initial commit

master
lurchi 2018-08-16 22:26:53 +02:00
commit 48af85e4a6
No known key found for this signature in database
GPG Key ID: 8241BE099775A097
1 changed files with 285 additions and 0 deletions

285
parse.nim Normal file
View File

@ -0,0 +1,285 @@
import strutils, unittest
type
ParseState = enum
RoutingModifierName,
RoutingModifierValue,
ContentLength,
EntityModifierName,
EntityModifierValue,
Method,
Data,
Complete
Modifier = object
op: char
name: string
value: string
PsycPacket = object
routingHeader*: seq[Modifier]
entityHeader*: seq[Modifier]
entityHeaderLen*: int
contentLength*: int
methodName*: string
data*: string
state*: ParseState
cursor: int
remainingPartLen: int
Validator = proc(c: char): bool
proc validateTrue(c: char): bool = true
proc getUntil(input: string, delimiters: openArray[char], packet: var PsycPacket,
validate: Validator = validateTrue): tuple[complete: bool,
value: string] =
assert(packet.cursor < input.len())
result.complete = false
for i in packet.cursor .. input.high():
if input[i] in delimiters:
result.complete = true
if i > packet.cursor:
result.value = input[packet.cursor .. i - 1]
packet.cursor = i
break
if not validate(input[i]):
raise new(ValueError)
proc parseKeyword(input: string,
packet: var PsycPacket): tuple[complete: bool, value: string] =
assert(packet.cursor < input.len())
proc validate(c: char): bool = c == '_' or c.isAlphaNumeric()
result = input.getUntil([' ', '\t'], packet, validate)
if result.complete:
packet.cursor.inc()
if result.value.isNil() or result.value.len() == 0 or result.value[0] != '_':
raise new(ValueError) # name does not start with '_'
proc parseModifierValueLength(input: string,
packet: var PsycPacket): tuple[complete: bool,
value: int] =
let (complete, value) = input.getUntil(['\t'], packet)
if not complete:
return (false, -1)
packet.cursor.inc()
if value.len() == 0:
raise new(ValueError) # no value length
result = (true, value.parseInt()) # may throw ValueError if invalid number
if result.value < 0:
raise new(ValueError) # negative value length
proc parseModifierName(input: string,
packet: var PsycPacket): tuple[complete: bool,
op: char,
name: string] =
assert(packet.cursor < input.len())
const operators = ['=', ':', '+', '-', '?']
if packet.cursor == input.high():
result.complete = false
return
if input[packet.cursor] == '\n':
packet.cursor.inc()
return (true, '\0', nil)
result.op = input[packet.cursor]
if not (result.op in operators):
raise new(ValueError) # invalid operator
packet.cursor.inc()
if result.op == '?':
if input[packet.cursor] != '\n':
raise new(ValueError) # '?' must be on a line by itself
result.complete = true
return
(result.complete, result.name) = parseKeyword(input, packet)
if not result.complete:
return
# the following separator must either be ' ' (binary-arg) or '\t' (simple-arg)
if input[packet.cursor - 1] == ' ':
(result.complete, packet.remainingPartLen) = parseModifierValueLength(input,
packet)
elif input[packet.cursor - 1] != '\t':
raise new(ValueError) # invalid separator
proc parseMethod(input: string, packet: var PsycPacket): tuple[complete: bool,
value: string] =
assert(packet.cursor < input.len())
if input[packet.cursor] == '|':
return (true, nil)
return parseKeyword(input, packet)
proc parseModifierSimpleValue(input: string,
packet: var PsycPacket): tuple[complete: bool,
value: string] =
assert(packet.cursor < input.len())
result = input.getUntil(['\n'], packet)
if result.complete:
packet.cursor.inc()
proc parseModifierValue(input: string,
packet: var PsycPacket): tuple[complete: bool,
value: string] =
assert(packet.cursor < input.len())
if packet.remainingPartLen < 0:
return parseModifierSimpleValue(input, packet)
result.complete = packet.remainingPartLen < input.high() - packet.cursor
if result.complete:
result.value = input[packet.cursor .. packet.cursor + packet.remainingPartLen - 1]
if input[packet.cursor + packet.remainingPartLen] != '\n':
raise new(ValueError) # missing '\n' after binary-arg
packet.cursor += packet.remainingPartLen + 1
packet.remainingPartLen = -1
proc parseContentLength(input: string,
packet: var PsycPacket): tuple[complete: bool, value: int] =
assert(packet.cursor < input.len())
let (complete, value) = input.getUntil(['\n'], packet)
result.complete = complete
if result.complete:
packet.cursor.inc()
if value.len() == 0:
result.value = -1
else:
result.value = value.parseInt() # parseInt may throw ValueError if invalid number
if result.value < 0:
raise new(ValueError) # negative content length
proc parseData(input: string,
packet: var PsycPacket): tuple[complete: bool, value: string] =
assert(packet.cursor < input.len())
if packet.remainingPartLen < 0:
if packet.contentLength < 0:
if input.continuesWith("|\n", packet.cursor):
return (true, "")
let findResult = input.find("\n|\n")
if findResult < 0:
return (false, input[packet.cursor .. input.high()])
return (true, input[packet.cursor .. findResult - 1])
packet.remainingPartLen = packet.contentLength -
packet.entityHeaderLen -
packet.methodName.len() -
sizeof('\n')
result.complete = packet.remainingPartLen <= input.high() - packet.cursor
if result.complete:
result.value = input[packet.cursor .. packet.cursor + packet.remainingPartLen - 1]
packet.remainingPartLen = -1
proc newPacket*(): PsycPacket =
PsycPacket(routingHeader: newSeq[Modifier](),
entityHeader: newSeq[Modifier](),
state: ParseState.RoutingModifierName,
remainingPartLen: -1)
proc consumePart*(packet: var PsycPacket, slice: Slice[int]): string =
assert(packet.remainingPartLen >= slice.len())
case packet.state
of ParseState.RoutingModifierValue:
if packet.routingHeader.len() > 0: swap(result, packet.routingHeader[^1].value)
of ParseState.EntityModifierValue:
if packet.entityHeader.len() > 0: swap(result, packet.entityHeader[^1].value)
of ParseState.Data:
swap(result, packet.data)
else:
return
packet.remainingPartLen -= result.len()
proc parse*(input: string,
packet: var PsycPacket): tuple[needMoreInput: bool,
unparsed: Slice[int]] =
#proc addedPart(packet: var PsycPacket, slice: Slice[int]) =
# packet.remainingPartLen -= slice.len()
# packet.cursor += slice.len()
result.needMoreInput = true
case packet.state:
of ParseState.RoutingModifierName:
let (complete, op, name) = parseModifierName(input, packet)
if complete:
result.needMoreInput = false
packet.state = ParseState.RoutingModifierValue
let modifier = Modifier(op: op, name: name)
packet.routingHeader.add(modifier)
of ParseState.RoutingModifierValue:
let (complete, value) = parseModifierSimpleValue(input, packet)
result.needMoreInput = not complete
if complete and value.len() > 0:
packet.state = ParseState.ContentLength
if packet.routingHeader[^1].value.isNil():
packet.routingHeader[^1].value = ""
packet.routingHeader[^1].value.add(value)
#packet.addedPart(slice)
of ParseState.ContentLength:
let (complete, value) = parseContentLength(input, packet)
if complete:
result.needMoreInput = false
packet.state = ParseState.EntityModifierName
packet.contentLength = value
of ParseState.EntityModifierName:
let oldCursor = packet.cursor
let (complete, op, name) = parseModifierName(input, packet)
packet.entityHeaderLen += packet.cursor - oldCursor
if complete:
result.needMoreInput = false
packet.state = ParseState.EntityModifierValue
let modifier = Modifier(op: op, name: name)
packet.entityHeader.add(modifier)
of ParseState.EntityModifierValue:
let oldCursor = packet.cursor
let (complete, value) = parseModifierValue(input, packet)
result.needMoreInput = not complete
packet.entityHeaderLen += (packet.cursor - oldCursor)
if complete:
packet.state = ParseState.Method
if not value.isNil():
if packet.entityHeader[^1].value.isNil():
packet.entityHeader[^1].value = ""
packet.entityHeader[^1].value.add(value)
#packet.addedPart(slice)
of ParseState.Method:
let (complete, value) = parseMethod(input, packet)
if complete:
result.needMoreInput = false
packet.state = ParseState.Data
packet.methodName = value
packet.data = ""
of ParseState.Data:
let (complete, value) = parseData(input, packet)
result.needMoreInput = not complete
if complete:
packet.state = ParseState.Complete
packet.data.add(value)
#packet.addedPart(slice)
of ParseState.Complete:
assert(false)
result.unparsed = packet.cursor .. input.high()
if result.needMoreInput:
packet.cursor = 0
suite "parser tests":
setup:
var packet = newPacket()
test "state sync":
let input = ":_target\tpsyc://ve.symlynx.com/@blog\n\n?\n|\n"
while packet.state != ParseState.Complete:
discard parse(input, packet)
check(packet.routingHeader.len() == 1)
check(packet.routingHeader[0] == Modifier(op: ':',
name: "_target",
value: "psyc://ve.symlynx.com/@blog"))
check(packet.entityHeader.len() == 1)
check(packet.entityHeader[0] == Modifier(op: '?', name: nil, value: nil))
test "simple-arg":
let input = ":_target\tpsyc://ve.symlynx.com/@blog\n\n:_test\thello\n|\n"
while packet.state != ParseState.Complete:
discard parse(input, packet)
check(packet.entityHeader.len() == 1)
check(packet.entityHeader[0] == Modifier(op: ':', name: "_test", value: "hello"))
test "binary-arg":
let input = ":_target\tpsyc://ve.symlynx.com/@blog\n\n:_test 5\thello\n|\n"
while packet.state != ParseState.Complete:
discard parse(input, packet)
check(packet.entityHeader.len() == 1)
check(packet.entityHeader[0] == Modifier(op: ':', name: "_test", value: "hello"))
echo "parser tests completed"