unicode-analyzer: Display friendly general category name

eliandoran · Jul 9, 2024 · a01057e · a01057e
1 parent 94512cb
commit a01057e
Showing 1 changed file with 111 additions and 34 deletions.
diff --git a/src/routes/unicode-analyzer/analyzer.js b/src/routes/unicode-analyzer/analyzer.js
@@ -2,35 +2,105 @@ import { get_unicode_by_decimal } from "unicode-information";
 import he from "he";
 
 const generalCategoryMappings = {
-    Cc: true,
-    Cf: true,
-    Co: true,
-    Cs: true,
-    Ll: false,
-    Lm: false,
-    Lo: false,
-    Lt: false,
-    Lu: false,
-    Mc: true,
-    Me: true,
-    Mn: true,
-    Nd: false,
-    Nl: true,
-    No: true,
-    Pc: false,
-    Pd: false,
-    Pe: false,
-    Pf: false,
-    Pi: false,
-    Po: false,
-    Ps: false,
-    Sc: false,
-    Sk: false,
-    Sm: false,
-    So: false,
-    Zl: true,
-    Zp: true,
-    Zs: true
+    Cc: {
+        name: "Control",
+        control: true
+    },
+    Cf: {
+        name: "Format",
+        control: true,
+    },
+    Co: {
+        name: "Private Use",
+        control: true,
+    },
+    Cs: {
+        name: "Surrogate",
+        control: true,
+    },
+    Ll: {
+        name: "Lowercase Letter"
+    },
+    Lm: {
+        name: "Modifier Letter"
+    },
+    Lo: {
+        name: "Other Letter"
+    },
+    Lt: {
+        name: "Titlecase Letter"
+    },
+    Lu: {
+        name: "Uppercase Letter"
+    },
+    Mc: {
+        name: "Spacing Mark",
+        control: true
+    },
+    Me: {
+        name: "Enclosing Mark",
+        control: true
+    },
+    Mn: {
+        name: "Nonspacing Mark",
+        control: true,
+    },
+    Nd: {
+        name: "Decimal Number"
+    },
+    Nl: {
+        name: "Letter Number",
+        control: true
+    },
+    No: {
+        name: "Other Number",
+        control: true
+    },
+    Pc: {
+        name: "Connector Punctuation"
+    },
+    Pd: {
+        name: "Dash Punctuation"
+    },
+    Pe: {
+        name: "Close Punctuation"
+    },
+    Pf: {
+        name: "Final Punctuation"
+    },
+    Pi: {
+        name: "Initial Punctuation"
+    },
+    Po: {
+        name: "Other Punctuation"
+    },
+    Ps: {
+        name: "Open Punctuation"
+    },
+    Sc: {
+        name: "Currency Symbol"
+    },
+    Sk: {
+        name: "Modifier Symbol"
+    },
+    Sm: {
+        name: "Math Symbol"
+    },
+    So: {
+        name: "Other Symbol"
+    },
+    Zl: {
+        name: "Line Separator",
+        control: true
+    },
+    Zp: {
+        name: "Paragraph Separator",
+        control: true,
+    },
+    Zs: {
+        name: "Space Character",
+        control: true,
+    }
 };
 
 export class UnicodeAnalyzer {
@@ -47,10 +117,17 @@ export class UnicodeAnalyzer {
         }
 
         const info = get_unicode_by_decimal(ch.charCodeAt(0))
-        info.htmlEntity = he.encode(ch, {
-            useNamedReferences: true
-        });
-        this.#dataCodeLookup[code] = info;
+        if (info) {
+            this.#dataCodeLookup[code] = {
+                cp: info.cp,            
+                name: info.name,
+                gc: generalCategoryMappings[info.gc]?.name || info.gc,
+                blk: info.blk,
+                htmlEntity: he.encode(ch, {
+                    useNamedReferences: true
+                })
+            };
+        }
 
         const builtData = this.#buildNewCharacter(ch, info);        
         if (typeof builtData === "object") {
@@ -88,7 +165,7 @@ export class UnicodeAnalyzer {
                 return { symbol: "␣", noBackground: true }
             }
 
-        if (info && generalCategoryMappings[info.gc]) {
+        if (info && generalCategoryMappings[info.gc].control) {
             const unicodeHexValue = `U+${ch.charCodeAt(0).toString(16).toUpperCase().padStart(4, "0")}`;
             return { symbol: unicodeHexValue }
         }