From 7e59f9159f193b8052cdf24efece805127c216bb Mon Sep 17 00:00:00 2001
From: Casey <casey@test.com>
Date: Wed, 14 Aug 2024 21:50:20 +1000
Subject: [PATCH] need to add multi-column join criteria

---
 scratchpad.ipynb | 176 +++++++++++++++++++++++++++++++++--------------
 1 file changed, 124 insertions(+), 52 deletions(-)

diff --git a/scratchpad.ipynb b/scratchpad.ipynb
index bd480cb..7e19a8f 100644
--- a/scratchpad.ipynb
+++ b/scratchpad.ipynb
@@ -53,7 +53,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 254,
+   "execution_count": 329,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -90,6 +90,100 @@
     "    return dynamic_code"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 334,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'joinParams': ('Col_3_renamed', 'Column 3'), 'SelectFields': {'Right_Column 3': ('True', 'Right_Column 3', 'Right_', None), 'Right_Column 4': ('True', 'Right_Column 4', 'Right_', None), '*Unknown': ('True', None, None, None)}}\n",
+      "None\n"
+     ]
+    }
+   ],
+   "source": [
+    "def tool_join(join_spec: dict, df_L: pl.DataFrame, df_R: pl.DataFrame):\n",
+    "    \n",
+    "\n",
+    "\n",
+    "xml_join_tool = results[6][1]\n",
+    "join_spec = getConf_Join(xml_join_tool)\n",
+    "\n",
+    "print(join_spec)\n",
+    "\n",
+    "print(tool_join(join_spec))\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 349,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "L in: shape: (3, 3)\n",
+      "┌─────┬─────┬─────┐\n",
+      "│ foo ┆ bar ┆ ham │\n",
+      "│ --- ┆ --- ┆ --- │\n",
+      "│ i64 ┆ f64 ┆ str │\n",
+      "╞═════╪═════╪═════╡\n",
+      "│ 1   ┆ 6.0 ┆ a   │\n",
+      "│ 2   ┆ 7.0 ┆ b   │\n",
+      "│ 3   ┆ 8.0 ┆ c   │\n",
+      "└─────┴─────┴─────┘\n",
+      "R in: shape: (3, 2)\n",
+      "┌───────┬─────┐\n",
+      "│ apple ┆ ham │\n",
+      "│ ---   ┆ --- │\n",
+      "│ str   ┆ str │\n",
+      "╞═══════╪═════╡\n",
+      "│ x     ┆ a   │\n",
+      "│ y     ┆ b   │\n",
+      "│ z     ┆ d   │\n",
+      "└───────┴─────┘\n",
+      "L: shape: (1, 3)\n",
+      "┌─────┬─────┬─────┐\n",
+      "│ foo ┆ bar ┆ ham │\n",
+      "│ --- ┆ --- ┆ --- │\n",
+      "│ i64 ┆ f64 ┆ str │\n",
+      "╞═════╪═════╪═════╡\n",
+      "│ 3   ┆ 8.0 ┆ c   │\n",
+      "└─────┴─────┴─────┘\n"
+     ]
+    }
+   ],
+   "source": [
+    "df = pl.DataFrame(\n",
+    "    {\n",
+    "        \"foo\": [1, 2, 3],\n",
+    "        \"bar\": [6.0, 7.0, 8.0],\n",
+    "        \"ham\": [\"a\", \"b\", \"c\"],\n",
+    "    }\n",
+    ")\n",
+    "\n",
+    "other_df = pl.DataFrame(\n",
+    "    {\n",
+    "        \"apple\": [\"x\", \"y\", \"z\"],\n",
+    "        \"ham\": [\"a\", \"b\", \"d\"],\n",
+    "    }\n",
+    ")\n",
+    "\n",
+    "print(\"L in:\", df)\n",
+    "\n",
+    "print(\"R in:\", other_df)\n",
+    "\n",
+    "print(\"L:\", df.join(other_df, left_on=\"ham\", right_on=\"ham\", how='anti'))\n",
+    "\n",
+    "# print(\"J: \", df.join(other_df, on=\"ham\", how='inner'))\n",
+    "\n",
+    "# print(\"R:\", other_df.join(df, on=\"ham\", how='anti'))"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -146,60 +240,16 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 313,
+   "execution_count": 330,
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "<Node ToolID=\"7\">\n",
-      "      <GuiSettings Plugin=\"AlteryxBasePluginsGui.Join.Join\">\n",
-      "        <Position x=\"318\" y=\"258\" />\n",
-      "      </GuiSettings>\n",
-      "      <Properties>\n",
-      "        <Configuration joinByRecordPos=\"False\">\n",
-      "          <JoinInfo connection=\"Left\">\n",
-      "            <Field field=\"Col_3_renamed\" />\n",
-      "          </JoinInfo>\n",
-      "          <JoinInfo connection=\"Right\">\n",
-      "            <Field field=\"Column 3\" />\n",
-      "          </JoinInfo>\n",
-      "          <SelectConfiguration>\n",
-      "            <Configuration outputConnection=\"Join\">\n",
-      "              <OrderChanged value=\"False\" />\n",
-      "              <CommaDecimal value=\"False\" />\n",
-      "              <SelectFields>\n",
-      "                <SelectField field=\"Right_Column 3\" selected=\"True\" rename=\"Right_Column 3\" input=\"Right_\" />\n",
-      "                <SelectField field=\"Right_Column 4\" selected=\"True\" rename=\"Right_Column 4\" input=\"Right_\" />\n",
-      "                <SelectField field=\"*Unknown\" selected=\"True\" />\n",
-      "              </SelectFields>\n",
-      "            </Configuration>\n",
-      "          </SelectConfiguration>\n",
-      "        </Configuration>\n",
-      "        <Annotation DisplayMode=\"0\">\n",
-      "          <Name />\n",
-      "          <DefaultAnnotationText />\n",
-      "          <Left value=\"False\" />\n",
-      "        </Annotation>\n",
-      "      </Properties>\n",
-      "      <EngineSettings EngineDll=\"AlteryxBasePluginsEngine.dll\" EngineDllEntryPoint=\"AlteryxJoin\" />\n",
-      "    </Node>\n",
-      "    \n",
-      "joinByRecordPos: False\n",
-      "{'joinParams': ('Col_3_renamed', 'Column 3')}\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
-    "### WIP\n",
-    "\n",
     "def getConf_Join(xml_string):\n",
-    "    print(xml_string)\n",
+    "    # print(xml_string)\n",
     "    root = ET.fromstring(xml_string)\n",
     "    dict_JoinTool = {}\n",
     "\n",
-    "    print(\"joinByRecordPos:\", root.find(\".//Properties/Configuration\").attrib['joinByRecordPos'])\n",
+    "    # print(\"joinByRecordPos:\", root.find(\".//Properties/Configuration\").attrib['joinByRecordPos'])\n",
     "\n",
     "    # Join parameters\n",
     "    for joinField in root.findall(\".//Configuration/JoinInfo\"):\n",
@@ -213,13 +263,35 @@
     "    else:\n",
     "        dict_JoinTool['joinParams'] = (left_on, right_on)\n",
     "\n",
+    "    fieldConfig = {}\n",
     "    # Select parameters\n",
-    "    for joinField in root.findall(\".//Configuration/SelectConfiguration\"):\n",
+    "    for joinField in root.findall(\".//SelectField\"):\n",
+    "        field = joinField.attrib['field']\n",
+    "        fieldSelected = joinField.attrib['selected']\n",
     "\n",
+    "        if field != \"*Unknown\":\n",
+    "            try:\n",
+    "                fieldRename = joinField.attrib['rename']\n",
+    "            except:\n",
+    "                fieldRename = None\n",
+    "            try:\n",
+    "                fieldType = joinField.attrib['type']\n",
+    "            except:\n",
+    "                fieldType = None\n",
+    "\n",
+    "            fieldInput = joinField.attrib['input']\n",
+    "\n",
+    "        fieldConfig[field] = (fieldSelected, fieldRename, fieldInput, fieldType)\n",
+    "\n",
+    "        fieldRename = None\n",
+    "        fieldType = None \n",
+    "        fieldInput  = None\n",
+    "\n",
+    "    dict_JoinTool[\"SelectFields\"] = fieldConfig\n",
     "    return dict_JoinTool\n",
     "\n",
-    "xml_join_tool = results[6][1]\n",
-    "print(getConf_Join(xml_join_tool))\n"
+    "# xml_join_tool = results[6][1]\n",
+    "# print(getConf_Join(xml_join_tool))\n"
    ]
   },
   {