I have a Phing project that you pass in a parameter. I want to perform simple string manipulation on this parameter such a strtolower() or ucwords() etc. Any ideas how I can go about this?
How about using the PhpEvaLTask:
<project name="StringTest" default="all" basedir=".">
<target name="stringtest" description="test">
<php expression="strtolower(${param})" returnProperty="paramToLower"/>
<php expression="ucwords(${param})" returnProperty="paramUcwords"/>
<echo>To lower ${paramToLower}</echo>
<echo>UcWords ${paramUcwords}</echo>
</target>
Running it with:
phing -Dparam=BLAH stringtest
Yields:
Buildfile: /export/users/marcelog/build.xml
StringTest > stringtest:
[php] Evaluating PHP expression: strtolower(BLAH)
[php] Evaluating PHP expression: ucwords(BLAH)
[echo] To lower blah
[echo] UcWords BLAH
BUILD FINISHED
Another way to do this:
<php function="strtolower" returnProperty="paramToLower">
<param value="${param}" />
</php>
<php function="ucwords" returnProperty="paramUcwords">
<param value="${param}" />
</php>
Related
As part of a CI process I am trying to create a buildlabel which consists of the content of an xml element within an xml structure. For this purpose I am using nant and xmlpeek. My problem is that I get an odd error stating:
"Nodeindex '0' is out of range"
This is only the case if the xml file I am xmlpeeking contains a namespace definition in the root node.
Removing the namespace from the xml file gives me the output I expect.
The nant target that generates the error can be boild down to:
<target name="TDSLabel">
<property name="element" value=""/>
<echo message="Getting element" />
<xmlpeek file="C:\xxx\test1.xml" xpath="//Project/PropertyGroup/ProductVersion" property="element"/>
<echo message="The found element value was: ${element}" />
</target>
and the test1.xml file looks like this:
<Project ToolsVersion="3.5" DefaultTargets="Build" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
<PropertyGroup>
<ProductVersion>9.0.21022</ProductVersion>
</PropertyGroup>
</Project>
You already gave the right hint yourself. It's about the namespace. This should fix it:
<target name="TDSLabel">
<property name="element" value=""/>
<echo message="Getting element" />
<xmlpeek
file="C:\xxx\test1.xml"
xpath="//x:Project/x:PropertyGroup/x:ProductVersion"
property="element"
verbose="true">
<namespaces>
<namespace prefix="x" uri="http://schemas.microsoft.com/developer/msbuild/2003" />
</namespaces>
</xmlpeek>
<echo message="The found element value was: ${element}" />
</target>
Found a similar problem and the anwser to my problem here: XmlPoke and unique nodes. The problem was that I did not include the namespace definition within the xmlpeek element and afterwards omitted the necessary reference to the namespace in my xpath statement:
<xmlpeek file="C:\xxx\test1.xml" xpath="//x:Project/x:PropertyGroup/x:ProductVersion" property="element">
<namespaces>
<namespace prefix="x" uri="http://schemas.microsoft.com/developer/msbuild/2003" />
</namespaces>
</xmlpeek>
What I am trying to do, is to find a file with NAnt. This file could by anywhere in a directory structure of a given folder.
I tried to this with the NAnt-foreach task (this works) but I am not quite convinced of that:
<target name="find-file">
<fail message="Property param.dir must be set" unless="${property::exists('param.dir')}" />
<fail message="Property param.pattern must be set" unless="${property::exists('param.pattern')}" />
<property name="return.file" value="" />
<foreach item="File" property="iterator.file">
<in>
<items>
<include name="${param.dir}\**\${param.pattern}" />
</items>
</in>
<do>
<property name="return.file" value="${iterator.file}" if="${string::get-length(return.file) == 0}" />
</do>
</foreach>
</target>
Is there anybody aware of a better approach? If not how can I accomplish to exit the foreach-loop after the first element is found?
This nantcontrib function will put the matching filenames into a delimited string..
If you know that only one matching file will exist then it may get you what you want. If there are several then you could use the nant substring function to just get the first match by taking the substring up to the first delimiter.
The following nant script:
<?xml version="1.0" encoding="utf-8"?>
<project default="find-file2">
<property name="NantContrib.dir" value="C:\Program Files\nantcontrib-0.85\" readonly="true" />
<target name="LoadNantContrib">
<loadtasks assembly="${NantContrib.dir}bin\NAnt.Contrib.Tasks.dll" />
</target>
<target name="find-file2" depends="LoadNantContrib">
<fileset id="find.set">
<include name="${param.dir}\**\${param.pattern}" />
</fileset>
<property name="return.file" value="${fileset::to-string('find.set', ' | ')}" />
<echo message="return.file=${return.file}"/>
<echo message="Found ${fileset::get-file-count('find.set')} files"/>
</target>
</project>
...and the following folder structure:
\---folderroot
+---folder1
| dontfindme.txt
| findme.txt
|
+---folder2
| dontfindme.txt
|
\---folderempty
...works as expected. Searching for findme.txt finds one file. Searching for dontfindme.txt finds two files. Searching for *.txt finds three files.
Example call:
nant -D:param.dir=folderroot -D:param.pattern=findme.txt
Example output:
find-file2:
[echo] return.file=C:\Documents and Settings\rbaker\My Documents\nantfindfile\folderroot\folder1\findme.txt
[echo] Found 1 files
BUILD SUCCEEDED
I'm new to ant, and I want to require a file name if something other than the default target is used, so the calling syntax would be something like this:
ant specifictarget -Dfile=myfile
I'm using the ant contrib package to give me additional functionality, so I have this:
<if>
<equals arg1="${file}" arg2="" />
<then>
<!-- fail here -->
</then>
</if>
My thinking is that if file was not specified it might be equal to the empty string. Obviously, this didn't work, and I'm not finding any examples on google or the right syntax in the manual.
So what syntax should I use?
You don't really need the contrib package. This is more conveniently done using built-in ant capabilities like if/unless and depends. See below:
<target name="check" unless="file" description="check that the file property is set" >
<fail message="set file property in the command line (e.g. -Dfile=someval)"/>
</target>
<target name="specifictarget" if="file" depends="check" description=" " >
<echo message="do something ${file}"/>
</target>
You've got the right idea. The
ant specifictarget -Dfile=myfile
sets Ant Properties from the command line. All you really need is
<property name="file" value=""/>
for your default value. That way if file is not specified, it will be equal to the empty string.
Since properties are not mutable in Ant, you can add this:
<property name="file" value="" />
This will set the property file to an empty string if it hasn't already been set on the command line. Then your equality test will work as you intended.
Alternately, you can use escape the value since ant just spits out the actual text when it can't do a property substitution.
<if>
<equals arg1="${file}" arg2="$${file}" />
<then>
<echo>BARF!</echo>
</then>
</if>
Summary:
How do I expand a property with value "download\${bulidmode}\project\setup.msi" to "download\Debug\project\setup.msi" if the property buildmode contained debug so I can use it as the file="" part of < copy >
Detail:
I have a bit of a requirement to be able to expand properties within a string in nant.
For example I have a target that is copying file A to B. A and B both come from a simple two field CSV file which I'm iterating through using
<foreach item="Line" in="filelist.csv" delim="," property="source.file,target.file">
<property name="sourcefile" value="${path::combine(source.dir,source)}" />
<property name="targetfile" value="${path::combine(download.dir,destination)}" />
<echo message="Copy ${sourcefile} to ${targetfile}" />
<copy file="${sourcefile" tofile="${destination}" />
</foreach>
and the filelist.csv will be
build\manifest.xml
solutiondirectory\setup-proj-directory\Release\setupproj.msi,ProductA\ProductA.msi
solutiondirectory\another-proj-dir\Release\setupproj.msi,ProductB\ProductB.msi
(The reason we split these out is that we write multi-tiered applications and deploy by MSI to each tier - so one product has multiple msi's all built with the same version numbers)
Anyway - I want to change this to that I no longer have "Release" in the filelist.csv file but something like ${build.mode}. I would wrap the above code with a
<foreach item="String" in="Release,Debug" delim="," property="build.mode">
....as above
</foreach>
and the property embedded within the string in the file gets expanded.
I've been beating my head against a brick wall for a few hours, but just can't figure it out.
Thanks
It is possible with a custom function :
<?xml version="1.0"?>
<project>
<script language="C#" prefix="vbfox" >
<code>
<![CDATA[
[Function("expand")]
public string ExpandString(string str)
{
return Project.Properties.ExpandProperties(str, Location.UnknownLocation);
}
]]>
</code>
</script>
<property name="hello" value="{path::combine('_hello_', '_world_')}" />
<property name="hello" value="${'$' + hello}" />
<echo message="${hello}" />
<echo message="${vbfox::expand(hello)}" />
</project>
When I do a ReadLinesFromFile on a file in MSBUILD and go to output that file again, I get all the text on one line. All the Carriage returns and LineFeeds are stripped out.
<Project DefaultTargets = "Deploy"
xmlns="http://schemas.microsoft.com/developer/msbuild/2003" >
<Import Project="$(MSBuildExtensionsPath)\MSBuildCommunityTasks\MSBuild.Community.Tasks.Targets"/>
<ItemGroup>
<MyTextFile Include="$(ReleaseNotesDir)$(NewBuildNumber).txt"/>
</ItemGroup>
<Target Name="ReadReleaseNotes">
<ReadLinesFromFile
File="#(MyTextFile)" >
<Output
TaskParameter="Lines"
ItemName="ReleaseNoteItems"/>
</ReadLinesFromFile>
</Target>
<Target Name="MailUsers" DependsOnTargets="ReadReleaseNotes" >
<Mail SmtpServer="$(MailServer)"
To="$(MyEMail)"
From="$(MyEMail)"
Subject="Test Mail Task"
Body="#(ReleaseNoteItems)" />
</Target>
<Target Name="Deploy">
<CallTarget Targets="MailUsers" />
</Target>
</Project>
I get the text from the file which normally looks like this
- New Deployment Tool for BLAH
- Random other stuff()""
Coming out like this
- New Deployment Tool for BLAH;- Random other stuff()""
I know that the code for ReadLinesFromFile will pull the data in one line at a time and strip out the carriage returns.
Is there a way to put them back in?
So my e-mail looks all nicely formatted?
Thanks
The problem here is you are using the ReadLinesFromFile task in a manner it wasn't intended.
ReadLinesFromFile Task
Reads a list of items from a text file.
So it's not just reading all the text from a file, it's reading individual items from a file and returning an item group of ITaskItems. Whenever you output a list of items using the #() syntax you will get a separated list, the default of which is a semicolon. This example illustrates this behavior:
<?xml version="1.0" encoding="utf-8"?>
<Project DefaultTargets="Build" xmlns="http://schemas.microsoft.com/developer/msbuild/2003" ToolsVersion="3.5">
<ItemGroup>
<Color Include="Red" />
<Color Include="Blue" />
<Color Include="Green" />
</ItemGroup>
<Target Name="Build">
<Message Text="ItemGroup Color: #(Color)" />
</Target>
</Project>
And the output looks like this:
ItemGroup Color: Red;Blue;Green
So while the best solution to your problem is to write an MSBuild task that reads a file into a property as a string an not a list of items, that's really not what you asked for. You asked if there was a way to put them back, and there is using MSBuild Transforms.
Transforms are used to create one list from another and also have the ability to transform using a custom separator. So the answer is to transform your list read in using ReadItemsFromFile into another list with newlines. Here is an example that does just that:
<?xml version="1.0" encoding="utf-8"?>
<Project DefaultTargets="Build" xmlns="http://schemas.microsoft.com/developer/msbuild/2003" ToolsVersion="3.5">
<ItemGroup>
<File Include="$(MSBuildProjectDirectory)\Test.txt" />
</ItemGroup>
<Target Name="Build">
<ReadLinesFromFile File="#(File)">
<Output TaskParameter="Lines" ItemName="FileContents" />
</ReadLinesFromFile>
<Message Text="FileContents: #(FileContents)" />
<Message Text="FileContents Transformed: #(FileContents->'%(Identity)', '%0a%0d')" />
</Target>
</Project>
Test.text looks like:
Red
Green
Blue
And the output looks like this:
[C:\temp]:: msbuild test.proj
Microsoft (R) Build Engine Version 3.5.21022.8
[Microsoft .NET Framework, Version 2.0.50727.1433]
Copyright (C) Microsoft Corporation 2007. All rights reserved.
Build started 11/8/2008 8:16:59 AM.
Project "C:\temp\test.proj" on node 0 (default targets).
FileContents: Red;Green;Blue
FileContents Transformed: Red
Green
Blue
Done Building Project "C:\temp\test.proj" (default targets).
Build succeeded.
0 Warning(s)
0 Error(s)
Time Elapsed 00:00:00.03
What's going on here is two things.
#(FileContents->'%(Identity)', '%0a%0d')
We are transforming the list from one type to another using the same values (Identity) but a custom separator '%0a%0d'
We are using MSBuild Escaping to escape the line feed (%0a) and carriage return (%0d)
If you are using MSBuild 4.0, you can do the following instead, to get the contents of a file:
$([System.IO.File]::ReadAllText($FilePath))
Instead of #(FileContents->'%(Identity)', '%0a%0d') I believe you can do #(FileContents, '%0a%0d')
You can use WriteLinesToFile combined with
$([System.IO.File]::ReadAllText($(SourceFilePath))):
< WriteLinesToFile File="$(DestinationFilePath)" Lines="$([System.IO.File]::ReadAllText($(SourceFilePath)))"
Overwrite="true"
/>
This will copy your file exactly at it is.