wip

src/utils/qdf/core.rs

@@ -37,21 +37,11 @@ pub fn is_quantamental_dataframe(df: &DataFrame) -> bool {
     check_quantamental_dataframe(df).is_ok()
 }
 
-/// Sort the columns of a Quantamental DataFrame.
-/// The first columns are `real_date`, `cid`, and `xcat`.
-/// These are followed by any available JPMAQS metrics, 'value', 'grading', 'eop_lag', 'mop_lag',
-/// (**in that order**), followed by any other metrics (in alphabetical order).
-pub fn sort_qdf_columns(qdf: &mut DataFrame) -> Result<(), Box<dyn Error>> {
+pub fn get_sorted_qdf_columns(columns: Vec<String>) -> Vec<String> {
     let index_columns = ["real_date", "cid", "xcat"];
     let known_metrics = ["value", "grading", "eop_lag", "mop_lag"];
 
-    let df_columns = qdf
-        .get_column_names()
-        .into_iter()
-        .map(|s| s.clone().into_string())
-        .collect::<Vec<String>>();
-
-    let mut unknown_metrics: Vec<String> = df_columns
+    let mut unknown_metrics: Vec<String> = columns
         .iter()
         .filter(|&m| !known_metrics.contains(&m.as_str()))
         .filter(|&m| !index_columns.contains(&m.as_str()))
@@ -61,13 +51,30 @@ pub fn sort_qdf_columns(qdf: &mut DataFrame) -> Result<(), Box<dyn Error>> {
     let mut new_columns: Vec<String> = vec![];
     new_columns.extend(index_columns.iter().map(|s| s.to_string()));
     for &colname in &known_metrics {
-        if df_columns.contains(&colname.into()) {
+        if columns.contains(&colname.into()) {
             new_columns.push(colname.to_string());
         }
     }
 
     unknown_metrics.sort();
     new_columns.extend(unknown_metrics);
+
+    new_columns
+}
+
+/// Sort the columns of a Quantamental DataFrame.
+/// The first columns are `real_date`, `cid`, and `xcat`.
+/// These are followed by any available JPMAQS metrics, 'value', 'grading', 'eop_lag', 'mop_lag',
+/// (**in that order**), followed by any other metrics (in alphabetical order).
+pub fn sort_qdf_columns(qdf: &mut DataFrame) -> Result<(), Box<dyn Error>> {
+    let df_columns = qdf
+        .get_column_names()
+        .into_iter()
+        .map(|s| s.to_string())
+        .collect::<Vec<String>>();
+
+    let new_columns = get_sorted_qdf_columns(df_columns);
+
     *qdf = qdf
         .select(new_columns.clone())
         .expect("Failed to select columns");
@@ -142,21 +149,23 @@ pub fn get_unique_xcats(df: &DataFrame) -> Result<Vec<String>, Box<dyn Error>> {
     get_unique_from_str_column(df, "xcat")
 }
 
+pub fn get_unique_metrics(df: &DataFrame) -> Result<Vec<String>, Box<dyn Error>> {
+    // return a list of all columns that are not 'real_date', 'cid', 'xcat'
+    let columns = df
+        .get_column_names()
+        .iter()
+        .map(|s| s.as_str().to_string())
+        .collect();
+
+    let sorted_cols = get_sorted_qdf_columns(columns);
+
+    // return sorted_cols[3..].to_vec()
+    Ok(sorted_cols[3..].to_vec())
+}
+
 /// Get the unique dates as a polars Column from a Quantamental DataFrame.
 pub fn get_unique_dates(df: &DataFrame) -> Result<Column, Box<dyn Error>> {
     let date_col = df.column("real_date")?;
     let unique_dates = date_col.unique()?.sort(SortOptions::default())?;
     Ok(unique_dates)
 }
-
-
-
-
-
-
-
-
-
-
-
-

src/utils/qdf/pivots.rs

@@ -27,8 +27,8 @@ pub fn pivot_dataframe_by_ticker(
         // set metric to the first non-index column
         metric = df.get_column_names()[3].to_string();
     }
+    let unique_dates: Column = get_unique_dates(&df)?;
 
-    // let mut new_df = df.clone();
     // keep only the index columns and the metric column
     let mut keep_cols = QDF_INDEX_COLUMNS
         .to_vec()
@@ -37,23 +37,24 @@ pub fn pivot_dataframe_by_ticker(
         .collect::<Vec<String>>();
     keep_cols.push(metric.clone());
 
-    return Err("Not implemented".into());
+    let out_dfs = split_df_by_tickers(&df, Some(vec![metric.clone()]))?;
+
+    // create a new dataframe with the unique dates, and iteratively add add the metric columns
+    let mut new_df = DataFrame::new(vec![unique_dates])?;
+
+    
 
-    // new_df = new_df.select(keep_cols)?;
-    // let ticker_col = get_ticker_column_for_quantamental_dataframe(&new_df)?;
 
-    // new_df.with_column(ticker_col)?;
-    // // drop the cid and xcat columns
-    // new_df = new_df.drop_many(&["cid".to_string(), "xcat".to_string()]);
 
-    // let dates_col = df.column("real_date")?;
 
-    // Ok(df)
 }
 
 /// Splits a dataframe by ticker.
 #[allow(dead_code)]
-fn split_df_by_tickers(df: &DataFrame) -> Result<HashMap<String, DataFrame>, Box<dyn Error>> {
+fn split_df_by_tickers(
+    df: &DataFrame,
+    metrics: Option<Vec<String>>,
+) -> Result<HashMap<String, DataFrame>, Box<dyn Error>> {
     check_quantamental_dataframe(df)?;
 
     let mut df_outs = HashMap::new();
@@ -66,6 +67,15 @@ fn split_df_by_tickers(df: &DataFrame) -> Result<HashMap<String, DataFrame>, Box
         icids.push(cid);
         ixcats.push(xcat);
     }
+    let metrics = metrics.unwrap_or_else(|| get_unique_metrics(df).unwrap());
+    // let keep_cols:Vec<String> = QDF_INDEX_COLUMNS + metrics;
+    let mut keep_cols = QDF_INDEX_COLUMNS
+        .to_vec()
+        .iter()
+        .map(|s| s.to_string())
+        .collect::<Vec<String>>();
+
+    keep_cols.extend(metrics);
 
     for (cid, xcat) in icids.iter().zip(ixcats.iter()) {
         // Apply filter while borrowing df and avoid moving ownership.
@@ -74,17 +84,39 @@ fn split_df_by_tickers(df: &DataFrame) -> Result<HashMap<String, DataFrame>, Box
             .and(col("xcat").eq(lit(xcat.clone())));
 
         let df_out = df.clone().lazy().filter(filter).collect()?;
+        // select keep_cols
+        let df_out = df_out.select(keep_cols.clone())?;
         df_outs.insert(format!("{}_{}", cid, xcat), df_out);
     }
 
     Ok(df_outs)
 }
 
-fn single_ticker_qdf_to_timeseries(df: &DataFrame) -> Result<Vec<DataFrame>, Box<dyn Error>> {
+fn single_ticker_qdf_to_timeseries(mut df: DataFrame) -> Result<Vec<DataFrame>, Box<dyn Error>> {
     let mut df_vec = Vec::new();
 
-    // copy the date col
-    // let date_col =
+    // Since we own `df`, we can remove the "real_date" column, consuming it.
+    let date_col = df.drop_in_place("real_date")?;
 
-    Ok(vec![df.to_owned()])
+    // Assuming `get_unique_metrics` only needs a reference to `df`
+    let metrics = get_unique_metrics(&df)?;
+
+    for metric in metrics.iter() {
+        // Remove the metric column from `df`, consuming it.
+        let metric_col = df.drop_in_place(metric)?;
+
+        // Create a new DataFrame with the date column and the metric column.
+        let new_df = DataFrame::new(vec![
+            date_col.clone(), // Clone because `date_col` is used multiple times
+            metric_col,       // Move the metric column (no clone needed)
+        ])?;
+        df_vec.push(new_df);
+    }
+    // At this point, `df` has had its metric columns removed and is effectively consumed.
+    // The "date_col" has been moved out of `df` as well.
+
+    // If you need to use `df` after this function, consider redesigning the function
+    // to avoid consuming it.
+
+    Ok(df_vec)
 }