jyapayne/Nim
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

lib_ropes: better buffering in equalsFile, cleanups

Browse source
This commit is contained in:
Jacek Sieka 2015-04-08 23:26:44 +08:00
commit 76a316bd35
1 changed files with 81 additions and 64 deletions
Download patch file Download diff file Expand all files Collapse all files

145
lib/pure/ropes.nim
View file

@ -52,9 +52,9 @@ proc len*(a: Rope): int {.rtl, extern: "nro$1".} =
## the rope's length ## the rope's length
if a == nil: result = 0 if a == nil: result = 0
else: result = a.length else: result = a.length
proc newRope(): Rope = new(result) proc newRope(): Rope = new(result)
proc newRope(data: string): Rope = proc newRope(data: string): Rope =
new(result) new(result)
result.length = len(data) result.length = len(data)
result.data = data result.data = data
@ -65,18 +65,18 @@ var
when countCacheMisses: when countCacheMisses:
var misses, hits: int var misses, hits: int
proc splay(s: string, tree: Rope, cmpres: var int): Rope = proc splay(s: string, tree: Rope, cmpres: var int): Rope =
var c: int var c: int
var t = tree var t = tree
N.left = nil N.left = nil
N.right = nil # reset to nil N.right = nil # reset to nil
var le = N var le = N
var r = N var r = N
while true: while true:
c = cmp(s, t.data) c = cmp(s, t.data)
if c < 0: if c < 0:
if (t.left != nil) and (s < t.left.data): if (t.left != nil) and (s < t.left.data):
var y = t.left var y = t.left
t.left = y.right t.left = y.right
y.right = t y.right = t
@ -85,8 +85,8 @@ proc splay(s: string, tree: Rope, cmpres: var int): Rope =
r.left = t r.left = t
r = t r = t
t = t.left t = t.left
elif c > 0: elif c > 0:
if (t.right != nil) and (s > t.right.data): if (t.right != nil) and (s > t.right.data):
var y = t.right var y = t.right
t.right = y.left t.right = y.left
y.left = t y.left = t
@ -95,8 +95,8 @@ proc splay(s: string, tree: Rope, cmpres: var int): Rope =
le.right = t le.right = t
le = t le = t
t = t.right t = t.right
else: else:
break break
cmpres = c cmpres = c
le.right = t.left le.right = t.left
r.left = t.right r.left = t.right
@ -104,50 +104,50 @@ proc splay(s: string, tree: Rope, cmpres: var int): Rope =
t.right = N.left t.right = N.left
result = t result = t
proc insertInCache(s: string, tree: Rope): Rope = proc insertInCache(s: string, tree: Rope): Rope =
var t = tree var t = tree
if t == nil: if t == nil:
result = newRope(s) result = newRope(s)
when countCacheMisses: inc(misses) when countCacheMisses: inc(misses)
return return
var cmp: int var cmp: int
t = splay(s, t, cmp) t = splay(s, t, cmp)
if cmp == 0: if cmp == 0:
# We get here if it's already in the Tree # We get here if it's already in the Tree
# Don't add it again # Don't add it again
result = t result = t
when countCacheMisses: inc(hits) when countCacheMisses: inc(hits)
else: else:
when countCacheMisses: inc(misses) when countCacheMisses: inc(misses)
result = newRope(s) result = newRope(s)
if cmp < 0: if cmp < 0:
result.left = t.left result.left = t.left
result.right = t result.right = t
t.left = nil t.left = nil
else: else:
# i > t.item: # i > t.item:
result.right = t.right result.right = t.right
result.left = t result.left = t
t.right = nil t.right = nil
proc rope*(s: string): Rope {.rtl, extern: "nro$1Str".} = proc rope*(s: string): Rope {.rtl, extern: "nro$1Str".} =
## Converts a string to a rope. ## Converts a string to a rope.
if s.len == 0: if s.len == 0:
result = nil result = nil
elif cacheEnabled: elif cacheEnabled:
result = insertInCache(s, cache) result = insertInCache(s, cache)
cache = result cache = result
else: else:
result = newRope(s) result = newRope(s)
proc rope*(i: BiggestInt): Rope {.rtl, extern: "nro$1BiggestInt".} = proc rope*(i: BiggestInt): Rope {.rtl, extern: "nro$1BiggestInt".} =
## Converts an int to a rope. ## Converts an int to a rope.
result = rope($i) result = rope($i)
proc rope*(f: BiggestFloat): Rope {.rtl, extern: "nro$1BiggestFloat".} = proc rope*(f: BiggestFloat): Rope {.rtl, extern: "nro$1BiggestFloat".} =
## Converts a float to a rope. ## Converts a float to a rope.
result = rope($f) result = rope($f)
proc enableCache*() {.rtl, extern: "nro$1".} = proc enableCache*() {.rtl, extern: "nro$1".} =
## Enables the caching of leaves. This reduces the memory footprint at ## Enables the caching of leaves. This reduces the memory footprint at
## the cost of runtime efficiency. ## the cost of runtime efficiency.
@ -160,9 +160,9 @@ proc disableCache*() {.rtl, extern: "nro$1".} =
proc `&`*(a, b: Rope): Rope {.rtl, extern: "nroConcRopeRope".} = proc `&`*(a, b: Rope): Rope {.rtl, extern: "nroConcRopeRope".} =
## the concatenation operator for ropes. ## the concatenation operator for ropes.
if a == nil: if a == nil:
result = b result = b
elif b == nil: elif b == nil:
result = a result = a
else: else:
result = newRope() result = newRope()
@ -177,16 +177,16 @@ proc `&`*(a, b: Rope): Rope {.rtl, extern: "nroConcRopeRope".} =
else: else:
result.left = a result.left = a
result.right = b result.right = b
proc `&`*(a: Rope, b: string): Rope {.rtl, extern: "nroConcRopeStr".} = proc `&`*(a: Rope, b: string): Rope {.rtl, extern: "nroConcRopeStr".} =
## the concatenation operator for ropes. ## the concatenation operator for ropes.
result = a & rope(b) result = a & rope(b)
proc `&`*(a: string, b: Rope): Rope {.rtl, extern: "nroConcStrRope".} = proc `&`*(a: string, b: Rope): Rope {.rtl, extern: "nroConcStrRope".} =
## the concatenation operator for ropes. ## the concatenation operator for ropes.
result = rope(a) & b result = rope(a) & b
proc `&`*(a: openArray[Rope]): Rope {.rtl, extern: "nroConcOpenArray".} = proc `&`*(a: openArray[Rope]): Rope {.rtl, extern: "nroConcOpenArray".} =
## the concatenation operator for an openarray of ropes. ## the concatenation operator for an openarray of ropes.
for i in countup(0, high(a)): result = result & a[i] for i in countup(0, high(a)): result = result & a[i]
@ -219,7 +219,7 @@ iterator leaves*(r: Rope): string =
## iterates over any leaf string in the rope `r`. ## iterates over any leaf string in the rope `r`.
if r != nil: if r != nil:
var stack = @[r] var stack = @[r]
while stack.len > 0: while stack.len > 0:
var it = stack.pop var it = stack.pop
while isConc(it): while isConc(it):
stack.add(it.right) stack.add(it.right)
@ -227,7 +227,7 @@ iterator leaves*(r: Rope): string =
assert(it != nil) assert(it != nil)
assert(it.data != nil) assert(it.data != nil)
yield it.data yield it.data
iterator items*(r: Rope): char = iterator items*(r: Rope): char =
## iterates over any character in the rope `r`. ## iterates over any character in the rope `r`.
for s in leaves(r): for s in leaves(r):
@ -237,7 +237,7 @@ proc write*(f: File, r: Rope) {.rtl, extern: "nro$1".} =
## writes a rope to a file. ## writes a rope to a file.
for s in leaves(r): write(f, s) for s in leaves(r): write(f, s)
proc `$`*(r: Rope): string {.rtl, extern: "nroToString".}= proc `$`*(r: Rope): string {.rtl, extern: "nroToString".}=
## converts a rope back to a string. ## converts a rope back to a string.
result = newString(r.len) result = newString(r.len)
setLen(result, 0) setLen(result, 0)
@ -251,25 +251,25 @@ when false:
new(result) new(result)
result.length = -idx result.length = -idx
proc compileFrmt(frmt: string): Rope = proc compileFrmt(frmt: string): Rope =
var i = 0 var i = 0
var length = len(frmt) var length = len(frmt)
result = nil result = nil
var num = 0 var num = 0
while i < length: while i < length:
if frmt[i] == '$': if frmt[i] == '$':
inc(i) inc(i)
case frmt[i] case frmt[i]
of '$': of '$':
add(result, "$") add(result, "$")
inc(i) inc(i)
of '#': of '#':
inc(i) inc(i)
add(result, compiledArg(num+1)) add(result, compiledArg(num+1))
inc(num) inc(num)
of '0'..'9': of '0'..'9':
var j = 0 var j = 0
while true: while true:
j = j * 10 + ord(frmt[i]) - ord('0') j = j * 10 + ord(frmt[i]) - ord('0')
inc(i) inc(i)
if frmt[i] notin {'0'..'9'}: break if frmt[i] notin {'0'..'9'}: break
@ -285,10 +285,10 @@ when false:
add(s, compiledArg(j)) add(s, compiledArg(j))
else: raise newException(EInvalidValue, "invalid format string") else: raise newException(EInvalidValue, "invalid format string")
var start = i var start = i
while i < length: while i < length:
if frmt[i] != '$': inc(i) if frmt[i] != '$': inc(i)
else: break else: break
if i - 1 >= start: if i - 1 >= start:
add(result, substr(frmt, start, i-1)) add(result, substr(frmt, start, i-1))
proc `%`*(frmt: string, args: openArray[Rope]): Rope {. proc `%`*(frmt: string, args: openArray[Rope]): Rope {.
@ -340,29 +340,46 @@ proc addf*(c: var Rope, frmt: string, args: openArray[Rope]) {.
## shortcut for ``add(c, frmt % args)``. ## shortcut for ``add(c, frmt % args)``.
add(c, frmt % args) add(c, frmt % args)
const
bufSize = 1024 # 1 KB is reasonable
proc equalsFile*(r: Rope, f: File): bool {.rtl, extern: "nro$1File".} = proc equalsFile*(r: Rope, f: File): bool {.rtl, extern: "nro$1File".} =
## returns true if the contents of the file `f` equal `r`. ## returns true if the contents of the file `f` equal `r`.
var bufSize = 1024 # reasonable start value var
var buf = alloc(bufSize) buf: array[bufSize, char]
for s in leaves(r): bpos = buf.len
if s.len > bufSize: blen = buf.len
bufSize = max(bufSize * 2, s.len)
buf = realloc(buf, bufSize)
var readBytes = readBuffer(f, buf, s.len)
result = readBytes == s.len and equalMem(buf, cstring(s), s.len)
if not result: break
if result:
result = readBuffer(f, buf, 1) == 0 # really at the end of file?
dealloc(buf)
proc equalsFile*(r: Rope, f: string): bool {.rtl, extern: "nro$1Str".} = for s in leaves(r):
var spos = 0
let slen = s.len
while spos < slen:
if bpos == blen:
# Read more data
bpos = 0
blen = readBuffer(f, addr(buf[0]), buf.len)
if blen == 0: # no more data in file
result = false
return
let n = min(blen - bpos, slen - spos)
# TODO There's gotta be a better way of comparing here...
if not equalMem(addr(buf[bpos]),
cast[pointer](cast[int](cstring(s))+spos), n):
result = false
return
spos += n
bpos += n
result = readBuffer(f, addr(buf[0]), 1) == 0 # check that we've read all
proc equalsFile*(r: Rope, filename: string): bool {.rtl, extern: "nro$1Str".} =
## returns true if the contents of the file `f` equal `r`. If `f` does not ## returns true if the contents of the file `f` equal `r`. If `f` does not
## exist, false is returned. ## exist, false is returned.
var bin: File var f: File
result = open(bin, f) result = open(f, filename)
if result: if result:
result = equalsFile(r, bin) result = equalsFile(r, f)
close(bin) close(f)
new(N) # init dummy node for splay algorithm new(N) # init dummy node for splay algorithm