dont use unsafeNew in ropes.nim for now

This commit is contained in:
Araq 2012-12-02 11:11:54 +01:00
commit d7adc7c328
3 changed files with 497 additions and 231 deletions

View file

@ -66,54 +66,49 @@ type
# performance of the code generator (assignments # performance of the code generator (assignments
# copy the format strings # copy the format strings
# though it is not necessary) # though it is not necessary)
PRope* = ref TRope
TRope*{.acyclic.} = object of TObject # the empty rope is represented
# by nil to safe space
left, right: PRope
L: int # < 0 if a leaf
d: array [0..PayloadSize, char] # != nil if a leaf
TRopeSeq* = seq[PRope]
proc con*(a, b: PRope): PRope when true:
proc con*(a: PRope, b: string): PRope # working version:
proc con*(a: string, b: PRope): PRope type
proc con*(a: varargs[PRope]): PRope PRope* = ref TRope
proc app*(a: var PRope, b: PRope) TRope*{.acyclic.} = object of TObject # the empty rope is represented
proc app*(a: var PRope, b: string) # by nil to safe space
proc prepend*(a: var PRope, b: PRope) left*, right*: PRope
proc toRope*(s: string): PRope length*: int
proc toRope*(i: BiggestInt): PRope data*: string # != nil if a leaf
proc ropeLen*(a: PRope): int
proc writeRopeIfNotEqual*(r: PRope, filename: string): bool TRopeSeq* = seq[PRope]
proc ropeToStr*(p: PRope): string
proc ropef*(frmt: TFormatStr, args: varargs[PRope]): PRope
proc appf*(c: var PRope, frmt: TFormatStr, args: varargs[PRope])
proc RopeEqualsFile*(r: PRope, f: string): bool
# returns true if the rope r is the same as the contents of file f
proc RopeInvariant*(r: PRope): bool
# exported for debugging
# implementation
proc ropeLen(a: PRope): int = proc con*(a, b: PRope): PRope
if a == nil: result = 0 proc con*(a: PRope, b: string): PRope
else: result = a.L.abs proc con*(a: string, b: PRope): PRope
proc con*(a: varargs[PRope]): PRope
proc newRope(data: string = nil): PRope = proc app*(a: var PRope, b: PRope)
if data != nil: proc app*(a: var PRope, b: string)
unsafeNew(result, sizeof(TRope)-PayloadSize+len(data)) proc prepend*(a: var PRope, b: PRope)
result.L = -len(data) proc toRope*(s: string): PRope
# copy including '\0': proc toRope*(i: BiggestInt): PRope
copyMem(addr result.d, cstring(data), len(data)) proc ropeLen*(a: PRope): int
else: proc writeRopeIfNotEqual*(r: PRope, filename: string): bool
unsafeNew(result, sizeof(TRope)-PayloadSize) proc ropeToStr*(p: PRope): string
proc ropef*(frmt: TFormatStr, args: varargs[PRope]): PRope
proc appf*(c: var PRope, frmt: TFormatStr, args: varargs[PRope])
proc RopeEqualsFile*(r: PRope, f: string): bool
# returns true if the rope r is the same as the contents of file f
proc RopeInvariant*(r: PRope): bool
# exported for debugging
# implementation
proc eqContent(r: PRope, s: string): bool = proc ropeLen(a: PRope): int =
assert r.L < 0 if a == nil: result = 0
if -r.L == s.len: else: result = a.length
result = equalMem(addr(r.d), cstring(s), s.len)
proc newRope(data: string = nil): PRope =
new(result)
if data != nil:
result.length = len(data)
result.data = data
when false:
proc newMutableRope*(capacity = 30): PRope = proc newMutableRope*(capacity = 30): PRope =
## creates a new rope that supports direct modifications of the rope's ## creates a new rope that supports direct modifications of the rope's
## 'data' and 'length' fields. ## 'data' and 'length' fields.
@ -123,204 +118,476 @@ when false:
proc freezeMutableRope*(r: PRope) {.inline.} = proc freezeMutableRope*(r: PRope) {.inline.} =
r.length = r.data.len r.length = r.data.len
var var
cache: array[0..2048*2 -1, PRope] cache: array[0..2048*2 -1, PRope]
proc RopeInvariant(r: PRope): bool = proc RopeInvariant(r: PRope): bool =
if r == nil: if r == nil:
result = true result = true
else: else:
result = true # result = true #
# if r.data <> snil then # if r.data <> snil then
# result := true # result := true
# else begin # else begin
# result := (r.left <> nil) and (r.right <> nil); # result := (r.left <> nil) and (r.right <> nil);
# if result then result := ropeInvariant(r.left); # if result then result := ropeInvariant(r.left);
# if result then result := ropeInvariant(r.right); # if result then result := ropeInvariant(r.right);
# end # end
proc insertInCache(s: string): PRope = proc insertInCache(s: string): PRope =
var h = hash(s) and high(cache) var h = hash(s) and high(cache)
result = cache[h] result = cache[h]
if isNil(result) or not eqContent(result, s): if isNil(result) or result.data != s:
result = newRope(s) result = newRope(s)
cache[h] = result cache[h] = result
proc toRope(s: string): PRope = proc toRope(s: string): PRope =
if s.len == 0: if s.len == 0:
result = nil result = nil
else: else:
result = insertInCache(s) result = insertInCache(s)
assert(RopeInvariant(result)) assert(RopeInvariant(result))
proc RopeSeqInsert(rs: var TRopeSeq, r: PRope, at: Natural) = proc RopeSeqInsert(rs: var TRopeSeq, r: PRope, at: Natural) =
var length = len(rs) var length = len(rs)
if at > length: if at > length:
setlen(rs, at + 1) setlen(rs, at + 1)
else: else:
setlen(rs, length + 1) # move old rope elements: setlen(rs, length + 1) # move old rope elements:
for i in countdown(length, at + 1): for i in countdown(length, at + 1):
rs[i] = rs[i - 1] # this is correct, I used pen and paper to validate it rs[i] = rs[i - 1] # this is correct, I used pen and paper to validate it
rs[at] = r rs[at] = r
proc newRecRopeToStr(result: var string, resultLen: var int, r: PRope) = proc newRecRopeToStr(result: var string, resultLen: var int, r: PRope) =
var stack = @[r] var stack = @[r]
while len(stack) > 0: while len(stack) > 0:
var it = pop(stack) var it = pop(stack)
while it.L >= 0: while it.data == nil:
add(stack, it.right) add(stack, it.right)
it = it.left it = it.left
assert(it.L < 0) assert(it.data != nil)
CopyMem(addr(result[resultLen]), addr(it.d[0]), -it.L) CopyMem(addr(result[resultLen]), addr(it.data[0]), it.length)
Inc(resultLen, -it.L) Inc(resultLen, it.length)
assert(resultLen <= len(result)) assert(resultLen <= len(result))
proc ropeToStr(p: PRope): string = proc ropeToStr(p: PRope): string =
if p == nil: if p == nil:
result = "" result = ""
else: else:
result = newString(p.L.abs) result = newString(p.length)
var resultLen = 0 var resultLen = 0
newRecRopeToStr(result, resultLen, p) newRecRopeToStr(result, resultLen, p)
proc con(a, b: PRope): PRope = proc con(a, b: PRope): PRope =
if a == nil: result = b if a == nil: result = b
elif b == nil: result = a elif b == nil: result = a
else: else:
result = newRope() result = newRope()
result.L = a.L.abs + b.L.abs result.length = a.length + b.length
result.left = a result.left = a
result.right = b result.right = b
proc con(a: PRope, b: string): PRope = result = con(a, toRope(b)) proc con(a: PRope, b: string): PRope = result = con(a, toRope(b))
proc con(a: string, b: PRope): PRope = result = con(toRope(a), b) proc con(a: string, b: PRope): PRope = result = con(toRope(a), b)
proc con(a: varargs[PRope]): PRope = proc con(a: varargs[PRope]): PRope =
for i in countup(0, high(a)): result = con(result, a[i]) for i in countup(0, high(a)): result = con(result, a[i])
proc toRope(i: BiggestInt): PRope = result = toRope($i) proc toRope(i: BiggestInt): PRope = result = toRope($i)
proc app(a: var PRope, b: PRope) = a = con(a, b) proc app(a: var PRope, b: PRope) = a = con(a, b)
proc app(a: var PRope, b: string) = a = con(a, b) proc app(a: var PRope, b: string) = a = con(a, b)
proc prepend(a: var PRope, b: PRope) = a = con(b, a) proc prepend(a: var PRope, b: PRope) = a = con(b, a)
proc writeRope*(f: TFile, c: PRope) = proc writeRope*(f: TFile, c: PRope) =
var stack = @[c] var stack = @[c]
while len(stack) > 0: while len(stack) > 0:
var it = pop(stack) var it = pop(stack)
while it.L >= 0: while it.data == nil:
add(stack, it.right) add(stack, it.right)
it = it.left it = it.left
assert(it != nil) assert(it != nil)
assert(it.L < 0) assert(it.data != nil)
write(f, cstring(it.d), -it.L) write(f, it.data)
proc WriteRope*(head: PRope, filename: string, useWarning = false) = proc WriteRope*(head: PRope, filename: string, useWarning = false) =
var f: tfile var f: tfile
if open(f, filename, fmWrite): if open(f, filename, fmWrite):
if head != nil: WriteRope(f, head) if head != nil: WriteRope(f, head)
close(f) close(f)
else: else:
rawMessage(if useWarning: warnCannotOpenFile else: errCannotOpenFile, rawMessage(if useWarning: warnCannotOpenFile else: errCannotOpenFile,
filename) filename)
proc ropef(frmt: TFormatStr, args: varargs[PRope]): PRope = proc ropef(frmt: TFormatStr, args: varargs[PRope]): PRope =
var i = 0
var length = len(frmt)
result = nil
var num = 0
while i <= length - 1:
if frmt[i] == '$':
inc(i) # skip '$'
case frmt[i]
of '$':
app(result, "$")
inc(i)
of '#':
inc(i)
app(result, args[num])
inc(num)
of '0'..'9':
var j = 0
while true:
j = (j * 10) + Ord(frmt[i]) - ord('0')
inc(i)
if (i > length + 0 - 1) or not (frmt[i] in {'0'..'9'}): break
num = j
if j > high(args) + 1:
internalError("ropes: invalid format string $" & $(j))
else:
app(result, args[j - 1])
of 'n':
if optLineDir notin gOptions: app(result, tnl)
inc i
of 'N':
app(result, tnl)
inc(i)
else: InternalError("ropes: invalid format string $" & frmt[i])
var start = i
while i < length:
if frmt[i] != '$': inc(i)
else: break
if i - 1 >= start:
app(result, substr(frmt, start, i - 1))
assert(RopeInvariant(result))
proc appf(c: var PRope, frmt: TFormatStr, args: varargs[PRope]) =
app(c, ropef(frmt, args))
const
bufSize = 1024 # 1 KB is reasonable
proc auxRopeEqualsFile(r: PRope, bin: var tfile, buf: Pointer): bool =
if r.L < 0:
if -r.L > bufSize:
internalError("ropes: token too long")
return
var readBytes = readBuffer(bin, buf, -r.L)
result = readBytes == -r.L and
equalMem(buf, addr(r.d[0]), readBytes)
else:
result = auxRopeEqualsFile(r.left, bin, buf)
if result: result = auxRopeEqualsFile(r.right, bin, buf)
proc RopeEqualsFile(r: PRope, f: string): bool =
var bin: tfile
result = open(bin, f)
if not result:
return # not equal if file does not exist
var buf = alloc(BufSize)
result = auxRopeEqualsFile(r, bin, buf)
if result:
result = readBuffer(bin, buf, bufSize) == 0 # really at the end of file?
dealloc(buf)
close(bin)
proc newCrcFromRopeAux(r: PRope, startVal: TCrc32): TCrc32 =
# XXX profiling shows this is actually expensive
var stack: TRopeSeq = @[r]
result = startVal
while len(stack) > 0:
var it = pop(stack)
while it.L >= 0:
add(stack, it.right)
it = it.left
assert(it.L < 0)
var i = 0 var i = 0
var L = -it.L var length = len(frmt)
while i < L: result = nil
result = updateCrc32(it.d[i], result) var num = 0
inc(i) while i <= length - 1:
if frmt[i] == '$':
inc(i) # skip '$'
case frmt[i]
of '$':
app(result, "$")
inc(i)
of '#':
inc(i)
app(result, args[num])
inc(num)
of '0'..'9':
var j = 0
while true:
j = (j * 10) + Ord(frmt[i]) - ord('0')
inc(i)
if (i > length + 0 - 1) or not (frmt[i] in {'0'..'9'}): break
num = j
if j > high(args) + 1:
internalError("ropes: invalid format string $" & $(j))
else:
app(result, args[j - 1])
of 'n':
if optLineDir notin gOptions: app(result, tnl)
inc i
of 'N':
app(result, tnl)
inc(i)
else: InternalError("ropes: invalid format string $" & frmt[i])
var start = i
while i < length:
if frmt[i] != '$': inc(i)
else: break
if i - 1 >= start:
app(result, substr(frmt, start, i - 1))
assert(RopeInvariant(result))
proc crcFromRope(r: PRope): TCrc32 = proc appf(c: var PRope, frmt: TFormatStr, args: varargs[PRope]) =
result = newCrcFromRopeAux(r, initCrc32) app(c, ropef(frmt, args))
proc writeRopeIfNotEqual(r: PRope, filename: string): bool = const
# returns true if overwritten bufSize = 1024 # 1 KB is reasonable
var c: TCrc32
c = crcFromFile(filename) proc auxRopeEqualsFile(r: PRope, bin: var tfile, buf: Pointer): bool =
if c != crcFromRope(r): if r.data != nil:
writeRope(r, filename) if r.length > bufSize:
result = true internalError("ropes: token too long")
else: return
result = false var readBytes = readBuffer(bin, buf, r.length)
result = readBytes == r.length and
equalMem(buf, addr(r.data[0]), r.length) # BUGFIX
else:
result = auxRopeEqualsFile(r.left, bin, buf)
if result: result = auxRopeEqualsFile(r.right, bin, buf)
proc RopeEqualsFile(r: PRope, f: string): bool =
var bin: tfile
result = open(bin, f)
if not result:
return # not equal if file does not exist
var buf = alloc(BufSize)
result = auxRopeEqualsFile(r, bin, buf)
if result:
result = readBuffer(bin, buf, bufSize) == 0 # really at the end of file?
dealloc(buf)
close(bin)
proc crcFromRopeAux(r: PRope, startVal: TCrc32): TCrc32 =
if r.data != nil:
result = startVal
for i in countup(0, len(r.data) - 1):
result = updateCrc32(r.data[i], result)
else:
result = crcFromRopeAux(r.left, startVal)
result = crcFromRopeAux(r.right, result)
proc newCrcFromRopeAux(r: PRope, startVal: TCrc32): TCrc32 =
# XXX profiling shows this is actually expensive
var stack: TRopeSeq = @[r]
result = startVal
while len(stack) > 0:
var it = pop(stack)
while it.data == nil:
add(stack, it.right)
it = it.left
assert(it.data != nil)
var i = 0
var L = len(it.data)
while i < L:
result = updateCrc32(it.data[i], result)
inc(i)
proc crcFromRope(r: PRope): TCrc32 =
result = newCrcFromRopeAux(r, initCrc32)
proc writeRopeIfNotEqual(r: PRope, filename: string): bool =
# returns true if overwritten
var c: TCrc32
c = crcFromFile(filename)
if c != crcFromRope(r):
writeRope(r, filename)
result = true
else:
result = false
else:
# optimized but broken version:
type
PRope* = ref TRope
TRope*{.acyclic.} = object of TObject # the empty rope is represented
# by nil to safe space
left, right: PRope
L: int # < 0 if a leaf
d: array [0..PayloadSize, char] # != nil if a leaf
TRopeSeq* = seq[PRope]
proc con*(a, b: PRope): PRope
proc con*(a: PRope, b: string): PRope
proc con*(a: string, b: PRope): PRope
proc con*(a: varargs[PRope]): PRope
proc app*(a: var PRope, b: PRope)
proc app*(a: var PRope, b: string)
proc prepend*(a: var PRope, b: PRope)
proc toRope*(s: string): PRope
proc toRope*(i: BiggestInt): PRope
proc ropeLen*(a: PRope): int
proc writeRopeIfNotEqual*(r: PRope, filename: string): bool
proc ropeToStr*(p: PRope): string
proc ropef*(frmt: TFormatStr, args: varargs[PRope]): PRope
proc appf*(c: var PRope, frmt: TFormatStr, args: varargs[PRope])
proc RopeEqualsFile*(r: PRope, f: string): bool
# returns true if the rope r is the same as the contents of file f
proc RopeInvariant*(r: PRope): bool
# exported for debugging
# implementation
proc ropeLen(a: PRope): int =
if a == nil: result = 0
else: result = a.L.abs
proc newRope(data: string = nil): PRope =
if data != nil:
unsafeNew(result, sizeof(TRope)-PayloadSize+len(data)+1)
result.L = -len(data)
# copy including '\0':
copyMem(addr result.d, cstring(data), len(data))
else:
unsafeNew(result, sizeof(TRope)-PayloadSize+1)
proc eqContent(r: PRope, s: string): bool =
assert r.L < 0
if -r.L == s.len:
result = equalMem(addr(r.d), cstring(s), s.len)
when false:
proc newMutableRope*(capacity = 30): PRope =
## creates a new rope that supports direct modifications of the rope's
## 'data' and 'length' fields.
new(result)
result.data = newStringOfCap(capacity)
proc freezeMutableRope*(r: PRope) {.inline.} =
r.length = r.data.len
var
cache: array[0..2048*2 -1, PRope]
proc RopeInvariant(r: PRope): bool =
if r == nil:
result = true
else:
result = true #
# if r.data <> snil then
# result := true
# else begin
# result := (r.left <> nil) and (r.right <> nil);
# if result then result := ropeInvariant(r.left);
# if result then result := ropeInvariant(r.right);
# end
proc insertInCache(s: string): PRope =
var h = hash(s) and high(cache)
result = cache[h]
if isNil(result) or not eqContent(result, s):
result = newRope(s)
cache[h] = result
proc toRope(s: string): PRope =
if s.len == 0:
result = nil
else:
result = insertInCache(s)
assert(RopeInvariant(result))
proc RopeSeqInsert(rs: var TRopeSeq, r: PRope, at: Natural) =
var length = len(rs)
if at > length:
setlen(rs, at + 1)
else:
setlen(rs, length + 1) # move old rope elements:
for i in countdown(length, at + 1):
rs[i] = rs[i - 1] # this is correct, I used pen and paper to validate it
rs[at] = r
proc newRecRopeToStr(result: var string, resultLen: var int, r: PRope) =
var stack = @[r]
while len(stack) > 0:
var it = pop(stack)
while it.L >= 0:
add(stack, it.right)
it = it.left
assert(it.L < 0)
CopyMem(addr(result[resultLen]), addr(it.d[0]), -it.L)
Inc(resultLen, -it.L)
assert(resultLen <= len(result))
proc ropeToStr(p: PRope): string =
if p == nil:
result = ""
else:
result = newString(p.L.abs)
var resultLen = 0
newRecRopeToStr(result, resultLen, p)
proc con(a, b: PRope): PRope =
if a == nil: result = b
elif b == nil: result = a
else:
result = newRope()
result.L = a.L.abs + b.L.abs
result.left = a
result.right = b
proc con(a: PRope, b: string): PRope = result = con(a, toRope(b))
proc con(a: string, b: PRope): PRope = result = con(toRope(a), b)
proc con(a: varargs[PRope]): PRope =
for i in countup(0, high(a)): result = con(result, a[i])
proc toRope(i: BiggestInt): PRope = result = toRope($i)
proc app(a: var PRope, b: PRope) = a = con(a, b)
proc app(a: var PRope, b: string) = a = con(a, b)
proc prepend(a: var PRope, b: PRope) = a = con(b, a)
proc writeRope*(f: TFile, c: PRope) =
var stack = @[c]
while len(stack) > 0:
var it = pop(stack)
while it.L >= 0:
add(stack, it.right)
it = it.left
assert(it != nil)
assert(it.L < 0)
write(f, cstring(it.d), -it.L)
proc WriteRope*(head: PRope, filename: string, useWarning = false) =
var f: tfile
if open(f, filename, fmWrite):
if head != nil: WriteRope(f, head)
close(f)
else:
rawMessage(if useWarning: warnCannotOpenFile else: errCannotOpenFile,
filename)
proc ropef(frmt: TFormatStr, args: varargs[PRope]): PRope =
var i = 0
var length = len(frmt)
result = nil
var num = 0
while i <= length - 1:
if frmt[i] == '$':
inc(i) # skip '$'
case frmt[i]
of '$':
app(result, "$")
inc(i)
of '#':
inc(i)
app(result, args[num])
inc(num)
of '0'..'9':
var j = 0
while true:
j = (j * 10) + Ord(frmt[i]) - ord('0')
inc(i)
if (i > length + 0 - 1) or not (frmt[i] in {'0'..'9'}): break
num = j
if j > high(args) + 1:
internalError("ropes: invalid format string $" & $(j))
else:
app(result, args[j - 1])
of 'n':
if optLineDir notin gOptions: app(result, tnl)
inc i
of 'N':
app(result, tnl)
inc(i)
else: InternalError("ropes: invalid format string $" & frmt[i])
var start = i
while i < length:
if frmt[i] != '$': inc(i)
else: break
if i - 1 >= start:
app(result, substr(frmt, start, i - 1))
assert(RopeInvariant(result))
proc appf(c: var PRope, frmt: TFormatStr, args: varargs[PRope]) =
app(c, ropef(frmt, args))
const
bufSize = 1024 # 1 KB is reasonable
proc auxRopeEqualsFile(r: PRope, bin: var tfile, buf: Pointer): bool =
if r.L < 0:
if -r.L > bufSize:
internalError("ropes: token too long")
return
var readBytes = readBuffer(bin, buf, -r.L)
result = readBytes == -r.L and
equalMem(buf, addr(r.d[0]), readBytes)
else:
result = auxRopeEqualsFile(r.left, bin, buf)
if result: result = auxRopeEqualsFile(r.right, bin, buf)
proc RopeEqualsFile(r: PRope, f: string): bool =
var bin: tfile
result = open(bin, f)
if not result:
return # not equal if file does not exist
var buf = alloc(BufSize)
result = auxRopeEqualsFile(r, bin, buf)
if result:
result = readBuffer(bin, buf, bufSize) == 0 # really at the end of file?
dealloc(buf)
close(bin)
proc newCrcFromRopeAux(r: PRope, startVal: TCrc32): TCrc32 =
# XXX profiling shows this is actually expensive
var stack: TRopeSeq = @[r]
result = startVal
while len(stack) > 0:
var it = pop(stack)
while it.L >= 0:
add(stack, it.right)
it = it.left
assert(it.L < 0)
var i = 0
var L = -it.L
while i < L:
result = updateCrc32(it.d[i], result)
inc(i)
proc crcFromRope(r: PRope): TCrc32 =
result = newCrcFromRopeAux(r, initCrc32)
proc writeRopeIfNotEqual(r: PRope, filename: string): bool =
# returns true if overwritten
var c: TCrc32
c = crcFromFile(filename)
if c != crcFromRope(r):
writeRope(r, filename)
result = true
else:
result = false

View file

@ -15,8 +15,6 @@ version 0.9.2
- ``hoist`` pragma for loop hoisting: can be easily done with - ``hoist`` pragma for loop hoisting: can be easily done with
AST overloading + global AST overloading + global
- implement ``system.unsafeNew``.
version 0.9.X version 0.9.X
============= =============

View file

@ -17,6 +17,7 @@ Library Additions
- Added ``system.onRaise`` to support a condition system. - Added ``system.onRaise`` to support a condition system.
- Added ``macros.quote`` for AST quasi-quoting. - Added ``macros.quote`` for AST quasi-quoting.
- Added ``system.unsafeNew`` to support hacky variable length objects.
Changes affecting backwards compatibility Changes affecting backwards compatibility