I am working in Linux and I have a library written in Fortran 90 (written by 3rd party), that is reading from a file in the current working directory. I would like to be able to call the resulting library from other folders, hence I need to read the path where the library is installed. How can I know the path to the compiled library within the Fortran code?
I need to store in a variable the path within the code.
For who knows python, I want to achieve the same as
import os
os.path.dirname(os.path.abspath(__file__))
but in f90 (see Get location of the .py source file)
Using the suggestions in the comment I have done the following:
export DATAPATH=`pwd`
make
in the Makefile
ifort -O3 -fpic -fpp -DDATAPATH -c mysource.f90
in mysource.f90
subroutine mysub
character(len=100)::dpath
#ifdef DATAPATH
dpath=DATAPATH
#endif
open(10,file=trim(dpath)//'initialise.dat')
....
....
the problem is that at compile time I get
mysource.f90(42): error #6054: A CHARACTER data type is required in this context. [1]
dpath=1
----------^
compilation aborted for mysource.f90 (code 1)
If you wish you can fix the path at compile time. Something like
gfortran -cpp mylib.f90 -DPREFIX=\"/usr/local/\"
open(newunit=u,file=PREFIX//'mylib/initialise.dat')
You must than make sure the library is indeed installed in that place PREFIX/mylib/
You can create an environment variable containing the path of your data. This variable can be set by hand, in your .bashrc or .bash_profile or in the system /etc/profile.d/ or /etc/bash.bashrc, there are manyways and they depend if the library is just for one user or for all users of some large computer.
For example
export MYLIB_PATH='/usr/local/mylib'
Then you can read the variable in Fortran as
CALL get_environment_variable("MYLIB_PATH", mylib_path, status=stat)
and the path is now in variable mylib_path. You can check the success by checking if stat==0.
This is not the only possible approach. You can also have a configuration file for your library in your home directory:
mkdir $HOME/.config/mylib/
echo "/usr/local/mylib" > $HOME/.config/mylib/path
and then you can try to read the path from this file if the environment variable was not set
if (stat/=0) then
CALL get_environment_variable("HOME", home_dir)
open(newunit=path_unit, file=home_dir//'/.config/mylib/path',status='old',action='read',iostat=stat)
if (stat/=0) complain
read(path_unit,'(a)',iostat=stat) mylib_path
if (stat/=0) complain
end if
So when you compiled with -DDATAPATH you have not passed the variable DATAPATH into your code only declared a symbol called DATAPATH as being true, so ifort will substitute DATAPATH as 1. What you need to do is pass it as a value:
-DDATAPATH="$(DATAPATH)"
For the compilation to work.
Sorry for asking this newbie question but I can't get off this s***t...
In the same directory I have 4 files : ctacs.ini; ct_api.h; libctacs.a and main.cpp.
My cpp file contains #include "ct_api.h" and when I try to compile with :
g++ -lctacs main.cpp -o main
I got undefined references to the functions which are defined in my library -__-
Could you please tell what did I wrong ? I search on the internet but the option -lctacs seems to be the right way to proceed...
Thank you very much
Some compilers and linkers resolve references to functions by searching the object files/sources files/libraries left-to-right on the command line. This means that files that call externally-defined functions should appear before the object files/libraries/source files that contain their definitions.
You happen to have a linker that does indeed depend on this ordering. You need to put the library after main.cpp so that the function definitions can be found:
g++ main.cpp -lctacs -o main
How do you import CommonCrypto in a Swift framework for iOS?
I understand how to use CommonCrypto in a Swift app:
You add #import <CommonCrypto/CommonCrypto.h> to the bridging header.
However, Swift frameworks don't support bridging headers. The documentation says:
You can import external frameworks that have a pure Objective-C codebase, a pure Swift codebase, or a mixed-language codebase. The
process for importing an external framework is the same whether the
framework is written in a single language or contains files from both
languages. When you import an external framework, make sure the
Defines Module build setting for the framework you’re importing is set
to Yes.
You can import a framework into any Swift file within a different
target using the following syntax:
import FrameworkName
Unfortunately, import CommonCrypto doesn't work. Neither does adding #import <CommonCrypto/CommonCrypto.h> to the umbrella header.
Something a little simpler and more robust is to create an Aggregate target called "CommonCryptoModuleMap" with a Run Script phase to generate the module map automatically and with the correct Xcode/SDK path:
The Run Script phase should contain this bash:
# This if-statement means we'll only run the main script if the CommonCryptoModuleMap directory doesn't exist
# Because otherwise the rest of the script causes a full recompile for anything where CommonCrypto is a dependency
# Do a "Clean Build Folder" to remove this directory and trigger the rest of the script to run
if [ -d "${BUILT_PRODUCTS_DIR}/CommonCryptoModuleMap" ]; then
echo "${BUILT_PRODUCTS_DIR}/CommonCryptoModuleMap directory already exists, so skipping the rest of the script."
exit 0
fi
mkdir -p "${BUILT_PRODUCTS_DIR}/CommonCryptoModuleMap"
cat <<EOF > "${BUILT_PRODUCTS_DIR}/CommonCryptoModuleMap/module.modulemap"
module CommonCrypto [system] {
header "${SDKROOT}/usr/include/CommonCrypto/CommonCrypto.h"
export *
}
EOF
Using shell code and ${SDKROOT} means you don't have to hard code the Xcode.app path which can vary system-to-system, especially if you use xcode-select to switch to a beta version, or are building on a CI server where multiple versions are installed in non-standard locations. You also don't need to hard code the SDK so this should work for iOS, macOS, etc. You also don't need to have anything sitting in your project's source directory.
After creating this target, make your library/framework depend on it with a Target Dependencies item:
This will ensure the module map is generated before your framework is built.
macOS note: If you're supporting macOS as well, you'll need to add macosx to the Supported Platforms build setting on the new aggregate target you just created, otherwise it won't put the module map in the correct Debug derived data folder with the rest of the framework products.
Next, add the module map's parent directory, ${BUILT_PRODUCTS_DIR}/CommonCryptoModuleMap, to the "Import Paths" build setting under the Swift section (SWIFT_INCLUDE_PATHS):
Remember to add a $(inherited) line if you have search paths defined at the project or xcconfig level.
That's it, you should now be able to import CommonCrypto
Update for Xcode 10
Xcode 10 now ships with a CommonCrypto module map making this workaround unnecessary. If you would like to support both Xcode 9 and 10 you can do a check in the Run Script phase to see if the module map exists or not, e.g.
COMMON_CRYPTO_DIR="${SDKROOT}/usr/include/CommonCrypto"
if [ -f "${COMMON_CRYPTO_DIR}/module.modulemap" ]
then
echo "CommonCrypto already exists, skipping"
else
# generate the module map, using the original code above
fi
You can actually build a solution that "just works" (no need to copy a module.modulemap and SWIFT_INCLUDE_PATHS settings over to your project, as required by other solutions here), but it does require you to create a dummy framework/module that you'll import into your framework proper. We can also ensure it works regardless of platform (iphoneos, iphonesimulator, or macosx).
Add a new framework target to your project and name it after the system library, e.g., "CommonCrypto". (You can delete the umbrella header, CommonCrypto.h.)
Add a new Configuration Settings File and name it, e.g., "CommonCrypto.xcconfig". (Don't check any of your targets for inclusion.) Populate it with the following:
MODULEMAP_FILE[sdk=iphoneos*] = \
$(SRCROOT)/CommonCrypto/iphoneos.modulemap
MODULEMAP_FILE[sdk=iphonesimulator*] = \
$(SRCROOT)/CommonCrypto/iphonesimulator.modulemap
MODULEMAP_FILE[sdk=macosx*] = \
$(SRCROOT)/CommonCrypto/macosx.modulemap
Create the three referenced module map files, above, and populate them with the following:
iphoneos.modulemap
module CommonCrypto [system] {
header "/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS.sdk/usr/include/CommonCrypto/CommonCrypto.h"
export *
}
iphonesimulator.modulemap
module CommonCrypto [system] {
header "/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator.sdk/usr/include/CommonCrypto/CommonCrypto.h"
export *
}
macosx.modulemap
module CommonCrypto [system] {
header "/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.11.sdk/usr/include/CommonCrypto/CommonCrypto.h"
export *
}
(Replace "Xcode.app" with "Xcode-beta.app" if you're running a beta version. Replace 10.11 with your current OS SDK if not running El Capitan.)
On the Info tab of your project settings, under Configurations, set the Debug and Release configurations of CommonCrypto to CommonCrypto (referencing CommonCrypto.xcconfig).
On your framework target's Build Phases tab, add the CommonCrypto framework to Target Dependencies. Additionally add libcommonCrypto.dylib to the Link Binary With Libraries build phase.
Select CommonCrypto.framework in Products and make sure its Target Membership for your wrapper is set to Optional.
You should now be able to build, run and import CommonCrypto in your wrapper framework.
For an example, see how SQLite.swift uses a dummy sqlite3.framework.
I found a GitHub project that successfully uses CommonCrypto in a Swift framework: SHA256-Swift. Also, this article about the same problem with sqlite3 was useful.
Based on the above, the steps are:
1) Create a CommonCrypto directory inside the project directory. Within, create a module.map file. The module map will allow us to use the CommonCrypto library as a module within Swift. Its contents are:
module CommonCrypto [system] {
header "/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator8.0.sdk/usr/include/CommonCrypto/CommonCrypto.h"
link "CommonCrypto"
export *
}
2) In Build Settings, within Swift Compiler - Search Paths, add the CommonCrypto directory to Import Paths (SWIFT_INCLUDE_PATHS).
3) Finally, import CommonCrypto inside your Swift files as any other modules. For example:
import CommonCrypto
extension String {
func hnk_MD5String() -> String {
if let data = self.dataUsingEncoding(NSUTF8StringEncoding)
{
let result = NSMutableData(length: Int(CC_MD5_DIGEST_LENGTH))
let resultBytes = UnsafeMutablePointer<CUnsignedChar>(result.mutableBytes)
CC_MD5(data.bytes, CC_LONG(data.length), resultBytes)
let resultEnumerator = UnsafeBufferPointer<CUnsignedChar>(start: resultBytes, length: result.length)
let MD5 = NSMutableString()
for c in resultEnumerator {
MD5.appendFormat("%02x", c)
}
return MD5
}
return ""
}
}
Limitations
Using the custom framework in another project fails at compile time with the error missing required module 'CommonCrypto'. This is because the CommonCrypto module does not appear to be included with the custom framework. A workaround is to repeat step 2 (setting Import Paths) in the project that uses the framework.
The module map is not platform independent (it currently points to a specific platform, the iOS 8 Simulator). I don't know how to make the header path relative to the current platform.
Updates for iOS 8 <= We should remove the line link "CommonCrypto", to get the successful compilation.
UPDATE / EDIT
I kept getting the following build error:
ld: library not found for -lCommonCrypto for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
Unless I removed the line link "CommonCrypto" from the module.map file I created. Once I removed this line it built ok.
This answer discusses how to make it work inside a framework, and with Cocoapods and Carthage
🐟 modulemap approach
I use modulemap in my wrapper around CommonCrypto https://github.com/onmyway133/arcane, https://github.com/onmyway133/Reindeer
For those getting header not found, please take a look https://github.com/onmyway133/Arcane/issues/4 or run xcode-select --install
Make a folder CCommonCrypto containing module.modulemap
module CCommonCrypto {
header "/usr/include/CommonCrypto/CommonCrypto.h"
export *
}
Go to Built Settings -> Import Paths
${SRCROOT}/Sources/CCommonCrypto
🌳 Cocoapods with modulemap approach
Here is the podspec https://github.com/onmyway133/Arcane/blob/master/Arcane.podspec
s.source_files = 'Sources/**/*.swift'
s.xcconfig = { 'SWIFT_INCLUDE_PATHS' =>
'$(PODS_ROOT)/CommonCryptoSwift/Sources/CCommonCrypto' }
s.preserve_paths = 'Sources/CCommonCrypto/module.modulemap'
Using module_map does not work, see https://github.com/CocoaPods/CocoaPods/issues/5271
Using Local Development Pod with path does not work, see https://github.com/CocoaPods/CocoaPods/issues/809
That's why you see that my Example Podfile https://github.com/onmyway133/CommonCrypto.swift/blob/master/Example/CommonCryptoSwiftDemo/Podfile points to the git repo
target 'CommonCryptoSwiftDemo' do
pod 'CommonCryptoSwift', :git => 'https://github.com/onmyway133/CommonCrypto.swift'
end
🐘 public header approach
Ji is a wrapper around libxml2, and it uses public header approach
It has a header file https://github.com/honghaoz/Ji/blob/master/Source/Ji.h with Target Membership set to Public
It has a list of header files for libxml2 https://github.com/honghaoz/Ji/tree/master/Source/Ji-libxml
It has Build Settings -> Header Search Paths
$(SDKROOT)/usr/include/libxml2
It has Build Settings -> Other Linker Flags
-lxml2
🐏 Cocoapods with public header approach
Take a look at the podspec https://github.com/honghaoz/Ji/blob/master/Ji.podspec
s.libraries = "xml2"
s.xcconfig = { 'HEADER_SEARCH_PATHS' => '$(SDKROOT)/usr/include/libxml2', 'OTHER_LDFLAGS' => '-lxml2' }
🐝 Interesting related posts
How to call C from Swift?
https://spin.atomicobject.com/2015/02/23/c-libraries-swift/
Good news! Swift 4.2 (Xcode 10) finally provides CommonCrypto!
Just add import CommonCrypto in your swift file.
WARNING: iTunesConnect may reject apps that are using this method.
New member on my team accidentally broke the solution given by one of the top answers, so I decided to consolidate it in a small wrapper project called CommonCryptoModule. You can install it manually or via Cocoapods:
pod 'CommonCryptoModule', '~> 1.0.2'
Then, all you have to do is to import the module where you need CommonCrypto, like so:
import CommonCryptoModule
Hope someone else finds this useful.
I think I have an improvement to Mike Weller's excellent work.
Add a Run Script phase before the Compile Sources phase containing this bash:
# This if-statement means we'll only run the main script if the
# CommonCrypto.framework directory doesn't exist because otherwise
# the rest of the script causes a full recompile for anything
# where CommonCrypto is a dependency
# Do a "Clean Build Folder" to remove this directory and trigger
# the rest of the script to run
FRAMEWORK_DIR="${BUILT_PRODUCTS_DIR}/CommonCrypto.framework"
if [ -d "${FRAMEWORK_DIR}" ]; then
echo "${FRAMEWORK_DIR} already exists, so skipping the rest of the script."
exit 0
fi
mkdir -p "${FRAMEWORK_DIR}/Modules"
cat <<EOF > "${FRAMEWORK_DIR}/Modules/module.modulemap"
module CommonCrypto [system] {
header "${SDKROOT}/usr/include/CommonCrypto/CommonCrypto.h"
export *
}
EOF
ln -sf "${SDKROOT}/usr/include/CommonCrypto" "${FRAMEWORK_DIR}/Headers"
This script constructs a bare bones framework with the module.map in the correct place and then relies on Xcode's automatic search of BUILT_PRODUCTS_DIR for frameworks.
I linked the original CommonCrypto include folder as the framework's Headers folder so the result should also function for Objective C projects.
For anyone using swift 4.2 with Xcode 10:
CommonCrypto module is now provided by the system, so you can directly import it like any other system framework.
import CommonCrypto
#mogstad has been kind enough to wrap #stephencelis solution in a Cocoapod:
pod 'libCommonCrypto'
The other pods available did not work for me.
The modulemap solutions can be good, and are robust against SDK changes, but I've found them awkward to use in practice, and not as reliable as I'd like when handing things out to others. To try to make it all more foolproof, I went a different way:
Just copy the headers.
I know, fragile. But Apple almost never makes significant changes to CommonCrypto and I'm living the dream that they will not change it in any significant way without also finally making CommonCrypto a modular header.
By "copy the headers" I mean "cut and paste all of the headers you need into one massive header in your project just like the preprocessor would do." As an example of this that you can copy or adapt, see RNCryptor.h.
Note that all of these files are licensed under APSL 2.0, and this approach intentionally maintains the copyright and license notices. My concatenation step is licensed under MIT, and that only applies up to the next license notice).
I am not saying this is a beautiful solution, but so far it seems to have been an incredibly simple solution to both implement and support.
I know this is an old question. But I figure out an alternative way to use the library in Swift project, which might be helpful for those who don't want to import framework introduced in these answers.
In Swift project, create a Objective-C bridging header, create NSData category (or custom class that to use the library) in Objective-C. The only drawback would be that you have to write all implementation code in Objective-C.
For example:
#import "NSData+NSDataEncryptionExtension.h"
#import <CommonCrypto/CommonCryptor.h>
#implementation NSData (NSDataEncryptionExtension)
- (NSData *)AES256EncryptWithKey:(NSString *)key {
//do something
}
- (NSData *)AES256DecryptWithKey:(NSString *)key {
//do something
}
And then in your objective-c bridging header, add this
#import "NSData+NSDataEncryptionExtension.h"
And then in Swift class do similar thing:
public extension String {
func encryp(withKey key:String) -> String? {
if let data = self.data(using: .utf8), let encrypedData = NSData(data: data).aes256Encrypt(withKey: key) {
return encrypedData.base64EncodedString()
}
return nil
}
func decryp(withKey key:String) -> String? {
if let data = NSData(base64Encoded: self, options: []), let decrypedData = data.aes256Decrypt(withKey: key) {
return decrypedData.UTF8String
}
return nil
}
}
It works as expected.
I've added some cocoapods magic to jjrscott's answer in case you need to use CommonCrypto in your cocoapods library.
1) Add this line to your podspec:
s.script_phase = { :name => 'CommonCrypto', :script => 'sh $PROJECT_DIR/../../install_common_crypto.sh', :execution_position => :before_compile }
2) Save this in your library folder or wherever you like (however don't forget to change the script_phase accordingly ...)
# This if-statement means we'll only run the main script if the
# CommonCrypto.framework directory doesn't exist because otherwise
# the rest of the script causes a full recompile for anything
# where CommonCrypto is a dependency
# Do a "Clean Build Folder" to remove this directory and trigger
# the rest of the script to run
FRAMEWORK_DIR="${BUILT_PRODUCTS_DIR}/CommonCrypto.framework"
if [ -d "${FRAMEWORK_DIR}" ]; then
echo "${FRAMEWORK_DIR} already exists, so skipping the rest of the script."
exit 0
fi
mkdir -p "${FRAMEWORK_DIR}/Modules"
echo "module CommonCrypto [system] {
header "${SDKROOT}/usr/include/CommonCrypto/CommonCrypto.h"
export *
}" >> "${FRAMEWORK_DIR}/Modules/module.modulemap"
ln -sf "${SDKROOT}/usr/include/CommonCrypto" "${FRAMEWORK_DIR}/Headers"
Works like a charm :)
I'm not sure if something's changed with Xcode 9.2 but it's now much simpler to achieve this. The only things I had to do are create a folder called "CommonCrypto" in my framework project directory and create two files inside it, one called "cc.h" as follows:
#include <CommonCrypto/CommonCrypto.h>
#include <CommonCrypto/CommonRandom.h>
And another called module.modulemap:
module CommonCrypto {
export *
header "cc.h"
}
(I don't know why you can't reference header files from the SDKROOT area directly in a modulemap file but I couldn't get it to work)
The third thing is to find the "Import Paths" setting and set to $(SRCROOT).
In fact you can set it to whatever folder you want the CommonCrypto folder to be under, if you don't want it at the root level.
After this you should be able to use
import CommonCrypto
In any swift file and all the types/functions/etc. are available.
A word of warning though - if your app uses libCommonCrypto (or libcoreCrypto) it's exceptionally easy for a not-too-sophisticated hacker to attach a debugger to your app and find out what keys are being passed to these functions.
In case you have the below issue :
ld: library not found for -lapple_crypto
clang: error: linker command failed with exit code 1 (use -v to see invocation)
In Xcode 10, Swift 4.0. CommonCrypto is a part of the framework.
Add
import CommonCrypto
Remove
CommonCrpto lib file from link binary with libraries from Build
phases
import CommonCrypto from Bridging header
This worked for me!
It happened the same to me after updating Xcode.
I tried everything I can do such as reinstalling cocoapods and cleaning the project, but it didn't work.
Now it's been solved after restart the system.
It's very simple. Add
#import <CommonCrypto/CommonCrypto.h>
to a .h file (the bridging header file of your project). As a convention you can call it YourProjectName-Bridging-Header.h.
Then go to your project Build Settings and look for Swift Compiler - Code Generation. Under it, add the name of your bridging header to the entry "Objetive-C Bridging Header".
You're done. No imports required in your Swift code. Any public Objective-C headers listed in this bridging header file will be visible to Swift.
I'm making a project which uses lots of it's own shared libraries; my intention is to create a library name within a directory, so, e.g., instead of -lfoo (to find /usr/lib/libfoo.so or /opt/lib/libfoo.so or so on), I would use -lfoo/bar (to find /usr/lib/libfoo/bar.so or /opt/lib/libfoo/bar.so or so on).
I made a real small code to test:
const char *mylib(void) {
return "it woooorks! =D";
};
And compiled it with: gcc -fPIC -shared -Wl,-rpath,libfoo/ lib.c -o /usr/lib/libfoo/bar.so.
Then, in the test program, I use gcc -lfoo/bar test.c, and it compiles (it finds the mylib() symbol from my library), but when I try to run the program (./a.out), the dynamic linker complains that it can't find the library. In my case, using Mac OS X Lion:
dyld: Library not loaded: bar.so
Referenced from: /Users/takanuva/tmp/lib/./a.out
Reason: image not found
Trace/BPT trap: 5
What am I doing wrong? Maybe the answer is "everything", so... how should I accomplish the desired effect, to look for libfoo/bar.so instead of libfoo.so on the library paths?
Thanks in advance. :)
If the library is not located in a standard path (and /opt is such a path that is normally not searched), you must specify the paths with -L/-rpath, as in gcc blah.c -L/opt/whatever -Wl,-rpath,/opt/whatever -lbar. Do NOT attempt to use path separators in the -l argument.
I would like to use the -rpath linker option to set the .so search path. I'd like it to search [app_path]/../lib. I've tried add this to my qmake .pro file:
QMAKE_LFLAGS += -Wl,-rpath=$ORIGIN/../lib/
But qmake links my project this way:
g++ -Wl,-rpath=RIGIN/../lib/ -Wl,-O1 -o myoutput main.o [...]
How can I escape the $ORIGIN?
I found here a way to properly escape it:
QMAKE_LFLAGS += '-Wl,-rpath,\'\$$ORIGIN/../mylibs\''
If you want $ORIGIN to be (properly) evaluated while building you can simply add this to your .pro file:
QMAKE_RPATHDIR += $ORIGIN/../mylibs
This is a really old question, but for people getting here through a search: the methods described in the old answers are not needed anymore. Modern Qt versions (5.9 in my case), allow you to just use this:
QMAKE_RPATHDIR += lib
This will add the needed entries (including $ORIGIN and -Wl,-z,origin where needed) automatically to the makefile, as long as you're using a relative directory. Meaning that lib will produce the needed "origin" entries, while something like /lib will not. Any relative directory you add to QMAKE_RPATHDIR will be made relative to $ORIGIN automatically.
DOLLAR=$
QMAKE_LFLAGS += -Wl,-rpath=$${DOLLAR}$${DOLLAR}ORIGIN/../myLibs