MRo Blog

UILabel with a (custom) CGFont

2010-09-03T11:45:44Z

UILabel’s font property accepts UIFonts – but strange enough there’s no way to get a custom loaded CGFont (from a ttf or otf file) converted into such an UIFont. You’re stuck with the iPhone’s pre-installed fonts – at least when you have to support iOS 3.0 devices.

After googling a bit and searching Stackoverflow I found the solutions presented there not ideal or great, but too heavy weight.

So I inherited UILabel with a very lean custom class UILabelWithCGFont and overloaded it’s drawTextInRect: method like this:

-(void)drawTextInRect:(CGRect)rect
{
	MRLogD(@"(%f,%f) (%f,%f)", rect.origin.x, rect.origin.y, rect.size.width, rect.size.height);
	if ( _CGFont == NULL ) {
		[super drawTextInRect:rect];
		return;
	}
	NSAssert(_mapping != NULL, @"Mapping function pointer not set.");
 
	// prepare the target graphics context.
	const CGContextRef ctx = UIGraphicsGetCurrentContext();
	CGContextSaveGState(ctx);
	{
		// prepare the glyphs array to draw
		const NSString *txt = self.text;
		const size_t glyphCount = txt.length;
		CGGlyph glyphs[glyphCount];
		{
			// turn the string txt into glyphs (indices into the font):
			// give non-allocating unicode character retrieval a try:
			const UniChar *raw_unichars = CFStringGetCharactersPtr( (CFStringRef)txt );
			const UniChar *unichars = raw_unichars == NULL ? malloc( glyphCount * sizeof(UniChar) ) : raw_unichars;
			NSAssert(unichars != NULL, @"unichars not allocated");
			if ( raw_unichars == NULL )
				CFStringGetCharacters( (CFStringRef)txt, CFRangeMake(0, txt.length), (UniChar *)unichars );
			for ( int i = glyphCount - 1; i &gt;= 0; i-- )
				glyphs[i] = _mapping(unichars[i]);
			if ( raw_unichars == NULL )
				free( (void *)unichars );
		}
 
		CGContextSetFont(ctx, _CGFont);
		CGContextSetFontSize(ctx, self.font.pointSize);
		CGContextSetTextMatrix( ctx, CGAffineTransformMake(1.0, 0.0, 0.0, -1.0, 0.0, 0.0) );
 
		// first print 'invisible' to measure size:
		CGContextSetTextDrawingMode(ctx, kCGTextInvisible);
		const CGPoint pre = CGContextGetTextPosition(ctx);
		CGContextShowGlyphs(ctx, glyphs, glyphCount);
		const CGPoint post = CGContextGetTextPosition(ctx);
		// restore text position
		CGContextSetTextPosition(ctx, pre.x, pre.y);
 
		// centered horizontal + vertical:
		NSAssert( (int)rect.origin.x == 0, @"origin.x not zero" );
		NSAssert( (int)rect.origin.y == 0, @"origin.y not zero" );
		NSAssert(self.baselineAdjustment == UIBaselineAdjustmentAlignCenters, @"vertical alignment not 'center'");
		NSAssert(self.textAlignment == UITextAlignmentCenter, @"horizontal alignment not 'center'");
		const CGPoint p = CGPointMake( ( rect.size.width - (post.x - pre.x) ) / 2, (rect.size.height + self.font.pointSize + pre.y) / 2 );
 
		// finally render it to the graphics context:
		CGContextSetTextDrawingMode(ctx, kCGTextFill);
		CGContextSetFillColorWithColor(ctx, self.textColor.CGColor);
		CGContextShowGlyphsAtPoint(ctx, p.x, p.y, glyphs, glyphCount);
	}
	CGContextRestoreGState(ctx);
}

Usage: Just turn the UILabel instances in Interface Builder into UILabelWithCGFont and implement the UIViewController::viewDidLoad method like this:

CGGlyph unicode2glyphDeutscheDruckschrift(UniChar c)
{
	if ( '0' < = c && c <= '9' )
		return c + (16 - '0');
	if ( 'A' <= c && c <= 'Z' )
		return c + (32 - 'A');
	if ( 'a' <= c && c <= 'z' )
		return c + (58 - 'a');
	return 0;
}
 
-(void)viewDidLoad
{
	[super viewDidLoad];
	...
	[fontLabel setFontFromFile:@"DeutscheDruckschrift" ofType:@"ttf" mapping:unicode2glyphDeutscheDruckschrift];
	...
}

See this github gist for the complete implementation.

The mapping from Unicode character codes to glyph indices (inside the font description) currently is done via a C mapping function you have to provide a function pointer for. A later implementation could map the unicode character code to the glyph name and leverage CGFontGetGlyphWithGlyphName and render the custom mapping function obsolete.

Visualise macports dependencies

2010-08-15T16:19:46Z

to clean up your installed macports and remove cruft you need to uninstall them in the correct order – according to their dependencies.

A graphical visualisation might help doing so:

Call

$ ./port-deps2dot.rb | dot -Tpdf -o port-deps.pdf ; open port-deps.pdf

with the ruby script port-deps2dot.rb as follows:

#!/usr/bin/ruby -w
 
# visualize macports dependencies.
# pipe the result through graphviz, e.g.
# $ ./port-deps2dot.rb | dot -Tpdf -o port-deps.pdf ; open port-deps.pdf
 
def scan_deps
	pat = /^([^:]+):(.+)$/
	name = ''
	deps = []
	IO.popen('port info --name --pretty --depends installed') do |f|
		f.each_line do |l|
			case l
				when /^--$/
					yield name, deps
					name = ''
					deps = []
				when /^([^:]+):(.+)$/
					if 'name' == "#$1"
						name = "#$2".strip
					else
						deps.concat("#$2".split(/\s*,\s*/))
					end
				else
					raise "Fall-through for '#{l}'"
			end
		end
	end
end
 
all = {}
 
scan_deps do |name,deps|
	d = all[name]
	all[name] = d = [] if d.nil?
	deps.collect! {|i| i.strip}
	d.concat deps
	d.sort!
	d.uniq!
end
 
head = < <END_OF_STRING
#!/usr/bin/dot -Tpdf -o port-deps.pdf
/*
	See http://www.graphviz.org/Documentation.php
*/
digraph "port deps" {
	rankdir=LR;
    label="port deps";
    node [style=filled,fillcolor=lightblue,shape=ellipse];
END_OF_STRING
 
puts head
 
all.keys.sort.each do |name|
	deps = all[name]
	if deps.count > 0
		deps.each {|d| puts "\t\"#{name}\" -> \"#{d}\";" }
	else
		puts "\t\"#{name}\";"
	end
end
 
foot = < <END_OF_STRING
}
END_OF_STRING
 
puts foot

Develop with iOS 4 SDK on an OS 3.0 Device

2010-06-23T11:32:27Z

After upgrading to iOS 4.0 SDK, iPhone OS 3.0 is no longer a valid “Base SDK”. Naively switching to iPhone 3.2 prevents deployment to a 3.0 device.

But such “Project -> Edit Project Settings” work out fine:

iPhoneOS 3.0 development device with iOS 4.0 SDK

“Deployment Target” hint found here.

iOS 4 Simulator crash (when installed to custom folder)

2010-06-23T11:23:41Z

just downloaded and installed the iOS 4 SDK and as my root OS X partition is rather (too) small, I put it into a custom location /Users/Developer.Snowleopard/.

This causes the iPhone Simulator to crash and compiling gives an error like:

ibtool failed with exception: Interface Builder encountered an error communicating with the iPhone Simulator. If you choose to file a crash report or radar for this issue, please check Console.app for crash reports for "Interface Builder Cocoa Touch Tool" and include their content in your crash report.
...
dyld: Library not loaded: /Developer/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator4.0.sdk/System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
  Referenced from: /Users/Developer.SnowLeopard/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator4.0.sdk/System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libLAPACK.dylib
  Reason: image not found

The cure – a softlink:

$ ln -s /Users/Developer.SnowLeopard/Platforms /Developer/Platforms

XCode: missing “deprecated” warnings

2010-06-23T09:23:34Z

when developing for long-term use, you want to use APIs that aren’t likely to be removed soon, a.k.a. “deprecated”.

So, don’t use downward compatible calls below a point you really aim for.

XCode helps with compiler warnings about “deprecated” calls – if “Project -> Edit Project Settings -> GCC_WARN_ABOUT_DEPRECATED_FUNCTIONS” is set:

Compiler setting: Warn about deprecated calls

But be careful, this complains e.g. here only once while initWithFrame: is also deprecated (compiling for iPhone OS 3.0):

No "deprecated" warning on initXY

As Kevin Ballard pointed out at Stackoverflow, this is because [AnyClass alloc] returns a type id – which doesn’t know about it’s interface.

To get this kind of compiler warnings, you have to type-cast the [AnyClass alloc] like this:

[(AnyClass*)[AnyClass alloc] initXY];

Maybe time for a macro?

The macro could look like

// http://blog.mro.name/2010/06/xcode-missing-deprecated-warnings/
#define alloc(c)	((c*)[c alloc])

Migrate your codebase via XCode “Edit -> Find in Project” with search pattern \[\s*([^\[\]]+)\s+alloc\s*\] and replacement pattern alloc(\1).

Simple HTTP Access Authorisation

2010-06-14T10:56:49Z

sometimes you may want to lock down RESTful APIs or plain HTTP GET resources for authorised access by your own client software only, without requiring authentication. You don’t know who (not authenticated), but you know she may access (is authorised).

If the server has a valid SSL certificate based on a root certificate pre-installed on the iPhone among the simplest ways to do it are:

HTTP Basic Authentication with static username + password. This requires just a .htaccess configuration setting and you’re done.

send a custom HTTP Request Header with a secret token, also just a .htaccess rewrite setting required:

RewriteEngine On
RewriteCond %{HTTP:My-Secret-Token} !=WRdsWXwwTZjEIRrgD5tODVf0U
RewriteRule ^.*$ - [forbidden,last]
# Test: $  curl --header "My-Secret-Token:WRdsWXwwTZjEIRrgD5tODVf0U" http://myserver.example.com/demo/

Decompiling an App may raise the bar high enough though hard-coded secrets surely aren’t bulletproof Secret Service grade quality. If you don’t want the password or secret token as literal string inside the App, synthesize it at runtime.

If your transport channel isn’t confidential (e.g. plain HTTP, not HTTPS) you might think about Digest Authentication or a custom implemented CRAMish mechanism which I will not go into in this post.

P.S.: Here are some really nice .htaccess examples.

Automatic gzip compression for Apache2 Webservers

2010-06-13T11:55:25Z

after failing and failing again in the last months, I finally got it with the help of http://www.debian-administration.org/articles/137

The .htaccess configuration

AddOutputFilterByType DEFLATE text/html text/plain text/xml text/css application/javascript

requires Apache’s mod_deflate enabled via

$ a2enmod deflate
Module deflate installed; run /etc/init.d/apache2 force-reload to enable.

Check the result with http://www.gidnetwork.com/tools/gzip-test.php

Caution: There seems to be an If-Modified-Since/Last-Modified HTTP 304 bug in mod_deflate, so better don’t use it for large, rarely changing files.

iPhone: libxml2 & RELAX NG validation

2010-06-14T12:37:40Z

Having a validating parser in place can reduce the required code to parse XML a lot – you know very well what you actually get. As mentioned in my last post about RELAX NG & trang, I prefer RELAX NG over W3C XML Schema – which doesn’t matter anyway because Apple’s suggested XML parser doesn’t validate at all.

So we have to go one level deeper and have a look at libxml2.

Apple’s example “XmlPerformance” helped to get started, but didn’t do the trick because libxml2 allows validation for xmlDocPtr or xmlTextReader but not for SAX parsers as used in the example.

The libxml2 examples didn’t help me too much either, but luckily there’s xmllint available in source (OSS just rocks) which does almost what we want. It first parses the XML into a xmlDocPtr and validates afterwards – and it does so for a reason:

You can have a validating xmlTextReader (via xmlTextReaderRelaxNGSetSchema), but it won’t detect IDREFs missing their referred to ID and the error messages lack the name of the failing item. BTW – when validating against a W3C schema this ID/IDREF check isn’t available yet.

I finally discarded streaming XML parsing in favour of validation and “push” parsing (nice for data coming in over the wire) and did:

load the RELAX NG regular form schema (watch out for the assignment of relaxngschemas) – similar to xmllint schema loading,
push the raw XML data into a xmlDocPtr (xmlCreatePushParserCtxt) exactly like xmllint,
validate the in-memory document (xmlRelaxNGValidateDoc),
turn it into a xmlTextReader,
process the reader.

Wrap up:

if you want full RELAX NG validation with libxml2 v2.7.3, forget about streamed parsing,
wrap the document into a xmlTextReader if you want a SAXish programming model.

I may prepare and publish a MroLibxml2Parser inheriting NSXMLParser and firing it’s callbacks in order to easily switch validating and non-validating parser implementations, but this has to wait a bit. Stay tuned.

XML Toolbox: RELAX NG & trang

2010-06-14T20:50:24Z

e.g. when handling RESTful APIs you may want to validate the response XML – a custom one in most cases.

I typically use tools already installed on every Mac and fire a http GET request with curl and immediately check it with xmllint like

$ curl http://www.heise.de/newsticker/heise-atom.xml | xmllint --format --schema myschema.xsd -

But I just don’t like to create and edit W3C XML Schemas – the notorious angle brackets hurt my eyes and the redundant element names hide the real stuff in tons of ever same text. Neither do I like to click through graphical schema editors and getting lost hunting for hidden settings and property dialogs.

A minimal and naive schema validating the above example Atom feed (and simply created from the feed itself with trang, see below) as W3C Schema looks like this:

Naive Atom W3C Schema

Here comes in RELAX NG, especially it’s “compact form“, which is just what I like – a concise, BNF-ish syntax. It was designed by Murata Makoto and James Clark, Technical Lead of the XML Working Group back when XML was created and father of the famous expat parser.

The very same schema as above as RELAX NG boils down to ½ the lines and about ⅓ of the characters without a single angle bracket:

default namespace = "http://www.w3.org/2005/Atom"
 
start =
  element feed {
    title,
    element subtitle { text },
    link+,
    updated,
    element author {
      element name { text }
    },
    id,
    element entry { title, link, id, updated }+
  }
title = element title { text }
link =
  element link {
    attribute href { xsd:anyURI },
    attribute rel { xsd:NCName }?
  }
updated = element updated { xsd:dateTime }
id = element id { xsd:anyURI }

And as libxml2 and therefore xmllint supports RELAX NG, you can use the regular syntax to validate like in the beginning, but with a much more editable schema:

$ curl http://www.heise.de/newsticker/heise-atom.xml | xmllint --format --relaxng myschema.rng -

trang

is a schema converter for RELAX NG written in Java which I wrapped inside a bash script:

#!/bin/sh
java -jar `dirname $0`/trang-20090818/trang.jar $@

Writing a new schema from scratch can be much more convenient if you have a bunch of XML files you can feed into trang:

$ trang *.xml myschema.rnc

then refine the resulting schema in compact form and finally turn it into the regular form:

$ trang myschema.rnc myschema.rng

Trang also serves me as a schema indenter by converting from compact to regular and back.

BUT: trang converts RELAX NG into W3C but not vice versa.

Deep validation

Validating XML documents shouldn’t stop with elements and attributes but rather leverage XML Schema Datatypes and apply e.g. regular expressions

  element uuid {
    xsd:string {
 
      ## A UUID
      pattern =
        "[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}"
    }
  }

or range constraints

        element year {
          xsd:unsignedShort { minInclusive = "1900" maxInclusive = "2100" }
        }

P.S.: For a more complete Atom RELAX NG schema see here or ask your search engine of choice.

TextWrangler + tidy

2010-05-21T21:59:42Z

as I didn’t get TidyService to work correctly with UTF8 umlauts, I created a UNIX Shell Script wrapper for html tidy as it comes with OS X that does the job at least for TextWrangler:

open TextWranglers “Unix Filters Folder”
create a file named e.g. “Tidy Html.sh”,

paste the following lines into the file and save it:

#!/bin/sh
# run "tidy" on the file given as 1st (and only) parameter.
#
/usr/bin/tidy -utf8 -asxhtml -indent -wrap 100 -quiet "$1" 2> /dev/null

now you can run tidy on files opened in TextWrangler, even remote files.
assign a keyboard shortcut (I used CTRL-T in the shot above) via the “Unix Filters” Palette:

Immo Betrug / Scam

2010-05-12T10:43:26Z

auf Wohnungssuche begegnet man lustigen Sachen. Da war dieses extrem billige Angebot (2 Zi, 80m² für 550€) mit spannender Adresse, aber fragen kostet ja nix.

Die erste Mail der Anbieterin war dann:

Hi,

Thanks for your interest. The apartment is still available. I moved recently with my job in London, United Kingdom, but the rent is high here, I decided to offer to rent the apartment I have in Munchen . I rented the apartment for maximum 10 years … This is the period that I have a contract here, but I can rent it for a shorter period also. I own the ground and is exactly as in the photos. The rent for 1 month is 550 EUR and does not include all the utilities you (water, electricity, Internet, cable). You can enter the apartment on the same day when receiving the keys .. The only problem is that i had to move with my job to United Kingdom, London where i am now and i left the keys and contract already signed by me at a company called Rent.com and they will handle the payment and delivery for us. If you want to know more about how this deal can work please get back to me ASAP and i will send you the details step by step. Thank you and hope to hear back from you.

My German is limited at just a few words so i wish a lot to continue our conversation in English.

dann ging’s 2x per Mail hin und her,

Hello,

If we decide to proceed with this transaction, I will have to contact Rent.com and provide them all the necessary information, so they can start the process right away. I will need your full name and address for that. You will receive a notification from Rent.com shortly after that, together with all the instructions to follow and the invoice as well.

Regarding the payment, you will be instructed to deposit the money to a Rent.com account. They will hold and insure your money until you check the apartment and decide if you want take it or not. That is how their buyer protection policy works. As far as my concerns, I will be glad to know that Rent.com has the possession of the money during the delivery period. That is my insurance.

As soon as the funds have been deposited into their account, they will immediately start the shipping process.

The keys and rental contract will be delivered at your address in no more than 3 working days. You will be given a ten days inspection period from the day you receive the keys and contract at home. If you decide to hold the apartment, then you will have to authorize Rent.com to release the funds to me, and the transaction will be completed. If you will not be satisfied with the apartment, you will be able to send the keys and contract back through the same service and ask Rent.com to return the funds to you.

Through I am sure you will love the apartment, it is good to know that you do have this second option available. If you wish to proceed with renting the apartment, please provide me your full name and address so I can initiate the deal through Rent.com right away.

I am looking forward to hearing from you.

Regards,

die Anbieterin wollte nicht anrufen aber eine Zahlungsanweisung kam, die mich dann richtig stutizg machte, weil

der ganze Ablauf eh schon recht seltsam war,
deren Absenderemail “rentcom@consultant.com” nicht mal von rent.com kam,
rent.com außerhalb USA gar nix macht,
die Einlieferungs-IP der Zahlungsanweisung die gleiche wie die der Mails von der Anbieterin war,
ja sogar “In-Reply” auf die vorherigen Mails,
das Geld an jemanden anderen gehen sollte,
an der Wohnungs-Adresse wohl gar keine Wohnungen sind,
ein Spezl mich auf die Warnungen bei rent.com und einen Bericht bei der BBC hinwies.

Nach meiner Frage nach dem Namen des Hausmeisters war dann Funkstille in London.

Zu guter letzt hat mich Immoscout24 heute nochmal per Mail vor Betrügern gewarnt.

WordPress Email Benachrichtigungen

2010-05-04T09:41:26Z

ojeoje, ist das so abseitig oder hab’ ich’s nicht verstanden? War jedenfalls komplizierter als erwartet:

Die Konto Einstellungen für den Email Absender müssen von Hand in die DB geschrieben werden – die 4 mailserver_* Zeilen in der Tabelle wordpress_options,
Absenderadresse und Nick dann noch per Plugin,

fertig. Aber wieso so kompliziert?

Tooltip: doxygen & graphviz (OS X)

2010-04-27T15:42:12Z

ever seen error messages like this when generating doxygen docs for an iPhone/Mac Objective C project:

.../build/doxygen/html/a00136.png' produced by dot is not a valid PNG!
You should either select a different format (DOT_IMAGE_FORMAT in the config file) or install a more recent version of graphviz (1.7+)

Happened to me this morning when running doxygen on another box. The doxygen dmg was installed in /Applications, I even added the graphviz macport, so WTF.

It turned out, that

the doxygen dmg brings everything it needs, including graphviz,
but: the full path to the dot executable has to be set inside the doxygen config.

So, when using the dot from the doxygen dmg, it has to be set to

/Applications/Doxygen.app/Contents/Resources

I fixed my generic Makefile and make a good guess now while keeping things configurable per user as the Makefile and doxygen config are committed sources and should not contain user specific settings.

Radio Geeky

2010-03-28T09:41:59Z

wie schon in einem älteren Artikel angerissen möchte ich zeitgesteuert Radiosendungen aufnehmen – und zwar auf einem Rechner (mit Apache Webserver), der rund um die Uhr läuft.

Nun habe ich kürzlich einem großen Zündfunk-Fan, der selten Zeit hat ihn zu hören, diesen Mitschnitt als Podcast zum Geburtstag geschenkt. Das ganze wurde aufwändiger als ich dachte und setzt sich mittlerweile zusammen aus:

Screen Scraping der B2 Programmwebsite mit scrAPI – incl. Datumskorrektur weil das Programm nicht um Mitternacht umbricht sondern um 5 Uhr früh,
diese Daten in einer SQLite DB cachen um nicht mehrmals am Tag zu scrapen (wenn man z.B. noch andere Sendungen aufnimmt),
per cron einige Minuten vor Sendungsbeginn den streamripper starten (v1.62.3 wegen alter glib auf Debian Linux),
- da der einige Zeit zum Starten braucht triggere ich Sendungsbeginn und -ende zeitlich exakt über ein Script, das Titelwechsel in den laufenden Streamripper injiziert – der B2 Stream liefert selbst keine Sendungsinfos,
falls der Stream zwischendurch unterbrochen wurde setzen wir die verschiedenen mp3 Dateien zu einer zusammen (ohne Überblendung per Holzhammer: cat)
Sendungsinfos parallel zum mp3 als xml speichern, um nach Aufnahmeende von der DB unabhängig zu sein,
aus allen vorhandenen mp3+xml Paaren ein Podcast RSS bauen und mit gzip packen,
tägliche Gleitlöschung (mp3 älter als 3 Wochen) per cron+find damit die Platte nicht volläuft.

Den Podcast abonniert man schließlich z.B. mit iTunes und bekommt direkt nach der Sendung die frische Zündfunkfolge incl. komplettem Text zur Sendung frei Haus.

Das komplette Ruby Script gibts bei github.

iPhone Unit Test Coverage

2010-03-27T11:12:48Z

have a look at the CoverStory Howto,
download CoverStory,
open (with XCode) the XCode Project you want to measure,
run the script EnableGCov.scpt linked to from the howto,
run your tests and see a linker error, – dead end for the time being. http://groups.google.com/group/coverstory-discuss/browse_thread/thread/fbcbf5ed61d8d02b#

CocoaHeads Vortrag gestern: Unit Testing

2010-03-18T21:01:57Z

um ein wenig anzugeben und schlicht den Link hier zu platzieren:

iPhone: SQLite Fulltext Index

2010-03-18T22:13:04Z

Most blogposts I found on this topic don’t explain it from the beginning, so it took me quite a while to figure out how simple and powerful the sqlite fulltext search API (FTS3) actually is.

get the sqlite3 amalgamation sources (v3.6.22) and just unpack it into your iPhone project – sqlite3 as shipped on the iPhone doesn’t support FTS3,
activate FTS3 support,
create the index table programmatically inside your App as your Macs default sqlite3 doesn’t support FTS3 either,
use it like any other table except using the MATCH operator for index queries.

That’s it, and it’s hell quick.

Poignant: Xml meets Huffman

2010-02-02T20:28:41Z

There’s a spec at the w3c about compressing (XML) named “Efficient XML Interchange” Format taking into account the grammar and likelihood of atoms within the document. They indeed use something similar the the Huffman Coding.

The results are quite impressive – nice charts!

T-Mobile / MMS / iPhone

2010-01-29T09:36:32Z

klick ich den Link, den T-Mobile mir per SMS geschickt hat, kriege ich:

T-Mobile Scherz

NSCachedURLResponse / NSKeyedUnarchiver pain

2010-02-09T07:21:00Z

as the iPhone SDK comes with a rather dysfunctional NSURLCache — Apple suggests to implement it from scratch yourself in the code examples about caching — I went for just this.

Until I came across the [NSKeyedUnarchiver unarchiveObjectWithData:...] not restoring userInfo, storagePolicy and data of NSCachedURLResponse.

Couldn’t believe it and spent almost the whole day verifying that the error is not within my code but really the unarchiver treats those fields transient.

See yourself:

-(void)testiPhoneSDK312NSKeyedArchiver
{
//  prepare a NSCachedURLResponse
    NSCachedURLResponse *src = nil;
    {
        NSURL *i_url = [NSURL URLWithString:@"http://www.url.example/path/file.html"];
        NSString *i_mime = @"text/html";
        NSInteger expectedContentLength = 112;
        NSString *i_encoding = @"iso-8859-2";
 
        NSURLResponse *i_response = [[NSURLResponse alloc] initWithURL:i_url
            MIMEType:i_mime
            expectedContentLength:expectedContentLength
            textEncodingName:i_encoding];
        NSData *i_data = [@"Hello, world!" dataUsingEncoding:NSISOLatin2StringEncoding];
        NSURLCacheStoragePolicy i_storage =  NSURLCacheStorageAllowed;
        NSDictionary *i_userInfo = [NSDictionary dictionaryWithObject:
        [NSDate dateWithTimeIntervalSince1970:13] forKey:@"era"];
 
        NSCachedURLResponse *src = [[NSCachedURLResponse alloc] initWithResponse:i_response
            data:i_data
            userInfo:i_userInfo
            storagePolicy:i_storage];
 
//      ensure it's all in place:
        STAssertEqualObjects(i_url, src.response.URL, @"");
        STAssertEqualObjects(i_mime, src.response.MIMEType, @"");
        STAssertEquals((int)expectedContentLength, (int)src.response.expectedContentLength, @"");
        STAssertEqualObjects(@"file.html", src.response.suggestedFilename, @"");
        STAssertEqualObjects(i_encoding, src.response.textEncodingName, @"");
 
        STAssertEqualObjects(i_data, src.data, @"");
        STAssertEqualObjects(i_userInfo, src.userInfo, @"");
        STAssertEquals( 0u, src.storagePolicy, @"");
    }
 
//  archive + unarchive:
    NSCachedURLResponse *dst = [NSKeyedUnarchiver unarchiveObjectWithData:
      [NSKeyedArchiver archivedDataWithRootObject:src]];
 
//  check whether src == dst
    STAssertEqualObjects(src.response.URL, dst.response.URL, @"");
    STAssertEqualObjects(src.response.MIMEType, dst.response.MIMEType, @"");
    STAssertEquals(src.response.expectedContentLength, dst.response.expectedContentLength, @"");
    STAssertEqualObjects(src.response.suggestedFilename, dst.response.suggestedFilename, @"");
    STAssertEqualObjects(src.response.textEncodingName, dst.response.textEncodingName, @"");
 
//  !!!!!!!!!!
//  sad information loss after unarchiving:
    STAssertNil( dst.data, @"" );
    STAssertNil( dst.userInfo, @"" );
    STAssertEquals( 2u , dst.storagePolicy, @"" );
}

When my NSURLCache replacement is ready I think about publishing it a github — interested anyone?