From 6a5a77f670ec51093789f0a142a863a7150296d0 Mon Sep 17 00:00:00 2001
From: Sven Augustin <sven.augustin@psi.ch>
Date: Fri, 27 May 2022 19:40:04 +0200
Subject: [PATCH] associate DataFrameHolder to file

---
 stand.py          |  4 +---
 tableapi.py       |  2 +-
 utils/df_utils.py | 35 ++++++++++++++++++++++++++++++++---
 3 files changed, 34 insertions(+), 7 deletions(-)

diff --git a/stand.py b/stand.py
index 4eb5a11..22e23e8 100644
--- a/stand.py
+++ b/stand.py
@@ -47,9 +47,7 @@ if not new_df.equals(df) and not changed:
     restapi.data = df = new_df
 
 
-if not df.empty:
-    print("<<< write hdf")
-    df.to_hdf("output.h5", key="data")
+restapi.dfh.dump()
 
 
 print(">>> end of streamlit run")
diff --git a/tableapi.py b/tableapi.py
index 6dd9aab..b901102 100644
--- a/tableapi.py
+++ b/tableapi.py
@@ -8,7 +8,7 @@ from utils.st_utils import rerun
 class TableAPI:
 
     def __init__(self):
-        self.dfh = DateFrameHolder()
+        self.dfh = DateFrameHolder("output.h5")
         self.sid = None
         self.changed = True
 
diff --git a/utils/df_utils.py b/utils/df_utils.py
index 7114cf0..6b8ad89 100644
--- a/utils/df_utils.py
+++ b/utils/df_utils.py
@@ -3,9 +3,13 @@ import pandas as pd
 
 class DateFrameHolder:
 
-    def __init__(self, df=None):
-        self.df = df or pd.DataFrame()
-#        self.df = pd.DataFrame.from_records([dict(a=1, b=2, c=3)] * 4) #TODO: remove
+    def __init__(self, fn):
+        self.fn = fn
+        self.df = try_load_df(fn)
+
+    def dump(self):
+        dump_non_empty_df(self.df, self.fn)
+
 
     def append(self, data):
         data = pd.DataFrame.from_records([data])
@@ -15,6 +19,31 @@ class DateFrameHolder:
         self.df = pd.DataFrame()
 
 
+    def __repr__(self):
+        head = self.fn + ":"
+        line = "-" * len(head)
+        df = str(self.df)
+        return "\n".join((head, line, df))
+
+
+
+def try_load_df(fn):
+    try:
+        df = pd.read_hdf(fn)
+        print(f">>> loaded dataframe from {fn}")
+    except FileNotFoundError:
+        df = pd.DataFrame()
+        print(">>> created empty dataframe")
+    return df
+
+def dump_non_empty_df(df, fn, key="data"):
+    if df.empty:
+        print("<<< skip dumping empty dataframe")
+        return
+    print(f"<<< dump dataframe to {fn}")
+    df.to_hdf(fn, key)
+
+
 
 from io import BytesIO