Update app.py

app.py

@@ -356,372 +356,7 @@ def count_stats(text: str) -> dict:
         'words':   words,
     }
 
-# ============================================================
-# PDF生成
-# ============================================================
-
-def generate_pdf(src_orig: str, dst_orig: str, settings: dict,
-                 src_html_body: str, dst_html_body: str, has_diff: bool) -> str:
-
-    FONT       = "NotoSansJP"
-    PAGE_W, PAGE_H = A4
-    MARGIN     = 15 * mm
-    COL_GAP    = 4  * mm
-    COL_W      = (PAGE_W - MARGIN * 2 - COL_GAP) / 2
-    FONT_SIZE  = 8.5
-    LINE_HEIGHT= 13
-    COL_L      = MARGIN
-    COL_R      = MARGIN + COL_W + COL_GAP
-    DIV_X      = MARGIN + COL_W + COL_GAP / 2
-
-    color_mode = settings.get('color_mode', 'color1')
-    if color_mode == 'color1':
-        HL_COLOR_RL = (1.0, 0.6, 0.6)
-        HL_FG_RL    = (0.0, 0.0, 0.0)
-        HL_COLOR_MU = (1.0, 0.6, 0.6)
-    elif color_mode == 'color2':
-        HL_COLOR_RL = (0.6, 0.8, 1.0)
-        HL_FG_RL    = (0.0, 0.0, 0.0)
-        HL_COLOR_MU = (0.6, 0.8, 1.0)
-    else:  # monochrome
-        HL_COLOR_RL = (0.0, 0.0, 0.0)
-        HL_FG_RL    = (1.0, 1.0, 1.0)
-        HL_COLOR_MU = (0.5, 0.5, 0.5)
-
-    # ----------------------------------------------------------
-    # HTMLから差分範囲を復元
-    # ----------------------------------------------------------
-    def extract_diff_ranges(text: str, html_body: str) -> list:
-        """
-        html_bodyのspanタグからハイライト箇所の文字インデックスリストを復元する。
-        戻り値: [(start, end), ...] ※textのインデックス
-        """
-        ranges = []
-        # ハイライトspanを検索
-        pattern = re.compile(
-            r'<span style="[^"]*background-color[^"]*">(.*?)</span>',
-            re.DOTALL
-        )
-        # タグを除去してプレーンテキストを再構築しながら位置を追跡
-        plain = ''
-        pos   = 0
-        for m in re.finditer(
-            r'<span(?:\s+style="([^"]*)")?>(.*?)</span>',
-            html_body, re.DOTALL
-        ):
-            style   = m.group(1) or ''
-            content = m.group(2)
-            # <br>を改行に戻す
-            content = content.replace('<br>', '\n')
-            # HTMLエスケープを戻す
-            content = html.unescape(content)
-            is_hl   = 'background-color' in style
-            start   = len(plain)
-            plain  += content
-            end     = len(plain)
-            if is_hl:
-                ranges.append((start, end))
-        return ranges
-
-    src_diff_ranges = extract_diff_ranges(src_orig, src_html_body)
-    dst_diff_ranges = extract_diff_ranges(dst_orig, dst_html_body)
-
-    # ----------------------------------------------------------
-    # テキストを折り返し済み描画単位に変換
-    # ----------------------------------------------------------
-    def to_sublines(text: str, col_w: float, font_size: float):
-        """
-        戻り値: [(char_offset, line_text), ...]
-        char_offset: テキスト先頭からの文字インデックス
-        """
-        result = []
-        lines  = text.split('\n')
-        offset = 0
-        for line in lines:
-            cur        = ''
-            cur_start  = offset
-            for ci, ch in enumerate(line):
-                test = cur + ch
-                if pdfmetrics.stringWidth(test, FONT, font_size) > col_w - 2 and cur:
-                    result.append((cur_start, cur))
-                    cur       = ch
-                    cur_start = offset + ci
-                else:
-                    cur = test
-            result.append((cur_start, cur))
-            offset += len(line) + 1  # +1 for '\n'
-        return result
-
-    # ----------------------------------------------------------
-    # ハイライト判定
-    # ----------------------------------------------------------
-    def is_highlighted(char_idx: int, diff_ranges: list) -> bool:
-        for s, e in diff_ranges:
-            if s <= char_idx < e:
-                return True
-        return False
-
-    # ----------------------------------------------------------
-    # 1行描画（背景矩形 → 文字）
-    # 戻り値: ハイライトした文字の座標リスト [(x0,y0,x1,y1), ...]
-    # ※ PyMuPDF注釈付与用
-    # ----------------------------------------------------------
-    def draw_subline(c, sub_text: str, sub_offset: int, diff_ranges: list,
-                     x_start: float, y_rl: float, font_size: float) -> list:
-        """
-        y_rl: reportlabのy座標（下原点）
-        戻り値: ハイライト文字のrect座標リスト (reportlab座標系)
-        """
-        if not sub_text:
-            return []
-
-        hl_rects = []
-
-        # パス1: 背景矩形
-        dx = x_start
-        for ci, ch in enumerate(sub_text):
-            cw = pdfmetrics.stringWidth(ch, FONT, font_size)
-            if is_highlighted(sub_offset + ci, diff_ranges):
-                c.saveState()
-                c.setFillColorRGB(*HL_COLOR_RL)
-                c.rect(dx, y_rl - 2, cw, LINE_HEIGHT - 1, fill=1, stroke=0)
-                c.restoreState()
-                hl_rects.append((dx, y_rl - 2, dx + cw, y_rl + LINE_HEIGHT - 1))
-            dx += cw
-
-        # パス2: 文字描画
-        dx      = x_start
-        seg     = ''
-        seg_hl  = None
-        seg_x   = x_start
-
-        for ci, ch in enumerate(sub_text):
-            hl = is_highlighted(sub_offset + ci, diff_ranges)
-            cw = pdfmetrics.stringWidth(ch, FONT, font_size)
-            if seg_hl is None:
-                seg_hl, seg, seg_x = hl, ch, dx
-            elif hl == seg_hl:
-                seg += ch
-            else:
-                c.saveState()
-                c.setFont(FONT, font_size)
-                c.setFillColorRGB(*(HL_FG_RL if seg_hl else (0.0, 0.0, 0.0)))
-                c.drawString(seg_x, y_rl, seg)
-                c.restoreState()
-                seg_x += pdfmetrics.stringWidth(seg, FONT, font_size)
-                seg_hl, seg = hl, ch
-            dx += cw
-
-        if seg:
-            c.saveState()
-            c.setFont(FONT, font_size)
-            c.setFillColorRGB(*(HL_FG_RL if seg_hl else (0.0, 0.0, 0.0)))
-            c.drawString(seg_x, y_rl, seg)
-            c.restoreState()
-
-        return hl_rects
-
-    # ----------------------------------------------------------
-    # reportlabでベースPDF生成
-    # ----------------------------------------------------------
-    def build_base_pdf() -> tuple:
-        """
-        戻り値: (pdf_bytes, all_hl_rects)
-        all_hl_rects: {page_no: {'src': [(x0,y0_rl,x1,y1_rl),...], 'dst': [...]}}
-        ※ y座標はreportlab座標系（下原点）
-        """
-        buffer   = io.BytesIO()
-        c        = rl_canvas.Canvas(buffer, pagesize=A4)
-        all_hl   = {}  # page_no -> {'src':[], 'dst':[]}
-        page_no  = 0
-
-        def hline(y, lw=0.5):
-            c.saveState()
-            c.setStrokeColorRGB(0.5, 0.5, 0.5)
-            c.setLineWidth(lw)
-            c.line(MARGIN, y, PAGE_W - MARGIN, y)
-            c.restoreState()
-
-        def new_page():
-            nonlocal page_no
-            c.showPage()
-            page_no += 1
-
-        def ensure_page_space(y, needed=LINE_HEIGHT):
-            if y < MARGIN + needed:
-                new_page()
-                return PAGE_H - MARGIN
-            return y
-
-        y = PAGE_H - MARGIN
-
-        # タイトル
-        c.saveState()
-        c.setFont(FONT, 18)
-        c.setFillColorRGB(0, 0, 0)
-        title = "テキスト比較ツール"
-        c.drawString((PAGE_W - pdfmetrics.stringWidth(title, FONT, 18)) / 2, y, title)
-        c.restoreState()
-        y -= 24
-
-        hline(y)
-        y -= 10
-
-        # 比較設定
-        c.saveState()
-        c.setFont(FONT, 10)
-        c.setFillColorRGB(0, 0, 0)
-        c.drawString(MARGIN, y, "■ 比較設定")
-        c.restoreState()
-        y -= 14
-
-        ignore_labels = {
-            'ignore_alpha':    '英字(全半角)無視',
-            'ignore_digit':    '数字(全半角)無視',
-            'ignore_kana':     'カナ(全半角)無視',
-            'ignore_yakumono': '約物(全半角)無視',
-            'ignore_ligature': 'リガチャ(合字)無視',
-            'ignore_kangxi':   '康煕部首(CJK部首補助)無視',
-            'ignore_brackets_with_content': '括弧付き文字無視',
-            'ignore_kanjinum': '漢数字(算用数字)無視',
-            'ignore_hyphen':   'ハイフン無視',
-            'ignore_kumimoji': '組み文字無視',
-            'ignore_space':    'スペース(空白)削除',
-            'ignore_newline':  '改行削除',
-        }
-        checked = [v for k, v in ignore_labels.items() if settings.get(k)]
-        c.saveState()
-        c.setFont(FONT, 8)
-        c.setFillColorRGB(0, 0, 0)
-        c.drawString(MARGIN, y, "有効な無視設定: " + ("、".join(checked) if checked else "なし"))
-        c.restoreState()
-        y -= 12
-
-        hline(y)
-        y -= 14
-
-        # 結果ヘッダー
-        c.saveState()
-        c.setFont(FONT, 14)
-        c.setFillColorRGB(0.85, 0, 0)
-        c.drawString(MARGIN, y, "相違が見つかりました" if has_diff else "相違はありません")
-        c.restoreState()
-        y -= 20
-
-        # カラムヘッダー
-        c.saveState()
-        c.setFont(FONT, 9)
-        c.setFillColorRGB(0, 0, 0)
-        c.drawString(COL_L, y, "【比較元】")
-        c.drawString(COL_R, y, "【比較先】")
-        c.restoreState()
-        y -= 14
-
-        # 本文描画
-        sublines_src = to_sublines(src_orig, COL_W, FONT_SIZE)
-        sublines_dst = to_sublines(dst_orig, COL_W, FONT_SIZE)
-        max_lines    = max(len(sublines_src), len(sublines_dst))
-        while len(sublines_src) < max_lines:
-            sublines_src.append((len(src_orig), ''))
-        while len(sublines_dst) < max_lines:
-            sublines_dst.append((len(dst_orig), ''))
-
-        seg_y_top = y
-        all_hl[page_no] = {'src': [], 'dst': []}
-
-        for (off_s, txt_s), (off_d, txt_d) in zip(sublines_src, sublines_dst):
-            y = ensure_page_space(y)
-            if y == PAGE_H - MARGIN:
-                # ページ送り発生 → 縦区切り線を前ページに描画済み想定
-                # 新ページの縦線開始位置を更新
-                seg_y_top = y
-                all_hl[page_no] = {'src': [], 'dst': []}
-
-            rects_s = draw_subline(c, txt_s, off_s, src_diff_ranges, COL_L, y, FONT_SIZE)
-            rects_d = draw_subline(c, txt_d, off_d, dst_diff_ranges, COL_R, y, FONT_SIZE)
-            all_hl[page_no]['src'].extend(rects_s)
-            all_hl[page_no]['dst'].extend(rects_d)
-            y -= LINE_HEIGHT
-
-        # 縦区切り線
-        c.saveState()
-        c.setStrokeColorRGB(0.5, 0.5, 0.5)
-        c.setLineWidth(0.5)
-        c.line(DIV_X, seg_y_top, DIV_X, y + LINE_HEIGHT)
-        c.restoreState()
-
-        y -= 6
-
-        # 統計
-        hline(y, lw=0.3)
-        y -= 10
-
-        src_stats = count_stats(src_orig)
-        dst_stats = count_stats(dst_orig)
-
-        def draw_stats(x, stats):
-            nonlocal y
-            for line in [
-                f"文字数: {stats['char']}",
-                f"空白数: {stats['space']}  空白込み文字数: {stats['space_incl']}",
-                f"改行数: {stats['newline']}  改行込み文字数: {stats['newline_incl']}",
-                f"単語数: {stats['words']}",
-            ]:
-                c.saveState()
-                c.setFont(FONT, 7.5)
-                c.setFillColorRGB(0.2, 0.2, 0.2)
-                c.drawString(x, y, line)
-                c.restoreState()
-                y -= 10
-
-        y_stats = y
-        draw_stats(COL_L, src_stats)
-        y = y_stats
-        draw_stats(COL_R, dst_stats)
-
-        c.save()
-        return buffer.getvalue(), all_hl
-
-    # ----------------------------------------------------------
-    # PyMuPDFでHighlight Annotation付与
-    # ----------------------------------------------------------
-    def add_highlight_annots(pdf_bytes: bytes, all_hl: dict) -> bytes:
-        doc = fitz.open(stream=pdf_bytes, filetype="pdf")
-
-        for page_no, hl_data in all_hl.items():
-            if page_no >= len(doc):
-                continue
-            page   = doc[page_no]
-            page_h = page.rect.height  # PyMuPDF座標系の高さ（上原点）
-
-            rects_all = hl_data.get('src', []) + hl_data.get('dst', [])
-            for (x0, y0_rl, x1, y1_rl) in rects_all:
-                # reportlab座標(下原点) → PyMuPDF座標(上原点)変換
-                # reportlab: y0_rl=下端, y1_rl=上端
-                # PyMuPDF:   y_top = page_h - y1_rl, y_bottom = page_h - y0_rl
-                mu_y_top    = page_h - y1_rl
-                mu_y_bottom = page_h - y0_rl
-                rect  = fitz.Rect(x0, mu_y_top, x1, mu_y_bottom)
-                annot = page.add_highlight_annot(rect)
-                annot.set_colors(stroke=HL_COLOR_MU)
-                annot.update()
-
-        result = io.BytesIO()
-        doc.save(result)
-        doc.close()
-        return result.getvalue()
-
-    # ----------------------------------------------------------
-    # 実行
-    # ----------------------------------------------------------
-    base_pdf_bytes, all_hl_rects = build_base_pdf()
-    final_pdf_bytes = add_highlight_annots(base_pdf_bytes, all_hl_rects)
-
-    tmp = tempfile.NamedTemporaryFile(delete=False, suffix='.pdf')
-    tmp.write(final_pdf_bytes)
-    tmp.close()
-    return tmp.name
-
+undefined
 # ============================================================
 # Gradio メイン処理
 # ============================================================