From cc0d9ffc5d55f544e10110ae2e82f9bb3ea48cff Mon Sep 17 00:00:00 2001
From: tcpsyn <tcpsyn@gmail.com>
Date: Fri, 14 Aug 2026 03:34:14 -0500
Subject: [PATCH] Add transcript parser for episode page generation

---
 tests/test_transcript_parser.py | 37 +++++++++++++++++++++++++++++++++
 website_gen/__init__.py         |  0
 website_gen/transcript.py       | 29 ++++++++++++++++++++++++++
 3 files changed, 66 insertions(+)
 create mode 100644 tests/test_transcript_parser.py
 create mode 100644 website_gen/__init__.py
 create mode 100644 website_gen/transcript.py

diff --git a/tests/test_transcript_parser.py b/tests/test_transcript_parser.py
new file mode 100644
index 0000000..8b40f07
--- /dev/null
+++ b/tests/test_transcript_parser.py
@@ -0,0 +1,37 @@
+import sys
+from pathlib import Path
+sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
+
+from website_gen.transcript import parse_transcript
+
+
+def test_splits_speaker_turns():
+    raw = "LUKE: Welcome back.\n\nSLIM: Hey Luke, thanks for taking my call."
+    turns = parse_transcript(raw)
+    assert turns == [("LUKE", "Welcome back."),
+                     ("SLIM", "Hey Luke, thanks for taking my call.")]
+
+
+def test_unlabeled_paragraph_carries_previous_speaker():
+    raw = "LUKE: First thing.\n\nStill Luke talking."
+    assert parse_transcript(raw) == [("LUKE", "First thing."),
+                                     ("LUKE", "Still Luke talking.")]
+
+
+def test_ignores_blank_and_whitespace_paragraphs():
+    raw = "LUKE: One.\n\n   \n\nSLIM: Two."
+    assert len(parse_transcript(raw)) == 2
+
+
+def test_speaker_name_with_spaces_is_not_treated_as_label():
+    """A colon mid-sentence must not be mistaken for a speaker label."""
+    raw = "LUKE: Here's the thing: it was the alternator."
+    turns = parse_transcript(raw)
+    assert len(turns) == 1
+    assert turns[0][0] == "LUKE"
+    assert "the thing: it was" in turns[0][1]
+
+
+def test_empty_input_returns_empty_list():
+    assert parse_transcript("") == []
+    assert parse_transcript("   \n\n  ") == []
diff --git a/website_gen/__init__.py b/website_gen/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/website_gen/transcript.py b/website_gen/transcript.py
new file mode 100644
index 0000000..913ff5c
--- /dev/null
+++ b/website_gen/transcript.py
@@ -0,0 +1,29 @@
+import re
+
+SPEAKER_RE = re.compile(r"^([A-Z][A-Z0-9 .'-]{0,30}):\s*(.*)$", re.DOTALL)
+
+
+def parse_transcript(raw: str) -> list[tuple[str, str]]:
+    """Split a transcript into (speaker, text) turns.
+
+    Paragraphs are blank-line separated. A paragraph that does not open with a
+    SPEAKER: label is attributed to whoever spoke last, which is how the
+    transcriber emits long turns that wrap.
+    """
+    turns: list[tuple[str, str]] = []
+    current = None
+    for para in re.split(r"\n\s*\n", raw or ""):
+        para = para.strip()
+        if not para:
+            continue
+        m = SPEAKER_RE.match(para)
+        if m:
+            current = m.group(1).strip()
+            text = m.group(2).strip()
+        else:
+            text = para
+        if current is None:
+            current = "LUKE"
+        if text:
+            turns.append((current, text))
+    return turns