14 files changed, 321 insertions, 100 deletions

diff --git a/lib/python/qmk/cli/__init__.py b/lib/python/qmk/cli/__init__.py
index 1e1c26671..91d42bb3a 100644
--- a/lib/python/qmk/cli/__init__.py
+++ b/lib/python/qmk/cli/__init__.py
@@ -50,6 +50,7 @@ subcommands = [
     'qmk.cli.generate.layouts',
     'qmk.cli.generate.rgb_breathe_table',
     'qmk.cli.generate.rules_mk',
+    'qmk.cli.generate.version_h',
     'qmk.cli.hello',
     'qmk.cli.info',
     'qmk.cli.json2c',
diff --git a/lib/python/qmk/cli/doctor/__init__.py b/lib/python/qmk/cli/doctor/__init__.py
new file mode 100755
index 000000000..272e04202
--- /dev/null
+++ b/lib/python/qmk/cli/doctor/__init__.py
@@ -0,0 +1,5 @@
+"""QMK Doctor
+
+Check out the user's QMK environment and make sure it's ready to compile.
+"""
+from .main import doctor
diff --git a/lib/python/qmk/os_helpers/__init__.py b/lib/python/qmk/cli/doctor/check.py
index 3e98db3c3..a0bbb2816 100644
--- a/lib/python/qmk/os_helpers/__init__.py
+++ b/lib/python/qmk/cli/doctor/check.py
@@ -1,4 +1,4 @@
-"""OS-agnostic helper functions
+"""Check for specific programs.
 """
 from enum import Enum
 import re
@@ -30,7 +30,7 @@ ESSENTIAL_BINARIES = {
 }
 
 
-def parse_gcc_version(version):
+def _parse_gcc_version(version):
     m = re.match(r"(\d+)(?:\.(\d+))?(?:\.(\d+))?", version)
 
     return {
@@ -40,7 +40,7 @@ def parse_gcc_version(version):
     }
 
 
-def check_arm_gcc_version():
+def _check_arm_gcc_version():
     """Returns True if the arm-none-eabi-gcc version is not known to cause problems.
     """
     if 'output' in ESSENTIAL_BINARIES['arm-none-eabi-gcc']:
@@ -50,7 +50,7 @@ def check_arm_gcc_version():
     return CheckStatus.OK  # Right now all known arm versions are ok
 
 
-def check_avr_gcc_version():
+def _check_avr_gcc_version():
     """Returns True if the avr-gcc version is not known to cause problems.
     """
     rc = CheckStatus.ERROR
@@ -60,7 +60,7 @@ def check_avr_gcc_version():
         cli.log.info('Found avr-gcc version %s', version_number)
         rc = CheckStatus.OK
 
-        parsed_version = parse_gcc_version(version_number)
+        parsed_version = _parse_gcc_version(version_number)
         if parsed_version['major'] > 8:
             cli.log.warning('{fg_yellow}We do not recommend avr-gcc newer than 8. Downgrading to 8.x is recommended.')
             rc = CheckStatus.WARNING
@@ -68,7 +68,7 @@ def check_avr_gcc_version():
     return rc
 
 
-def check_avrdude_version():
+def _check_avrdude_version():
     if 'output' in ESSENTIAL_BINARIES['avrdude']:
         last_line = ESSENTIAL_BINARIES['avrdude']['output'].split('\n')[-2]
         version_number = last_line.split()[2][:-1]
@@ -77,7 +77,7 @@ def check_avrdude_version():
     return CheckStatus.OK
 
 
-def check_dfu_util_version():
+def _check_dfu_util_version():
     if 'output' in ESSENTIAL_BINARIES['dfu-util']:
         first_line = ESSENTIAL_BINARIES['dfu-util']['output'].split('\n')[0]
         version_number = first_line.split()[1]
@@ -86,7 +86,7 @@ def check_dfu_util_version():
     return CheckStatus.OK
 
 
-def check_dfu_programmer_version():
+def _check_dfu_programmer_version():
     if 'output' in ESSENTIAL_BINARIES['dfu-programmer']:
         first_line = ESSENTIAL_BINARIES['dfu-programmer']['output'].split('\n')[0]
         version_number = first_line.split()[1]
@@ -111,7 +111,7 @@ def check_binary_versions():
     """Check the versions of ESSENTIAL_BINARIES
     """
     versions = []
-    for check in (check_arm_gcc_version, check_avr_gcc_version, check_avrdude_version, check_dfu_util_version, check_dfu_programmer_version):
+    for check in (_check_arm_gcc_version, _check_avr_gcc_version, _check_avrdude_version, _check_dfu_util_version, _check_dfu_programmer_version):
         versions.append(check())
     return versions
 
diff --git a/lib/python/qmk/os_helpers/linux/__init__.py b/lib/python/qmk/cli/doctor/linux.py
index 008654ab0..c0b77216a 100644
--- a/lib/python/qmk/os_helpers/linux/__init__.py
+++ b/lib/python/qmk/cli/doctor/linux.py
@@ -1,11 +1,13 @@
 """OS-specific functions for: Linux
 """
-from pathlib import Path
+import platform
 import shutil
+from pathlib import Path
 
 from milc import cli
+
 from qmk.constants import QMK_FIRMWARE
-from qmk.os_helpers import CheckStatus
+from .check import CheckStatus
 
 
 def _udev_rule(vid, pid=None, *args):
@@ -138,3 +140,23 @@ def check_modem_manager():
         """(TODO): Add check for non-systemd systems
         """
     return False
+
+
+def os_test_linux():
+    """Run the Linux specific tests.
+    """
+    # Don't bother with udev on WSL, for now
+    if 'microsoft' in platform.uname().release.lower():
+        cli.log.info("Detected {fg_cyan}Linux (WSL){fg_reset}.")
+
+        # https://github.com/microsoft/WSL/issues/4197
+        if QMK_FIRMWARE.as_posix().startswith("/mnt"):
+            cli.log.warning("I/O performance on /mnt may be extremely slow.")
+            return CheckStatus.WARNING
+
+        return CheckStatus.OK
+    else:
+        cli.log.info("Detected {fg_cyan}Linux{fg_reset}.")
+        from .linux import check_udev_rules
+
+        return check_udev_rules()
diff --git a/lib/python/qmk/cli/doctor/macos.py b/lib/python/qmk/cli/doctor/macos.py
new file mode 100644
index 000000000..00fb27285
--- /dev/null
+++ b/lib/python/qmk/cli/doctor/macos.py
@@ -0,0 +1,13 @@
+import platform
+
+from milc import cli
+
+from .check import CheckStatus
+
+
+def os_test_macos():
+    """Run the Mac specific tests.
+    """
+    cli.log.info("Detected {fg_cyan}macOS %s{fg_reset}.", platform.mac_ver()[0])
+
+    return CheckStatus.OK
diff --git a/lib/python/qmk/cli/doctor.py b/lib/python/qmk/cli/doctor/main.py
index 327bc9cb3..6a31ccdfd 100755
--- a/lib/python/qmk/cli/doctor.py
+++ b/lib/python/qmk/cli/doctor/main.py
@@ -7,9 +7,11 @@ from subprocess import DEVNULL
 
 from milc import cli
 from milc.questions import yesno
+
 from qmk import submodules
-from qmk.constants import QMK_FIRMWARE
-from qmk.os_helpers import CheckStatus, check_binaries, check_binary_versions, check_submodules, check_git_repo
+from qmk.constants import QMK_FIRMWARE, QMK_FIRMWARE_UPSTREAM
+from .check import CheckStatus, check_binaries, check_binary_versions, check_submodules
+from qmk.commands import git_check_repo, git_get_branch, git_is_dirty, git_get_remotes, git_check_deviation, in_virtualenv
 
 
 def os_tests():
@@ -18,51 +20,48 @@ def os_tests():
     platform_id = platform.platform().lower()
 
     if 'darwin' in platform_id or 'macos' in platform_id:
+        from .macos import os_test_macos
         return os_test_macos()
     elif 'linux' in platform_id:
+        from .linux import os_test_linux
         return os_test_linux()
     elif 'windows' in platform_id:
+        from .windows import os_test_windows
         return os_test_windows()
     else:
         cli.log.warning('Unsupported OS detected: %s', platform_id)
         return CheckStatus.WARNING
 
 
-def os_test_linux():
-    """Run the Linux specific tests.
+def git_tests():
+    """Run Git-related checks
     """
-    # Don't bother with udev on WSL, for now
-    if 'microsoft' in platform.uname().release.lower():
-        cli.log.info("Detected {fg_cyan}Linux (WSL){fg_reset}.")
-
-        # https://github.com/microsoft/WSL/issues/4197
-        if QMK_FIRMWARE.as_posix().startswith("/mnt"):
-            cli.log.warning("I/O performance on /mnt may be extremely slow.")
-            return CheckStatus.WARNING
+    status = CheckStatus.OK
 
-        return CheckStatus.OK
+    # Make sure our QMK home is a Git repo
+    git_ok = git_check_repo()
+    if not git_ok:
+        cli.log.warning("{fg_yellow}QMK home does not appear to be a Git repository! (no .git folder)")
+        status = CheckStatus.WARNING
     else:
-        cli.log.info("Detected {fg_cyan}Linux{fg_reset}.")
-        from qmk.os_helpers.linux import check_udev_rules
-
-        return check_udev_rules()
-
-
-def os_test_macos():
-    """Run the Mac specific tests.
-    """
-    cli.log.info("Detected {fg_cyan}macOS %s{fg_reset}.", platform.mac_ver()[0])
-
-    return CheckStatus.OK
-
-
-def os_test_windows():
-    """Run the Windows specific tests.
-    """
-    win32_ver = platform.win32_ver()
-    cli.log.info("Detected {fg_cyan}Windows %s (%s){fg_reset}.", win32_ver[0], win32_ver[1])
-
-    return CheckStatus.OK
+        git_branch = git_get_branch()
+        if git_branch:
+            cli.log.info('Git branch: %s', git_branch)
+            git_dirty = git_is_dirty()
+            if git_dirty:
+                cli.log.warning('{fg_yellow}Git has unstashed/uncommitted changes.')
+                status = CheckStatus.WARNING
+            git_remotes = git_get_remotes()
+            if 'upstream' not in git_remotes.keys() or QMK_FIRMWARE_UPSTREAM not in git_remotes['upstream'].get('url', ''):
+                cli.log.warning('{fg_yellow}The official repository does not seem to be configured as git remote "upstream".')
+                status = CheckStatus.WARNING
+            else:
+                git_deviation = git_check_deviation(git_branch)
+                if git_branch in ['master', 'develop'] and git_deviation:
+                    cli.log.warning('{fg_yellow}The local "%s" branch contains commits not found in the upstream branch.', git_branch)
+                    status = CheckStatus.WARNING
+
+    return status
 
 
 @cli.argument('-y', '--yes', action='store_true', arg_only=True, help='Answer yes to all questions.')
@@ -82,12 +81,11 @@ def doctor(cli):
 
     status = os_tests()
 
-    # Make sure our QMK home is a Git repo
-    git_ok = check_git_repo()
+    status = git_tests()
 
-    if git_ok == CheckStatus.WARNING:
-        cli.log.warning("QMK home does not appear to be a Git repository! (no .git folder)")
-        status = CheckStatus.WARNING
+    venv = in_virtualenv()
+    if venv:
+        cli.log.info('CLI installed in virtualenv.')
 
     # Make sure the basic CLI tools we need are available and can be executed.
     bin_ok = check_binaries()
diff --git a/lib/python/qmk/cli/doctor/windows.py b/lib/python/qmk/cli/doctor/windows.py
new file mode 100644
index 000000000..381ab36fd
--- /dev/null
+++ b/lib/python/qmk/cli/doctor/windows.py
@@ -0,0 +1,14 @@
+import platform
+
+from milc import cli
+
+from .check import CheckStatus
+
+
+def os_test_windows():
+    """Run the Windows specific tests.
+    """
+    win32_ver = platform.win32_ver()
+    cli.log.info("Detected {fg_cyan}Windows %s (%s){fg_reset}.", win32_ver[0], win32_ver[1])
+
+    return CheckStatus.OK
diff --git a/lib/python/qmk/cli/format/json.py b/lib/python/qmk/cli/format/json.py
index 1358c70e7..19d504491 100755
--- a/lib/python/qmk/cli/format/json.py
+++ b/lib/python/qmk/cli/format/json.py
@@ -8,7 +8,7 @@ from jsonschema import ValidationError
 from milc import cli
 
 from qmk.info import info_json
-from qmk.json_schema import json_load, keyboard_validate
+from qmk.json_schema import json_load, validate
 from qmk.json_encoders import InfoJSONEncoder, KeymapJSONEncoder
 from qmk.path import normpath
 
@@ -23,14 +23,13 @@ def format_json(cli):
 
     if cli.args.format == 'auto':
         try:
-            keyboard_validate(json_file)
+            validate(json_file, 'qmk.keyboard.v1')
             json_encoder = InfoJSONEncoder
 
         except ValidationError as e:
             cli.log.warning('File %s did not validate as a keyboard:\n\t%s', cli.args.json_file, e)
             cli.log.info('Treating %s as a keymap file.', cli.args.json_file)
             json_encoder = KeymapJSONEncoder
-
     elif cli.args.format == 'keyboard':
         json_encoder = InfoJSONEncoder
     elif cli.args.format == 'keymap':
diff --git a/lib/python/qmk/cli/generate/version_h.py b/lib/python/qmk/cli/generate/version_h.py
new file mode 100644
index 000000000..b8e52588c
--- /dev/null
+++ b/lib/python/qmk/cli/generate/version_h.py
@@ -0,0 +1,28 @@
+"""Used by the make system to generate version.h for use in code.
+"""
+from milc import cli
+
+from qmk.commands import create_version_h
+from qmk.path import normpath
+
+
+@cli.argument('-o', '--output', arg_only=True, type=normpath, help='File to write to')
+@cli.argument('-q', '--quiet', arg_only=True, action='store_true', help="Quiet mode, only output error messages")
+@cli.argument('--skip-git', arg_only=True, action='store_true', help='Skip Git operations')
+@cli.argument('--skip-all', arg_only=True, action='store_true', help='Use placeholder values for all defines (implies --skip-git)')
+@cli.subcommand('Used by the make system to generate version.h for use in code', hidden=True)
+def generate_version_h(cli):
+    """Generates the version.h file.
+    """
+    if cli.args.skip_all:
+        cli.args.skip_git = True
+
+    version_h = create_version_h(cli.args.skip_git, cli.args.skip_all)
+
+    if cli.args.output:
+        cli.args.output.write_text(version_h)
+
+        if not cli.args.quiet:
+            cli.log.info('Wrote version.h to %s.', cli.args.output)
+    else:
+        print(version_h)
diff --git a/lib/python/qmk/commands.py b/lib/python/qmk/commands.py
index 3a35c1103..8ff8501bf 100644
--- a/lib/python/qmk/commands.py
+++ b/lib/python/qmk/commands.py
@@ -2,6 +2,7 @@
 """
 import json
 import os
+import sys
 import shutil
 from pathlib import Path
 from subprocess import DEVNULL
@@ -10,7 +11,7 @@ from time import strftime
 from milc import cli
 
 import qmk.keymap
-from qmk.constants import KEYBOARD_OUTPUT_PREFIX
+from qmk.constants import QMK_FIRMWARE, KEYBOARD_OUTPUT_PREFIX
 from qmk.json_schema import json_load
 
 time_fmt = '%Y-%m-%d-%H:%M:%S'
@@ -86,11 +87,17 @@ def create_make_command(keyboard, keymap, target=None, parallel=1, **env_vars):
     return create_make_target(':'.join(make_args), parallel, **env_vars)
 
 
-def get_git_version(repo_dir='.', check_dir='.'):
+def get_git_version(current_time, repo_dir='.', check_dir='.'):
     """Returns the current git version for a repo, or the current time.
     """
     git_describe_cmd = ['git', 'describe', '--abbrev=6', '--dirty', '--always', '--tags']
 
+    if repo_dir != '.':
+        repo_dir = Path('lib') / repo_dir
+
+    if check_dir != '.':
+        check_dir = repo_dir / check_dir
+
     if Path(check_dir).exists():
         git_describe = cli.run(git_describe_cmd, stdin=DEVNULL, cwd=repo_dir)
 
@@ -100,23 +107,40 @@ def get_git_version(repo_dir='.', check_dir='.'):
         else:
             cli.log.warn(f'"{" ".join(git_describe_cmd)}" returned error code {git_describe.returncode}')
             print(git_describe.stderr)
-            return strftime(time_fmt)
+            return current_time
 
-    return strftime(time_fmt)
+    return current_time
 
 
-def write_version_h(git_version, build_date, chibios_version, chibios_contrib_version):
-    """Generate and write quantum/version.h
+def create_version_h(skip_git=False, skip_all=False):
+    """Generate version.h contents
     """
-    version_h = [
-        f'#define QMK_VERSION "{git_version}"',
-        f'#define QMK_BUILDDATE "{build_date}"',
-        f'#define CHIBIOS_VERSION "{chibios_version}"',
-        f'#define CHIBIOS_CONTRIB_VERSION "{chibios_contrib_version}"',
-    ]
+    if skip_all:
+        current_time = "1970-01-01-00:00:00"
+    else:
+        current_time = strftime(time_fmt)
+
+    if skip_git:
+        git_version = "NA"
+        chibios_version = "NA"
+        chibios_contrib_version = "NA"
+    else:
+        git_version = get_git_version(current_time)
+        chibios_version = get_git_version(current_time, "chibios", "os")
+        chibios_contrib_version = get_git_version(current_time, "chibios-contrib", "os")
+
+    version_h_lines = f"""/* This file was automatically generated. Do not edit or copy.
+ */
+
+#pragma once
+
+#define QMK_VERSION "{git_version}"
+#define QMK_BUILDDATE "{current_time}"
+#define CHIBIOS_VERSION "{chibios_version}"
+#define CHIBIOS_CONTRIB_VERSION "{chibios_contrib_version}"
+"""
 
-    version_h_file = Path('quantum/version.h')
-    version_h_file.write_text('\n'.join(version_h))
+    return version_h_lines
 
 
 def compile_configurator_json(user_keymap, bootloader=None, parallel=1, **env_vars):
@@ -149,13 +173,8 @@ def compile_configurator_json(user_keymap, bootloader=None, parallel=1, **env_va
     keymap_dir.mkdir(exist_ok=True, parents=True)
     keymap_c.write_text(c_text)
 
-    # Write the version.h file
-    git_version = get_git_version()
-    build_date = strftime('%Y-%m-%d-%H:%M:%S')
-    chibios_version = get_git_version("lib/chibios", "lib/chibios/os")
-    chibios_contrib_version = get_git_version("lib/chibios-contrib", "lib/chibios-contrib/os")
-
-    write_version_h(git_version, build_date, chibios_version, chibios_contrib_version)
+    version_h = Path('quantum/version.h')
+    version_h.write_text(create_version_h())
 
     # Return a command that can be run to make the keymap and flash if given
     verbose = 'true' if cli.config.general.verbose else 'false'
@@ -181,10 +200,6 @@ def compile_configurator_json(user_keymap, bootloader=None, parallel=1, **env_va
         make_command.append(f'{key}={value}')
 
     make_command.extend([
-        f'GIT_VERSION={git_version}',
-        f'BUILD_DATE={build_date}',
-        f'CHIBIOS_VERSION={chibios_version}',
-        f'CHIBIOS_CONTRIB_VERSION={chibios_contrib_version}',
         f'KEYBOARD={user_keymap["keyboard"]}',
         f'KEYMAP={user_keymap["keymap"]}',
         f'KEYBOARD_FILESAFE={keyboard_filesafe}',
@@ -223,3 +238,71 @@ def parse_configurator_json(configurator_file):
             user_keymap['layout'] = aliases[orig_keyboard]['layouts'][user_keymap['layout']]
 
     return user_keymap
+
+
+def git_check_repo():
+    """Checks that the .git directory exists inside QMK_HOME.
+
+    This is a decent enough indicator that the qmk_firmware directory is a
+    proper Git repository, rather than a .zip download from GitHub.
+    """
+    dot_git_dir = QMK_FIRMWARE / '.git'
+
+    return dot_git_dir.is_dir()
+
+
+def git_get_branch():
+    """Returns the current branch for a repo, or None.
+    """
+    git_branch = cli.run(['git', 'branch', '--show-current'])
+    if not git_branch.returncode != 0 or not git_branch.stdout:
+        # Workaround for Git pre-2.22
+        git_branch = cli.run(['git', 'rev-parse', '--abbrev-ref', 'HEAD'])
+
+    if git_branch.returncode == 0:
+        return git_branch.stdout.strip()
+
+
+def git_is_dirty():
+    """Returns 1 if repo is dirty, or 0 if clean
+    """
+    git_diff_staged_cmd = ['git', 'diff', '--quiet']
+    git_diff_unstaged_cmd = [*git_diff_staged_cmd, '--cached']
+
+    unstaged = cli.run(git_diff_staged_cmd)
+    staged = cli.run(git_diff_unstaged_cmd)
+
+    return unstaged.returncode != 0 or staged.returncode != 0
+
+
+def git_get_remotes():
+    """Returns the current remotes for a repo.
+    """
+    remotes = {}
+
+    git_remote_show_cmd = ['git', 'remote', 'show']
+    git_remote_get_cmd = ['git', 'remote', 'get-url']
+
+    git_remote_show = cli.run(git_remote_show_cmd)
+    if git_remote_show.returncode == 0:
+        for name in git_remote_show.stdout.splitlines():
+            git_remote_name = cli.run([*git_remote_get_cmd, name])
+            remotes[name.strip()] = {"url": git_remote_name.stdout.strip()}
+
+    return remotes
+
+
+def git_check_deviation(active_branch):
+    """Return True if branch has custom commits
+    """
+    cli.run(['git', 'fetch', 'upstream', active_branch])
+    deviations = cli.run(['git', '--no-pager', 'log', f'upstream/{active_branch}...{active_branch}'])
+    return bool(deviations.returncode)
+
+
+def in_virtualenv():
+    """Check if running inside a virtualenv.
+    Based on https://stackoverflow.com/a/1883251
+    """
+    active_prefix = getattr(sys, "base_prefix", None) or getattr(sys, "real_prefix", None) or sys.prefix
+    return active_prefix != sys.prefix
diff --git a/lib/python/qmk/constants.py b/lib/python/qmk/constants.py
index 49e5e0eb4..1078f4ad5 100644
--- a/lib/python/qmk/constants.py
+++ b/lib/python/qmk/constants.py
@@ -6,11 +6,14 @@ from pathlib import Path
 # The root of the qmk_firmware tree.
 QMK_FIRMWARE = Path.cwd()
 
+# Upstream repo url
+QMK_FIRMWARE_UPSTREAM = 'qmk/qmk_firmware'
+
 # This is the number of directories under `qmk_firmware/keyboards` that will be traversed. This is currently a limitation of our make system.
 MAX_KEYBOARD_SUBFOLDERS = 5
 
 # Supported processor types
-CHIBIOS_PROCESSORS = 'cortex-m0', 'cortex-m0plus', 'cortex-m3', 'cortex-m4', 'MKL26Z64', 'MK20DX128', 'MK20DX256', 'MK66F18', 'STM32F042', 'STM32F072', 'STM32F103', 'STM32F303', 'STM32F401', 'STM32F411', 'STM32F446', 'STM32G431', 'STM32G474', 'STM32L433', 'STM32L443'
+CHIBIOS_PROCESSORS = 'cortex-m0', 'cortex-m0plus', 'cortex-m3', 'cortex-m4', 'MKL26Z64', 'MK20DX128', 'MK20DX256', 'MK66F18', 'STM32F042', 'STM32F072', 'STM32F103', 'STM32F303', 'STM32F401', 'STM32F411', 'STM32F446', 'STM32G431', 'STM32G474', 'STM32L412', 'STM32L422', 'STM32L433', 'STM32L443'
 LUFA_PROCESSORS = 'at90usb162', 'atmega16u2', 'atmega32u2', 'atmega16u4', 'atmega32u4', 'at90usb646', 'at90usb647', 'at90usb1286', 'at90usb1287', None
 VUSB_PROCESSORS = 'atmega32a', 'atmega328p', 'atmega328', 'attiny85'
 
diff --git a/lib/python/qmk/info.py b/lib/python/qmk/info.py
index 47c8bff7a..bcb4d81ef 100644
--- a/lib/python/qmk/info.py
+++ b/lib/python/qmk/info.py
@@ -9,7 +9,7 @@ from milc import cli
 
 from qmk.constants import CHIBIOS_PROCESSORS, LUFA_PROCESSORS, VUSB_PROCESSORS
 from qmk.c_parse import find_layouts
-from qmk.json_schema import deep_update, json_load, keyboard_validate, keyboard_api_validate
+from qmk.json_schema import deep_update, json_load, validate
 from qmk.keyboard import config_h, rules_mk
 from qmk.keymap import list_keymaps
 from qmk.makefile import parse_rules_mk_file
@@ -64,9 +64,12 @@ def info_json(keyboard):
     info_data = _extract_config_h(info_data)
     info_data = _extract_rules_mk(info_data)
 
+    # Ensure that we have matrix row and column counts
+    info_data = _matrix_size(info_data)
+
     # Validate against the jsonschema
     try:
-        keyboard_api_validate(info_data)
+        validate(info_data, 'qmk.api.keyboard.v1')
 
     except jsonschema.ValidationError as e:
         json_path = '.'.join([str(p) for p in e.absolute_path])
@@ -90,6 +93,9 @@ def info_json(keyboard):
         if layout_name not in info_data.get('layouts', {}) and layout_name not in info_data.get('layout_aliases', {}):
             _log_error(info_data, 'Claims to support community layout %s but no %s() macro found' % (layout, layout_name))
 
+    # Check that the reported matrix size is consistent with the actual matrix size
+    _check_matrix(info_data)
+
     return info_data
 
 
@@ -143,10 +149,7 @@ def _pin_name(pin):
     elif pin == 'NO_PIN':
         return None
 
-    elif pin[0] in 'ABCDEFGHIJK' and pin[1].isdigit():
-        return pin
-
-    raise ValueError(f'Invalid pin: {pin}')
+    return pin
 
 
 def _extract_pins(pins):
@@ -341,6 +344,46 @@ def _extract_rules_mk(info_data):
     return info_data
 
 
+def _matrix_size(info_data):
+    """Add info_data['matrix_size'] if it doesn't exist.
+    """
+    if 'matrix_size' not in info_data and 'matrix_pins' in info_data:
+        info_data['matrix_size'] = {}
+
+        if 'direct' in info_data['matrix_pins']:
+            info_data['matrix_size']['cols'] = len(info_data['matrix_pins']['direct'][0])
+            info_data['matrix_size']['rows'] = len(info_data['matrix_pins']['direct'])
+        elif 'cols' in info_data['matrix_pins'] and 'rows' in info_data['matrix_pins']:
+            info_data['matrix_size']['cols'] = len(info_data['matrix_pins']['cols'])
+            info_data['matrix_size']['rows'] = len(info_data['matrix_pins']['rows'])
+
+    return info_data
+
+
+def _check_matrix(info_data):
+    """Check the matrix to ensure that row/column count is consistent.
+    """
+    if 'matrix_pins' in info_data and 'matrix_size' in info_data:
+        actual_col_count = info_data['matrix_size'].get('cols', 0)
+        actual_row_count = info_data['matrix_size'].get('rows', 0)
+        col_count = row_count = 0
+
+        if 'direct' in info_data['matrix_pins']:
+            col_count = len(info_data['matrix_pins']['direct'][0])
+            row_count = len(info_data['matrix_pins']['direct'])
+        elif 'cols' in info_data['matrix_pins'] and 'rows' in info_data['matrix_pins']:
+            col_count = len(info_data['matrix_pins']['cols'])
+            row_count = len(info_data['matrix_pins']['rows'])
+
+        if col_count != actual_col_count and col_count != (actual_col_count / 2):
+            # FIXME: once we can we should detect if split is enabled to do the actual_col_count/2 check.
+            _log_error(info_data, f'MATRIX_COLS is inconsistent with the size of MATRIX_COL_PINS: {col_count} != {actual_col_count}')
+
+        if row_count != actual_row_count and row_count != (actual_row_count / 2):
+            # FIXME: once we can we should detect if split is enabled to do the actual_row_count/2 check.
+            _log_error(info_data, f'MATRIX_ROWS is inconsistent with the size of MATRIX_ROW_PINS: {row_count} != {actual_row_count}')
+
+
 def _merge_layouts(info_data, new_info_data):
     """Merge new_info_data into info_data in an intelligent way.
     """
@@ -493,7 +536,7 @@ def merge_info_jsons(keyboard, info_data):
             continue
 
         try:
-            keyboard_validate(new_info_data)
+            validate(new_info_data, 'qmk.keyboard.v1')
         except jsonschema.ValidationError as e:
             json_path = '.'.join([str(p) for p in e.absolute_path])
             cli.log.error('Not including data from file: %s', info_file)
diff --git a/lib/python/qmk/json_schema.py b/lib/python/qmk/json_schema.py
index 077dfcaa9..3e5663a29 100644
--- a/lib/python/qmk/json_schema.py
+++ b/lib/python/qmk/json_schema.py
@@ -24,9 +24,10 @@ def json_load(json_file):
 
 def load_jsonschema(schema_name):
     """Read a jsonschema file from disk.
-
-    FIXME(skullydazed/anyone): Refactor to make this a public function.
     """
+    if Path(schema_name).exists():
+        return json_load(schema_name)
+
     schema_path = Path(f'data/schemas/{schema_name}.jsonschema')
 
     if not schema_path.exists():
@@ -35,28 +36,33 @@ def load_jsonschema(schema_name):
     return json_load(schema_path)
 
 
-def keyboard_validate(data):
-    """Validates data against the keyboard jsonschema.
+def create_validator(schema):
+    """Creates a validator for the given schema id.
     """
-    schema = load_jsonschema('keyboard')
-    validator = jsonschema.Draft7Validator(schema).validate
+    schema_store = {}
 
-    return validator(data)
+    for schema_file in Path('data/schemas').glob('*.jsonschema'):
+        schema_data = load_jsonschema(schema_file)
+        if not isinstance(schema_data, dict):
+            cli.log.debug('Skipping schema file %s', schema_file)
+            continue
+        schema_store[schema_data['$id']] = schema_data
+
+    resolver = jsonschema.RefResolver.from_schema(schema_store['qmk.keyboard.v1'], store=schema_store)
+
+    return jsonschema.Draft7Validator(schema_store[schema], resolver=resolver).validate
 
 
-def keyboard_api_validate(data):
-    """Validates data against the api_keyboard jsonschema.
+def validate(data, schema):
+    """Validates data against a schema.
     """
-    base = load_jsonschema('keyboard')
-    relative = load_jsonschema('api_keyboard')
-    resolver = jsonschema.RefResolver.from_schema(base)
-    validator = jsonschema.Draft7Validator(relative, resolver=resolver).validate
+    validator = create_validator(schema)
 
     return validator(data)
 
 
 def deep_update(origdict, newdict):
-    """Update a dictionary in place, recursing to do a deep copy.
+    """Update a dictionary in place, recursing to do a depth-first deep copy.
     """
     for key, value in newdict.items():
         if isinstance(value, Mapping):
diff --git a/lib/python/qmk/tests/test_cli_commands.py b/lib/python/qmk/tests/test_cli_commands.py
index afdbc8142..b341e1c91 100644
--- a/lib/python/qmk/tests/test_cli_commands.py
+++ b/lib/python/qmk/tests/test_cli_commands.py
@@ -258,6 +258,12 @@ def test_generate_rules_mk():
     assert 'MCU ?= atmega32u4' in result.stdout
 
 
+def test_generate_version_h():
+    result = check_subcommand('generate-version-h')
+    check_returncode(result)
+    assert '#define QMK_VERSION' in result.stdout
+
+
 def test_generate_layouts():
     result = check_subcommand('generate-layouts', '-kb', 'handwired/pytest/basic')
     check_returncode(result)