From 9b8d9e0672f62849decfbf2d2e05d749cc621ce8 Mon Sep 17 00:00:00 2001
From: Songchen Tan <i@tansongchen.com>
Date: Fri, 28 Jun 2024 14:47:00 -0400
Subject: [PATCH] Add analyze

---
 src/pages/[id]/statistics.tsx | 226 ++++++++++++++++++++++++----------
 1 file changed, 163 insertions(+), 63 deletions(-)

diff --git a/src/pages/[id]/statistics.tsx b/src/pages/[id]/statistics.tsx
index af21bd8..2046b66 100644
--- a/src/pages/[id]/statistics.tsx
+++ b/src/pages/[id]/statistics.tsx
@@ -1,8 +1,10 @@
-import { Flex, Table } from "antd";
+import { Flex, Select, Table } from "antd";
 import {
   alphabetAtom,
   displayAtom,
   frequencyAtom,
+  repertoireAtom,
+  sequenceAtom,
   useChaifenTitle,
 } from "~/atoms";
 import {
@@ -14,11 +16,9 @@ import {
 } from "@ant-design/pro-components";
 import { Form, Space, Typography } from "antd";
 import { useAtomValue } from "jotai";
-import type { Frequency } from "~/atoms";
 import { maxLengthAtom } from "~/atoms";
-import type { AnalyzerForm, AssemblyResult } from "~/lib";
+import type { AnalyzerForm, AssemblyResult, Frequency } from "~/lib";
 import { renderIndexed } from "~/lib";
-import { Select } from "~/components/Utils";
 import { useState } from "react";
 import { assemblyResultAtom } from "~/atoms/cache";
 import type { ColumnsType } from "antd/es/table";
@@ -45,6 +45,7 @@ const analyzePrimitiveDuplication = (
   result: AssemblyResult,
   display: (d: string) => string,
   maxLength: number,
+  replacer: (d: string) => string = (d) => d,
 ) => {
   const reverseMap = new Map<string, string[]>();
   let relevant = result;
@@ -59,11 +60,11 @@ const analyzePrimitiveDuplication = (
     relevant = relevant.slice(0, analyzer.top);
   }
   for (const assembly of relevant) {
-    const { name, sequence: elements } = assembly;
+    const { name, sequence } = assembly;
     const sliced = range(maxLength).map((i) =>
-      analyzer.position.includes(i) ? elements[i] : "*",
+      analyzer.position.includes(i) ? sequence[i] : "*",
     );
-    const summary = `(${sliced.map((x) => renderIndexed(x, display)).join(", ")})`;
+    const summary = `(${sliced.map((x) => replacer(renderIndexed(x, display))).join(", ")})`;
     reverseMap.set(summary, (reverseMap.get(summary) || []).concat(name));
   }
 
@@ -75,11 +76,76 @@ interface Density {
   items: string[];
 }
 
+const AnalyzerConfig = ({
+  analyzer,
+  setAnalyzer,
+}: {
+  analyzer: AnalyzerForm;
+  setAnalyzer: (a: AnalyzerForm) => void;
+}) => {
+  const [form] = Form.useForm<AnalyzerForm>();
+  const maxLength = useAtomValue(maxLengthAtom);
+  return (
+    <ProForm<AnalyzerForm>
+      form={form}
+      layout="horizontal"
+      submitter={false}
+      initialValues={analyzer}
+      onValuesChange={(_, values) => setAnalyzer(values)}
+      autoFocusFirstInput={false}
+    >
+      <ProFormGroup>
+        <ProFormSelect
+          mode="multiple"
+          name="position"
+          label="取码"
+          options={range(maxLength).map((d) => ({
+            label: `第 ${d + 1} 码`,
+            value: d,
+          }))}
+          allowClear={false}
+        />
+        <ProFormSelect
+          name="type"
+          label="类型"
+          width="xs"
+          options={[
+            { label: "全部", value: "all" },
+            { label: "一字词", value: "single" },
+            { label: "多字词", value: "multi" },
+          ]}
+        />
+        <ProFormDependency name={["top"]}>
+          {({ top }) => (
+            <Space>
+              <Form.Item label="范围">
+                <Select
+                  value={top === 0 ? 0 : 1}
+                  options={[
+                    { label: "全部", value: 0 },
+                    { label: "前", value: 1 },
+                  ]}
+                  onChange={(value) => {
+                    const top = value === 0 ? 0 : 500;
+                    form.setFieldValue("top", top);
+                    setAnalyzer({ ...form.getFieldsValue(), top });
+                  }}
+                  style={{ width: 96 }}
+                />
+              </Form.Item>
+              <ProFormDigit name="top" width="xs" disabled={top === 0} />
+            </Space>
+          )}
+        </ProFormDependency>
+      </ProFormGroup>
+    </ProForm>
+  );
+};
+
 const SubStatistics = ({ init }: { init: AnalyzerForm }) => {
   const maxLength = useAtomValue(maxLengthAtom);
   const [analyzer, setAnalyzer] = useState(init);
-  const [form] = Form.useForm<AnalyzerForm>();
-  const assemblyResult = useAtomValue(assemblyResultAtom) ?? [];
+  const assemblyResult = useAtomValue(assemblyResultAtom);
   const frequency = useAtomValue(frequencyAtom);
   const display = useAtomValue(displayAtom);
   const reverseMap = analyzePrimitiveDuplication(
@@ -127,59 +193,7 @@ const SubStatistics = ({ init }: { init: AnalyzerForm }) => {
       <Typography.Title level={3}>
         {order}元分布（{coorder}阶重码估计：{Math.round(estimation)}）
       </Typography.Title>
-      <ProForm<AnalyzerForm>
-        form={form}
-        layout="horizontal"
-        submitter={false}
-        initialValues={analyzer}
-        onValuesChange={(_, values) => setAnalyzer(values)}
-        autoFocusFirstInput={false}
-      >
-        <ProFormGroup>
-          <ProFormSelect
-            mode="multiple"
-            name="position"
-            label="取码"
-            options={range(maxLength).map((d) => ({
-              label: `第 ${d + 1} 码`,
-              value: d,
-            }))}
-            allowClear={false}
-          />
-          <ProFormSelect
-            name="type"
-            label="类型"
-            width="xs"
-            options={[
-              { label: "全部", value: "all" },
-              { label: "一字词", value: "single" },
-              { label: "多字词", value: "multi" },
-            ]}
-          />
-          <ProFormDependency name={["top"]}>
-            {({ top }) => (
-              <Space>
-                <Form.Item label="范围">
-                  <Select
-                    value={top === 0 ? 0 : 1}
-                    options={[
-                      { label: "全部", value: 0 },
-                      { label: "前", value: 1 },
-                    ]}
-                    onChange={(value) => {
-                      const top = value === 0 ? 0 : 500;
-                      form.setFieldValue("top", top);
-                      setAnalyzer({ ...form.getFieldsValue(), top });
-                    }}
-                    style={{ width: 96 }}
-                  />
-                </Form.Item>
-                <ProFormDigit name="top" width="xs" disabled={top === 0} />
-              </Space>
-            )}
-          </ProFormDependency>
-        </ProFormGroup>
-      </ProForm>
+      <AnalyzerConfig analyzer={analyzer} setAnalyzer={setAnalyzer} />
       <Table
         dataSource={dataSource}
         columns={columns}
@@ -190,12 +204,98 @@ const SubStatistics = ({ init }: { init: AnalyzerForm }) => {
   );
 };
 
+const MarginalFirstOrderDuplication = () => {
+  const assemblyResult = useAtomValue(assemblyResultAtom);
+  const frequency = useAtomValue(frequencyAtom);
+  const maxLength = useAtomValue(maxLengthAtom);
+  const display = useAtomValue(displayAtom);
+  const sequenceMap = useAtomValue(sequenceAtom);
+  const repertoire = useAtomValue(repertoireAtom);
+  const hashedElements = new Set<string>();
+  for (const { sequence } of assemblyResult) {
+    sequence.forEach((x) => {
+      hashedElements.add(renderIndexed(x, display));
+    });
+  }
+  const allElements = [...hashedElements].sort();
+
+  const [elements, setElements] = useState(allElements.slice(0, 2));
+  const [analyzer, setAnalyzer] = useState<AnalyzerForm>({
+    type: "single",
+    position: range(0, maxLength),
+    top: 0,
+  });
+
+  const rmBefore = analyzePrimitiveDuplication(
+    analyzer,
+    frequency,
+    assemblyResult,
+    display,
+    maxLength,
+  );
+  const szBefore = new Set<string>();
+  rmBefore.forEach((items) => {
+    if (items.length > 1) items.forEach((x) => szBefore.add(x));
+  });
+  const rmAfter = analyzePrimitiveDuplication(
+    analyzer,
+    frequency,
+    assemblyResult,
+    display,
+    maxLength,
+    (d) => (elements.includes(d) ? "^" : d),
+  );
+  const szAfter = new Set<string>();
+  rmAfter.forEach((items) => {
+    if (items.length > 1) items.forEach((x) => szAfter.add(x));
+  });
+  return (
+    <>
+      <Typography.Title level={3}>边际一阶重码计算</Typography.Title>
+      <AnalyzerConfig analyzer={analyzer} setAnalyzer={setAnalyzer} />
+      <Flex gap="middle" align="baseline">
+        <Form.Item label="合并">
+          <Select
+            value={elements}
+            onChange={setElements}
+            style={{ minWidth: 128 }}
+            mode="multiple"
+            options={allElements.map((x) => ({ label: x, value: x }))}
+            filterOption={(input, option) => {
+              if (option === undefined) return false;
+              const value = option.value.replace(/[⁰¹²³⁴⁵⁶⁷⁸⁹]/g, "");
+              if (repertoire[value] !== undefined) {
+                return sequenceMap.get(value)!.startsWith(input);
+              } else {
+                return value.includes(input);
+              }
+            }}
+            filterSort={(a, b) => {
+              return (
+                (sequenceMap.get(a.value.replace(/[⁰¹²³⁴⁵⁶⁷⁸⁹]/g, "")) ?? "")
+                  .length -
+                (sequenceMap.get(b.value.replace(/[⁰¹²³⁴⁵⁶⁷⁸⁹]/g, "")) ?? "")
+                  .length
+              );
+            }}
+          />
+        </Form.Item>
+        <Typography.Paragraph>
+          将增加重码：
+          {[...szAfter].filter((x) => !szBefore.has(x)).join("、")}
+        </Typography.Paragraph>
+      </Flex>
+    </>
+  );
+};
+
 export default function Statistics() {
   useChaifenTitle("统计");
   const maxLength = useAtomValue(maxLengthAtom);
   return (
     <Flex vertical gap="middle">
-      <Typography.Title level={2}>离散性</Typography.Title>
+      <Typography.Title level={2}>离散性分析</Typography.Title>
+      <MarginalFirstOrderDuplication />
       {range(maxLength, 0).map((x) => {
         const type = "single";
         return (