preview md

Taguette-Preprocess.py

@@ -34,40 +34,44 @@ def _(INPUT_DIR, mo):
     return (file_dropdown,)
 
 
+@app.function(hide_code=True)
+def csv_to_markdown(df):
+    """Convert transcript DataFrame to markdown, merging consecutive same-speaker turns."""
+    lines = ["# Interview Transcript"]
+
+    # Track previous speaker to detect when speaker changes
+    prev_speaker = None
+    # Accumulate text from consecutive turns by same speaker
+    merged_text = []
+
+    for _, row in df.iterrows():
+        speaker = row["Speaker"]
+        text = str(row["Transcript"]).strip()
+
+        if speaker == prev_speaker:
+            # Same speaker continues — append text to current block
+            merged_text.append(text)
+        else:
+            # New speaker detected — flush previous speaker's block
+            if prev_speaker is not None:
+                # Format: **Speaker**: text-part-1\n\ntext-part-2
+                # Use \n\n to ensure distinct paragraphs for readability
+                lines.append(f"**{prev_speaker}**: {'\n\n'.join(merged_text)}")
+
+            # Start new block for current speaker
+            prev_speaker = speaker
+            merged_text = [text]
+
+    # Flush final speaker's block
+    if prev_speaker is not None:
+        lines.append(f"**{prev_speaker}**: {'\n\n'.join(merged_text)}")
+
+    # Join all blocks with double newlines for clear separation
+    return "\n\n".join(lines)
+
+
 @app.cell
 def _(file_dropdown, mo, pd):
-    def csv_to_markdown(df):
-        """Convert transcript DataFrame to markdown, merging consecutive same-speaker turns."""
-        lines = [f"# Interview Transcript\n"]
-
-        # Track previous speaker to detect when speaker changes
-        prev_speaker = None
-        # Accumulate text from consecutive turns by same speaker
-        merged_text = []
-
-        for _, row in df.iterrows():
-            speaker = row["Speaker"]
-            text = str(row["Transcript"]).strip()
-
-            if speaker == prev_speaker:
-                # Same speaker continues — append text to current block
-                merged_text.append(text)
-            else:
-                # New speaker detected — flush previous speaker's block
-                if prev_speaker is not None:
-                    # Format: **Speaker**: text-part-1\ntext-part-2 + blank line
-                    lines.append(f"**{prev_speaker}**: {'\n'.join(merged_text)}\n\n")
-
-                # Start new block for current speaker
-                prev_speaker = speaker
-                merged_text = [text]
-
-        # Flush final speaker's block
-        if prev_speaker is not None:
-            lines.append(f"**{prev_speaker}**: {'\n'.join(merged_text)}\n\n")
-
-        return "\n".join(lines)
-
     # Preview
     preview = mo.md("")
     if file_dropdown.value:
@@ -76,7 +80,7 @@ def _(file_dropdown, mo, pd):
         preview = mo.md(md_content)
 
     preview
-    return (csv_to_markdown,)
+    return
 
 
 @app.cell
@@ -87,17 +91,30 @@ def _(mo):
 
 
 @app.cell
-def _(OUTPUT_DIR, Path, convert_btn, csv_to_markdown, file_dropdown, mo, pd):
+def _(OUTPUT_DIR, Path, convert_btn, file_dropdown, mo, pd):
     result = mo.md("")
+    saved_md = None
 
     if convert_btn.value and file_dropdown.value:
         _df = pd.read_csv(file_dropdown.value)
-        _md = csv_to_markdown(_df)
+        saved_md = csv_to_markdown(_df)
         _out_path = OUTPUT_DIR / (Path(file_dropdown.value).stem + ".md")
-        _out_path.write_text(_md)
+        _out_path.write_text(saved_md)
         result = mo.callout(f"✅ Saved to `{_out_path}`", kind="success")
 
     result
+    return (saved_md,)
+
+
+@app.cell
+def _(mo, saved_md):
+    saved_preview = mo.md("")
+    if saved_md:
+        saved_preview = mo.vstack([
+            mo.md("### Saved Markdown Preview"),
+            mo.md(saved_md)
+        ])
+    saved_preview
     return