thank you Q&A, now theres proper unicode support n shit

1 files changed, 55 insertions, 0 deletions

diff --git a/src/utils.zig b/src/utils.zig
index c6e8508..631e464 100644
--- a/src/utils.zig
+++ b/src/utils.zig
@@ -2,6 +2,9 @@ const std = @import("std");
 
 const Allocator = std.mem.Allocator;
 const ArrayList = std.ArrayList;
+const CaseData = @import("CaseData");
+const GenCatData = @import("GenCatData");
+const Utf8View = std.unicode.Utf8View;
 
 pub fn escapeXml(writer: anytype, text: []const u8) !void {
     for (text) |ch| {
@@ -15,6 +18,26 @@ pub fn escapeXml(writer: anytype, text: []const u8) !void {
     }
 }
 
+var gcd_global: ?GenCatData = null;
+
+pub fn getGCD() !GenCatData {
+    if (gcd_global) |gcd| {
+        return gcd;
+    }
+    gcd_global = try GenCatData.init(std.heap.page_allocator);
+    return gcd_global.?;
+}
+
+var cd_global: ?CaseData = null;
+
+pub fn getCD() !CaseData {
+    if (cd_global) |cd| {
+        return cd;
+    }
+    cd_global = try CaseData.init(std.heap.page_allocator);
+    return cd_global.?;
+}
+
 pub inline fn isNull(value: anytype) bool {
     return switch (@typeInfo(@TypeOf(value))) {
         .Null => true,
@@ -22,3 +45,35 @@ pub inline fn isNull(value: anytype) bool {
         else => false,
     };
 }
+
+pub fn trim(str: []const u8) ![]const u8 {
+    const view = try Utf8View.init(str);
+    const gcd = try getGCD();
+
+    var it = view.iterator();
+    var idx: usize = 0;
+    const first = while (it.nextCodepoint()) |cp| {
+        if (!isTrimmable(gcd, cp)) {
+            break idx;
+        }
+        idx = it.i;
+    } else {
+        return "";
+    };
+
+    idx = it.i;
+
+    var last = first;
+    while (it.nextCodepoint()) |cp| {
+        if (!isTrimmable(gcd, cp)) {
+            last = idx + (std.unicode.utf8CodepointSequenceLength(cp) catch unreachable) - 1;
+        }
+        idx = it.i;
+    }
+
+    return str[first .. last + 1];
+}
+
+inline fn isTrimmable(gcd: GenCatData, cp: u21) bool {
+    return gcd.isSeparator(cp) or gcd.isControl(cp);
+}