SMI 字幕格式

文件

<SAMI>
<HEAD>
<STYLE TYPE="text/css">
<!--
P { text-align: center; font-family: Arial; }
.ENCC { Name: English; lang: en-US; }
-->
</STYLE>
</HEAD>
<BODY>
<SYNC Start=1000>
<P Class=ENCC>Hello there.
<SYNC Start=3000>
<P Class=ENCC>&nbsp;
<SYNC Start=3500>
<P Class=ENCC>How are you?<br>Fine, thanks.
<SYNC Start=6000>
<P Class=ENCC>&nbsp;
</BODY>
</SAMI>

从中读出了什么

2 cues, 3 lines of text

  Start=1000  =  00:00:01.000 → 00:00:03.000
    Hello there.
    [Class=ENCC]
  Start=3500  =  00:00:03.500 → 00:00:06.000
    How are you?
    Fine, thanks.
    [Class=ENCC]

Outside the cues:
  head       a <STYLE> block with CSS for the paragraphs, kept as it came in
  cue ends   the next <SYNC> tag — a SAMI file writes no end time of its own
  clearing   a <P>&nbsp; block blanks the screen; the two here are kept as clearing blocks, not turned into cues

文件

<SAMI>
<BODY>
<SYNC Start=abc>
<P Class=ENCC>this block's start cannot be read
<SYNC Start=2000>
<SYNC Start=4000>
<P>no attributes here<br>second line
<SYNC Start="6000">
some loose text before any P
</BODY>
</SAMI>

从中读出了什么

3 cues, 3 lines of text

  Start=2000  =  00:00:02.000 → 00:00:04.000
    (no text)
  Start=4000  =  00:00:04.000 → 00:00:06.000
    no attributes here
    second line
  Start=6000  =  00:00:06.000 → 00:00:08.000
    some loose text before any P

Outside the cues:
  head       no <STYLE> block and no <TITLE> — just the <SAMI> and <BODY> tags
  cue ends   the next <SYNC> tag; the last cue has none after it, so its end time was given a default
  loose text text between two <SYNC> tags that belongs to no <P> is kept and attached to the cue open at that point

Reported while reading — repaired or skipped, never silently:
  1 × SYNC blocks whose start time could not be read (they were skipped)
  1 × the last cue has no following timestamp, so its end time was generated