)", text.strip())
if m:
return int(m.group(1))
return None
def extract_chapter_title(tag: Tag) -> tuple:
"""Extract chapter number and title from a p223 element.
The p223 contains something like:
Chapter 1Observing the Armies...
Returns (chapter_number, title_text) or None.
"""
text = decode_text(tag)
# Try to parse "Chapter N\nTitle" or "Chapter N Title"
m = re.match(r"Chapter\s+(\d+)\s*(.*)", text, re.DOTALL)
if m:
ch_num = int(m.group(1))
title = m.group(2).strip()
# Clean up any leading/trailing whitespace or newlines
title = re.sub(r"\s+", " ", title).strip()
return ch_num, title
return None
def extract_verse_content(tag: Tag) -> str:
"""Extract Bengali verse content from a p24 tag.
The content is italic with
line breaks. We wrap it as a centered
italic div:
content
ScaGoudy decoding is applied.
"""
html = decode_html_content(tag)
if not html:
return ""
# Strip outer tags if present — we'll re-wrap
stripped = re.sub(r"^(.*)$", r"\1", html, flags=re.DOTALL)
# Also remove any remaining /<\/i> pairs within
stripped = re.sub(r"", "", stripped)
if stripped.strip():
return f'{stripped.strip()}
'
return ""
def extract_text_heading(tag: Tag) -> tuple:
"""Extract text heading from a p224 tag.
Content is bold, e.g. Text 1 or Texts 16-18
or closing: Thus Bhaktivedānta sings...
Some headings have the translation merged in, like:
Text 30: All these performers who know...
These are split: the text number is returned as the heading,
and the translation is returned separately as a "deferred" paragraph
to be emitted AFTER the next verse block.
Returns (heading_html, deferred_translation_html_or_None).
"""
html = decode_html_content(tag)
if not html:
return ("", None)
stripped = html.strip()
# Remove outer to inspect inner text
inner = stripped
if inner.startswith("") and inner.endswith(""):
inner = inner[3:-4].strip()
# Check if this is a merged "Text N: translation..." heading
m = re.match(r"^(Texts?\s+[\d\-–]+)\s*[:\.]\s*(.+)$", inner, re.DOTALL)
if m:
text_num = m.group(1).strip()
translation = m.group(2).strip()
heading = f'{text_num}
'
deferred = f'{translation}
'
return (heading, deferred)
# Plain text heading (e.g. "Text 1") or closing statement — center it
if not stripped.startswith(""):
stripped = f"{stripped}"
return (f'{stripped}
', None)
def extract_chapter_heading(tag: Tag) -> str:
"""Extract chapter heading from a p223 tag as a bold paragraph.
Returns something like: Chapter 1
Observing the Armies...
"""
html = decode_html_content(tag)
if not html:
return ""
# Clean up the HTML — remove inner bold tags and re-wrap the whole thing
inner = re.sub(r"", "", html)
inner = inner.strip()
# Try to add a
between "Chapter N" and the title if not already there
inner = re.sub(r"(Chapter\s+\d+)\s+", r"\1
", inner, count=1)
return f"{inner}"
def extract_paragraphs_for_intro(
all_paragraphs: list,
start_idx: int,
end_idx: int,
) -> list:
"""Extract paragraphs for the introduction section.
Handles p27 (heading), p6 (body text), p222 (signature block).
"""
result = []
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 bookmark tags
if cls == BOOKMARK_CLASS:
continue
# Skip book header (p12)
if cls == BOOK_HEADER_CLASS:
continue
# Skip subtitle (p31)
if cls == SUBTITLE_CLASS:
continue
# Introduction heading (p27)
if cls == HEADING_CLASS:
html = decode_html_content(p)
if html:
result.append(html)
continue
# Body text (p6)
if cls == BODY_CLASS:
html = decode_html_content(p)
if html:
result.append(html)
continue
# Signature block (p222)
if cls == SIGNATURE_CLASS:
html = decode_html_content(p)
if html:
result.append(html)
continue
# Any other class in intro range — try to include
html = decode_html_content(p)
if html:
result.append(html)
return result
def extract_paragraphs_for_chapter(
all_paragraphs: list,
start_idx: int,
end_idx: int,
) -> list:
"""Extract paragraphs for a chapter.
Handles:
p223 — Chapter heading (bold paragraph)
p24 — Bengali verse text (centered italic div)
p224 — Text number heading (bold paragraph)
p225, p226, p15 — Skip (empty/break)
"""
result = []
deferred_translation = None
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 bookmark tags
if cls == BOOKMARK_CLASS:
continue
# Chapter title heading (p223)
if cls == CHAPTER_TITLE_CLASS:
heading = extract_chapter_heading(p)
if heading:
result.append(heading)
continue
# Bengali verse text (p24)
if cls == VERSE_CLASS:
verse = extract_verse_content(p)
if verse:
result.append(verse)
# Emit deferred translation AFTER the verse
if deferred_translation:
result.append(deferred_translation)
deferred_translation = None
continue
# Text number heading (p224)
if cls == TEXT_HEADING_CLASS:
# If there's a deferred translation from a previous heading,
# emit it now (before this new heading)
if deferred_translation:
result.append(deferred_translation)
deferred_translation = None
heading, deferred = extract_text_heading(p)
if heading:
result.append(heading)
if deferred:
# Hold the translation — it will be emitted after the next verse block
deferred_translation = deferred
continue
# Heading class (p27) — may appear within chapters
if cls == HEADING_CLASS:
html = decode_html_content(p)
if html:
if not html.strip().startswith(""):
html = f"{html}"
result.append(html)
continue
# Body text (p6) — may appear within chapters
if cls == BODY_CLASS:
html = decode_html_content(p)
if html:
result.append(html)
continue
# Any other class — try to extract content (be permissive)
html = decode_html_content(p)
if html:
result.append(html)
# Flush any remaining deferred translation at end of chapter
if deferred_translation:
result.append(deferred_translation)
deferred_translation = None
return result
# ---------------------------------------------------------------------------
# Main extraction logic
# ---------------------------------------------------------------------------
def extract_gg(soup: BeautifulSoup) -> dict:
"""Extract the entire Gītār Gān 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 GG boundaries
gg_start, gg_end = find_gg_boundaries(all_paragraphs)
print(f" GG content: paragraphs {gg_start} to {gg_end}")
print(f" Total paragraphs in range: {gg_end - gg_start}")
# Step 2: Find all bookmarks
bookmarks = find_all_bookmarks(all_paragraphs, gg_start, gg_end)
print(f" Found {len(bookmarks)} GG bookmarks")
for pos, text in bookmarks:
print(f" Para ~{pos}: {text}")
# Step 3: Categorize bookmarks
chapter_bookmarks = [] # (para_idx, chapter_num)
for pos, text in bookmarks:
ch_num = parse_chapter_bookmark(text)
if ch_num is not None:
chapter_bookmarks.append((pos, ch_num))
print(f" Chapter bookmarks: {len(chapter_bookmarks)}")
# Sort chapter bookmarks by position
chapter_bookmarks.sort(key=lambda x: x[0])
# Build an ordered list of all section start positions for boundary computation
all_section_starts = sorted([pos for pos, _ in chapter_bookmarks])
def next_section_after(pos):
"""Return the next section start after `pos`, or gg_end."""
for s in all_section_starts:
if s > pos:
return s
return gg_end
# Step 4: Extract Introduction
# The introduction is everything between the book header and the first chapter bookmark
introduction = None
if chapter_bookmarks:
first_chapter_pos = chapter_bookmarks[0][0]
# Introduction starts after the book header (gg_start + 1) and ends at first bookmark
intro_content_start = gg_start + 1
intro_content_end = first_chapter_pos
intro_paras = extract_paragraphs_for_intro(
all_paragraphs, intro_content_start, intro_content_end
)
if intro_paras:
introduction = {
"slug": "introduction",
"title": "Introduction",
"type": "intro-segment",
"book": "gg",
"order": 1,
"paragraphs": intro_paras,
}
# Step 5: Extract chapters 1–18
chapters = []
for idx, (ch_pos, ch_num) in enumerate(chapter_bookmarks):
content_start = ch_pos + 1
content_end = next_section_after(ch_pos)
paragraphs = extract_paragraphs_for_chapter(
all_paragraphs, content_start, content_end
)
# Get title from CHAPTER_TITLES mapping, or try to extract from p223
title = CHAPTER_TITLES.get(ch_num, f"Chapter {ch_num}")
# Try to extract title from the first p223 in the range
for i in range(content_start, min(content_start + 10, content_end)):
p = all_paragraphs[i]
cls = get_css_class(p)
if cls == CHAPTER_TITLE_CLASS:
extracted = extract_chapter_title(p)
if extracted:
_, extracted_title = extracted
if extracted_title:
title = extracted_title
break
ch = {
"book": "gg",
"chapter": ch_num,
"title": title,
"type": "chapter",
"paragraphs": paragraphs,
}
chapters.append(ch)
# Step 6: Build intro segments list
intro_segments = []
if introduction:
intro_segments.append(introduction)
# Step 7: 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", "gg")
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 Gītār Gān...\n")
result = extract_gg(soup)
intro_segments = result["intro_segments"]
chapters = result["chapters"]
chapter_listing = result["chapter_listing"]
# --- Write intro.json ---
intro_meta = {
"book": "gg",
"title": "Gītār Gān",
"author": "His Divine Grace A.C. Bhaktivedanta Swami Prabhupāda",
"year": 1995,
"description": "A Bengali poetic rendering of Bhagavad-gītā As It Is by Śrīla Prabhupāda.",
"totalChapters": 18,
"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)}")
for seg in intro_segments:
print(
f" {seg['order']}. {seg['title']}: {len(seg['paragraphs'])} paragraphs"
)
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: Chapter 1 ---")
if chapters:
ch1 = chapters[0]
for i, p in enumerate(ch1["paragraphs"][:5]):
preview = re.sub(r"<[^>]+>", "", p)[:120]
print(f" [{i}] {preview}...")
if len(chapters) >= 9:
print(f"\n--- Sample: Chapter 9 ---")
ch9 = [c for c in chapters if c["chapter"] == 9]
if ch9:
for i, p in enumerate(ch9[0]["paragraphs"][:3]):
preview = re.sub(r"<[^>]+>", "", p)[:120]
print(f" [{i}] {preview}...")
if len(chapters) >= 18:
print(f"\n--- Sample: Chapter 18 (last) ---")
last = chapters[-1]
for i, p in enumerate(last["paragraphs"][:3]):
preview = re.sub(r"<[^>]+>", "", p)[:120]
print(f" [{i}] {preview}...")
# --- Quality Checks ---
print(f"\n--- Quality Checks ---")
# Check all 18 chapters present
expected_chapters = set(range(1, 19))
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 18 chapters present")
# Check intro segments
expected_intros = {"introduction"}
actual_intros = {s["slug"] for s in intro_segments}
missing_intros = expected_intros - actual_intros
if missing_intros:
print(f" ⚠ Missing intro segments: {missing_intros}")
else:
print(f" ✅ Introduction segment 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 ScaGoudy decoding — look for common encoded characters
encoded_chars = set("äåöëòçüéàùè")
suspect_count = 0
all_sections = chapters + intro_segments
for section in all_sections:
for p in section["paragraphs"]:
plain = re.sub(r"<[^>]+>", "", p)
for c in encoded_chars:
if c in plain:
suspect_count += 1
break
if suspect_count == 0:
print(f" ✅ No suspected undecoded ScaGoudy characters found")
else:
print(
f" ℹ {suspect_count} paragraphs contain characters that may be undecoded ScaGoudy"
)
# Check verse blocks
verse_block_count = 0
for ch in chapters:
for p in ch["paragraphs"]:
if '' in p:
verse_block_count += 1
print(f" ℹ {verse_block_count} verse blocks in chapters")
# Check text headings
text_heading_count = 0
for ch in chapters:
for p in ch["paragraphs"]:
if re.match(r"Texts?\s+\d+", p):
text_heading_count += 1
print(f" ℹ {text_heading_count} text number headings in chapters")
# ScaGoudy decoding spot check
print(f"\n--- ScaGoudy Decoding Spot Check ---")
if chapters:
ch1_text = " ".join(chapters[0]["paragraphs"][:15])
if "Kṛṣṇa" in ch1_text:
print(f" ✅ 'Kṛṣṇa' correctly decoded in Chapter 1")
elif "Kåñëa" in ch1_text:
print(f" ⚠ 'Kåñëa' still encoded — ScaGoudy decoding may have failed!")
else:
print(
f" ℹ 'Kṛṣṇa' not found in Chapter 1 first 15 paragraphs (may be normal)"
)
if "Kurukṣetra" in ch1_text:
print(f" ✅ 'Kurukṣetra' correctly decoded in Chapter 1")
elif "Kurukñetra" in ch1_text:
print(f" ⚠ 'Kurukñetra' still encoded")
else:
print(f" ℹ 'Kurukṣetra' not found in Chapter 1 first 15 paragraphs")
# Sample verse blocks
print(f"\n--- Sample Verse Blocks ---")
shown = 0
for ch in chapters:
for p in ch["paragraphs"]:
if '' in p:
preview = re.sub(r"<[^>]+>", "", p)[:120]
print(f" Ch {ch['chapter']}: {preview}...")
shown += 1
if shown >= 3:
break
if shown >= 3:
break
# Check chapter headings
print(f"\n--- Chapter Headings ---")
heading_count = 0
for ch in chapters:
for p in ch["paragraphs"]:
if p.startswith("Chapter"):
heading_count += 1
preview = re.sub(r"<[^>]+>", "", p)[:80]
if heading_count <= 5:
print(f" Ch {ch['chapter']}: {preview}")
break
print(f" Total chapter headings: {heading_count}")
# Check closing statements
print(f"\n--- Closing Statements ---")
closing_count = 0
for ch in chapters:
for p in ch["paragraphs"]:
plain = re.sub(r"<[^>]+>", "", p)
if "Bhaktivedānta" in plain and "sings" in plain.lower():
closing_count += 1
break
print(f" Chapters with closing statement: {closing_count} / {len(chapters)}")
print(f"\nDone! ✨")
if __name__ == "__main__":
main()