Code Point Mapper; #529

the-moisrex · Apr 7, 2024 · 15152c0 · 15152c0
1 parent d6c3ae4
commit 15152c0
Show file tree

Hide file tree

Showing 2 changed files with 92 additions and 14 deletions.
diff --git a/webpp/uri/idna/details/generate_idna_mapping_table.js b/webpp/uri/idna/details/generate_idna_mapping_table.js
@@ -51,17 +51,54 @@ const splitLine = line => line.split(';').map(seg => seg.trim());
 const findVersion = fileContent =>
     fileContent.match(/Version: (\d+\.\d+\.\d+)/)[1];
 const findDate = fileContent => fileContent.match(/Date: ([^\n\r]+)/)[1];
-const parseCodePoints =
-    codePoints => {
-      let [rangeStart, rangeEnd] =
-          codePoints.split('..').map(codePoint => parseInt(codePoint, 16));
-      rangeEnd = rangeEnd || rangeStart;
-      return [ rangeStart, rangeEnd ];
-    }
-
+const parseCodePoints = codePoints => {
+  let [rangeStart, rangeEnd] =
+      codePoints.split('..').map(codePoint => parseInt(codePoint, 16));
+  rangeEnd = rangeEnd || rangeStart;
+  return [ rangeStart, rangeEnd ];
+};
 const parseMappedCodePoints = codePoints =>
     codePoints.split(" ").map(codePoint => parseInt(codePoint, 16));
 
+class codePointMapper {
+  constructor(max) {
+    this.bytes = new Uint8Array(max);
+    this.index = 0;
+  }
+
+  append(start, end, isMapped = true) {
+    for (; start !== end; ++start) {
+      const arrayIndex = Math.floor(this.index / 8);
+      const byteIndex = this.index % 8;
+      const mask = (isMapped ? 0x1 : 0x0) << byteIndex;
+
+      // activating the bit:
+      this.bytes[arrayIndex] = this.bytes[arrayIndex] | mask;
+      // console.log(arrayIndex, byteIndex, mask, isMapped,
+      // this.bytes[arrayIndex] | mask)
+
+      ++this.index;
+    }
+  }
+
+  /// Get the length of the table
+  get length() { return Math.ceil(this.index / 8); }
+
+  /// Get how many bits are in the whole table
+  get bitLength() { return this.index; }
+
+  serializeTable(appendFunc, cols = 16) {
+    let pos = 0;
+    for (; pos !== this.length;) {
+      appendFunc(`${this.bytes[pos]}, `);
+      ++pos;
+      if (pos % cols === 0) {
+        appendFunc('\n');
+      }
+    }
+  }
+}
+
 function processCachedFile(fileContent) {
   const lines = fileContent.split('\n');
 
@@ -78,7 +115,8 @@ function processCachedFile(fileContent) {
   console.log(`Version: ${version}`);
   console.log(`Creation Date: ${creationDate}`);
 
-  const table = lines.map((line, index) => {
+  const table = new codePointMapper(200000);
+  lines.forEach((line, index) => {
     line = cleanComments(line)
 
     // ignore empty lines
@@ -92,18 +130,22 @@ function processCachedFile(fileContent) {
 
     switch (status) {
     case 'disallowed_STD3_valid':
-      break;
     case 'deviation': // https://www.unicode.org/reports/tr46/#Deviations
     // Deviations are considered valid in IDNA2008 and UTS #46.
     case 'valid':
+      table.append(rangeStart, rangeEnd, false);
       break;
     case 'disallowed_STD3_mapped':
+      table.append(rangeStart, rangeEnd, true);
       break;
     case 'mapped':
+      table.append(rangeStart, rangeEnd, true);
       break;
     case 'ignored':
+      table.append(rangeStart, rangeEnd, true);
       break;
     case 'disallowed':
+      table.append(rangeStart, rangeEnd, true);
       break;
     default:
       console.error(`Invalid 'status' found: ${status}; line: ${line}`);
@@ -117,14 +159,17 @@ function processCachedFile(fileContent) {
     return `${codePoints}`;
   });
 
+  console.log(`Table Length: ${table.length}`);
+  console.log(`Table Bit Length: ${table.bitLength}`);
+
   createTableFile(version, creationDate, table);
 
   console.log('File processing complete.');
 }
 
 const createTableFile =
     async (version, creationDate, table) => {
-  const fileContent = `
+  const begContent = `
 /**
  * Attention: Auto-generated file, don't modify.
  * 
@@ -141,14 +186,33 @@ const createTableFile =
 #ifndef WEBPP_URI_IDNA_MAPPING_TABLE_HPP
 #define WEBPP_URI_IDNA_MAPPING_TABLE_HPP
 
+#include <array>
+#include <cstdint>
+
 namespace webpp::uri::idna::details {
-    
+    static constexpr std::array<std::uint8_t, ${
+      table.length}> idna_mapping_table{
+`;
+
+  const endContent = `
+    };
 } // webpp::uri::idna::details
 
 #endif // WEBPP_URI_IDNA_MAPPING_TABLE_HPP
     `;
 
-  await fs.writeFile(outFilePath, fileContent);
+  await fs.writeFile(outFilePath, begContent);
+  let content = "";
+  table.serializeTable(line => content += line);
+  await fs.appendFile(outFilePath, content);
+  await fs.appendFile(outFilePath, endContent);
+
+  // Reformat the file
+  require('child_process').exec(`clang-format -i "${outFilePath}"`, err => {
+    if (err) {
+      console.error("Could not re-format the file.", err);
+    }
+  });
 }
 
 start();
diff --git a/webpp/uri/idna/idna_ascii.hpp b/webpp/uri/idna/idna_ascii.hpp
@@ -3,6 +3,20 @@
 #ifndef WEBPP_URI_IDNA_ASCII_HPP
 #define WEBPP_URI_IDNA_ASCII_HPP
 
-namespace webpp::uri::idna {}
+namespace webpp::uri::idna {
+
+    enum struct domain_to_ascii_status {
+        success = 0,
+        invalid_code_point,
+        dissallowed_code_point_found
+    };
+
+    /**
+     *
+     * RFC: https://www.rfc-editor.org/rfc/rfc3490.html#section-4.1
+     */
+    static constexpr auto domain_to_ascii() {}
+
+} // namespace webpp::uri::idna
 
 #endif // WEBPP_URI_IDNA_ASCII_HPP