taguette pre-process

Taguette-Preprocess.py

@@ -0,0 +1,120 @@
+import marimo
+
+__generated_with = "0.18.0"
+app = marimo.App(width="medium")
+
+
+@app.cell
+def _():
+    import marimo as mo
+    import pandas as pd
+    from pathlib import Path
+    return Path, mo, pd
+
+
+@app.cell
+def _(Path):
+    INPUT_DIR = Path("data/transcripts/raw")
+    OUTPUT_DIR = Path("data/transcripts/clean")
+    OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
+    return INPUT_DIR, OUTPUT_DIR
+
+
+@app.cell
+def _(INPUT_DIR, mo):
+    csv_files = list(INPUT_DIR.glob("*.csv"))
+    file_options = {f.stem: str(f) for f in csv_files}
+
+    file_dropdown = mo.ui.dropdown(
+        options=file_options,
+        label="Select CSV Transcript",
+        full_width=True
+    )
+    file_dropdown
+    return (file_dropdown,)
+
+
+@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:
+        df = pd.read_csv(file_dropdown.value)
+        md_content = csv_to_markdown(df)
+        preview = mo.md(md_content)
+
+    preview
+    return (csv_to_markdown,)
+
+
+@app.cell
+def _(mo):
+    convert_btn = mo.ui.run_button(label="Convert to Markdown")
+    convert_btn
+    return (convert_btn,)
+
+
+@app.cell
+def _(OUTPUT_DIR, Path, convert_btn, csv_to_markdown, file_dropdown, mo, pd):
+    result = mo.md("")
+
+    if convert_btn.value and file_dropdown.value:
+        _df = pd.read_csv(file_dropdown.value)
+        _md = csv_to_markdown(_df)
+        _out_path = OUTPUT_DIR / (Path(file_dropdown.value).stem + ".md")
+        _out_path.write_text(_md)
+        result = mo.callout(f"✅ Saved to `{_out_path}`", kind="success")
+
+    result
+    return
+
+
+@app.cell(hide_code=True)
+def _(mo):
+    mo.md(r"""
+    # Taguette
+
+    Upload and process using taguette: http://taguette.tail44fa00.ts.net/
+    """)
+    return
+
+
+@app.cell
+def _():
+    return
+
+
+if __name__ == "__main__":
+    app.run()