Python difflib生成HTML差异：行内变更与整行增删的差异原因

问题 #

我创建了两个文件（内容如下），并使用difflib.make_file对它们进行比对。在生成的HTML文件中，为什么第1-3行显示每行内的具体变更，而第4行却显示为从diff_text1中删除、在diff_text2中添加？

两个文件内容 #

diff_text1:

The world is a big place.
The galaxy is a big place.
The planet is a big place.
This is in file1 and not file2.

diff_text2:

This world is a large place.
This galaxy is a large place.
This planet is a large place.
This is in file2 and not file1.

对比代码 #

with open("diff_text1.txt", "r") as d1:
    diff_text1 = d1.readlines()
    
with open("diff_text2.txt", "r") as d2:
    diff_text2 = d2.readlines()

# 假设h是difflib.HtmlDiff实例
result4 = h.make_file(diff_text1, diff_text2)
print(result4)

生成的HTML关键差异行 #

<tbody>
    <tr><td class="diff_next" id="difflib_chg_to14__0"><a href="#difflib_chg_to14__top">t</a></td><td class="diff_header" id="from14_1">1</td><td nowrap="nowrap">Th<span class="diff_chg">e</span>&nbsp;world&nbsp;is&nbsp;a&nbsp;<span class="diff_chg">bi</span>g&nbsp;place.</td><td class="diff_next"><a href="#difflib_chg_to14__top">t</a></td><td class="diff_header" id="to14_1">1</td><td nowrap="nowrap">Th<span class="diff_chg">is</span>&nbsp;world&nbsp;is&nbsp;a&nbsp;<span class="diff_chg">lar</span>g<span class="diff_add">e</span>&nbsp;place.</td></tr>
    <tr><td class="diff_next"></td><td class="diff_header" id="from14_2">2</td><td nowrap="nowrap">Th<span class="diff_chg">e</span>&nbsp;galaxy&nbsp;is&nbsp;a&nbsp;<span class="diff_chg">bi</span>g&nbsp;place.</td><td class="diff_next"></td><td class="diff_header" id="to14_2">2</td><td nowrap="nowrap">Th<span class="diff_chg">is</span>&nbsp;galaxy&nbsp;is&nbsp;a&nbsp;<span class="diff_chg">lar</span>g<span class="diff_add">e</span>&nbsp;place.</td></tr>
    <tr><td class="diff_next"></td><td class="diff_header" id="from14_3">3</td><td nowrap="nowrap">Th<span class="diff_chg">e</span>&nbsp;planet&nbsp;is&nbsp;a&nbsp;<span class="diff_chg">bi</span>g&nbsp;place.</td><td class="diff_next"></td><td class="diff_header" id="to14_3">3</td><td nowrap="nowrap">Th<span class="diff_chg">is</span>&nbsp;planet&nbsp;is&nbsp;a&nbsp;<span class="diff_chg">lar</span>g<span class="diff_add">e</span>&nbsp;place.</td></tr>
    <tr><td class="diff_next"></td><td class="diff_header" id="from14_4">4</td><td nowrap="nowrap"><span class="diff_sub">This&nbsp;is&nbsp;in&nbsp;file1&nbsp;and&nbsp;not&nbsp;file2.</span></td><td class="diff_next"></td><td class="diff_header" id="to14_4">4</td><td nowrap="nowrap"><span class="diff_add">This&nbsp;is&nbsp;in&nbsp;file2&nbsp;and&nbsp;not&nbsp;file1.</span></td></tr>
</tbody>

原因分析 #

这是difflib的行相似度匹配逻辑在起作用：

• difflib对比两行时，会先判断它们的相似程度是否达标（核心看两行的最长公共内容占比，由默认参数控制阈值）。
• 第1-3行：对应行只有局部改动——比如把The改成This、big改成large，大部分内容完全一致，相似度足够高，所以被判定为同一行的修改，会高亮显示行内具体变更的字符。
• 第4行：两行开头虽然相同，但关键内容（file1和file2的位置互换）导致整体相似度低于阈值，difflib直接判定这是完全不同的两行，于是把原文件的行标为删除（diff_sub样式），新文件的行标为添加（diff_add样式），不会再去解析行内的局部差异。

内容的提问来源于stack exchange，提问作者Dimurof82