add new functions from the python math module (#6258)

* add new functions from the python math module * Update .shed.yml * Update tools/column_maker/column_maker.xml Co-authored-by: Nicola Soranzo <nicola.soranzo@gmail.com> * latest updates * fix linting * remove e * update test data --------- Co-authored-by: Nicola Soranzo <nicola.soranzo@gmail.com>
galaxyproject · Aug 22, 2024 · 06a9256 · 06a9256
1 parent ab5e118
commit 06a9256
Show file tree

Hide file tree

Showing 4 changed files with 146 additions and 15 deletions.
diff --git a/tools/column_maker/.shed.yml b/tools/column_maker/.shed.yml
@@ -7,4 +7,5 @@ long_description: |
 name: column_maker
 owner: devteam
 remote_repository_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/column_maker
+homepage_url: https://github.com/galaxyproject/tools-iuc/tree/master/tools/column_maker
 type: unrestricted
diff --git a/tools/column_maker/column_maker.py b/tools/column_maker/column_maker.py
@@ -13,12 +13,61 @@
 import sys
 # Functions that may be used in the compute expression
 from math import (  # noqa: F401
+    acos,
+    acosh,
+    asin,
+    asinh,
+    atan,
+    atan2,
+    atanh,
+    cbrt,
     ceil,
+    comb,
+    copysign,
+    cos,
+    cosh,
+    degrees,
+    dist,
+    erf,
+    erfc,
     exp,
+    exp2,
+    expm1,
+    fabs,
+    factorial,
     floor,
+    fmod,
+    frexp,
+    fsum,
+    gamma,
+    gcd,
+    hypot,
+    inf,
+    isclose,
+    isfinite,
+    isinf,
+    isnan,
+    isqrt,
+    ldexp,
+    lgamma,
     log,
     log10,
+    log1p,
+    log2,
+    modf,
+    nextafter,
+    perm,
+    pi,
+    pow,
+    prod,
+    remainder,
+    sin,
     sqrt,
+    tan,
+    tanh,
+    tau,
+    trunc,
+    ulp,
 )
 
 from numpy import format_float_positional
@@ -121,10 +170,10 @@ def to_str(obj):
         'lambda fields: [from_str(s, t) for s, t in zip(fields, [%s])]'
         % args.column_types
     )
-except Exception as e:
+except Exception as err:
     sys.exit(
         'While parsing column types, the following problem occured: "%s"'
-        % e
+        % err
     )
 
 # Get and parse actions
@@ -150,11 +199,19 @@ def to_str(obj):
 # COL_NAME is required with the --header option and specifies the name of the
 # new column; without --header, any COL_NAME gets ignored.
 operators = 'is|not|or|and'
-builtin_and_math_functions = (
+builtin_functions = (
     'abs|all|any|ascii|bin|bool|chr|complex|divmod|float|format|hex|int|len|'
-    'list|map|max|min|oct|ord|pow|range|reversed|round|set|sorted|str|sum|type|'
-    'log|log10|exp|sqrt|ceil|floor'
+    'list|map|max|min|oct|ord|pow|range|reversed|round|set|sorted|str|sum|type'
 )
+
+math_functions = (
+    'acos|acosh|asin|asinh|atan|atan2|atanh|cbrt|ceil|comb|copysign|cos|cosh|degrees|'
+    'dist|erf|erfc|exp|exp2|expm1|fabs|factorial|floor|fmod|frexp|fsum|gamma|gcd|'
+    'hypot|inf|isclose|isfinite|isinf|isnan|isqrt|ldexp|lgamma|log|log10|log1p|'
+    'log2|modf|nextafter|perm|pi|pow|prod|remainder|sin|'
+    'sqrt|tan|tanh|tau|trunc|ulp'
+)
+builtin_and_math_functions = builtin_functions + '|' + math_functions
 imported_numpy_function = 'format_float_positional'
 string_and_list_methods = [
     name for name in dir('') + dir([]) if not name.startswith('_')
@@ -171,6 +228,7 @@ def to_str(obj):
 for ac in actions:
     try:
         expr_string, col_add_spec, new_col_name = ac.split(';')
+        print(expr_string)
     except ValueError:
         sys.exit(
             'Invalid Action: "%s".  '

diff --git a/tools/column_maker/column_maker.xml b/tools/column_maker/column_maker.xml
@@ -1,4 +1,4 @@
-<tool id="Add_a_column1" name="Compute" version="2.0">
+<tool id="Add_a_column1" name="Compute" version="2.1" profile="23.0">
     <description>on rows</description>
     <macros>
         <xml name="compute_repeat">
@@ -34,8 +34,8 @@
         </xml>
     </macros>
     <requirements>
-        <requirement type="package" version="3.8">python</requirement>
-        <requirement type="package" version="1.23.1">numpy</requirement>
+        <requirement type="package" version="3.12">python</requirement>
+        <requirement type="package" version="2.1.0">numpy</requirement>
     </requirements>
     <command detect_errors="aggressive"><![CDATA[
 python '$__tool_directory__/column_maker.py'
@@ -193,6 +193,11 @@ ${expr.cond};${expr.add_column.pos}${expr.add_column.mode};
             </repeat>
             <output name="out_file1" file="column_maker_out4.tab" />
         </test>
+        <test>
+            <param name="cond" value="cos(pi)+sin(tau)*c2"/>
+            <param name="input" value="1.bed" ftype="bed" />
+            <output name="out_file1" file="column_maker_out5.interval"/>
+        </test>
         <!-- Test list column type in input -->
         <test>
             <param name="input" value="bed12.bed" ftype="bed12" />
@@ -311,13 +316,13 @@ Several expressions can be specified and will be applied sequentially to each ro
 
 - The following built-in Python functions are available for use in expressions::
 
-    abs | all | any | ascii | bin | bool | chr | ceil | complex | divmod
-
-    exp | float | floor | format | hex | int | len | list | log | log10
-
-    list | map | max | min | oct | ord | pow | range | reversed
+    abs | all | any | ascii | bin | bool | chr | complex | divmod | float | format | hex | int | len | list 
+    map | max | min | oct | ord | pow | range | reversed | round | set | sorted | str | sum | type
 
-    round | set | sorted | sqrt | str | sum | type
+    acos | acosh | asin | asinh | atan | atan2 | atanh | cbrt | ceil | comb | copysign | cos | cosh | degrees 
+    dist | erf | erfc | exp | exp2 | expm1 | fabs | factorial | floor | fmod | frexp | fsum | gamma | gcd
+    hypot | inf | isclose | isfinite | isinf | isnan | isqrt | ldexp | lgamma | log | log10 | log1p | log2
+    modf | nextafter | perm | pow | prod | remainder | sin | sqrt | tan | tanh | tau | trunc | ulp
 
 - In addition the numpy function ``format_float_positional`` is available to
   control the formatting of floating point numbers.
@@ -434,5 +439,7 @@ You can fix the file with the following tool run:
 
 **Replacement value**: `.`
     ]]></help>
-    <citations />
+    <citations>
+        <citation type="doi">10.1093/nar/gkae410</citation>
+    </citations>
 </tool>
diff --git a/tools/column_maker/test-data/column_maker_out5.interval b/tools/column_maker/test-data/column_maker_out5.interval
@@ -0,0 +1,65 @@
+chr1	147962192	147962580	CCDS989.1_cds_0_0_chr1_147962193_r	0	-	-1.000000036240285
+chr1	147984545	147984630	CCDS990.1_cds_0_0_chr1_147984546_f	0	+	-1.00000003624576
+chr1	148078400	148078582	CCDS993.1_cds_0_0_chr1_148078401_r	0	-	-1.0000000362687478
+chr1	148185136	148185276	CCDS996.1_cds_0_0_chr1_148185137_f	0	+	-1.0000000362948904
+chr10	55251623	55253124	CCDS7248.1_cds_0_0_chr10_55251624_r	0	-	-1.0000000135327447
+chr11	116124407	116124501	CCDS8374.1_cds_0_0_chr11_116124408_r	0	-	-1.0000000284422768
+chr11	116206508	116206563	CCDS8377.1_cds_0_0_chr11_116206509_f	0	+	-1.0000000284623856
+chr11	116211733	116212337	CCDS8378.1_cds_0_0_chr11_116211734_r	0	-	-1.0000000284636654
+chr11	1812377	1812407	CCDS7726.1_cds_0_0_chr11_1812378_f	0	+	-1.0000000004439042
+chr12	38440094	38440321	CCDS8736.1_cds_0_0_chr12_38440095_r	0	-	-1.0000000094151076
+chr13	112381694	112381953	CCDS9526.1_cds_0_0_chr13_112381695_f	0	+	-1.0000000275255763
+chr14	98710240	98712285	CCDS9949.1_cds_0_0_chr14_98710241_r	0	-	-1.000000024177036
+chr15	41486872	41487060	CCDS10096.1_cds_0_0_chr15_41486873_r	0	-	-1.000000010161353
+chr15	41673708	41673857	CCDS10097.1_cds_0_0_chr15_41673709_f	0	+	-1.0000000102071147
+chr15	41679161	41679250	CCDS10098.1_cds_0_0_chr15_41679162_r	0	-	-1.0000000102084503
+chr15	41826029	41826196	CCDS10101.1_cds_0_0_chr15_41826030_f	0	+	-1.0000000102444224
+chr16	142908	143003	CCDS10397.1_cds_0_0_chr16_142909_f	0	+	-1.0000000000350024
+chr16	179963	180135	CCDS10401.1_cds_0_0_chr16_179964_r	0	-	-1.0000000000440783
+chr16	244413	244681	CCDS10402.1_cds_0_0_chr16_244414_f	0	+	-1.000000000059864
+chr16	259268	259383	CCDS10403.1_cds_0_0_chr16_259269_r	0	-	-1.0000000000635023
+chr18	23786114	23786321	CCDS11891.1_cds_0_0_chr18_23786115_r	0	-	-1.0000000058259177
+chr18	59406881	59407046	CCDS11985.1_cds_0_0_chr18_59406882_f	0	+	-1.0000000145504893
+chr18	59455932	59456337	CCDS11986.1_cds_0_0_chr18_59455933_r	0	-	-1.0000000145625034
+chr18	59600586	59600754	CCDS11988.1_cds_0_0_chr18_59600587_f	0	+	-1.0000000145979333
+chr19	59068595	59069564	CCDS12866.1_cds_0_0_chr19_59068596_f	0	+	-1.000000014467633
+chr19	59236026	59236146	CCDS12872.1_cds_0_0_chr19_59236027_r	0	-	-1.0000000145086418
+chr19	59297998	59298008	CCDS12877.1_cds_0_0_chr19_59297999_f	0	+	-1.0000000145238206
+chr19	59302168	59302288	CCDS12878.1_cds_0_0_chr19_59302169_r	0	-	-1.000000014524842
+chr2	118288583	118288668	CCDS2120.1_cds_0_0_chr2_118288584_f	0	+	-1.000000028972347
+chr2	118394148	118394202	CCDS2121.1_cds_0_0_chr2_118394149_r	0	-	-1.000000028998203
+chr2	220190202	220190242	CCDS2441.1_cds_0_0_chr2_220190203_f	0	+	-1.0000000539310452
+chr2	220229609	220230869	CCDS2443.1_cds_0_0_chr2_220229610_r	0	-	-1.0000000539406972
+chr20	33330413	33330423	CCDS13249.1_cds_0_0_chr20_33330414_r	0	-	-1.0000000081635967
+chr20	33513606	33513792	CCDS13255.1_cds_0_0_chr20_33513607_f	0	+	-1.0000000082084661
+chr20	33579500	33579527	CCDS13256.1_cds_0_0_chr20_33579501_r	0	-	-1.0000000082246054
+chr20	33593260	33593348	CCDS13257.1_cds_0_0_chr20_33593261_f	0	+	-1.0000000082279756
+chr21	32707032	32707192	CCDS13614.1_cds_0_0_chr21_32707033_f	0	+	-1.0000000080109124
+chr21	32869641	32870022	CCDS13615.1_cds_0_0_chr21_32869642_r	0	-	-1.00000000805074
+chr21	33321040	33322012	CCDS13620.1_cds_0_0_chr21_33321041_f	0	+	-1.000000008161301
+chr21	33744994	33745040	CCDS13625.1_cds_0_0_chr21_33744995_r	0	-	-1.0000000082651397
+chr22	30120223	30120265	CCDS13897.1_cds_0_0_chr22_30120224_f	0	+	-1.000000007377327
+chr22	30160419	30160661	CCDS13898.1_cds_0_0_chr22_30160420_r	0	-	-1.0000000073871722
+chr22	30665273	30665360	CCDS13901.1_cds_0_0_chr22_30665274_f	0	+	-1.0000000075108257
+chr22	30939054	30939266	CCDS13903.1_cds_0_0_chr22_30939055_r	0	-	-1.0000000075778828
+chr5	131424298	131424460	CCDS4149.1_cds_0_0_chr5_131424299_f	0	+	-1.0000000321896692
+chr5	131556601	131556672	CCDS4151.1_cds_0_0_chr5_131556602_r	0	-	-1.0000000322220741
+chr5	131621326	131621419	CCDS4152.1_cds_0_0_chr5_131621327_f	0	+	-1.0000000322379272
+chr5	131847541	131847666	CCDS4155.1_cds_0_0_chr5_131847542_r	0	-	-1.0000000322933338
+chr6	108299600	108299744	CCDS5061.1_cds_0_0_chr6_108299601_r	0	-	-1.0000000265257516
+chr6	108594662	108594687	CCDS5063.1_cds_0_0_chr6_108594663_f	0	+	-1.0000000265980211
+chr6	108640045	108640151	CCDS5064.1_cds_0_0_chr6_108640046_r	0	-	-1.0000000266091367
+chr6	108722976	108723115	CCDS5067.1_cds_0_0_chr6_108722977_f	0	+	-1.0000000266294489
+chr7	113660517	113660685	CCDS5760.1_cds_0_0_chr7_113660518_f	0	+	-1.0000000278387977
+chr7	116512159	116512389	CCDS5771.1_cds_0_0_chr7_116512160_r	0	-	-1.0000000285372486
+chr7	116714099	116714152	CCDS5773.1_cds_0_0_chr7_116714100_f	0	+	-1.0000000285867094
+chr7	116945541	116945787	CCDS5774.1_cds_0_0_chr7_116945542_r	0	-	-1.0000000286433965
+chr8	118881131	118881317	CCDS6324.1_cds_0_0_chr8_118881132_r	0	-	-1.0000000291174793
+chr9	128764156	128764189	CCDS6914.1_cds_0_0_chr9_128764157_f	0	+	-1.0000000315381223
+chr9	128787519	128789136	CCDS6915.1_cds_0_0_chr9_128787520_r	0	-	-1.0000000315438446
+chr9	128882427	128882523	CCDS6917.1_cds_0_0_chr9_128882428_f	0	+	-1.0000000315670903
+chr9	128937229	128937445	CCDS6919.1_cds_0_0_chr9_128937230_r	0	-	-1.0000000315805129
+chrX	122745047	122745924	CCDS14606.1_cds_0_0_chrX_122745048_f	0	+	-1.0000000300638658
+chrX	152648964	152649196	CCDS14733.1_cds_0_0_chrX_152648965_r	0	-	-1.000000037388213
+chrX	152691446	152691471	CCDS14735.1_cds_0_0_chrX_152691447_f	0	+	-1.0000000373986182
+chrX	152694029	152694263	CCDS14736.1_cds_0_0_chrX_152694030_r	0	-	-1.0000000373992508