#!/usr/bin/python3
# Check that xpopple's code and documentation are consistent.

import os
import re
import subprocess
import sys

testdir = "testdata"
old_versions = ("3.03", "3.04")

# Check we're in the right dir
assert os.path.exists("test-options")

# Download the old source if it's not already there
old_srcdirs = {}
for version in old_versions:
    distfile = "xpdf-" + version + ".tar.gz"
    tarball = os.path.join(testdir, "archives", distfile)
    srcdir = os.path.join(testdir, "src", "xpdf-" + version)
    old_srcdirs[version] = srcdir

    os.makedirs(os.path.dirname(tarball), exist_ok=True)
    if not os.path.exists(tarball):
        subprocess.check_call(["wget", "-O", tarball,
                               "https://dl.xpdfreader.com/old/" + distfile])

    os.makedirs(os.path.dirname(srcdir), exist_ok=True)
    if not os.path.exists(srcdir):
        subprocess.check_call(["tar", "-C", os.path.dirname(srcdir),
                               "-xf", tarball])

class XPDFTree:
    """Information extracted from an Xpdf/xpopple source tree."""

    def __init__(self, dirname):
        self.dirname = dirname

        # Command-line option arguments
        self.code_args = set()
        self.code_args_obsolete = set()
        self.man_args = set()
        self.man_args_obsolete = set()

        # X resources
        self.code_resources = set()
        self.man_resources = set()

        # Config file options
        self.code_config = set()
        self.code_config_obsolete = set()
        self.man_config = set()
        self.man_config_obsolete = set()

        # Remote commands
        self.code_commands = set()
        self.man_commands = set()

        # Key bindings (only, since these are documented individually)
        self.code_keys = set()
        self.man_keys = set()

        # Default bindings
        self.code_bindings = set()
        self.man_bindings = set()

        self.read_xpdf_code()
        self.read_app_code()
        self.read_params_code()
        self.read_viewer_code()
        self.read_xpdf_man()
        self.read_xpdfrc_man()

        # code_keys are just the keys that have something bound to them
        for binding in self.code_bindings:
            key = binding.split()[1]
            if not key.startswith("mouse"):
                self.code_keys.add(key)

    # Each of these read_ methods is a little state machine that identifies and
    # searches the relevant sections of a source file or man page. It's not a
    # proper parser but it'll do the job!

    def read_xpdf_code(self):
        with open(os.path.join(self.dirname, "xpdf", "xpdf.cc")) as f:
            state = "before"
            for line in f.readlines():
                line = line.rstrip()

                if state == "before":
                    if re.search(r'argDesc\[\] =', line):
                        state = "tab"

                elif state == "tab":
                    m = re.search(r'^ *{ *"([^"]*)", *([^ ,]*)', line)
                    if m:
                        if m.group(2).startswith("argObsolete"):
                            self.code_args_obsolete.add(m.group(1))
                        else:
                            self.code_args.add(m.group(1))
                    if re.search(r'^}', line):
                        break

        assert self.code_args != set()
        # Old versions didn't handle obsolete args
        if self.dirname == ".":
            assert self.code_args_obsolete != set()

    def read_app_code(self):
        args = set()

        with open(os.path.join(self.dirname, "xpdf", "XPDFApp.cc")) as f:
            state = "before"
            for line in f.readlines():
                line = line.rstrip()

                if state == "before":
                    if re.search(r'xOpts\[\] =', line):
                        state = "tab"
                    if re.search(r'xResources\[\] =', line):
                        state = "resources"

                elif state == "tab":
                    m = re.search(r'^ *{ *"([^"]*)", *"[.*]*([^"]*)",', line)
                    if m:
                        args.add(m.group(1))
                        self.code_resources.add(m.group(2))
                    if re.search(r'^}', line):
                        state = "before"

                elif state == "resources":
                    m = re.search(r'^ *{ *"([^"]*)",', line)
                    if m:
                        self.code_resources.add(m.group(1))
                    if re.search(r'^}', line):
                        state = "before"

        assert args != set()
        self.code_args.update(args)
        assert self.code_resources != set()

    def read_params_code(self):
        fn = os.path.join(self.dirname, "xpdf", "XPDFParams.cc")
        if not os.path.exists(fn):
            fn = os.path.join(self.dirname, "xpdf", "GlobalParams.cc")

        with open(fn) as f:
            state = "before"
            lines = ""
            for line in f.readlines():
                line = line.rstrip()

                if state == "before":
                    if re.search(r'::createDefaultKeyBindings', line):
                        state = "bindings"
                    if re.search(r'::parseLine', line):
                        state = "config"

                elif state == "bindings":
                    # Bindings are broken across multiple lines.
                    if re.search(r'^ *//', line):
                        pass
                    elif re.search(r'^}', line):
                        state = "before"
                    else:
                        lines += " " + line
                        if re.search(r';$', lines):
                            m = re.search(r'(?:new KeyBinding|emplace_back)\((.*")\)', lines)
                            if m:
                                binding = convert_binding(m.group(1))
                                self.code_bindings.add(binding)
                            lines = ""

                elif state == "config":
                    for s in re.findall(r'(?:cmd->cmp\(|cmd == )"([^"]*)"', line):
                        self.code_config.add(s)
                    if re.search(r'error.*Unknown config file command', line):
                        state = "obsolete"

                elif state == "obsolete":
                    for s in re.findall(r'(?:cmd->cmp\(|cmd == )"([^"]*)"', line):
                        self.code_config_obsolete.add(s)
                    if re.search(r'^\}', line):
                        state = "before"

        assert self.code_config != set()
        assert self.code_config_obsolete != set()

    def read_viewer_code(self):
        with open(os.path.join(self.dirname, "xpdf", "XPDFViewer.cc")) as f:
            state = "before"
            for line in f.readlines():
                line = line.rstrip()

                if state == "before":
                    if re.search(r'::cmdTab\[\] =', line):
                        state = "tab"

                elif state == "tab":
                    m = re.search(r'^ *{ *"([^"]*)",', line)
                    if m:
                        self.code_commands.add(m.group(1))
                    if re.search(r'^}', line):
                        break

        assert self.code_commands != set()

    def read_xpdf_man(self):
        with open(os.path.join(self.dirname, "doc", "xpdf.1")) as f:
            state = "before"
            for line in f.readlines():
                line = line.rstrip()

                if state == "before":
                    if re.search(r'^\.SH OPTIONS', line):
                        state = "args"
                    if re.search(r'^\.SS Key bindings', line):
                        state = "keys"
                    if re.search(r'^\.SH COMMANDS', line):
                        state = "commands"

                elif state == "args":
                    m = re.search(r'^(?:\.BI?|\.RB \() \\(-[^ ]*)', line)
                    if m:
                        self.man_args.add(m.group(1).replace('\\', ''))
                    m = re.search(r'config file: " ([^ ]*)', line)
                    if m:
                        self.man_config.add(m.group(1))
                    m = re.search(r'(?:X resource: "|^\.B) [Xx]pdf[.*]([^ ]*)', line)
                    # toolTipEnable is handled by Motif.
                    if m and m.group(1) != "toolTipEnable":
                        self.man_resources.add(m.group(1))
                    if re.search(r'^\.SH OBSOLETE OPTIONS', line):
                        state = "obsolete"
                    elif re.search(r'^\.SH', line):
                        state = "before"

                elif state == "obsolete":
                    m = re.search(r'^(?:\.BI?|\.RB \() \\(-[^ ]*)', line)
                    if m:
                        self.man_args_obsolete.add(m.group(1).replace('\\', ''))
                    if re.search(r'^\.SH', line):
                        state = "before"

                elif state == "keys":
                    m = re.search(r'^\.BR? (.*)$', line)
                    if m:
                        for key in m.group(1).split(' " or " '):
                            # Convert the key name to bind command form
                            if key.startswith("control-"):
                                key = "ctrl-" + key[8:]
                            i = key.find("<")
                            if i != -1:
                                key = key[:i] + key[i + 1:-1]
                            key = key.lower()
                            if key == "pagedown":
                                key = "pgdn"
                            if key == "pageup":
                                key = "pgup"
                            if key == "arrows":
                                for key in ("up", "down", "left", "right"):
                                    self.man_keys.add(key)
                            else:
                                self.man_keys.add(key)
                                if len(key) == 1:
                                    # Letters should be accepted in
                                    # both lower and upper case
                                    self.man_keys.add(key.upper())
                    if re.search(r'^\.SH', line):
                        state = "before"

                elif state == "commands":
                    m = re.search(r'^\.BI? ([^ (]*)', line)
                    if m:
                        self.man_commands.add(m.group(1))
                    if re.search(r'^\.SS Default Bindings', line):
                        state = "bindings"

                elif state == "bindings":
                    m = re.search(r'^ *(bind .*)$', line)
                    if m:
                        binding = m.group(1)
                        binding = binding.replace("\\&", "")
                        binding = re.sub(r' +', r' ', binding)
                        self.man_bindings.add(binding)
                    if re.search(r'Previous versions of xpdf', line):
                        state = "before"

        assert self.man_args != set()
        assert self.man_resources != set()
        assert self.man_commands != set()
        assert self.man_config != set()
        assert self.man_keys != set()
        assert self.man_bindings != set()

    def read_xpdfrc_man(self):
        config = set()

        with open(os.path.join(self.dirname, "doc", "xpdfrc.5")) as f:
            state = "before"
            for line in f.readlines():
                line = line.rstrip()

                if state == "before":
                    if re.search(r'^The following sections list all', line):
                        state = "config"

                elif state == "config":
                    m = re.search(r'^\.B[IR]? ([^ ]*)', line)
                    if m and m.group(1) != "xpdf":
                        config.add(m.group(1))
                    if re.search(r'^\.SH OBSOLETE CONFIG', line):
                        state = "obsolete"
                    if re.search(r'^\.SH EXAMPLES', line):
                        break

                elif state == "obsolete":
                    m = re.search(r'^\.B[IR]? ([^ ]*)', line)
                    if m:
                        self.man_config_obsolete.add(m.group(1))
                    if re.search(r'^\.SH EXAMPLES', line):
                        break

        assert config != set()
        self.man_config.update(config)
        # Old versions didn't list obsolete options
        if self.dirname == ".":
            assert self.man_config_obsolete != set()

def convert_binding(arg_string):
    """Convert KeyBinding constructor args to a bind command."""
    args = [arg.strip() for arg in arg_string.split(",")]

    def lower_first(s):
        return s[:1].lower() + s[1:]
    def unquote(s, quote):
        assert s[0] == quote and s[-1] == quote
        return s[1:-1]

    cmd = ["bind"]

    if args[0].startswith("xpdfKeyCode"):
        key = lower_first(args[0][11:])
        if key in ("pgUp", "pgDn"):
            key = key.lower()
    else:
        key = unquote(args[0], "'")
        if key == " ":
            key = "space"

    assert args[1].startswith("xpdfKeyMod")
    mod = lower_first(args[1][10:])
    if mod != "none":
        key = mod + "-" + key
    cmd.append(key)

    assert args[2].startswith("xpdfKeyContext")
    cmd.append(lower_first(args[2][14:]))

    for arg in args[3:]:
        cmd.append(unquote(arg, '"'))

    return " ".join(cmd)

rc = 0
def warn(*s):
    """Complain about something and ensure we exit non-zero."""
    print(*s, file=sys.stderr)
    global rc
    rc = 1

# The lists of documentable things we've gathered above
item_types = {
    "args": "Command-line arg",
    "args_obsolete": "Obsolete command-line arg",
    "resources": "X resource",
    "config": "Config file option",
    "config_obsolete": "Obsolete config file option",
    "commands": "Command",
    "keys": "Key",
    "bindings": "Default binding",
    }

# Check current documentation is consistent with code
current = XPDFTree(".")
for item_type, desc in sorted(item_types.items()):
    from_code = getattr(current, "code_" + item_type)
    from_man = getattr(current, "man_" + item_type)
    for item in sorted(from_code):
        if not item in from_man:
            warn(desc + " in code but not man:", item)
    for item in sorted(from_man):
        if not item in from_code:
            warn(desc + " in man but not code:", item)

# We don't care about bindings being to the same commands as older versions;
# checking keys will tell us if any have been removed.
del item_types["bindings"]

# Check code is consistent with older versions of the code
all_items = {}
for version, srcdir in old_srcdirs.items():
    old = XPDFTree(srcdir)
    for item_type, desc in item_types.items():
        from_old = all_items.setdefault(item_type, set())
        from_old.update(getattr(old, "code_" + item_type))
for item_type, desc in item_types.items():
    if item_type.endswith("_obsolete"):
        continue
    from_code = getattr(current, "code_" + item_type)
    from_obsolete = getattr(current, "code_" + item_type + "_obsolete", set())
    from_old = all_items[item_type]
    for item in sorted(from_old):
        # Each item must either be implemented, or be documented as obsolete
        # (and we checked the documentation matched the code above)
        if not (item in from_code or item in from_obsolete):
            warn(desc + " in old code but not current code:", item)

sys.exit(rc)