toutpt · jonata · Apr 30, 2020 · Apr 30, 2020 · Apr 30, 2020
diff --git a/captionstransformer/registry.py b/captionstransformer/registry.py
@@ -20,3 +20,8 @@
                     'writer': srt.Writer,
                     'mimetype': 'text/plain',
                     'extension': '.srt'}}
+            'SAMI': {'id': 'SAMI',
+                    'reader': sami.Reader,
+                    'writer': sami.Writer,
+                    'mimetype': 'text/plain',
+                    'extension': '.smi'}}
diff --git a/captionstransformer/sami.py b/captionstransformer/sami.py
@@ -0,0 +1,69 @@
+from captionstransformer import core
+from bs4 import BeautifulSoup
+from datetime import datetime, timedelta
+
+
+class Reader(core.Reader):
+    def text_to_captions(self):
+        soup = BeautifulSoup(self.rawcontent, 'html.parser')
+
+        # Check if there's a <SAMI> tag:
+        if not soup.sami:
+            return False
+        else:
+            samiparam = str(soup.sami.head.samiparam)
+            if 'Metrics {' in samiparam and 'duration: ' in samiparam.split('Metrics {',1)[-1]:
+                self.max_duration = int(samiparam.split('Metrics {',1)[-1].split('duration: ',1)[-1].split(';',1)[0])
+            else:
+                self.max_duration = 5000
+            empty_last_time = soup.new_tag("sync", start=str(self.max_duration))
+            soup.sami.body.append(empty_last_time)
+            prev_line_time = core.get_date(second=0)
+            prev_line_text = ''
+            for sync in soup.sami.body.find_all('sync'):
+                this_line_time = self.get_start(sync)
+                linerepr = repr(str(sync.p)[1:].split('<',1)[0].split('>',1)[-1])[1:-1]
+                this_line_text = linerepr.replace('<br>', '\n').strip()
+                if this_line_text.endswith('\\t'):
+                    this_line_text = this_line_text[:-2]
+                if this_line_text.endswith('\\n'):
+                    this_line_text = this_line_text[:-2]
+                if prev_line_text != '':
+                    caption = core.Caption()
+                    caption.start = prev_line_time
+                    caption.end = this_line_time
+                    caption.text = prev_line_text
+                    self.add_caption(caption)
+                prev_line_time = this_line_time
+                prev_line_text = this_line_text
+
+        return self.captions
+
+    def get_start(self, text):
+        return self.get_raw_time(str(float(text.get('start'))/1000), format="date")
+
+    def get_raw_time(self, utime, format="date"):
+        if '.' in utime:
+            second_f = float(utime)
+            second = int(second_f)
+            millisecond = int(1000 * (second_f - second))
+        else:
+            second = int(utime)
+            millisecond = 0
+        if format == "date":
+            return core.get_date(second=second, millisecond=millisecond)
+        else:
+            return timedelta(seconds=second, milliseconds=millisecond)
+
+
+class Writer(core.Writer):
+    DOCUMENT_TPL = u"""<SAMI>\n<Head>\n\t<Title></Title>\n\t<SAMIParam>\n\t\tCopyright {}\n\t\tMedia {}\n\t\tMetrics {time:ms; duration: 7200000;}\n\t\tSpec {MSFT:1.0;}\n\t</SAMIParam>\n</Head>\n\n<Body>\n%s</Body>\n</SAMI>"""
+    CAPTION_TPL = u"""\t<SYNC Start=%(start)s>\n\t\t<P Class=ENUSCC>%(text)s"""
+
+    def format_time(self, caption):
+        """Return start time for the given format"""
+        return {'start': self.get_utime(caption.start)}
+
+    def get_utime(self, dt):
+        start_milliseconds = ((3600 * dt.hour + 60 * dt.minute + dt.second) * 1000) + (dt.microsecond / 1000)
+        return u"%i" % start_milliseconds