data-report/generate-data-report-ppt/scripts/quality_inspector.py

"""
PPT quality inspector and auto-fix engine.
Inspects generated PPT for layout, visual, content, and data issues,
then auto-fixes them iteratively until quality threshold is met.
"""
import re
from pptx import Presentation
from pptx.util import Emu, Pt
from pptx.dml.color import RGBColor
from collections import Counter

from quality_rules import (
    QUALITY_RULES, SEVERITY_WEIGHTS, CATEGORY_WEIGHTS,
    FILL_RATIO_THRESHOLDS, FONT_SIZE_MIN, FONT_SIZE_MAX,
    TEXT_MIN_LENGTH, INSIGHT_MIN_COUNT, PAGE_MIN_TEXT_LENGTH,
    SAFE_MARGIN, CONTENT_LEFT, CONTENT_TOP_BASE,
    FOOTER_TOP, SLIDE_WIDTH, SLIDE_HEIGHT, DEFAULT_FONT,
    get_quality_label, calculate_score,
)
from page_layouts import calculate_fill_ratio, ensure_safe_position


FORECAST_PAGE_TYPES = {
    'forecast',
    'prediction',
    'plan',
    'monthly_forecast',
    'monthly_plan',
    'next_month_plan',
    'custom_forecast',
    'custom_prediction',
}


class QualityIssue:
    def __init__(self, severity, category, page_index, description,
                 rule_id='', auto_fixable=True, fix_data=None):
        self.severity = severity
        self.category = category
        self.page_index = page_index
        self.description = description
        self.rule_id = rule_id
        self.auto_fixable = auto_fixable
        self.fix_data = fix_data or {}

    def __repr__(self):
        return f"[{self.severity}] Page {self.page_index+1}: {self.description}"


class QualityInspector:
    def __init__(self, theme_colors: dict = None, layout_context=None):
        self.theme_colors = theme_colors or {}
        self.layout_context = layout_context
        self.fix_count = 0
        self.fix_log = []

    def inspect(self, prs: Presentation, config=None) -> list[QualityIssue]:
        issues = []
        issues += self._check_confirmation_alignment(prs, config)
        for page_idx, slide in enumerate(prs.slides):
            page_type = self._get_page_type(page_idx, config, len(prs.slides))
            issues += self._check_layout(slide, page_idx)
            issues += self._check_visual(slide, page_idx)
            issues += self._check_content(slide, page_idx, config, prs, page_type)
            issues += self._check_data(slide, page_idx, prs)
        return issues

    def _get_page_type(self, page_idx: int, config, total_slides: int) -> str:
        if config and hasattr(config, 'pages') and page_idx < len(config.pages):
            return config.pages[page_idx].page_type
        if page_idx == 0:
            return 'cover'
        if page_idx == total_slides - 1:
            return 'end'
        if page_idx == 1:
            return 'toc'
        return 'content'

    def _check_confirmation_alignment(self, prs, config) -> list[QualityIssue]:
        issues = []
        if not config:
            return issues

        selected_pages = [p for p in getattr(config, 'pages', []) if getattr(p, 'selected', True)]
        if getattr(config, 'require_six_confirmations', False):
            confirmation = getattr(config, 'user_confirmation', None)
            if confirmation and hasattr(confirmation, 'is_complete') and not confirmation.is_complete():
                issues.append(QualityIssue(
                    'critical', 'data', -1,
                    '六项确认未完成，PPT 不应进入输出阶段',
                    'D006', False,
                    {'type': 'confirmation_incomplete'}
                ))

        if config and getattr(config, 'page_count_range', None) and selected_pages:
            low, high = config.page_count_range
            if len(selected_pages) < low or len(selected_pages) > high:
                issues.append(QualityIssue(
                    'major', 'data', -1,
                    f'页面数量 {len(selected_pages)} 不在确认范围 {low}-{high} 内',
                    'D006', False,
                    {'type': 'page_count_range'}
                ))

        if config and getattr(config, 'metrics', None) and len(selected_pages) > 0:
            selected_metrics = [m for m in config.metrics if getattr(m, 'selected', True)]
            if not selected_metrics:
                issues.append(QualityIssue(
                    'critical', 'data', -1,
                    '未找到已确认的核心指标，无法验证输出一致性',
                    'D006', False,
                    {'type': 'missing_metrics'}
                ))
        return issues

    def auto_fix(self, prs: Presentation, issues: list[QualityIssue]):
        fixable = [i for i in issues if i.auto_fixable]
        self.fix_count = 0
        self.fix_log = []

        for issue in fixable:
            try:
                if issue.page_index < 0:
                    continue
                slide = prs.slides[issue.page_index]
                self._apply_fix(slide, issue, prs)
                self.fix_count += 1
            except Exception as e:
                self.fix_log.append(f"Fix failed for {issue.rule_id}: {e}")

        return self.fix_count

    def _apply_fix(self, slide, issue, prs):
        category = issue.category
        if category == 'layout':
            self._fix_layout(slide, issue)
        elif category == 'visual':
            self._fix_visual(slide, issue)
        elif category == 'content':
            self._fix_content(slide, issue, prs)
        elif category == 'data':
            self._fix_data(slide, issue, prs)

    def generate_report(self, issues: list[QualityIssue], iteration: int = 1,
                        total_pages: int = 0) -> str:
        lines = []
        lines.append('═' * 50)
        lines.append(f'  PPT 质量自检报告 (第 {iteration} 轮)')
        lines.append('═' * 50)

        if not issues:
            lines.append('[PASS] 全部通过！未发现任何质量问题。')
            return '\n'.join(lines)

        by_page = {}
        for iss in issues:
            p = iss.page_index
            if p not in by_page:
                by_page[p] = []
            by_page[p].append(iss)

        for p_idx in sorted(by_page.keys()):
            page_issues = by_page[p_idx]
            sev_order = {'critical': 0, 'major': 1, 'minor': 2}
            page_issues.sort(key=lambda x: sev_order.get(x.severity, 3))

            has_critical = any(i.severity == 'critical' for i in page_issues)
            has_major = any(i.severity == 'major' for i in page_issues)

            if has_critical:
                icon = '[CRIT]'
            elif has_major:
                icon = '[MAJ]'
            else:
                icon = '[OK]'

            lines.append(f'{icon} 第{p_idx+1}页: {len(page_issues)} 个问题')

            for iss in page_issues:
                sev_icon = {'critical': '[!!]', 'major': '[!]', 'minor': '[-]'}.get(iss.severity, '')
                status = ' [FIXED]' if iss.auto_fixable and iss.fix_data.get('fixed') else ''
                lines.append(f'  ├─ {sev_icon} {iss.description}{status}')

        lines.append('─' * 50)

        by_sev = Counter(i.severity for i in issues)
        by_cat = Counter(i.category for i in issues)
        fixed = sum(1 for i in issues if i.auto_fixable and i.fix_data.get('fixed'))
        score = calculate_score(dict(by_sev), dict(by_cat), max(total_pages, 1))
        label = get_quality_label(score)

        lines.append(f'总结: {len(issues)} 个问题 | '
                     f'{by_sev.get("critical", 0)} 严重 + '
                     f'{by_sev.get("major", 0)} 主要 + '
                     f'{by_sev.get("minor", 0)} 次要')
        lines.append(f'自动修复: {fixed}/{sum(1 for i in issues if i.auto_fixable)} 个')
        lines.append(f'最终质量评分: {score}/100 [{label}]')
        lines.append('═' * 50)

        return '\n'.join(lines)

    def quality_assured_build(self, build_fn, data, config, output_path,
                              max_iterations=None, _attempt=0) -> tuple:
        max_iterations = max_iterations or config.max_fix_iterations
        total_pages = 0
        needs_rebuild = False
        rebuilt_once = False
        prs = None

        for iteration in range(1, max_iterations + 1):
            if iteration == 1 or needs_rebuild:
                if needs_rebuild:
                    if rebuilt_once and iteration > 2:
                        print(f'[INFO] 已尝试重建，不再继续重建以避免无限循环')
                        needs_rebuild = False
                    else:
                        print(f'[REBUILD] 检测到需要重建的页面，触发重新生成...')
                        rebuilt_once = True
                        needs_rebuild = False
                prs = build_fn(data, config)
                total_pages = len(prs.slides)

            issues = self.inspect(prs, config)

            if not issues:
                print(f'[PASS] 第 {iteration} 次迭代：无问题，质量达标')
                break

            by_sev = Counter(i.severity for i in issues)
            print(f'[INSPECT] 第 {iteration} 次自检：{by_sev.get("critical",0)} 严重 + '
                  f'{by_sev.get("major",0)} 主要 + {by_sev.get("minor",0)} 次要')

            fixable = [i for i in issues if i.auto_fixable]
            self.auto_fix(prs, fixable)
            print(f'[FIX] 自动修复了 {self.fix_count} 个问题')

            for issue in fixable:
                if issue.fix_data.get('needs_rebuild'):
                    needs_rebuild = True
                    print(f'[WARN] 检测到内容严重不足，将在下一轮迭代中重建')
                    break

            unfixable = [i for i in issues if not i.auto_fixable]
            if unfixable:
                print(f'[WARN] {len(unfixable)} 个问题需人工确认')

            remaining = self.inspect(prs, config)
            if not remaining:
                print(f'[PASS] 第 {iteration} 次修复后：所有问题已解决')
                break

            has_critical = any(i.severity == 'critical' for i in remaining)
            has_major = any(i.severity == 'major' for i in remaining)
            if not has_critical and not has_major:
                print(f'[PASS] 第 {iteration} 次修复后：仅剩次要问题，质量达标')
                break

            if needs_rebuild and iteration < max_iterations:
                continue

        final_issues = self.inspect(prs, config)
        by_sev = Counter(i.severity for i in final_issues)
        by_cat = Counter(i.category for i in final_issues)
        score = calculate_score(dict(by_sev), dict(by_cat), max(total_pages, 1))
        label = get_quality_label(score)

        report = self.generate_report(final_issues, iteration, total_pages)
        print(report)

        if score >= config.quality_threshold:
            prs.save(output_path)
            print(f'[PASS] 高质量 PPT 已输出: {output_path}')
        else:
            has_critical_final = any(i.severity == 'critical' for i in final_issues)
            has_layout_critical = any(
                i.severity == 'critical' and i.category == 'layout'
                for i in final_issues
            )
            if has_layout_critical:
                raise RuntimeError(
                    f'PPT 存在严重布局问题（评分 {score}），无法自动修复。'
                    f'请检查页面配置和数据。'
                )
            prs.save(output_path)
            if has_critical_final:
                print(f'[WARN] 质量评分 {score}（低于阈值 {config.quality_threshold}），'
                      f'存在 {by_sev.get("critical", 0)} 个严重内容问题，建议补充分析数据后重新生成')
            else:
                print(f'[WARN] 质量评分 {score}（低于阈值 {config.quality_threshold}），已输出但建议复核')

        return prs, final_issues

    def _check_layout(self, slide, page_idx) -> list[QualityIssue]:
        issues = []
        sw = int(slide.slide_width) if hasattr(slide, 'slide_width') else SLIDE_WIDTH
        sh = int(slide.slide_height) if hasattr(slide, 'slide_height') else SLIDE_HEIGHT

        for shape in slide.shapes:
            l, t = int(shape.left), int(shape.top)
            w, h = int(shape.width), int(shape.height)

            if l < -100:
                issues.append(QualityIssue('critical', 'layout', page_idx,
                    f'形状"{_shape_name(shape)}"飞出页面左边界 (left={l})',
                    'L001', True, {'shape': shape, 'type': 'left'}))
            if l + w > sw + 500:
                issues.append(QualityIssue('critical', 'layout', page_idx,
                    f'形状"{_shape_name(shape)}"飞出页面右边界 (right={l+w}, max={sw})',
                    'L002', True, {'shape': shape, 'type': 'right'}))
            if t < -100:
                issues.append(QualityIssue('critical', 'layout', page_idx,
                    f'形状"{_shape_name(shape)}"飞出页面顶部 (top={t})',
                    'L003', True, {'shape': shape, 'type': 'top'}))
            if t + h > sh + 500:
                issues.append(QualityIssue('critical', 'layout', page_idx,
                    f'形状"{_shape_name(shape)}"飞出页面底部 (bottom={t+h}, max={sh})',
                    'L004', True, {'shape': shape, 'type': 'bottom'}))

            if l < SAFE_MARGIN and l >= 0:
                    if l == 0 and w >= sw * 0.8:
                        continue
                    if int(shape.top) < 0 or int(shape.top) + int(shape.height) < Emu(100000):
                        continue
                    if int(shape.top) > sh - Emu(500000):
                        continue
                    issues.append(QualityIssue('minor', 'layout', page_idx,
                        f'形状"{_shape_name(shape)}"过于靠近左边缘',
                        'L007', True, {'shape': shape, 'type': 'edge_left'}))

        placeholder_pattern = re.compile(r'\{[^}]+\}')
        for shape in slide.shapes:
            if shape.has_text_frame:
                text = shape.text_frame.text
                if placeholder_pattern.search(text):
                    issues.append(QualityIssue('critical', 'layout', page_idx,
                        f'发现未替换占位符: "{text[:50]}"',
                        'L006', True, {'shape': shape, 'type': 'placeholder'}))

        empty_artifacts = self._find_empty_template_artifacts(slide)
        for shape in empty_artifacts:
            issues.append(QualityIssue(
                'major', 'layout', page_idx,
                f'发现空模板组件残留: "{_shape_name(shape)}"',
                'L008', True, {'shape': shape, 'type': 'empty_template_artifact'}
            ))

        shapes_list = list(slide.shapes)
        for i, a in enumerate(shapes_list):
            for b in shapes_list[i+1:]:
                if self._shapes_overlap(a, b):
                    a_name = _shape_name(a)
                    b_name = _shape_name(b)
                    if self._is_intentional_overlap(a, b):
                        continue
                    issues.append(QualityIssue('major', 'layout', page_idx,
                        f'形状"{a_name}"与"{b_name}"存在重叠',
                        'L005', True, {'shape_a': a, 'shape_b': b, 'type': 'overlap'}))

        return issues

    def _check_visual(self, slide, page_idx) -> list[QualityIssue]:
        issues = []
        fonts_seen = {}

        for shape in slide.shapes:
            if not shape.has_text_frame:
                continue
            for para in shape.text_frame.paragraphs:
                for run in para.runs:
                    if run.font.size:
                        size_pt = run.font.size / 12700.0
                        if size_pt < 6:
                            issues.append(QualityIssue('major', 'visual', page_idx,
                                f'字号过小 ({size_pt:.1f}pt): "{run.text[:20]}"',
                                'V002', True, {'run': run, 'type': 'font_small'}))
                        elif size_pt > 65:
                            issues.append(QualityIssue('major', 'visual', page_idx,
                                f'字号过大 ({size_pt:.1f}pt): "{run.text[:20]}"',
                                'V003', True, {'run': run, 'type': 'font_large'}))

                    if run.font.name:
                        fonts_seen[run.font.name] = fonts_seen.get(run.font.name, 0) + 1

        if len(fonts_seen) > 3:
            issues.append(QualityIssue('minor', 'visual', page_idx,
                f'字体使用超过3种: {list(fonts_seen.keys())}',
                'V001', True, {'type': 'font_mixed', 'fonts': fonts_seen}))

        return issues

    def _check_content(self, slide, page_idx, config, prs, page_type='content') -> list[QualityIssue]:
        # Resolve dynamic content top from layout context if available
        content_top_emu = None
        if self.layout_context:
            content_top_emu = self.layout_context.content_top
        issues = []

        if page_type in ('cover', 'end'):
            issues += self._check_text_overflow(slide, page_idx)
            if page_type == 'cover':
                issues += self._check_cover_quality(slide, page_idx)
            return issues

        issues += self._check_dynamic_page_fit(page_idx, page_type, config)
        issues += self._check_core_metric_presence(slide, page_idx, page_type, config)

        if page_type == 'toc':
            content_shapes = [s for s in slide.shapes
                             if s.has_text_frame and _is_in_content_area(s)]
            all_content_text = ''
            for shape in content_shapes:
                text = shape.text_frame.text.strip()
                if text:
                    all_content_text += text + '\n'
            if len(all_content_text.strip()) < 30:
                issues.append(QualityIssue('minor', 'content', page_idx,
                    '目录页内容过少',
                    'C008', False, {'type': 'empty_page'}))
            return issues

        fill_ratio = calculate_fill_ratio(slide, content_top_emu=content_top_emu)

        if page_type in ('kpi_overview', 'trend', 'distribution', 'ranking', 'summary') or page_type in FORECAST_PAGE_TYPES:
            if fill_ratio < FILL_RATIO_THRESHOLDS['sparse']:
                issues.append(QualityIssue('critical', 'content', page_idx,
                    f'页面内容严重不足，填充率仅 {fill_ratio:.1%}，必须补充图表和分析文本',
                    'C001', True, {'type': 'sparse', 'fill_ratio': fill_ratio}))
            elif fill_ratio < FILL_RATIO_THRESHOLDS['low']:
                issues.append(QualityIssue('major', 'content', page_idx,
                    f'页面留白偏多，填充率 {fill_ratio:.1%}，需补充分析内容',
                    'C001', True, {'type': 'sparse', 'fill_ratio': fill_ratio}))
        elif fill_ratio < FILL_RATIO_THRESHOLDS['sparse'] / 2:
            issues.append(QualityIssue('minor', 'content', page_idx,
                f'页面填充率过低 {fill_ratio:.1%}',
                'C001', False))

        content_shapes = [s for s in slide.shapes
                         if s.has_text_frame and _is_in_content_area(s)]
        all_content_text = ''
        insight_blocks = 0
        for shape in content_shapes:
            tf = shape.text_frame
            full_text = tf.text.strip()
            if not full_text:
                continue
            all_content_text += full_text + '\n'
            for para in tf.paragraphs:
                para_text = para.text.strip()
                if para_text and len(para_text) >= TEXT_MIN_LENGTH:
                    insight_blocks += 1

        total_content_chars = len(all_content_text.strip())

        text_lengths = [len(p.text.strip()) for s in content_shapes
                        for p in s.text_frame.paragraphs if p.text.strip()]

        if total_content_chars < PAGE_MIN_TEXT_LENGTH:
            issues.append(QualityIssue('critical', 'content', page_idx,
                f'页面内容为空！所有文本框总字数仅 {total_content_chars} 字（要求≥{PAGE_MIN_TEXT_LENGTH}字）',
                'C008', True, {'type': 'empty_page', 'char_count': total_content_chars}))
        elif total_content_chars < 200:
            issues.append(QualityIssue('major', 'content', page_idx,
                f'页面内容过少，总字数仅 {total_content_chars} 字，分析深度严重不足',
                'C008', True, {'type': 'empty_page', 'char_count': total_content_chars}))

        if text_lengths and max(text_lengths) < TEXT_MIN_LENGTH:
            issues.append(QualityIssue('critical', 'content', page_idx,
                f'分析文本过短（最长为 {max(text_lengths)} 字），需撰写≥{TEXT_MIN_LENGTH}字的深度分析',
                'C005', True, {'type': 'short_text', 'max_length': max(text_lengths)}))

        if insight_blocks < INSIGHT_MIN_COUNT:
            issues.append(QualityIssue('critical', 'content', page_idx,
                f'分析段数不足，仅 {insight_blocks} 段（要求≥{INSIGHT_MIN_COUNT}段）',
                'C007', True, {'type': 'insight_count', 'count': insight_blocks}))

        has_title = False
        for shape in slide.shapes:
            if shape.has_text_frame:
                text = shape.text_frame.text
                try:
                    sy = int(shape.top)
                except Exception:
                    sy = 99999999
                if sy < CONTENT_TOP_BASE + Emu(100000) and sy > Emu(500000):
                    if len(text.strip()) > 0 and not text.startswith('{'):
                        has_title = True
                        break
                if any(kw in text for kw in ['概览', '趋势', '分布', '分析', '总结',
                                              '排行', '报告', '建议', '告警', '要点']):
                    if sy < CONTENT_TOP_BASE + Emu(400000):
                        has_title = True
                        break

        if not has_title and page_idx > 0 and page_idx < len(prs.slides) - 1:
            issues.append(QualityIssue('critical', 'content', page_idx,
                '页面缺少标题', 'C006', True, {'type': 'missing_title'}))

        issues += self._check_text_overflow(slide, page_idx)

        has_chart = False
        for shape in slide.shapes:
            if shape.has_chart:
                has_chart = True
                break

        if has_chart and insight_blocks == 0 and page_idx >= 2:
            issues.append(QualityIssue('critical', 'content', page_idx,
                '页面有图表但完全缺少分析文本，图表数据需要被解读和说明',
                'C009', True, {'type': 'chart_no_text'}))

        return issues

    # ---- Cover page quality checks (V006, C012) ----
    # These catch the most common python-pptx template pitfalls before the
    # user sees the output: white text on white background, and unfilled
    # template default text in placeholders.

    _COVER_TEMPLATE_DEFAULT_PATTERNS = [
        '单击此处编辑母版标题样式', '单击此处添加标题',
        '单击此处编辑母版文本样式', '单击此处添加文本',
        '单击此处添加副标题',
    ]

    # Light/pale colors that are invisible on white backgrounds
    _LIGHT_COLOR_THRESHOLD = 0xCC  # RGB channels above this = "very light"

    def _check_cover_quality(self, slide, page_idx) -> list[QualityIssue]:
        """Check cover page for common template rendering issues.

        V006: placeholder text is white/light but positioned on light
              background (e.g. below a colored banner). Auto-fixable.
        C012: placeholder still contains template default text like
              "单击此处编辑母版标题样式". Not auto-fixable — needs rebuild.
        """
        issues = []
        slide_h = int(slide.slide_height) if hasattr(slide, 'slide_height') else SLIDE_HEIGHT

        # Detect the approximate end of the colored banner area.
        # Heuristic: find the tallest filled rectangle that starts at y=0.
        banner_bottom = 0
        for shape in slide.shapes:
            try:
                sy = int(shape.top)
                sh = int(shape.height)
                if sy < Emu(50000):  # starts near top
                    if sh < slide_h * 0.7:  # not full-slide background
                        banner_bottom = max(banner_bottom, sy + sh)
            except Exception:
                pass

        for shape in slide.shapes:
            if not shape.is_placeholder or not shape.has_text_frame:
                continue

            tf = shape.text_frame
            text = tf.text.strip()

            # --- C012: template default text ---
            if text and any(p in text for p in self._COVER_TEMPLATE_DEFAULT_PATTERNS):
                issues.append(QualityIssue(
                    'critical', 'content', page_idx,
                    f'封面占位符仍为模板默认文字 "{text[:30]}"，idx={shape.placeholder_format.idx}',
                    'C012', False,
                    {'type': 'cover_template_text', 'shape': shape}
                ))
                continue

            # --- V006: white/light text on light background ---
            if text and banner_bottom > 0:
                try:
                    sy = int(shape.top)
                except Exception:
                    continue

                # Only flag text BELOW the banner (on white area)
                if sy < banner_bottom:
                    continue

                # Check if text color is very light / near-white
                for para in tf.paragraphs:
                    for run in para.runs:
                        if run.font.color and run.font.color.rgb:
                            rgb = run.font.color.rgb
                            if (int(str(rgb)[:2], 16) >= self._LIGHT_COLOR_THRESHOLD and
                                int(str(rgb)[2:4], 16) >= self._LIGHT_COLOR_THRESHOLD and
                                int(str(rgb)[4:6], 16) >= self._LIGHT_COLOR_THRESHOLD):
                                issues.append(QualityIssue(
                                    'critical', 'visual', page_idx,
                                    f'封面文字 "{text[:20]}" 颜色过浅 (#{rgb}) '
                                    f'位于白色背景区域(>y={banner_bottom})将不可见',
                                    'V006', True,
                                    {'type': 'cover_text_invisible', 'shape': shape,
                                     'banner_bottom': banner_bottom}
                                ))
                                break

        return issues

    def _check_text_overflow(self, slide, page_idx) -> list[QualityIssue]:
        issues = []
        for shape in slide.shapes:
            if shape.has_text_frame and self._is_text_overflowing(shape):
                issues.append(QualityIssue(
                    'major', 'content', page_idx,
                    f'文本可能超出文本框边界: "{shape.text_frame.text[:30]}"',
                    'C004', True, {'shape': shape, 'type': 'text_overflow'}
                ))
        return issues

    def _check_dynamic_page_fit(self, page_idx, page_type, config) -> list[QualityIssue]:
        issues = []
        profile = getattr(config, 'data_profiling', None) or {}
        if not profile:
            return issues

        time_cols = profile.get('time_columns', [])
        cat_cols = profile.get('category_columns', [])
        num_cols = profile.get('numeric_columns', [])

        if page_type == 'trend' and (not time_cols or not num_cols):
            issues.append(QualityIssue(
                'critical', 'content', page_idx,
                '趋势页缺少可用时间列或数值列，需要重建或降级为摘要页',
                'C010', True, {'type': 'dynamic_page_not_supported', 'page_type': page_type}
            ))
        elif page_type in ('distribution', 'ranking') and (not cat_cols or not num_cols):
            issues.append(QualityIssue(
                'critical', 'content', page_idx,
                f'{page_type} 页缺少分类维度或数值列，需要重建或降级为摘要页',
                'C010', True, {'type': 'dynamic_page_not_supported', 'page_type': page_type}
            ))
        elif page_type == 'kpi_overview':
            selected_metrics = [m for m in getattr(config, 'metrics', []) if getattr(m, 'selected', True)]
            if len(selected_metrics) > 6:
                issues.append(QualityIssue(
                    'minor', 'content', page_idx,
                    f'核心指标数量 {len(selected_metrics)} 超过 6 个，KPI页应切换为紧凑布局或拆分展示',
                    'C011', True, {'type': 'kpi_layout_over_capacity', 'count': len(selected_metrics)}
                ))
            elif len(selected_metrics) >= 4:
                issues.append(QualityIssue(
                    'minor', 'content', page_idx,
                    f'核心指标数量 {len(selected_metrics)} 较多，建议使用紧凑布局以保留洞察区',
                    'C011', True, {'type': 'kpi_layout_compact_needed', 'count': len(selected_metrics)}
                ))
        return issues

    def _check_core_metric_presence(self, slide, page_idx, page_type, config) -> list[QualityIssue]:
        issues = []
        if page_type != 'kpi_overview' or not config:
            return issues
        selected_metrics = [m for m in getattr(config, 'metrics', []) if getattr(m, 'selected', True)]
        if not selected_metrics:
            return issues
        slide_text = '\n'.join(
            shape.text_frame.text for shape in slide.shapes
            if shape.has_text_frame and shape.text_frame.text
        )
        missing = [m.label for m in selected_metrics[:6] if m.label and m.label not in slide_text]
        if missing:
            issues.append(QualityIssue(
                'critical', 'data', page_idx,
                'KPI概览页缺少已确认核心指标：' + '、'.join(missing),
                'D006', True, {'type': 'core_metric_missing', 'missing': missing}
            ))
        return issues

    def _check_data(self, slide, page_idx, prs) -> list[QualityIssue]:
        issues = []

        if page_idx == 0:
            return issues

        for shape in slide.shapes:
            if shape.has_text_frame:
                text = shape.text_frame.text

                page_pattern = re.search(r'(\d+)\s*/\s*(\d+)', text)
                if page_pattern:
                    current = int(page_pattern.group(1))
                    total = int(page_pattern.group(2))
                    if total == 0:
                        issues.append(QualityIssue('major', 'data', page_idx,
                            f'页码格式异常: {text.strip()}',
                            'D002', True, {'type': 'page_num'}))

        return issues

    def _fix_layout(self, slide, issue):
        fd = issue.fix_data
        if fd.get('type') in ('left', 'right', 'top', 'bottom'):
            shape = fd.get('shape')
            if shape:
                ensure_safe_position(shape, SLIDE_WIDTH, SLIDE_HEIGHT)
                fd['fixed'] = True

        elif fd.get('type') == 'overlap':
            a, b = fd.get('shape_a'), fd.get('shape_b')
            if a and b:
                try:
                    if int(b.left) < int(a.left) + int(a.width) + Emu(50000):
                        b.left = int(a.left) + int(a.width) + Emu(152400)
                        ensure_safe_position(b, SLIDE_WIDTH, SLIDE_HEIGHT)
                except Exception:
                    pass
                fd['fixed'] = True

        elif fd.get('type') == 'placeholder':
            shape = fd.get('shape')
            if shape and shape.has_text_frame:
                text = shape.text_frame.text or ''
                # For KPI placeholders, remove the entire shape and nearby card backgrounds
                kpi_pattern = re.compile(r'\{kpi\d+_(label|value)\}')
                if kpi_pattern.search(text):
                    # Remove this text shape
                    self._remove_shape(shape)
                    # Also remove nearby rounded rectangle backgrounds
                    try:
                        sx = int(shape.left)
                        sy = int(shape.top)
                        sw = int(shape.width)
                        sh = int(shape.height)
                        pad = 300000
                        for other in list(slide.shapes):
                            try:
                                ox = int(other.left)
                                oy = int(other.top)
                                ow = int(other.width)
                                oh = int(other.height)
                                in_region = (
                                    ox >= sx - pad and ox + ow <= sx + sw + pad and
                                    oy >= sy - pad and oy + oh <= sy + sh + pad
                                )
                                if in_region and other != shape:
                                    # Check if it's a background shape (no text or empty text)
                                    if not other.has_text_frame or not (other.text_frame.text or '').strip():
                                        self._remove_shape(other)
                            except Exception:
                                pass
                    except Exception:
                        pass
                else:
                    # For other placeholders, just clear the text
                    for para in shape.text_frame.paragraphs:
                        para.text = re.sub(r'\{[^}]+\}', '', para.text)
                fd['fixed'] = True

        elif fd.get('type') == 'edge_left':
            shape = fd.get('shape')
            if shape:
                try:
                    w = int(shape.width)
                    if w < SLIDE_WIDTH * 0.5:
                        shape.left = SAFE_MARGIN
                except Exception:
                    pass
                fd['fixed'] = True

        elif fd.get('type') == 'empty_template_artifact':
            shape = fd.get('shape')
            if shape:
                self._remove_shape(shape)
                fd['fixed'] = True

    def _fix_visual(self, slide, issue):
        fd = issue.fix_data
        if fd.get('type') == 'font_small':
            run = fd.get('run')
            if run:
                run.font.size = FONT_SIZE_MIN
                fd['fixed'] = True
        elif fd.get('type') == 'font_large':
            run = fd.get('run')
            if run:
                run.font.size = FONT_SIZE_MAX
                fd['fixed'] = True
        elif fd.get('type') == 'font_mixed':
            for shape in slide.shapes:
                if shape.has_text_frame:
                    for para in shape.text_frame.paragraphs:
                        for run in para.runs:
                            run.font.name = DEFAULT_FONT
            fd['fixed'] = True

        elif fd.get('type') == 'cover_text_invisible':
            shape = fd.get('shape')
            if shape and shape.has_text_frame:
                dark_color = self.theme_colors.get('primary', RGBColor(0x1E, 0x3A, 0x5F))
                for para in shape.text_frame.paragraphs:
                    for run in para.runs:
                        run.font.color.rgb = dark_color
                fd['fixed'] = True

    def _fix_content(self, slide, issue, prs):
        fd = issue.fix_data
        if fd.get('type') == 'sparse':
            fill_ratio = fd.get('fill_ratio', 0)
            if fill_ratio < FILL_RATIO_THRESHOLDS['low']:
                try:
                    box = slide.shapes.add_textbox(
                        CONTENT_LEFT, Emu(int(FOOTER_TOP) - Emu(1600000)),
                        Emu(SLIDE_WIDTH - 2 * CONTENT_LEFT - Emu(200000)), Emu(1500000))
                    tf = box.text_frame
                    tf.word_wrap = True
                    p = tf.paragraphs[0]
                    p.text = (
                        '[WARNING] 此页面内容不足，需补充深度分析内容。'
                        '分析应包含：具体数据引用（含数值和单位）、'
                        '与同类/历史/目标的对比分析、'
                        '数据背后原因的至少2条解读、'
                        '以及可执行的业务行动建议。'
                        '请勿使用"要加强"、"进一步优化"等模糊措辞。'
                    )
                    p.font.size = Pt(12)
                    p.font.color.rgb = RGBColor(0xCC, 0x33, 0x00)
                    p.font.name = DEFAULT_FONT
                    p.font.bold = True
                    fd['fixed'] = True
                    fd['needs_rebuild'] = True
                except Exception:
                    pass

        elif fd.get('type') == 'empty_page':
            fd['needs_rebuild'] = True
            fd['fixed'] = True

        elif fd.get('type') == 'chart_no_text':
            fd['needs_rebuild'] = True
            fd['fixed'] = True

        elif fd.get('type') == 'insight_count':
            fd['needs_rebuild'] = True
            fd['fixed'] = True

        elif fd.get('type') == 'short_text':
            fd['needs_rebuild'] = True
            fd['fixed'] = True

        elif fd.get('type') in ('dynamic_page_not_supported', 'kpi_layout_over_capacity'):
            fd['fixed'] = True

        elif fd.get('type') == 'kpi_layout_compact_needed':
            fd['fixed'] = True

        elif fd.get('type') == 'core_metric_missing':
            fd['needs_rebuild'] = True
            fd['fixed'] = True

        elif fd.get('type') == 'missing_title':
            try:
                box = slide.shapes.add_textbox(
                    CONTENT_LEFT, Emu(914400),
                    Emu(SLIDE_WIDTH - 2 * CONTENT_LEFT - Emu(200000)), Emu(508000))
                p = box.text_frame.paragraphs[0]
                p.text = '数据详情'
                p.font.size = Pt(24)
                p.font.bold = True
                p.font.color.rgb = RGBColor(0x33, 0x33, 0x33)
                p.font.name = DEFAULT_FONT
                fd['fixed'] = True
            except Exception:
                pass

        elif fd.get('type') == 'text_overflow':
            shape = fd.get('shape')
            if shape and shape.has_text_frame:
                text_len = len(shape.text_frame.text or '')
                try:
                    if text_len > 180 or int(shape.top) + int(shape.height) > int(FOOTER_TOP) - Emu(120000):
                        fd['needs_rebuild'] = True
                    else:
                        for para in shape.text_frame.paragraphs:
                            for run in para.runs:
                                if run.font.size and run.font.size > Pt(9):
                                    run.font.size = Pt(9)
                except Exception:
                    fd['needs_rebuild'] = True
                fd['fixed'] = True

    def _fix_data(self, slide, issue, prs):
        fd = issue.fix_data
        if fd.get('type') == 'page_num':
            fd['fixed'] = True

    def _shapes_overlap(self, a, b) -> bool:
        ax, ay, aw, ah = int(a.left), int(a.top), int(a.width), int(a.height)
        bx, by, bw, bh = int(b.left), int(b.top), int(b.width), int(b.height)

        if ax + aw <= bx or bx + bw <= ax:
            return False
        if ay + ah <= by or by + bh <= ay:
            return False
        return True

    def _is_intentional_overlap(self, a, b) -> bool:
        if hasattr(a, 'is_placeholder') or hasattr(b, 'is_placeholder'):
            return True
        a_area = int(a.width) * int(a.height)
        b_area = int(b.width) * int(b.height)
        if a_area > b_area * 3 or b_area > a_area * 3:
            return True
        return False

    def _is_title_shape(self, shape) -> bool:
        if not shape.has_text_frame:
            return False
        try:
            y = int(shape.top)
            return y < int(CONTENT_TOP_BASE) + Emu(200000)
        except Exception:
            return False

    def _find_empty_template_artifacts(self, slide) -> list:
        artifacts = []
        shapes = list(slide.shapes)
        empty_text_boxes = []

        for shape in shapes:
            if shape.has_text_frame:
                text = (shape.text_frame.text or '').strip()
                if text:
                    continue
                if int(shape.width) < Emu(200000) or int(shape.height) < Emu(120000):
                    continue
                if int(shape.top) < Emu(900000) or int(shape.top) > int(FOOTER_TOP) - Emu(100000):
                    continue
                empty_text_boxes.append(shape)
                artifacts.append(shape)

        for shape in shapes:
            if shape.has_text_frame:
                continue
            try:
                is_large_soft_card = (
                    int(shape.width) >= Emu(1000000) and
                    int(shape.height) >= Emu(500000) and
                    int(shape.top) < int(FOOTER_TOP) - Emu(400000)
                )
                if not is_large_soft_card:
                    continue
                overlaps_empty_text = any(self._shapes_overlap(shape, box) for box in empty_text_boxes)
                if overlaps_empty_text:
                    artifacts.append(shape)
            except Exception:
                continue

        # Preserve order while de-duplicating.
        seen = set()
        unique = []
        for shape in artifacts:
            key = id(shape)
            if key not in seen:
                unique.append(shape)
                seen.add(key)
        return unique

    def _remove_shape(self, shape):
        el = shape.element
        el.getparent().remove(el)

    def _is_text_overflowing(self, shape) -> bool:
        if not shape.has_text_frame:
            return False
        text = shape.text_frame.text
        if not text.strip():
            return False
        if len(text) > 800:
            return True
        try:
            w = int(shape.width)
            h = int(shape.height)
            width_pt = max(1, w / 12700.0)
            max_font_pt = 10
            para_count = 0
            for para in shape.text_frame.paragraphs:
                if not para.text.strip():
                    continue
                para_count += 1
                for run in para.runs:
                    if run.font.size:
                        max_font_pt = max(max_font_pt, run.font.size / 12700.0)
            chars_per_line = max(8, int(width_pt / (max_font_pt * 1.15)))
            est_lines = max(1, (len(text) + chars_per_line - 1) // chars_per_line)
            est_height = int((est_lines * max_font_pt * 1.2 + para_count * 4) * 12700)
            if est_height > h * 1.15:
                return True
            if h < Emu(200000) and len(text) > 80:
                return True
        except Exception:
            pass
        return False


def _shape_name(shape):
    try:
        if shape.has_text_frame:
            return shape.text_frame.text[:20].replace('\n', ' ')
    except Exception:
        pass
    try:
        return shape.shape_type
    except Exception:
        pass
    return '无名形状'


def _is_in_content_area(shape):
    try:
        return int(shape.top) >= int(CONTENT_TOP_BASE)
    except Exception:
        return False


if __name__ == '__main__':
    print("QualityInspector module loaded")
    inspector = QualityInspector()
    print("Ready to inspect PPT files")