feat: support multi-character input delimiters

README.md

@@ -173,13 +173,16 @@ Column names with spaces work — quote them in SQL:
 $ cat report.csv | sql-pipe 'SELECT "first name", "last name" FROM t WHERE "dept id" = "42"'
 ```
 
-Use a custom input delimiter with `-d` / `--delimiter` (single character), or `--tsv` for tab-separated files:
+Use a custom input delimiter with `-d` / `--delimiter` (1–8 characters), or `--tsv` for tab-separated files:
 
 ```sh
 $ cat data.psv | sql-pipe -d '|' 'SELECT * FROM t'
 $ cat data.tsv | sql-pipe --tsv 'SELECT * FROM t'
 # equivalent:
 $ cat data.tsv | sql-pipe --delimiter '\t' 'SELECT * FROM t'
+# multi-character delimiters:
+$ cat data.psv | sql-pipe -d '||' 'SELECT * FROM t'
+$ cat report.txt | sql-pipe --delimiter '  ' 'SELECT * FROM t'   # two spaces
 ```
 
 Output results as a JSON array of objects with `--json`:

build.zig

@@ -977,6 +977,40 @@ pub fn build(b: *std.Build) void {
     test_sample_fewer_rows.step.dependOn(b.getInstallStep());
     test_step.dependOn(&test_sample_fewer_rows.step);
 
+    // Integration test 95: 2-char delimiter (||) splits fields correctly
+    const test_delimiter_double_pipe = b.addSystemCommand(&.{
+        "bash", "-c",
+        \\printf 'name||age\nAlice||30\nBob||25\n' | ./zig-out/bin/sql-pipe --delimiter '||' 'SELECT name, age FROM t ORDER BY age' | diff - <(printf 'Bob,25\nAlice,30\n')
+    });
+    test_delimiter_double_pipe.step.dependOn(b.getInstallStep());
+    test_step.dependOn(&test_delimiter_double_pipe.step);
+
+    // Integration test 96: 3-char delimiter (;;;) splits fields correctly
+    const test_delimiter_three_char = b.addSystemCommand(&.{
+        "bash", "-c",
+        \\printf 'name;;;age\nAlice;;;30\nBob;;;25\n' | ./zig-out/bin/sql-pipe --delimiter ';;;' 'SELECT name, age FROM t ORDER BY age' | diff - <(printf 'Bob,25\nAlice,30\n')
+    });
+    test_delimiter_three_char.step.dependOn(b.getInstallStep());
+    test_step.dependOn(&test_delimiter_three_char.step);
+
+    // Integration test 97: empty delimiter string exits 1 with error
+    const test_delimiter_empty_error = b.addSystemCommand(&.{
+        "bash", "-c",
+        \\msg=$(printf 'a,b\n1,2\n' | ./zig-out/bin/sql-pipe -d '' 'SELECT * FROM t' 2>&1 >/dev/null; echo "EXIT:$?")
+        \\echo "$msg" | grep -q 'EXIT:1'
+    });
+    test_delimiter_empty_error.step.dependOn(b.getInstallStep());
+    test_step.dependOn(&test_delimiter_empty_error.step);
+
+    // Integration test 98: delimiter longer than 8 chars exits 1 with error
+    const test_delimiter_too_long_error = b.addSystemCommand(&.{
+        "bash", "-c",
+        \\msg=$(printf 'a,b\n1,2\n' | ./zig-out/bin/sql-pipe -d '123456789' 'SELECT * FROM t' 2>&1 >/dev/null; echo "EXIT:$?")
+        \\echo "$msg" | grep -q 'EXIT:1'
+    });
+    test_delimiter_too_long_error.step.dependOn(b.getInstallStep());
+    test_step.dependOn(&test_delimiter_too_long_error.step);
+
     // Unit tests for the RFC 4180 CSV parser (src/csv.zig)
     const unit_tests = b.addTest(.{
         .root_module = b.createModule(.{

docs/sql-pipe.1.scd

@@ -22,17 +22,18 @@ DESCRIPTION
 	behavior and treat all columns as TEXT.
 
 	By default, input fields are parsed as comma-separated values. Use
-	*--delimiter* (or *-d*) to parse other single-character delimiters, or *--tsv*
+	*--delimiter* (or *-d*) to parse other delimiters (1–8 characters), or *--tsv*
 	for tab-separated input.
 
 	Exit codes follow a structured convention for scripting integration (see EXIT CODES
 	section).
 
 OPTIONS
-	*-d, --delimiter* <char>
-		Use a custom single-character input field delimiter instead of comma.
+	*-d, --delimiter* <string>
+		Use a custom input field delimiter (1–8 bytes) instead of comma.
 		Examples: *-d '|'* for pipe-separated files, *--delimiter '\\t'* for
-		tab-separated files.
+		tab-separated files, *-d '||'* for double-pipe-separated files.
+		Produces a usage error if the value is empty or longer than 8 bytes.
 
 	*--tsv*
 		Alias for *--delimiter '\\t'*. Parses tab-separated input.