optimized parseurl module a bit

This commit is contained in:
Araq
2012-02-01 23:55:22 +01:00
parent e4a4588b48
commit 4203aef827

View File

@@ -12,59 +12,62 @@
import strutils
type
TURL* = tuple[ ## represents a *Uniform Resource Locator* (URL)
TUrl* = tuple[ ## represents a *Uniform Resource Locator* (URL)
## any optional component is "" if it does not exist
scheme, username, password,
hostname, port, path, query, anchor: string]
proc parseUrl*(url: string): TURL =
var i: int = 0
proc parseUrl*(url: string): TUrl =
var i = 0
var scheme, username, password: string = ""
var hostname, port, path, query, anchor: string = ""
var temp: string = ""
var temp = ""
if url[i] != '/': #url isn't a relative path
if url[i] != '/': # url isn't a relative path
while True:
#Scheme
# Scheme
if url[i] == ':':
if url[i+1] == '/' and url[i+2] == '/':
scheme = temp
temp = ""
inc(i, 3) #Skip the //
#Authority(username, password)
temp.setlen(0)
inc(i, 3) # Skip the //
# Authority(username, password)
if url[i] == '@':
username = temp.split(':')[0]
if temp.split(':').len() > 1:
password = temp.split(':')[1]
temp = ""
username = temp
let colon = username.find(':')
if colon >= 0:
password = username.substr(colon+1)
username = username.substr(0, colon-1)
temp.setlen(0)
inc(i) #Skip the @
#hostname(subdomain, domain, port)
# hostname(subdomain, domain, port)
if url[i] == '/' or url[i] == '\0':
hostname = temp
if hostname.split(':').len() > 1:
port = hostname.split(':')[1]
hostname = hostname.split(':')[0]
let colon = hostname.find(':')
if colon >= 0:
port = hostname.substr(colon+1)
hostname = hostname.substr(0, colon-1)
temp = ""
temp.setlen(0)
break
temp.add(url[i])
inc(i)
if url[i] == '/': inc(i) # Skip the '/'
#Path
# Path
while True:
if url[i] == '?':
path = temp
temp = ""
temp.setlen(0)
if url[i] == '#':
if temp[0] == '?':
query = temp
else:
path = temp
temp = ""
temp.setlen(0)
if url[i] == '\0':
if temp[0] == '?':
@@ -80,16 +83,25 @@ proc parseUrl*(url: string): TURL =
return (scheme, username, password, hostname, port, path, query, anchor)
proc `$`*(t: TURL): string =
proc `$`*(u: TUrl): string =
## turns the URL `u` into its string representation.
result = ""
if t.scheme != "": result.add(t.scheme & "://")
if t.username != "":
if t.password != "":
result.add(t.username & ":" & t.password & "@")
else:
result.add(t.username & "@")
result.add(t.hostname)
if t.port != "": result.add(":" & t.port)
if t.path != "": result.add("/" & t.path)
result.add(t.query)
result.add(t.anchor)
if u.scheme.len > 0:
result.add(u.scheme)
result.add("://")
if u.username.len > 0:
result.add(u.username)
if u.password.len > 0:
result.add(":")
result.add(u.password)
result.add("@")
result.add(u.hostname)
if u.port.len > 0:
result.add(":")
result.add(u.port)
if u.path.len > 0:
result.add("/")
result.add(u.path)
result.add(u.query)
result.add(u.anchor)