Re: [galaxy-dev] Tools that make datasets
These names have meaning in CollectedDatasetMatch: designation means designation if that exists else name name means name dbkey means dbkey ext means ext visible means visible On 21 October 2015 at 08:44, Steve Cassidy <steve.cassidy@mq.edu.au> wrote:
Thanks, yes that works, though I'm not really sure what the difference is between __name__ and __designation__. They both seem to have the same effect in my example.
Thanks for your help.
Steve
On 21 October 2015 at 16:49, Peter van Heusden <pvh@sanbi.ac.za> wrote:
Since you're using a directory, you can use one of the built in patterns:
DEFAULT_EXTRA_FILENAME_PATTERN = r"primary_DATASET_ID_(?P<designation>[^_]+)_(?P<visible>[^_]+)_(?P<ext>[^_]+)(_(?P<dbkey>[^_]+))?"
NAMED_PATTERNS = { "__default__": DEFAULT_EXTRA_FILENAME_PATTERN, "__name__": r"(?P<name>.*)", "__designation__": r"(?P<designation>.*)", "__name_and_ext__": r"(?P<name>.*)\.(?P<ext>[^\.]+)?", "__designation_and_ext__": r"(?P<designation>.*)\.(?P<ext>[^\._]+)?", }
In terms of docs, I don't know what the future is - the Galaxy wiki or http://galaxy.readthedocs.org/en/master/
On 21 October 2015 at 04:26, Steve Cassidy <steve.cassidy@mq.edu.au> wrote:
Ah, thankyou, yes, I can now get results by using patterns to match the output. I used your example but prepend 'simple' to the filename and then searches for that with:
<discover_datasets pattern="simple(?P<name>.*)" /> this solves the problem for the sample script but not generally since in general I can't predict the filenames that will be generated - this is a tool for downloading data from a repository which could be text, audio or video data.
If I don't use the 'simple' prefix and omit the file extension I still get my data but I also get three other files which are temporary scripts generated by galaxy and placed in the working directory. So, back to trying to put things in a subdirectory. It turns out that the issue I was having was as you pointed out earlier, the directory attribute to discover_datasets doesn't allow variables, so I need to write to a fixed directory name:
<discover_datasets pattern="(?P<name>.*)" directory="SampleDataset " /> This now works!
I had thought that I'd need to use a unique directory name but since galaxy runs each job in a separate directory, this isn't required. My real tool now works too after following the same pattern.
Thanks for your help. I'll see if I can write this up in a blog post.
Steve
On 21 October 2015 at 00:06, Peter van Heusden <pvh@sanbi.ac.za> wrote:
I poked around at your tool XML and the code a bit and the problem is directory="$job_name". Galaxy expects to collect files from the job's working directory - basically the current working directory the job runs in. The directory= argument doesn't have variables expanded as far as I can tell. In any event it is used in walk_over_extra_files() that is in lib/galaxy/tools/parameters/output_collect.py - if you look there you see that it is simply appended to the job's working directory.
So if you use:
<discover_datasets pattern="(?P<name>.*)\.txt" />
(note the < and > - this is effectively the regexp r"(?P<name>.*)\.txt" with the < and > escaped out)
And alter the code so that it just writes files to the current directory, then you'll pick up the files one.txt, two,txt and three.txt.
Peter
On 20 October 2015 at 12:28, Steve Cassidy <steve.cassidy@mq.edu.au> wrote:
Sorry, it was just an example of a tool that works - the file that it writes out is put into that directory, so I assume that's where my files should end up too.
Steve
On 20 October 2015 at 21:12, Peter van Heusden <pvh@sanbi.ac.za> wrote:
Sorry, I don't understand - what does the Upload File tool have to do with this?
On 20 October 2015 at 11:49, Steve Cassidy <steve.cassidy@mq.edu.au> wrote:
> Yes, I'm sure that's where the problem lies. Writing out to the > current directory doesn't work. The files get written to > 'job_working_directory/000/1/' but if I run the Upload File tool the result > is placed in 'files/000/'. I think I need to work out where to write the > files, I found some references to $__new_file_path__ but that doesn't seem > to help. > > Steve > > > > On 20 October 2015 at 19:57, Peter van Heusden <pvh@sanbi.ac.za> > wrote: > >> I suspect that the problem might be in the <discover_datasets> >> then. I'm not an export on this, but "__name_and_ext__" turns into the >> regexp r"(?P<name>.*)\.(?P<ext>[^\.]+)?" in >> lib/galaxy/tools/parameters/output_collect.py, and is used by the >> DatasetCollector (line 358). This looks like it should match the filenames >> you're creating, but I'm not 100% sure how that code works. One thing I >> notice is the "directory" argument. If you write jobs to the current >> directory instead of "output_path" can you get it to work? >> >> Peter >> >> On 20 October 2015 at 09:52, Steve Cassidy <steve.cassidy@mq.edu.au >> > wrote: >> >>> Thanks Peter, >>> I did see that proviso somewhere but no, refreshing doesn't help. >>> >>> That page was one of those that I referred to getting to this >>> point. >>> >>> Steve >>> >>> On 20 October 2015 at 18:33, Peter van Heusden <pvh@sanbi.ac.za> >>> wrote: >>> >>>> Just a quick check - did you refresh your history to confirm that >>>> the dataset *is* empty? We had the same thing at SANBI but it turns out >>>> that Galaxy creates an empty output collection and then only populates it >>>> sometime after job completion (this is a know UI bug). >>>> >>>> See: >>>> http://pvh.wp.sanbi.ac.za/2015/09/18/adventures-in-galaxy-output-collections... >>>> >>>> On 20 October 2015 at 08:48, Steve Cassidy < >>>> steve.cassidy@mq.edu.au> wrote: >>>> >>>>> Hi all, >>>>> I'm trying to understand how to write a tool that generates a >>>>> dataset rather than a single output file. I've tried following all of the >>>>> examples but I'm stuck, so I thought I would distil down the simplest >>>>> example I could write and ask for help here. >>>>> >>>>> So here's my example: >>>>> >>>>> https://gist.github.com/stevecassidy/0fa45ad5853faacb5f55 >>>>> >>>>> it's a simple python script that writes three files to a >>>>> directory named for the single input parameter. >>>>> >>>>> I think one of the problems I'm having is knowing where to write >>>>> the output to. I've run this under planemo serve and the job runs, creating >>>>> the output directory within the 'job_working_directory/000/1/SampleDataset' >>>>> directory, however my dataset doesn't contain anything so clearly my >>>>> outputs directive isn't working: >>>>> >>>>> <outputs> >>>>> <collection type="list" label="$job_name" name="output1"> >>>>> <discover_datasets pattern="__name_and_ext__" >>>>> directory="$job_name" /> >>>>> </collection> >>>>> </outputs> >>>>> >>>>> ($job_name is the name of the directory that is being written >>>>> to, SampleDataset in this case) >>>>> >>>>> Any help in getting this example working would be appreciated. >>>>> >>>>> Thanks, >>>>> >>>>> Steve >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> -- >>>>> Department of Computing, Macquarie University >>>>> http://web.science.mq.edu.au/~cassidy/ >>>>> >>>>> ___________________________________________________________ >>>>> Please keep all replies on the list by using "reply all" >>>>> in your mail client. To manage your subscriptions to this >>>>> and other Galaxy lists, please use the interface at: >>>>> https://lists.galaxyproject.org/ >>>>> >>>>> To search Galaxy mailing lists use the unified search at: >>>>> http://galaxyproject.org/search/mailinglists/ >>>>> >>>> >>>> >>>> ___________________________________________________________ >>>> Please keep all replies on the list by using "reply all" >>>> in your mail client. To manage your subscriptions to this >>>> and other Galaxy lists, please use the interface at: >>>> https://lists.galaxyproject.org/ >>>> >>>> To search Galaxy mailing lists use the unified search at: >>>> http://galaxyproject.org/search/mailinglists/ >>>> >>> >>> >>> >>> -- >>> Department of Computing, Macquarie University >>> http://web.science.mq.edu.au/~cassidy/ >>> >> >> >> ___________________________________________________________ >> Please keep all replies on the list by using "reply all" >> in your mail client. To manage your subscriptions to this >> and other Galaxy lists, please use the interface at: >> https://lists.galaxyproject.org/ >> >> To search Galaxy mailing lists use the unified search at: >> http://galaxyproject.org/search/mailinglists/ >> > > > > -- > Department of Computing, Macquarie University > http://web.science.mq.edu.au/~cassidy/ >
-- Department of Computing, Macquarie University http://web.science.mq.edu.au/~cassidy/
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: https://lists.galaxyproject.org/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
-- Department of Computing, Macquarie University http://web.science.mq.edu.au/~cassidy/
-- Department of Computing, Macquarie University http://web.science.mq.edu.au/~cassidy/
A brief writeup of my experiences: http://web.science.mq.edu.au/~cassidy/wordpress/2015/10/21/galaxy-tool-gener... Steve On 21 October 2015 at 18:27, Peter van Heusden <pvh@sanbi.ac.za> wrote:
These names have meaning in CollectedDatasetMatch:
designation means designation if that exists else name name means name dbkey means dbkey ext means ext visible means visible
On 21 October 2015 at 08:44, Steve Cassidy <steve.cassidy@mq.edu.au> wrote:
Thanks, yes that works, though I'm not really sure what the difference is between __name__ and __designation__. They both seem to have the same effect in my example.
Thanks for your help.
Steve
On 21 October 2015 at 16:49, Peter van Heusden <pvh@sanbi.ac.za> wrote:
Since you're using a directory, you can use one of the built in patterns:
DEFAULT_EXTRA_FILENAME_PATTERN = r"primary_DATASET_ID_(?P<designation>[^_]+)_(?P<visible>[^_]+)_(?P<ext>[^_]+)(_(?P<dbkey>[^_]+))?"
NAMED_PATTERNS = { "__default__": DEFAULT_EXTRA_FILENAME_PATTERN, "__name__": r"(?P<name>.*)", "__designation__": r"(?P<designation>.*)", "__name_and_ext__": r"(?P<name>.*)\.(?P<ext>[^\.]+)?", "__designation_and_ext__": r"(?P<designation>.*)\.(?P<ext>[^\._]+)?", }
In terms of docs, I don't know what the future is - the Galaxy wiki or http://galaxy.readthedocs.org/en/master/
On 21 October 2015 at 04:26, Steve Cassidy <steve.cassidy@mq.edu.au> wrote:
Ah, thankyou, yes, I can now get results by using patterns to match the output. I used your example but prepend 'simple' to the filename and then searches for that with:
<discover_datasets pattern="simple(?P<name>.*)" /> this solves the problem for the sample script but not generally since in general I can't predict the filenames that will be generated - this is a tool for downloading data from a repository which could be text, audio or video data.
If I don't use the 'simple' prefix and omit the file extension I still get my data but I also get three other files which are temporary scripts generated by galaxy and placed in the working directory. So, back to trying to put things in a subdirectory. It turns out that the issue I was having was as you pointed out earlier, the directory attribute to discover_datasets doesn't allow variables, so I need to write to a fixed directory name:
<discover_datasets pattern="(?P<name>.*)" directory=" SampleDataset" /> This now works!
I had thought that I'd need to use a unique directory name but since galaxy runs each job in a separate directory, this isn't required. My real tool now works too after following the same pattern.
Thanks for your help. I'll see if I can write this up in a blog post.
Steve
On 21 October 2015 at 00:06, Peter van Heusden <pvh@sanbi.ac.za> wrote:
I poked around at your tool XML and the code a bit and the problem is directory="$job_name". Galaxy expects to collect files from the job's working directory - basically the current working directory the job runs in. The directory= argument doesn't have variables expanded as far as I can tell. In any event it is used in walk_over_extra_files() that is in lib/galaxy/tools/parameters/output_collect.py - if you look there you see that it is simply appended to the job's working directory.
So if you use:
<discover_datasets pattern="(?P<name>.*)\.txt" />
(note the < and > - this is effectively the regexp r"(?P<name>.*)\.txt" with the < and > escaped out)
And alter the code so that it just writes files to the current directory, then you'll pick up the files one.txt, two,txt and three.txt.
Peter
On 20 October 2015 at 12:28, Steve Cassidy <steve.cassidy@mq.edu.au> wrote:
Sorry, it was just an example of a tool that works - the file that it writes out is put into that directory, so I assume that's where my files should end up too.
Steve
On 20 October 2015 at 21:12, Peter van Heusden <pvh@sanbi.ac.za> wrote:
> Sorry, I don't understand - what does the Upload File tool have to > do with this? > > On 20 October 2015 at 11:49, Steve Cassidy <steve.cassidy@mq.edu.au> > wrote: > >> Yes, I'm sure that's where the problem lies. Writing out to the >> current directory doesn't work. The files get written to >> 'job_working_directory/000/1/' but if I run the Upload File tool the result >> is placed in 'files/000/'. I think I need to work out where to write the >> files, I found some references to $__new_file_path__ but that doesn't seem >> to help. >> >> Steve >> >> >> >> On 20 October 2015 at 19:57, Peter van Heusden <pvh@sanbi.ac.za> >> wrote: >> >>> I suspect that the problem might be in the <discover_datasets> >>> then. I'm not an export on this, but "__name_and_ext__" turns into the >>> regexp r"(?P<name>.*)\.(?P<ext>[^\.]+)?" in >>> lib/galaxy/tools/parameters/output_collect.py, and is used by the >>> DatasetCollector (line 358). This looks like it should match the filenames >>> you're creating, but I'm not 100% sure how that code works. One thing I >>> notice is the "directory" argument. If you write jobs to the current >>> directory instead of "output_path" can you get it to work? >>> >>> Peter >>> >>> On 20 October 2015 at 09:52, Steve Cassidy < >>> steve.cassidy@mq.edu.au> wrote: >>> >>>> Thanks Peter, >>>> I did see that proviso somewhere but no, refreshing doesn't >>>> help. >>>> >>>> That page was one of those that I referred to getting to this >>>> point. >>>> >>>> Steve >>>> >>>> On 20 October 2015 at 18:33, Peter van Heusden <pvh@sanbi.ac.za> >>>> wrote: >>>> >>>>> Just a quick check - did you refresh your history to confirm >>>>> that the dataset *is* empty? We had the same thing at SANBI but it turns >>>>> out that Galaxy creates an empty output collection and then only populates >>>>> it sometime after job completion (this is a know UI bug). >>>>> >>>>> See: >>>>> http://pvh.wp.sanbi.ac.za/2015/09/18/adventures-in-galaxy-output-collections... >>>>> >>>>> On 20 October 2015 at 08:48, Steve Cassidy < >>>>> steve.cassidy@mq.edu.au> wrote: >>>>> >>>>>> Hi all, >>>>>> I'm trying to understand how to write a tool that generates a >>>>>> dataset rather than a single output file. I've tried following all of the >>>>>> examples but I'm stuck, so I thought I would distil down the simplest >>>>>> example I could write and ask for help here. >>>>>> >>>>>> So here's my example: >>>>>> >>>>>> https://gist.github.com/stevecassidy/0fa45ad5853faacb5f55 >>>>>> >>>>>> it's a simple python script that writes three files to a >>>>>> directory named for the single input parameter. >>>>>> >>>>>> I think one of the problems I'm having is knowing where to >>>>>> write the output to. I've run this under planemo serve and the job runs, >>>>>> creating the output directory within the >>>>>> 'job_working_directory/000/1/SampleDataset' directory, however my dataset >>>>>> doesn't contain anything so clearly my outputs directive isn't working: >>>>>> >>>>>> <outputs> >>>>>> <collection type="list" label="$job_name" >>>>>> name="output1"> >>>>>> <discover_datasets pattern="__name_and_ext__" >>>>>> directory="$job_name" /> >>>>>> </collection> >>>>>> </outputs> >>>>>> >>>>>> ($job_name is the name of the directory that is being written >>>>>> to, SampleDataset in this case) >>>>>> >>>>>> Any help in getting this example working would be appreciated. >>>>>> >>>>>> Thanks, >>>>>> >>>>>> Steve >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> -- >>>>>> Department of Computing, Macquarie University >>>>>> http://web.science.mq.edu.au/~cassidy/ >>>>>> >>>>>> ___________________________________________________________ >>>>>> Please keep all replies on the list by using "reply all" >>>>>> in your mail client. To manage your subscriptions to this >>>>>> and other Galaxy lists, please use the interface at: >>>>>> https://lists.galaxyproject.org/ >>>>>> >>>>>> To search Galaxy mailing lists use the unified search at: >>>>>> http://galaxyproject.org/search/mailinglists/ >>>>>> >>>>> >>>>> >>>>> ___________________________________________________________ >>>>> Please keep all replies on the list by using "reply all" >>>>> in your mail client. To manage your subscriptions to this >>>>> and other Galaxy lists, please use the interface at: >>>>> https://lists.galaxyproject.org/ >>>>> >>>>> To search Galaxy mailing lists use the unified search at: >>>>> http://galaxyproject.org/search/mailinglists/ >>>>> >>>> >>>> >>>> >>>> -- >>>> Department of Computing, Macquarie University >>>> http://web.science.mq.edu.au/~cassidy/ >>>> >>> >>> >>> ___________________________________________________________ >>> Please keep all replies on the list by using "reply all" >>> in your mail client. To manage your subscriptions to this >>> and other Galaxy lists, please use the interface at: >>> https://lists.galaxyproject.org/ >>> >>> To search Galaxy mailing lists use the unified search at: >>> http://galaxyproject.org/search/mailinglists/ >>> >> >> >> >> -- >> Department of Computing, Macquarie University >> http://web.science.mq.edu.au/~cassidy/ >> > >
-- Department of Computing, Macquarie University http://web.science.mq.edu.au/~cassidy/
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: https://lists.galaxyproject.org/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
-- Department of Computing, Macquarie University http://web.science.mq.edu.au/~cassidy/
-- Department of Computing, Macquarie University http://web.science.mq.edu.au/~cassidy/
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: https://lists.galaxyproject.org/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
-- Department of Computing, Macquarie University http://web.science.mq.edu.au/~cassidy/
participants (2)
-
Peter van Heusden
-
Steve Cassidy