paperfit-cli 1.0.0

package/skills/latex_fixers/fullwidth_fixers.py

@@ -0,0 +1,375 @@
+"""
+Full-Width and Reference Separation Fixers
+
+确保：
+1. 图片和表格使用满页（单栏）或满栏（双栏）宽度
+2. 参考文献与正文分离（另起一页）
+3. 正文末页要么满页要么缩到上一页
+
+核心原则：
+- 绝对禁用 \\resizebox - 该命令会暴力压缩表格，导致字体大小不一
+- 强制使用 tabularx 宏包配合 \\textwidth
+- 通过自动弹性列（X 格式）、动态字号、列间距微调实现满宽
+"""
+
+import re
+from typing import Any, Dict, Tuple
+
+from .utils import add_package_to_preamble, add_to_preamble
+
+
+def fix_figure_fullwidth(
+    tex_content: str,
+    template_layout: str = "two-column",
+) -> Tuple[str, Dict[str, Any]]:
+    """
+    修复图片为满宽格式。
+
+    策略：
+    1. 双栏模板：使用 figure* 跨双栏，图片宽度=\\textwidth
+    2. 单栏模板：使用 figure，图片宽度=\\textwidth
+    3. 移除所有非 \\textwidth/\\linewidth 的宽度设置
+
+    Args:
+        tex_content: .tex 文件内容
+        template_layout: 模板类型 ("two-column" | "single-column")
+
+    Returns:
+        (modified_content, change_record)
+    """
+    change_record = {
+        "defect_id": "B2-figure-fullwidth",
+        "action": "none",
+        "layout": template_layout,
+    }
+
+    includegraphics_pattern = r'\\includegraphics\[([^\]]*)\]\{([^}]+)\}'
+
+    def fix_width_to_full(match):
+        options = match.group(1)
+        filename = match.group(2)
+
+        # 替换为满宽
+        new_options = "width=\\textwidth"
+        change_record["action"] = f"set {filename} to full width (\\textwidth)"
+        return f"\\includegraphics[{new_options}]{{{filename}}}"
+
+    modified = re.sub(includegraphics_pattern, fix_width_to_full, tex_content)
+
+    if modified != tex_content:
+        change_record["count"] = len(re.findall(includegraphics_pattern, modified))
+
+    return modified, change_record
+
+
+def fix_table_fullwidth(
+    tex_content: str,
+    template_layout: str = "two-column",
+) -> Tuple[str, Dict[str, Any]]:
+    """
+    修复表格为满宽格式 - 原生自适应策略。
+
+    绝对禁用：\\resizebox
+
+    执行逻辑：
+    1. 强制使用 tabularx 宏包配合 \\textwidth
+    2. 通过自动弹性列（X 格式）实现自适应
+    3. 动态字号（如 \\small）以及列间距（\\tabcolsep）微调
+    4. 双栏模板使用 table* 跨双栏
+
+    Args:
+        tex_content: .tex 文件内容
+        template_layout: 模板类型 ("two-column" | "single-column")
+
+    Returns:
+        (modified_content, change_record)
+    """
+    change_record = {
+        "defect_id": "B2-table-fullwidth",
+        "action": "none",
+        "layout": template_layout,
+        "strategies_applied": [],
+    }
+
+    modified = tex_content
+
+    # 策略 1: 移除 \\resizebox 暴力缩放（最高优先级）
+    resizebox_pattern = r'\\resizebox\{[^}]*\}\{[^}]*\}\{\\begin\{tabular'
+    if re.search(resizebox_pattern, modified):
+        # 移除 \\resizebox，保留内部 tabular
+        modified = re.sub(
+            r'\\resizebox\{[^}]*\}\{[^}]*\}\{\\begin\{tabular\}\}',
+            '\\begin{tabular}',
+            modified
+        )
+        modified = re.sub(
+            r'\\end\{tabular\}\}\}',
+            '\\end{tabular}',
+            modified
+        )
+        change_record["strategies_applied"].append("removed_resizebox")
+        change_record["action"] = "removed \\resizebox hack"
+
+    # 策略 2: 添加 tabularx 宏包
+    if '\\usepackage{tabularx}' not in modified:
+        modified = add_package_to_preamble(modified, "tabularx")
+        change_record["strategies_applied"].append("added_tabularx_package")
+
+    # 策略 3: 将 tabular 转换为 tabularx 并设置 \\textwidth
+    tabular_pattern = r'\\begin\{tabular(\*)?\}\{([^}]+)\}'
+
+    def convert_to_tabularx(match):
+        star = match.group(1) or ""
+        col_spec = match.group(2)
+
+        # 将列规格转换为 X 列（弹性列）
+        new_col_spec = convert_cols_to_x(col_spec)
+
+        change_record["strategies_applied"].append(f"converted_to_tabularx_{new_col_spec}")
+        return f"\\begin{{tabularx}}{{\\textwidth}} {{{new_col_spec}}}"
+
+    modified = re.sub(tabular_pattern, convert_to_tabularx, modified)
+
+    # 策略 4: 双栏模板转换为 table*
+    if template_layout == "two-column":
+        if '\\begin{table*}' not in modified:
+            modified = re.sub(
+                r'\\begin\{table\}',
+                '\\begin{table*}',
+                modified
+            )
+            modified = re.sub(
+                r'\\end\{table\}',
+                '\\end{table*}',
+                modified
+            )
+            change_record["strategies_applied"].append("converted_to_table_star")
+            change_record["action"] = "converted table to table* for two-column layout"
+
+    # 策略 5: 优化列间距（如果表格仍然过窄）
+    if '\\tabcolsep' not in modified:
+        # 在导言区添加列间距微调
+        modified = add_to_preamble(modified, "\\setlength{\\tabcolsep}{4pt}")
+        change_record["strategies_applied"].append("reduced_tabcolsep")
+
+    if change_record["strategies_applied"]:
+        change_record["action"] = f"applied {len(change_record['strategies_applied'])} strategies for full-width table"
+
+    return modified, change_record
+
+
+def convert_cols_to_x(col_spec: str) -> str:
+    """
+    将列规格中的 l/c/r 转换为 X 列（弹性列）。
+
+    Args:
+        col_spec: 原始列规格（如 "l|c|r"）
+
+    Returns:
+        转换后的列规格（如 "X|X|X"）
+    """
+    # 保留 | 分隔符和其他格式控制符
+    result = []
+    for char in col_spec:
+        if char in ['l', 'c', 'r']:
+            result.append('X')
+        else:
+            result.append(char)
+    return ''.join(result)
+
+
+def fix_table_fullwidth_native(
+    tex_content: str,
+    template_layout: str = "two-column",
+) -> Tuple[str, Dict[str, Any]]:
+    """
+    表格原生自适应满宽 - 增强版。
+
+    此函数实现更激进的策略：
+    1. 优先使用 tabularx + \\textwidth
+    2. 自动调整字号（\\small, \\footnotesize）
+    3. 微调 \\tabcolsep
+    4. 双栏模板强制 table*
+
+    Args:
+        tex_content: .tex 文件内容
+        template_layout: 模板类型
+
+    Returns:
+        (modified_content, change_record)
+    """
+    change_record = {
+        "defect_id": "B2-table-native-fullwidth",
+        "action": "none",
+        "layout": template_layout,
+    }
+
+    # 调用主修复函数
+    modified, record = fix_table_fullwidth(tex_content, template_layout)
+    change_record.update(record)
+
+    # 额外策略：如果表格仍然溢出，添加字号调整
+    if 'Overfull' in tex_content or 'overflow' in change_record.get("note", ""):
+        # 在表格环境前添加 \\small
+        modified = re.sub(
+            r'\\begin\{tabularx\}',
+            '\\small\\begin{tabularx}',
+            modified,
+            count=1
+        )
+        change_record["action"] += " + added \\small for tighter fit"
+
+    return modified, change_record
+
+
+def fix_all_floats_fullwidth(
+    tex_content: str,
+    template_layout: str = "two-column",
+) -> Tuple[str, Dict[str, Any]]:
+    """
+    同时修复所有图片和表格为满宽格式。
+
+    Args:
+        tex_content: .tex 文件内容
+        template_layout: 模板类型 ("two-column" | "single-column")
+
+    Returns:
+        (modified_content, change_record)
+    """
+    change_record = {
+        "defect_id": "B2-all-floats-fullwidth",
+        "actions": [],
+    }
+
+    modified = tex_content
+
+    # 修复图片
+    modified, figure_record = fix_figure_fullwidth(modified, template_layout)
+    if figure_record["action"] != "none":
+        change_record["actions"].append(figure_record)
+
+    # 修复表格
+    modified, table_record = fix_table_fullwidth(modified, template_layout)
+    if table_record["action"] != "none":
+        change_record["actions"].append(table_record)
+
+    if not change_record["actions"]:
+        change_record["action"] = "none"
+    else:
+        change_record["action"] = f"fixed {len(change_record['actions'])} float types"
+
+    return modified, change_record
+
+
+def ensure_reference_newpage(
+    tex_content: str,
+) -> Tuple[str, Dict[str, Any]]:
+    """
+    确保参考文献另起一页，与正文分离。
+
+    策略：
+    1. 在 \\bibliography 或 \\printbibliography 前添加 \\newpage
+    2. 如果正文末页未满，尝试扩写结论段
+    3. 如果正文可以缩到上一页，压缩并分页
+
+    Args:
+        tex_content: .tex 文件内容
+
+    Returns:
+        (modified_content, change_record)
+    """
+    change_record = {
+        "defect_id": "A3-reference-separation",
+        "action": "none",
+    }
+
+    # 查找 bibliography 命令
+    biblio_patterns = [
+        (r'(\\bibliography\{[^}]*\})', '\\bibliography'),
+        (r'(\\printbibliography)', '\\printbibliography'),
+        (r'(\\begin\{thebibliography\})', '\\begin{thebibliography}'),
+    ]
+
+    for pattern, name in biblio_patterns:
+        matches = list(re.finditer(pattern, tex_content))
+        if matches:
+            for match in matches:
+                biblio_start = match.start()
+                # 检查前 50 字符内是否有 \\newpage
+                context_before = tex_content[max(0, biblio_start - 100):biblio_start]
+
+                # 检查是否已有 \\newpage 或 \\clearpage
+                if '\\newpage' not in context_before and '\\clearpage' not in context_before:
+                    # 在 bibliography 前添加 \\newpage
+                    # 找到 bibliography 前的最后一个空行或 section
+                    insert_pos = biblio_start
+
+                    # 向前查找合适位置（保持一些空白）
+                    for i in range(biblio_start - 1, max(0, biblio_start - 200), -1):
+                        if tex_content[i] == '\n':
+                            # 找到前一个空行
+                            if i > 0 and tex_content[i-1] == '\n':
+                                insert_pos = i + 1
+                                break
+
+                    modified = tex_content[:insert_pos] + "\\newpage\\section*{References}\n" + tex_content[insert_pos:]
+                    change_record["action"] = f"added \\newpage before {name}"
+                    return modified, change_record
+
+    change_record["note"] = "no bibliography command found"
+    return tex_content, change_record
+
+
+def fix_body_last_page(
+    tex_content: str,
+    target_section: str | None = None,
+) -> Tuple[str, Dict[str, Any]]:
+    """
+    修复正文末页：要么满页，要么缩到上一页页尾。
+
+    策略：
+    1. 检测正文最后一段（参考文献前的内容）
+    2. 如果末页留白超过 40%，扩写结论段
+    3. 如果末页内容少于 20%，压缩并添加到上一页
+    4. 确保参考文献从新页开始
+
+    Args:
+        tex_content: .tex 文件内容
+        target_section: 要扩写/缩写的节（如 "Conclusion"）
+
+    Returns:
+        (modified_content, change_record)
+    """
+    change_record = {
+        "defect_id": "A2-body-last-page",
+        "action": "none",
+    }
+
+    # 首先确保参考文献分页
+    modified, ref_record = ensure_reference_newpage(tex_content)
+    if ref_record["action"] != "none":
+        change_record["action"] = ref_record["action"]
+        change_record["ref_separation"] = ref_record
+
+    # 查找结论段并扩写（如果需要）
+    if target_section:
+        conclusion_pattern = rf'(\\section\*\{{{target_section}\}}|\\section\{{{target_section}\}})'
+        conclusion_match = re.search(conclusion_pattern, modified)
+
+        if conclusion_match:
+            conclusion_start = conclusion_match.end()
+            # 查找结论段内容
+            conclusion_end = modified.find('\\bibliography', conclusion_start)
+            if conclusion_end == -1:
+                conclusion_end = modified.find('\\end{document}', conclusion_start)
+
+            if conclusion_end > conclusion_start:
+                conclusion_content = modified[conclusion_start:conclusion_end]
+                lines = conclusion_content.strip().split('\n')
+
+                # 如果结论段少于 3 行，建议扩写
+                if len(lines) < 3:
+                    change_record["suggestion"] = "expand conclusion section to fill page"
+                    change_record["action"] = "identified short conclusion section"
+
+    return modified, change_record

package/skills/latex_fixers/overflow_fixers.py

@@ -0,0 +1,250 @@
+"""
+Overflow Repair Fixers - Category D 缺陷修复
+
+处理 overfull hbox、公式溢出、URL 溢出等问题。
+"""
+
+import re
+from pathlib import Path
+from typing import Any, Dict, List, Tuple
+
+from .utils import add_package_to_preamble, add_to_preamble, find_paragraph_end, find_paragraph_start
+
+
+def fix_overfull_hbox(
+    tex_content: str,
+    line_number: int,
+    overflow_type: str = "paragraph",
+    overflow_amount: float | None = None,
+) -> Tuple[str, Dict[str, Any]]:
+    """
+    修复指定行的 overfull hbox 问题。
+
+    Args:
+        tex_content: .tex 文件内容
+        line_number: 问题所在行号（从 1 开始）
+        overflow_type: 溢出类型 (paragraph/table/formula)
+        overflow_amount: 溢出量（pt）
+
+    Returns:
+        (modified_content, change_record)
+    """
+    lines = tex_content.split('\n')
+    if line_number < 1 or line_number > len(lines):
+        return tex_content, {"status": "failed", "reason": "行号超出范围"}
+
+    target_line = lines[line_number - 1]
+    change_record = {
+        "defect_id": "D1",
+        "line": line_number,
+        "type": overflow_type,
+        "overflow_amount": overflow_amount,
+    }
+
+    # 根据溢出类型选择修复策略
+    if overflow_type == "table":
+        return fix_table_overflow(tex_content, line_number)
+    elif overflow_type == "formula":
+        return fix_long_formula(tex_content, line_number)
+    else:
+        return fix_paragraph_overflow(tex_content, line_number)
+
+
+def fix_paragraph_overflow(tex_content: str, line_number: int | None = None) -> Tuple[str, Dict[str, Any]]:
+    """
+    修复段落文本溢出。
+
+    策略：
+    1. 在长单词中插入断词点 \-
+    2. 添加 \emergencystretch 允许额外拉伸
+    """
+    lines = tex_content.split('\n')
+    change_record = {
+        "defect_id": "D1-paragraph",
+        "action": "none",
+    }
+
+    # 策略 1: 查找长单词并添加断词点
+    if line_number:
+        target_idx = line_number - 1
+        if 0 <= target_idx < len(lines):
+            line = lines[target_idx]
+            # 查找长度超过 15 的单词
+            long_words = re.findall(r'\b[a-zA-Z]{15,}\b', line)
+            if long_words:
+                for word in long_words[:2]:  # 最多处理 2 个单词
+                    # 在元音后添加断词点
+                    hyphenated = add_hyphenation_points(word)
+                    line = line.replace(word, hyphenated, 1)
+                lines[target_idx] = line
+                change_record["action"] = f"insert_hyphenation in {long_words}"
+
+    # 策略 2: 如果仍未解决，在段落前添加\emergencystretch
+    if change_record["action"] == "none":
+        # 查找段落开始（简单启发式：找到包含文本的行）
+        if line_number:
+            # 向前查找段落开始
+            para_start = find_paragraph_start(lines, line_number - 1)
+            if para_start >= 0:
+                # 在段落前插入\emergencystretch
+                emergencystretch_line = "{\\emergencystretch=1.5em "
+                lines.insert(para_start, emergencystretch_line)
+                # 在段落结束添加闭合括号
+                para_end = find_paragraph_end(lines, para_start + 1)
+                if para_end >= 0:
+                    lines.insert(para_end + 1, "}")
+                change_record["action"] = f"add_emergencystretch at line {para_start}"
+
+    return '\n'.join(lines), change_record
+
+
+def add_hyphenation_points(word: str) -> str:
+    """
+    在单词中插入 LaTeX 断词点 \-。
+    简单规则：在元音后、辅音前断词。
+    """
+    vowels = "aeiouAEIOU"
+    result = []
+    for i, char in enumerate(word):
+        result.append(char)
+        # 在元音后且后面还有辅音时插入断词点
+        if char in vowels and i < len(word) - 2:
+            next_char = word[i + 1]
+            if next_char not in vowels and next_char.isalpha():
+                result.append("\\-")
+    return ''.join(result)
+
+
+def find_paragraph_start(lines: List[str], target_idx: int) -> int:
+    """向前查找段落开始（遇到空行或环境开始）"""
+    for i in range(target_idx, -1, -1):
+        line = lines[i].strip()
+        if not line or line.startswith('\\begin') or line.startswith('\\section'):
+            return i + 1 if i < target_idx else i
+    return 0
+
+
+def find_paragraph_end(lines: List[str], start_idx: int) -> int:
+    """向后查找段落结束（遇到空行或环境结束）"""
+    for i in range(start_idx, len(lines)):
+        line = lines[i].strip()
+        if not line or line.startswith('\\end') or line.startswith('\\section'):
+            return i - 1 if i > start_idx else i
+    return len(lines) - 1
+
+
+def fix_table_overflow(tex_content: str, line_number: int | None = None) -> Tuple[str, Dict[str, Any]]:
+    """
+    修复表格溢出。
+
+    策略：
+    1. 将 tabular 替换为 tabularx
+    2. 设置列宽为\linewidth
+    3. 使用 p{width}列类型
+    """
+    change_record = {
+        "defect_id": "D1-table",
+        "action": "none",
+    }
+
+    # 策略 1: 查找 tabular 环境并替换为 tabularx
+    tabular_pattern = r'\\begin\{tabular\}(\{[^}]*\})'
+
+    def replace_with_tabularx(match):
+        col_spec = match.group(1)
+        # 将 l/c/r列替换为 X 列（保留第一列）
+        cols = col_spec.strip('{}').split('|')
+        new_cols = []
+        for col in cols:
+            if col.strip() in ['l', 'c', 'r']:
+                new_cols.append('X')
+            else:
+                new_cols.append(col)
+        new_spec = '{' + '|'.join(new_cols) + '}'
+        change_record["action"] = f"replaced tabular with tabularx, col spec: {new_spec}"
+        return f"\\begin{{tabularx}}{{\\linewidth}}{new_spec}"
+
+    modified = re.sub(tabular_pattern, replace_with_tabularx, tex_content, count=1)
+
+    if modified != tex_content:
+        # 检查是否需要添加 tabularx 宏包
+        if '\\usepackage{tabularx}' not in modified:
+            # 在导言区添加宏包
+            modified = add_package_to_preamble(modified, "tabularx")
+            change_record["packages_added"] = ["tabularx"]
+
+    return modified, change_record
+
+
+def fix_long_formula(tex_content: str, line_number: int | None = None) -> Tuple[str, Dict[str, Any]]:
+    """
+    修复长公式溢出。
+
+    策略：
+    1. 将 equation 替换为 multline 或 split
+    2. 在运算符后添加换行\\\\
+    """
+    change_record = {
+        "defect_id": "D2-formula",
+        "action": "none",
+    }
+
+    # 策略 1: 查找单行 equation 环境并替换为 split
+    equation_pattern = r'\\begin\{equation\}([^\\]*?)\\end\{equation\}'
+
+    def replace_with_split(match):
+        formula = match.group(1)
+        # 查找等号或加减号位置
+        break_points = []
+        for i, char in enumerate(formula):
+            if char in '=+-' and i > len(formula) // 3 and i < 2 * len(formula) // 3:
+                break_points.append(i)
+
+        if break_points:
+            bp = break_points[0]
+            # 在断点处插入\\
+            part1 = formula[:bp+1].rstrip()
+            part2 = formula[bp+1:].lstrip()
+            new_formula = f"{part1}\n\\\\\n\\quad {part2}"
+            change_record["action"] = "split equation at operator"
+            return f"\\begin{{split}}\n{new_formula}\n\\end{{split}}"
+        return match.group(0)
+
+    modified = re.sub(equation_pattern, replace_with_split, tex_content, count=1, flags=re.DOTALL)
+
+    return modified, change_record
+
+
+def fix_url_overflow(tex_content: str, url: str | None = None) -> Tuple[str, Dict[str, Any]]:
+    """
+    修复 URL 溢出。
+
+    策略：
+    1. 使用\\url{}命令包裹
+    2. 添加\\urlbreaks 配置
+    """
+    change_record = {
+        "defect_id": "D3-url",
+        "action": "none",
+    }
+
+    # 策略 1: 查找裸 URL 并替换为\\url{}
+    url_pattern = r'(https?://[^\s\}\]\)]+)'
+
+    def wrap_with_url(match):
+        raw_url = match.group(1)
+        if not raw_url.startswith('\\url{'):
+            change_record["action"] = f"wrapped URL with \\url command"
+            return f"\\url{{{raw_url}}}"
+        return raw_url
+
+    modified = re.sub(url_pattern, wrap_with_url, tex_content)
+
+    # 策略 2: 添加 URL 断行配置
+    if change_record["action"] and '\\def\\UrlBreaks' not in modified:
+        config_line = "\\def\\UrlBreaks{\\do\\/\\do-}"
+        # 在导言区添加
+        modified = add_to_preamble(modified, config_line)
+        change_record["preamble_added"] = config_line
+
+    return modified, change_record