Git branches support

apps/api/drizzle/0021_boring_chat.sql

@@ -0,0 +1,156 @@
+-- create the generate_ulid function (https://github.com/chrhansen/postgres-ulid)
+CREATE EXTENSION IF NOT EXISTS pgcrypto;
+
+-- ULID according to the spec: https://github.com/ulid/spec
+  -- generate_ulid() => 1GND65M56YS10B35V121ZS4D4
+-- With prefix output in base58 (still sortable in Base 58)
+  -- generate_ulid('base58', 'user') => user_BtnhF9Zpi2zMFsUfMXTCR
+-- More usage: https://github.com/chrhansen/postgres-ulid#usage
+
+CREATE OR REPLACE FUNCTION generate_ulid(output_base text default 'base32', -- also base58 or uuid
+                                         prefix text default NULL,
+                                         delimiter text default '_') RETURNS text
+    LANGUAGE plpgsql
+    AS $$
+DECLARE
+    timestamp          BYTEA := E'\\000\\000\\000\\000\\000\\000';
+    unix_time          BIGINT;
+    ulid               BYTEA;
+    base32_alphabet    TEXT := '0123456789ABCDEFGHJKMNPQRSTVWXYZ'; -- Crockford's
+    base58_alphabet    TEXT := '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz'; -- Bitcoin
+    ulid_base32_length INT := 26;
+    return_string      TEXT := '';
+BEGIN
+    -- 6 timestamp bytes
+    unix_time := (EXTRACT(EPOCH FROM NOW()) * 1000)::BIGINT;
+    FOR i IN 0..5 LOOP
+        timestamp := SET_BYTE(timestamp, i, (unix_time >> (40 - i * 8))::BIT(8)::INTEGER);
+    END LOOP;
+
+    -- 10 entropy bytes
+    ulid := timestamp || gen_random_bytes(10);
+
+    -- Remove the leading '\x' and just keep the last 32 hex-characters
+    return_string := SUBSTRING(ulid::text, 3, 32);
+
+    CASE lower(output_base)
+    WHEN 'base32' THEN
+        return_string := hex_to_base(return_string, base32_alphabet);
+        IF ulid_base32_length - length(return_string) > 0 THEN
+           -- There will maximum be one (1) missing leading zero (0)
+           return_string := repeat('0', ulid_base32_length - length(return_string)) || return_string;
+        END IF;
+    WHEN 'base58' THEN
+        return_string := hex_to_base(return_string, base58_alphabet);
+    WHEN 'uuid' THEN
+       -- Do nothing, already a "hex-string". Can be cast to uuid: generate_ulid('uuid')::uuid
+    ELSE
+        RAISE EXCEPTION 'Unsupported output_base %', output_base
+        USING HINT = 'Should be base32, base58, or uuid.';
+    END CASE;
+
+    IF prefix IS NOT NULL THEN
+        return_string := prefix || delimiter || return_string;
+    END IF;
+
+    RETURN return_string;
+END;
+$$;
+
+CREATE OR REPLACE FUNCTION hex_to_base(hexstr TEXT, base_alphabet TEXT) RETURNS TEXT AS $$
+DECLARE
+    bytes          BYTEA := ('\x' || hexstr)::BYTEA;
+    leading_zeroes INT := 0;
+    -- There are max. 40 (base10) digits in a 32 digit hex-number
+    num            DECIMAL(40,0) := 0;
+    base           DECIMAL(40,0) := 1;
+    byte_value     INT;
+    byte_val       INT;
+    byte_values    INT[] DEFAULT ARRAY[]::INT[];
+    modulo         INT;
+    encoded_str    TEXT := '';
+BEGIN
+    -- This was built to convert 128-bit (32 hex-chars) UUIDs to another base,
+    -- so we'll only promise that. Postgres DECIMAL can be up to ~131000 digits
+    -- so bump the scale of 'num' and 'base' to convert bigger hex-numbers.
+    IF length(hexstr) != 32 THEN
+        RAISE EXCEPTION 'Hex-string >%< should be 32 characters long, but is % char(s).', hexstr, length(hexstr);
+    END IF;
+
+    -- Convert the 32-digit 'hexstr', to the base10 ('normal' digits) 'num'
+    FOR hex_index IN REVERSE ((length(hexstr) / 2) - 1)..0 LOOP
+        byte_value := get_byte(bytes, hex_index);
+        IF byte_value = 0 THEN
+            leading_zeroes := leading_zeroes + 1;
+        ELSE
+            leading_zeroes := 0;
+            num := num + (base * byte_value);
+        END IF;
+        base := base * 256; -- Two (2) hex-digits: 16 * 16 = 256
+    END LOOP;
+
+    -- Convert the up to 40-digit 'num', to the digits in 'base_alphabet'
+    WHILE num > 0 LOOP
+        modulo := num % length(base_alphabet);
+        num := div(num, length(base_alphabet));
+        byte_values := array_append(byte_values, modulo);
+    END LOOP;
+
+    -- Convert the 'byte_values' using characters from 'base_alphabet'. By
+    -- prepending to 'encoded_str' the order of 'byte_values' is reversed.
+    FOREACH byte_val IN ARRAY byte_values
+    LOOP
+        encoded_str := SUBSTRING(base_alphabet, byte_val + 1, 1) || encoded_str;
+    END LOOP;
+
+    -- Prepend first 'base_alphabet'-character to account for leading zeroes in 'hexstr'
+    encoded_str := repeat(SUBSTRING(base_alphabet, 1, 1), leading_zeroes) || encoded_str;
+
+    RETURN encoded_str;
+END;
+$$ LANGUAGE plpgsql IMMUTABLE STRICT;
+
+CREATE TABLE
+	"branches" (
+		"id" varchar(26) PRIMARY KEY NOT NULL,
+		"project_id" varchar(26) NOT NULL,
+		"name" text NOT NULL,
+		"time_spent" integer DEFAULT 0 NOT NULL,
+		"created_at" timestamp DEFAULT now () NOT NULL,
+		"updated_at" timestamp DEFAULT now () NOT NULL
+	);
+
+--> statement-breakpoint
+ALTER TABLE "files"
+ADD COLUMN "branch_id" varchar(26);
+
+--> statement-breakpoint
+ALTER TABLE "branches" ADD CONSTRAINT "branches_project_id_projects_id_fk" FOREIGN KEY ("project_id") REFERENCES "public"."projects" ("id") ON DELETE cascade ON UPDATE no action;
+
+--> statement-breakpoint
+CREATE INDEX "branch_project_id_index" ON "branches" USING btree ("project_id");
+
+--> statement-breakpoint
+CREATE INDEX "branch_name_index" ON "branches" USING btree ("name");
+
+--> statement-breakpoint
+ALTER TABLE "files" ADD CONSTRAINT "files_branch_id_branches_id_fk" FOREIGN KEY ("branch_id") REFERENCES "public"."branches" ("id") ON DELETE cascade ON UPDATE no action;
+
+INSERT INTO
+	"branches" (id, "project_id", name, "time_spent")
+SELECT
+	LOWER(generate_ulid()),
+	id,
+	'main',
+	time_spent
+FROM
+	"projects";
+
+UPDATE "files" f
+SET
+	"branch_id" = b.id
+FROM
+	"branches" b
+WHERE
+	b.project_id = f.project_id
+	AND b.name = 'main';

apps/api/drizzle/0022_sparkling_mephisto.sql

@@ -0,0 +1,14 @@
+ALTER TABLE "files"
+DROP CONSTRAINT "files_project_id_projects_id_fk";
+
+--> statement-breakpoint
+DROP INDEX "project_id_index";
+
+--> statement-breakpoint
+ALTER TABLE "files"
+DROP COLUMN "project_id";
+
+ALTER TABLE "files"
+ALTER COLUMN branch_id
+SET
+    NOT NULL