Support SRT data with leading whitespace

Closes #50.

srt.py

@@ -24,7 +24,7 @@
 RGX_POSSIBLE_CRLF = r"\r?\n"
 
 SRT_REGEX = re.compile(
-    r"({idx})\s*{eof}({ts}) +-[ -]> +({ts}) ?({proprietary}){eof}({content})"
+    r"\s*({idx})\s*{eof}({ts}) +-[ -]> +({ts}) ?({proprietary}){eof}({content})"
     # Many sub editors don't add a blank line to the end, and many editors and
     # players accept that. We allow it to be missing in input.
     #
@@ -374,6 +374,12 @@ def _raise_if_not_contiguous(srt, expected_start, actual_start):
     """
     if expected_start != actual_start:
         unmatched_content = srt[expected_start:actual_start]
+
+        if expected_start == 0 and unmatched_content.isspace():
+            # #50: Leading whitespace has nowhere to be captured like in an
+            # intermediate subtitle
+            return
+
         raise SRTParseError(expected_start, actual_start, unmatched_content)
 
 

tests/test_srt.py

@@ -7,7 +7,7 @@
 import string
 from io import StringIO
 
-from hypothesis import given, settings, HealthCheck
+from hypothesis import given, settings, HealthCheck, assume
 import hypothesis.strategies as st
 from nose.tools import (
     eq_ as eq,
@@ -215,6 +215,13 @@ def test_parsing_spaced_arrow(subs):
     subs_eq(reparsed_subtitles, subs)
 
 
+@given(st.text(string.whitespace), st.lists(subtitles()))
+def test_parsing_leading_whitespace(ws, subs):
+    prews_block = ws + srt.compose(subs, reindex=False, strict=False)
+    reparsed_subtitles = srt.parse(prews_block)
+    subs_eq(reparsed_subtitles, subs)
+
+
 @given(st.lists(subtitles()))
 def test_parsing_content_with_blank_lines(subs):
     for subtitle in subs:
@@ -372,6 +379,23 @@ def test_parser_noncontiguous(subs, fake_idx, garbage, fake_timedelta):
         list(srt.parse(composed))
 
 
+@given(st.lists(subtitles()), st.text(min_size=1))
+def test_parser_noncontiguous_leading(subs, garbage):
+    # Issue #50 permits leading whitespace, see test_parsing_leading_whitespace
+    assume(not garbage.isspace())
+
+    # It also shouldn't just be a number, because then we'd confuse it with our
+    # index...
+    assume(not garbage[-1].isdigit())
+
+    # Put some garbage at the beginning that should trigger our noncontiguity
+    # checks
+    composed = garbage + srt.compose(subs)
+
+    with assert_raises(srt.SRTParseError):
+        list(srt.parse(composed))
+
+
 @given(
     st.lists(subtitles(), min_size=1),
     st.integers(min_value=0),