""" Generate a bilingual (English + Chinese) paragraph-interleaved PDF for Jyotish-Prasana. Input: - progress/Jyotish-Prasana_pages.json (original English, 190 pages) - progress/Jyotish-Prasana_translated.json (DeepSeek Chinese translation, 190 pages) Output: - output/Jyotish-Prasana_bilingual.pdf Layout: EN paragraph → ZH paragraph → separator → repeat. Headings auto-detected and rendered in dark blue. Usage: python tools/generate_bilingual_pdf.py """ import json import re import sys from pathlib import Path from fpdf import FPDF # ---- Config ---- PROJECT_ROOT = Path(__file__).resolve().parent.parent ENGLISH_JSON = PROJECT_ROOT / "progress" / "Jyotish-Prasana_pages.json" CHINESE_JSON = PROJECT_ROOT / "progress" / "Jyotish-Prasana_translated.json" OUTPUT_DIR = PROJECT_ROOT / "output" OUTPUT_FILE = OUTPUT_DIR / "Jyotish-Prasana_bilingual.pdf" FONT_YAHEI = "C:/Windows/Fonts/msyh.ttc" FONT_YAHEI_BOLD = "C:/Windows/Fonts/msyhbd.ttc" FONT_TNR = "C:/Windows/Fonts/times.ttf" FONT_TNR_I = "C:/Windows/Fonts/timesi.ttf" # Page layout PAGE_W = 210 PAGE_H = 297 MARGIN_TOP = 18 MARGIN_BOTTOM = 18 MARGIN_SIDE = 20 BODY_W = PAGE_W - 2 * MARGIN_SIDE # 170mm # Colors COLOR_HEADING = (26, 82, 118) # dark blue COLOR_EN_BODY = (50, 50, 50) # dark gray COLOR_ZH_BODY = (0, 0, 0) # black COLOR_SEP = (180, 180, 180) # light gray # Font sizes SIZE_EN_BODY = 10.5 SIZE_ZH_BODY = 10 SIZE_HEADING_EN = 11.5 SIZE_HEADING_ZH = 11 LINE_H = 5.8 # ---- Character filtering ---- def strip_unsupported(text: str) -> str: """Keep only chars supported by YaHei/TNR: ASCII, Latin Ext (IAST), CJK, punctuation.""" result = [] for ch in text: cp = ord(ch) if ( cp < 0x7F or 0xA0 <= cp <= 0x24F or 0x2000 <= cp <= 0x206F or 0x2E00 <= cp <= 0x2E7F or 0x3000 <= cp <= 0x303F or 0x3400 <= cp <= 0x4DBF or 0x4E00 <= cp <= 0x9FFF or 0xFF00 <= cp <= 0xFFEF ): result.append(ch) return "".join(result) IAST_MAP = str.maketrans({ "Ā": "A", "ā": "a", "Ī": "I", "ī": "i", "Ū": "U", "ū": "u", "Ṛ": "R", "ṛ": "r", "Ṣ": "S", "ṣ": "s", "Ṭ": "T", "ṭ": "t", "Ḍ": "D", "ḍ": "d", "Ṇ": "N", "ṇ": "n", "Ṅ": "N", "ṅ": "n", "Ḷ": "L", "ḷ": "l", "Ḥ": "H", "ḥ": "h", "Ṃ": "M", "ṃ": "m", "Ś": "S", "ś": "s", "Ñ": "N", "ñ": "n", }) # ---- Text Cleaning ---- RE_IMG = re.compile(r"^\[(?:嵌入图片|插图)[^\]]*\]") RE_ROMAN = re.compile(r"^\([ivxlcdm]+\)\s*$", re.IGNORECASE) RE_HLINE = re.compile(r"^[\s\-_~=]+$") RE_HEADING_EN = re.compile( r"^(?:Chapter|CHAPTER)\s*-?\s*[IVXLCDM\d]+\b|" r"^(?:[IVXLCDM]+\.?\s+[A-Z])|" r"^[A-Z]{2,}(?:\s+[A-Z]+){0,4}\s*:?$|" r"^\d+\.\s+[A-Z][a-z]" ) RE_HEADING_ZH = re.compile( r"^第[一二三四五六七八九十百千\d]+[章节卷]|" r"^[一二三四五六七八九十]+[、．、]\s*\S|" r"^（[一二三四五六七八九十\d]+）" ) def is_heading_en(line: str) -> bool: """Detect if an English line is a heading/title.""" s = line.strip() if not s or len(s) > 80: return False if s.isupper() and len(s) > 3: return True if RE_HEADING_EN.match(s): return True return False def is_heading_zh(line: str) -> bool: """Detect if a Chinese line is a heading/title.""" s = line.strip() if not s or len(s) > 60: return False if RE_HEADING_ZH.match(s): return True # Line ending with ：that contains CJK chars (section heading) if s.endswith("：") and re.search(r"[一-鿿]", s): return True # Short line with Chinese + English in parens (e.g. "太阴年（Lunar Years）：") if re.match(r"^[一-鿿]{2,30}（[A-Za-z\s]+）[：:]?$", s): return True return False def clean_english(text: str) -> str: """Remove artifacts from English OCR text.""" lines = text.split("\n") out = [] for line in lines: stripped = line.strip() if RE_IMG.match(stripped): continue if RE_HLINE.match(stripped): continue if len(stripped) > 5: printable = sum(1 for c in stripped if 32 <= ord(c) <= 126) if printable / max(len(stripped), 1) < 0.4: continue # Detect garbled OCR: count actual English-like words (3+ alpha chars) words = re.findall(r"[A-Za-z]{3,}", stripped) word_chars = sum(len(w) for w in words) alpha_chars = sum(1 for c in stripped if c.isalpha()) if alpha_chars > 10 and word_chars / max(alpha_chars, 1) < 0.3: continue # mostly non-word garbage line = line.replace("~", "") line = strip_unsupported(line) # Clean up empty parenthetical left after Devanagari removal: "( / ... →" → "(..." line = re.sub(r"\(\s*/\s*", "(", line) out.append(line) return "\n".join(out) def clean_chinese(text: str) -> str: """Remove artifacts from Chinese translated text.""" lines = text.split("\n") out = [] for line in lines: stripped = line.strip() if RE_IMG.match(stripped): continue if "梵文诗句，内容待确认" in stripped: continue if RE_ROMAN.match(stripped): continue if RE_HLINE.match(stripped): continue line = line.replace("**", "") line = line.replace("~", "") line = strip_unsupported(line) line = re.sub(r"\(\s*/\s*", "(", line) out.append(line) return "\n".join(out) # ---- PDF Class ---- class BilingualPDF(FPDF): def __init__(self): super().__init__("P", "mm", "A4") self.set_auto_page_break(auto=True, margin=MARGIN_BOTTOM) self.add_font("YaHei", "", FONT_YAHEI) self.add_font("YaHei", "B", FONT_YAHEI_BOLD) self.add_font("TNR", "", FONT_TNR) self.add_font("TNR", "I", FONT_TNR_I) self._cover_done = False def footer(self): if not self._cover_done: return self.set_y(-12) self.set_font("TNR", "", 8) self.set_text_color(*COLOR_SEP) self.cell(0, 8, f"- {self.page_no()} -", align="C") def add_cover_page(self): """Bilingual cover page.""" self.add_page() self.ln(45) self.set_font("YaHei", "B", 28) self.set_text_color(*COLOR_HEADING) self.cell(0, 14, "MUHURTHA SINDHU", align="C", new_x="LMARGIN", new_y="NEXT") self.set_font("YaHei", "B", 20) self.cell(0, 12, "择时宝鉴", align="C", new_x="LMARGIN", new_y="NEXT") self.ln(4) self.set_font("TNR", "I", 14) self.set_text_color(80, 80, 80) self.cell(0, 10, "A Manual of Electional Astrology", align="C", new_x="LMARGIN", new_y="NEXT") self.set_font("YaHei", "", 12) self.cell(0, 10, "选举占星学手册", align="C", new_x="LMARGIN", new_y="NEXT") self.ln(8) self.set_font("YaHei", "", 11) self.set_text_color(0, 0, 0) self.cell(0, 8, "作者：Iranganti Rangacharya", align="C", new_x="LMARGIN", new_y="NEXT") self.ln(4) self.set_font("YaHei", "", 10) self.set_text_color(100, 100, 100) self.cell(0, 7, "英汉对照版 | Bilingual Edition", align="C", new_x="LMARGIN", new_y="NEXT") self.cell(0, 7, "Chinese translation by DeepSeek AI | Typeset with fpdf2", align="C", new_x="LMARGIN", new_y="NEXT") self.ln(14) # Rule self.set_draw_color(*COLOR_SEP) self.set_line_width(0.3) self.line(MARGIN_SIDE + 10, self.get_y(), PAGE_W - MARGIN_SIDE - 10, self.get_y()) self.ln(8) self.set_font("YaHei", "", 8.5) self.set_text_color(100, 100, 100) desc = ( "本书为 Iranganti Rangacharya 所著《Muhurtha Sindhu》的英汉对照版。" "英文原文通过 PDF 提取，中文翻译由 DeepSeek AI 完成。" "全书共 190 页，涵盖择时占星学的完整体系。" ) self.multi_cell(w=BODY_W, h=5, text=desc, align="C") self._cover_done = True # ---- Rendering ---- def write_styled_block(pdf: BilingualPDF, text: str, lang: str): """Write a text block with heading detection. Lines detected as headings are rendered in dark blue bold; body text in default color.""" lines = text.split("\n") # Normalize: merge consecutive non-heading lines for better wrapping i = 0 while i < len(lines): line = lines[i].strip() if not line: i += 1 continue heading = (is_heading_en(line) if lang == "en" else is_heading_zh(line)) if heading: # Render heading pdf.ln(1.5) if lang == "en": pdf.set_font("TNR", "I", SIZE_HEADING_EN) else: pdf.set_font("YaHei", "B", SIZE_HEADING_ZH) pdf.set_text_color(*COLOR_HEADING) pdf.multi_cell(w=BODY_W, h=LINE_H + 1, text=line, align="L") pdf.ln(0.5) i += 1 else: # Collect consecutive body lines body_lines = [] while i < len(lines): s = lines[i].strip() if not s: i += 1 continue if is_heading_en(s) if lang == "en" else is_heading_zh(s): break body_lines.append(s) i += 1 if body_lines: body_text = " ".join(body_lines) if lang == "en": pdf.set_font("TNR", "", SIZE_EN_BODY) pdf.set_text_color(*COLOR_EN_BODY) else: pdf.set_font("YaHei", "", SIZE_ZH_BODY) pdf.set_text_color(*COLOR_ZH_BODY) pdf.multi_cell(w=BODY_W, h=LINE_H, text=body_text, align="L") pdf.ln(1) # Check for page overflow after each element if pdf.get_y() > PAGE_H - MARGIN_BOTTOM - 20: break def write_lang_separator(pdf: BilingualPDF): """Bold separator marking the language switch.""" pdf.ln(3) pdf.set_draw_color(*COLOR_HEADING) pdf.set_line_width(0.5) y = pdf.get_y() pdf.line(MARGIN_SIDE + 30, y, PAGE_W - MARGIN_SIDE - 30, y) pdf.ln(2) pdf.set_font("YaHei", "B", 8) pdf.set_text_color(*COLOR_HEADING) pdf.cell(BODY_W, 4, "▼ 中文翻译 ▼", align="C") pdf.ln(6) def render_page_pair(pdf: BilingualPDF, en_text: str, cn_text: str, page_num: int): """Render one source page: English block → separator → Chinese block.""" # Page badge pdf.set_font("TNR", "I", 7) pdf.set_text_color(*COLOR_SEP) pdf.cell(BODY_W, 3, f"p. {page_num + 1}", align="R") pdf.ln(5) if not en_text.strip() and not cn_text.strip(): return # English block if en_text.strip(): write_styled_block(pdf, en_text, "en") # Language separator if en_text.strip() and cn_text.strip(): write_lang_separator(pdf) # Chinese block if cn_text.strip(): write_styled_block(pdf, cn_text, "zh") # ---- Main ---- def main(): print("Loading JSON files...") with open(ENGLISH_JSON, "r", encoding="utf-8") as f: pages_en = json.load(f) with open(CHINESE_JSON, "r", encoding="utf-8") as f: pages_cn = json.load(f) total = len(pages_en) print(f" Source pages: {total}") print(f" Output: {OUTPUT_FILE}") pdf = BilingualPDF() pdf.set_left_margin(MARGIN_SIDE) pdf.set_right_margin(MARGIN_SIDE) pdf.set_top_margin(MARGIN_TOP) pdf.set_auto_page_break(auto=True, margin=MARGIN_BOTTOM) pdf.add_cover_page() for pg in range(total): key = str(pg) en_raw = pages_en.get(key, "") cn_raw = pages_cn.get(key, "") en_text = clean_english(en_raw) cn_text = clean_chinese(cn_raw) # IAST → ASCII for Chinese only (YaHei lacks extended Latin) cn_text = cn_text.translate(IAST_MAP) if not en_text.strip() and not cn_text.strip(): continue pdf.add_page() render_page_pair(pdf, en_text, cn_text, pg) if (pg + 1) % 20 == 0: print(f" Processed {pg + 1}/{total} pages...") OUTPUT_DIR.mkdir(parents=True, exist_ok=True) pdf.output(str(OUTPUT_FILE)) print(f"\nDone: {OUTPUT_FILE}") print(f" PDF pages: {pdf.pages_count}") print(f" Content pages: {total}") if __name__ == "__main__": main()