Added more indexed types.

[didex.git] / didex / index.py

import struct, contextlib, math
from . import db, lib
from .db import bd, txnfun

deadlock = bd.DBLockDeadlockError
notfound = bd.DBNotFoundError

class simpletype(object):
    def __init__(self, encode, decode):
        self.enc = encode
        self.dec = decode

    def encode(self, ob):
        return self.enc(ob)
    def decode(self, dat):
        return self.dec(dat)
    def compare(self, a, b):
        if a < b:
            return -1
        elif a > b:
            return 1
        else:
            return 0

    @classmethod
    def struct(cls, fmt):
        return cls(lambda ob: struct.pack(fmt, ob),
                   lambda dat: struct.unpack(fmt, dat)[0])

class maybe(object):
    def __init__(self, bk):
        self.bk = bk

    def encode(self, ob):
        if ob is None: return b""
        return b"\0" + self.bk.encode(ob)
    def decode(self, dat):
        if dat == b"": return None
        return self.bk.dec(dat[1:])
    def compare(self, a, b):
        if a is b is None:
            return 0
        elif a is None:
            return -1
        elif b is None:
            return 1
        else:
            return self.bk.compare(a[1:], b[1:])

def floatcmp(a, b):
    if math.isnan(a) and math.isnan(b):
        return 0
    elif math.isnan(a):
        return -1
    elif math.isnan(b):
        return 1
    elif a < b:
        return -1
    elif a > b:
        return 1
    else:
        return 0

t_int = simpletype.struct(">q")
t_uint = simpletype.struct(">Q")
t_float = simpletype.struct(">d")
t_float.compare = floatcmp
t_str = simpletype((lambda ob: ob.encode("utf-8")), (lambda dat: dat.decode("utf-8")))

class index(object):
    def __init__(self, db, name, datatype):
        self.db = db
        self.nm = name
        self.typ = datatype

missing = object()

class ordered(index, lib.closable):
    def __init__(self, db, name, datatype, create=True):
        super().__init__(db, name, datatype)
        fl = bd.DB_THREAD | bd.DB_AUTO_COMMIT
        if create: fl |= bd.DB_CREATE
        def initdb(db):
            def compare(a, b):
                if a == b == "": return 0
                return self.typ.compare(self.typ.decode(a), self.typ.decode(b))
            db.set_flags(bd.DB_DUPSORT)
            db.set_bt_compare(compare)
        self.bk = db._opendb("i-" + name, bd.DB_BTREE, fl, initdb)
        self.bk.set_get_returns_none(False)

    def close(self):
        self.bk.close()

    class cursor(lib.closable):
        def __init__(self, idx, cur, item, stop):
            self.idx = idx
            self.cur = cur
            self.item = item
            self.stop = stop

        def close(self):
            if self.cur is not None:
                self.cur.close()

        def __iter__(self):
            return self

        def peek(self):
            if self.item is None:
                raise StopIteration()
            rk, rv = self.item
            rk = self.idx.typ.decode(rk)
            rv = struct.unpack(">Q", rv)[0]
            if self.stop(rk):
                self.item = None
                raise StopIteration()
            return rk, rv

        def __next__(self):
            rk, rv = self.peek()
            try:
                while True:
                    try:
                        self.item = self.cur.next()
                        break
                    except deadlock:
                        continue
            except notfound:
                self.item = None
            return rk, rv

        def skip(self, n=1):
            try:
                for i in range(n):
                    next(self)
            except StopIteration:
                return

    def get(self, *, match=missing, ge=missing, gt=missing, lt=missing, le=missing, all=False):
        while True:
            try:
                cur = self.bk.cursor()
                done = False
                try:
                    if match is not missing:
                        try:
                            k, v = cur.set(self.typ.encode(match))
                        except notfound:
                            return self.cursor(None, None, None, None)
                        else:
                            done = True
                            return self.cursor(self, cur, (k, v), lambda o: (self.typ.compare(o, match) != 0))
                    elif all:
                        try:
                            k, v = cur.first()
                        except notfound:
                            return self.cursor(None, None, None, None)
                        else:
                            done = True
                            return self.cursor(self, cur, (k, v), lambda o: False)
                    elif ge is not missing or gt is not missing or lt is not missing or le is not missing:
                        skip = False
                        try:
                            if ge is not missing:
                                k, v = cur.set_range(self.typ.encode(ge))
                            elif gt is not missing:
                                k, v = cur.set_range(self.typ.encode(gt))
                                skip = True
                            else:
                                k, v = cur.first()
                        except notfound:
                            return self.cursor(None, None, None, None)
                        if lt is not missing:
                            stop = lambda o: self.typ.compare(o, lt) >= 0
                        elif le is not missing:
                            stop = lambda o: self.typ.compare(o, le) > 0
                        else:
                            stop = lambda o: False
                        ret = self.cursor(self, cur, (k, v), stop)
                        if skip:
                            try:
                                while self.typ.compare(ret.peek()[0], gt) == 0:
                                    next(ret)
                            except StopIteration:
                                pass
                        done = True
                        return ret
                    else:
                        raise NameError("invalid get() specification")
                finally:
                    if not done:
                        cur.close()
            except deadlock:
                continue

    @txnfun(lambda self: self.db.env.env)
    def put(self, key, id, *, tx):
        obid = struct.pack(">Q", id)
        if not self.db.ob.has_key(obid, txn=tx.tx):
            raise ValueError("no such object in database: " + str(id))
        try:
            self.bk.put(self.typ.encode(key), obid, txn=tx.tx, flags=bd.DB_NODUPDATA)
        except bd.DBKeyExistError:
            return False
        return True

    @txnfun(lambda self: self.db.env.env)
    def remove(self, key, id, *, tx):
        obid = struct.pack(">Q", id)
        if not self.db.ob.has_key(obid, txn=tx.tx):
            raise ValueError("no such object in database: " + str(id))
        cur = self.bk.cursor(txn=tx.tx)
        try:
            try:
                cur.get_both(self.typ.encode(key), obid)
            except notfound:
                return False
            cur.delete()
        finally:
            cur.close()
        return True