update text processing tools (#1385)

* add header * ad conditional * move delimiter to 'select by field' branch * bump tool version * fix access to list parameter * Drop `-b` option Doesn't make sense given we restrict `input` to `txt` format. Also refactor Cheetah code. * add nl * update dependencies * add output number test * add styler R script * fix test data * add citation and creators * Trying to fix grep basic regex * Update tools/text_processing/text_processing/macros.xml --------- Co-authored-by: Marie Jossé <[email protected]> Co-authored-by: Matthias Bernt <[email protected]> Co-authored-by: Nicola Soranzo <[email protected]> Co-authored-by: Pavankumar Videm <[email protected]>
bgruening · Feb 29, 2024 · 70980e3 · 70980e3
1 parent 06dd963
commit 70980e3
Show file tree

Hide file tree

Showing 26 changed files with 254 additions and 125 deletions.
diff --git a/.github/styler.R b/.github/styler.R
@@ -0,0 +1,36 @@
+#!/usr/bin/env Rscript
+
+library("argparse")
+library("styler")
+
+parser <- ArgumentParser(description = "Call styler")
+parser$add_argument("dir",
+    metavar = "DIR", type = "character",
+    help = "File to parse"
+)
+parser$add_argument("--dry",
+    choices = c("off", "on"), default = "on"
+)
+args <- parser$parse_args()
+
+file_info <- file.info(args$dir)
+is_directory <- file_info$isdir
+
+if (is_directory) {
+    captured_output <- capture.output({
+        result <- style_dir(args$dir, indent_by = 4, dry = args$dry, recursive = TRUE)
+    })
+} else {
+    captured_output <- capture.output({
+        result <- style_file(args$dir, indent_by = 4, dry = args$dry)
+    })
+}
+
+n <- nrow(subset(result, changed == TRUE))
+if (n > 0) {
+    if (args$dry == "off") {
+        print(paste("Changed", n, "files"))
+    } else {
+        stop(paste("Linting failed for", n, "files"))
+    }
+}
diff --git a/tools/text_processing/text_processing/awk.xml b/tools/text_processing/text_processing/awk.xml
@@ -1,10 +1,11 @@
-<tool id="tp_awk_tool" name="Text reformatting" version="@[email protected]">
+<tool id="tp_awk_tool" name="Text reformatting" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
     <description>with awk</description>
     <macros>
         <import>macros.xml</import>
     </macros>
+    <expand macro="creator"/>
     <requirements>
-        <requirement type="package" version="4.2.0">gawk</requirement>
+        <requirement type="package" version="5.3.0">gawk</requirement>
     </requirements>
     <version_command>awk --version | head -n 1</version_command>
     <command>
@@ -15,9 +16,9 @@
             -v FS='	'
             -v OFS='	'
             --re-interval
-            -f "$awk_script"
-            "$infile"
-        > "$outfile"
+            -f '$awk_script'
+            '$infile'
+        > '$outfile'
 ]]>
     </command>
     <configfiles>
@@ -119,7 +120,6 @@ The select tool searches the data for lines containing or not containing a match
 - **$** matches the end of a line or string.
 - **\|** Separates alternate possibilities.
 
-@REFERENCES@
 ]]>
   </help>
   <expand macro="citations" />

diff --git a/tools/text_processing/text_processing/cat.xml b/tools/text_processing/text_processing/cat.xml
@@ -1,8 +1,9 @@
-<tool id="tp_cat" name="Concatenate datasets" version="0.1.1">
+<tool id="tp_cat" name="Concatenate datasets" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
     <description>tail-to-head (cat)</description>
     <macros>
         <import>macros.xml</import>
     </macros>
+    <expand macro="creator"/>
     <expand macro="requirements" />
     <version_command>
     <![CDATA[
@@ -89,7 +90,6 @@ will result in the following::
     chr2  100000030  200000955  P  0  +
     chr2  100000015  200000999  Q  0  +
 
-@REFERENCES@
 ]]>
     </help>
     <expand macro="citations" />

diff --git a/tools/text_processing/text_processing/cut.xml b/tools/text_processing/text_processing/cut.xml
@@ -1,4 +1,4 @@
-<tool id="tp_cut_tool" name="Advanced Cut" version="@[email protected]">
+<tool id="tp_cut_tool" name="Advanced Cut" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
     <description>columns from a table (cut)</description>
     <macros>
         <import>macros.xml</import>
@@ -8,14 +8,18 @@
     <command>
 <![CDATA[
         cut
-            #if str($delimiter) != '':
-                -d"${delimiter}"
+            $complement
+            $cut_type_options.cut_element
+            #if $cut_type_options.cut_element != '-f'
+                '$cut_type_options.list'
+            #else
+                #if str($cut_type_options.delimiter) != '':
+                    -d"$cut_type_options.delimiter"
+                #end if
+                '$cut_type_options.colnames_option.list'
             #end if
-            ${complement}
-            ${cut_type_options.cut_element}
-            '${cut_type_options.list}'
-            '${input}'
-        > '${output}'
+            '$input'
+        > '$output'
 ]]>
     </command>
     <inputs>
@@ -24,23 +28,33 @@
             <option value="">Keep</option>
             <option value="--complement">Discard</option>
         </param>
-        <param name="delimiter" type="select" label="Delimited by">
-            <option value="">Tab</option>
-            <option value=" ">Whitespace</option>
-            <option value=".">Dot</option>
-            <option value=",">Comma</option>
-            <option value="-">Dash</option>
-            <option value="_">Underscore</option>
-            <option value="|">Pipe</option>
-        </param>
         <conditional name="cut_type_options">
             <param name="cut_element" type="select" label="Cut by">
                 <option value="-f">fields</option>
                 <option value="-c">characters</option>
-                <option value="-b">bytes</option>
             </param>
             <when value="-f">
-                <param name="list" type="data_column" data_ref="input" multiple="true" label="List of Fields" help="(-f)" />
+                <param name="delimiter" type="select" label="Delimited by">
+                    <option value="">Tab</option>
+                    <option value=" ">Whitespace</option>
+                    <option value=".">Dot</option>
+                    <option value=",">Comma</option>
+                    <option value="-">Dash</option>
+                    <option value="_">Underscore</option>
+                    <option value="|">Pipe</option>
+                </param>
+                <conditional name="colnames_option">
+                    <param name="header" type="select" label="Is there a header for the data's columns ?">
+                        <option value="Y">Yes</option>
+                        <option value="N" selected="true">No</option>
+                    </param>
+                    <when value="Y">
+                        <param name="list" type="data_column" data_ref="input" use_header_names="true" multiple="true" label="List of Fields" help="(-f)" />
+                    </when>
+                    <when value="N">
+                        <param name="list" type="data_column" data_ref="input" multiple="true" label="List of Fields" help="(-f)" />
+                    </when>
+                </conditional>
             </when>
             <when value="-c">
                 <param name="list" type="text" value="" label="List of characters" help="These will be kept/discarded (depending on 'operation'). &lt;BR /&gt; Examples: 1,3,4 or 2-5">
@@ -51,15 +65,6 @@
                     </sanitizer>
                 </param>
             </when>
-            <when value="-b">
-                <param name="list" type="text" value="" label="List of Bytes" help="These will be kept/discarded (depending on 'operation'). &lt;BR /&gt; Examples: 1,3,4 or 2-5">
-                    <sanitizer>
-                        <valid initial="string.printable">
-                            <remove value="&apos;"/>
-                        </valid>
-                    </sanitizer>
-                </param>
-            </when>
         </conditional>
     </inputs>
     <outputs>
@@ -70,7 +75,7 @@
                 <conditional name="cut_type_options.cut_element">
                     <!-- fields -->
                     <when value="-f">
-                        <conditional name="delimiter">
+                        <conditional name="cut_type_options.delimiter">
                         <when value="T">
                             <conditional name="input">
                                 <when datatype_isinstance="interval">
@@ -184,20 +189,26 @@
         </data>
     </outputs>
     <tests>
-        <test>
+        <test expect_num_outputs="1">
             <param name="input" value="cut1.txt"/>
             <param name="list" value="1,3,4"/>
             <param name="delimiter" value=""/>
             <output name="output" file="cut_results1.txt"/>
         </test>
-        <test>
+        <test expect_num_outputs="1">
             <param name="input" value="cut1.txt"/>
-            <param name="list" value="2" />
-            <param name="delimiter" value="" />
+            <conditional name="cut_type_options">
+                <param name="delimiter" value="" />
+                <conditional name="colnames_option">
+                    <param name="header" value="Y"/>
+                    <!-- in the test we can (apparently) not select by header name -->
+                    <param name="list" value="2" />
+                </conditional>
+            </conditional>
             <param name="complement" value="--complement" />
             <output name="output" file="cut_results2.txt"/>
         </test>
-        <test>
+        <test expect_num_outputs="1">
             <param name="input" value="cut1.txt"/>
             <param name="list" value="-3" />
             <param name="delimiter" value="" />

diff --git a/tools/text_processing/text_processing/easyjoin.xml b/tools/text_processing/text_processing/easyjoin.xml
@@ -1,17 +1,18 @@
-<tool id="tp_easyjoin_tool" name="Join" version="@[email protected]">
+<tool id="tp_easyjoin_tool" name="Join" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
     <description>two files</description>
     <macros>
         <import>macros.xml</import>
     </macros>
+    <expand macro="creator"/>
     <expand macro="requirements">
-        <requirement type="package" version="5.22.0.1">perl</requirement>
+        <requirement type="package" version="5.32">perl</requirement>
     </expand>
     <version_command>join --version | head -n 1</version_command>
     <command>
 <![CDATA[
-        cp $__tool_directory__/sort-header ./ &&
+        cp '$__tool_directory__/sort-header' ./ &&
         chmod +x sort-header &&
-        perl $__tool_directory__/easyjoin
+        perl '$__tool_directory__/easyjoin'
             $jointype
             -t $'\t'
             $header
@@ -20,8 +21,8 @@
             $ignore_case
             -1 '$column1'
             -2 '$column2'
-            "$infile1"
-            "$infile2"
+            '$infile1'
+            '$infile2'
             > '$output'
 ]]>
     </command>
@@ -109,7 +110,6 @@ This tool joins two tabular files based on a common key column.
  * The header line (**Fruit  Color  Price**) was joined and kept as first line.
  * Missing values ( Avocado's color, missing from the first file ) are replaced with a period character.
 
-@REFERENCES@
 ]]>
     </help>
     <expand macro="citations" />

diff --git a/tools/text_processing/text_processing/find_and_replace.xml b/tools/text_processing/text_processing/find_and_replace.xml
@@ -1,10 +1,11 @@
-<tool id="tp_find_and_replace" name="Replace" version="@[email protected]">
+<tool id="tp_find_and_replace" name="Replace" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
     <description>parts of text</description>
     <macros>
         <import>macros.xml</import>
     </macros>
+    <expand macro="creator"/>
     <requirements>
-        <requirement type="package" version="5.22.0.1">perl</requirement>
+        <requirement type="package" version="5.32">perl</requirement>
     </requirements>
     <command>
 <![CDATA[
@@ -240,7 +241,6 @@ The Find & Replace tool searches the data for lines containing or not containing
 - **\\w** matches a single letter or digit or an underscore.
 - **\\s** matches a single white-space (space or tabs).
 
-@REFERENCES@
 ]]>
     </help>
     <expand macro="citations" />

diff --git a/tools/text_processing/text_processing/grep.xml b/tools/text_processing/text_processing/grep.xml
@@ -1,11 +1,12 @@
-<tool id="tp_grep_tool" name="Search in textfiles" version="@[email protected]">
+<tool id="tp_grep_tool" name="Search in textfiles" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
     <description>(grep)</description>
     <macros>
         <import>macros.xml</import>
     </macros>
+    <expand macro="creator"/>
     <requirements>
-        <requirement type="package" version="2.14">grep</requirement>
-        <requirement type="package" version="4.4">sed</requirement><!-- for ansi2html.sh -->
+        <requirement type="package" version="3.11">grep</requirement>
+        <requirement type="package" version="4.8">sed</requirement><!-- for ansi2html.sh -->
     </requirements>
     <version_command>grep --version | head -n 1</version_command>
     <command>
@@ -19,17 +20,17 @@
                 -B $lines_before
                 $invert
                 $case_sensitive
-                -- "${url_paste}"
-                '${infile}' | $__tool_directory__/ansi2html.sh > "${output}"
+                -- '${url_paste}'
+                '${infile}' | '$__tool_directory__/ansi2html.sh' > '${output}'
         #else:
             grep
                 $regex_type
                 -A $lines_after
                 -B $lines_before
                 $invert
                 $case_sensitive
-                 -- "${url_paste}"
-                '${infile}' | grep -v "^--$" > "${output}"
+                 -- '${url_paste}'
+                '${infile}' | grep -v "^--$" > '${output}'
         #end if
 ]]>
     </command>
@@ -42,9 +43,9 @@
         </param>
 
         <param name="regex_type" type="select" label="Type of regex">
-            <option value="-G">Basic</option>
-            <option value="-P" selected="true">Perl</option>
-            <option value="-E">Extended (egrep)</option>
+            <option value="-G">Basic (-G)</option>
+            <option value="-P" selected="true">Perl (-P)</option>
+            <option value="-E">Extended (egrep -E)</option>
         </param>
 
         <param name="url_paste" type="text" label="Regular Expression" help="See below for more details">
@@ -113,16 +114,16 @@
             <param name="color" value="NOCOLOR" />
             <output name="output" file="egrep_results1.txt" />
         </test>
-        <test><!-- same regex as egrep test, but different outcome with basic regex -->
+        <test><!-- tests basic regex; + must be backslashed to match -->
             <param name="infile" value="egrep1.txt" />
             <param name="case_sensitive" value="case sensitive" />
             <param name="regex_type" value="-G" />
             <param name="invert" value="" />
-            <param name="url_paste" value="[^ ]+" />
+            <param name="url_paste" value="[^ ]\+" />
             <param name="lines_before" value="0" />
             <param name="lines_after" value="0" />
             <param name="color" value="NOCOLOR" />
-            <output name="output" file="egrep_results2.txt" />
+            <output name="output" file="egrep_results1.txt" />
         </test>
     </tests>
     <help>
@@ -187,7 +188,6 @@ The select tool searches the data for lines containing or not containing a match
 - **$** matches the end of a line or string.
 - **\|** Separates alternate possibilities.
 
-@REFERENCES@
 ]]>
     </help>
     <expand macro="citations" />

diff --git a/tools/text_processing/text_processing/head.xml b/tools/text_processing/text_processing/head.xml
@@ -1,8 +1,9 @@
-<tool id="tp_head_tool" name="Select first" version="@[email protected]">
+<tool id="tp_head_tool" name="Select first" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
     <description>lines from a dataset (head)</description>
     <macros>
         <import>macros.xml</import>
     </macros>
+    <expand macro="creator"/>
     <expand macro="requirements" />
     <version_command>head --version | head -n 1</version_command>
     <command>
@@ -56,7 +57,6 @@ will produce::
     chr7  56632  56652  D17003_CTCF_R6  310  +
     chr7  56736  56756  D17003_CTCF_R7  354  +
 
-@REFERENCES@
 ]]>
     </help>
     <expand macro="citations" />