|
4
|
1
|
|
|
2 # see https://github.com/fubar2/toolfactory
|
|
|
3 #
|
|
|
4 # copyright ross lazarus (ross stop lazarus at gmail stop com) May 2012
|
|
|
5 #
|
|
|
6 # all rights reserved
|
|
|
7 # Licensed under the LGPL
|
|
|
8 # suggestions for improvement and bug fixes welcome at
|
|
|
9 # https://github.com/fubar2/toolfactory
|
|
|
10 #
|
|
|
11 # April 2021: Refactored into two tools - generate and test/install
|
|
|
12 # as part of GTN tutorial development and biocontainer adoption
|
|
|
13 # The tester runs planemo on a non-tested archive, creates the test outputs
|
|
|
14 # and returns a new proper tool with test.
|
|
|
15 # The tester was generated from the ToolFactory_tester.py script
|
|
|
16
|
|
|
17
|
|
|
18 import argparse
|
|
|
19 import copy
|
|
|
20 import json
|
|
|
21 import logging
|
|
|
22 import os
|
|
|
23 import re
|
|
|
24 import shlex
|
|
|
25 import shutil
|
|
|
26 import subprocess
|
|
|
27 import sys
|
|
|
28 import tarfile
|
|
|
29 import tempfile
|
|
|
30 import time
|
|
6
|
31 import urllib
|
|
4
|
32
|
|
|
33 from bioblend import ConnectionError
|
|
|
34 from bioblend import galaxy
|
|
|
35 from bioblend import toolshed
|
|
|
36
|
|
|
37 import galaxyxml.tool as gxt
|
|
|
38 import galaxyxml.tool.parameters as gxtp
|
|
|
39
|
|
|
40 import lxml.etree as ET
|
|
|
41
|
|
|
42 import yaml
|
|
|
43
|
|
|
44 myversion = "V2.3 April 2021"
|
|
|
45 verbose = True
|
|
|
46 debug = True
|
|
|
47 toolFactoryURL = "https://github.com/fubar2/toolfactory"
|
|
|
48 FAKEEXE = "~~~REMOVE~~~ME~~~"
|
|
|
49 # need this until a PR/version bump to fix galaxyxml prepending the exe even
|
|
|
50 # with override.
|
|
|
51
|
|
|
52
|
|
|
53 def timenow():
|
|
|
54 """return current time as a string"""
|
|
|
55 return time.strftime("%d/%m/%Y %H:%M:%S", time.localtime(time.time()))
|
|
|
56
|
|
|
57 cheetah_escape_table = {"$": "\\$", "#": "\\#"}
|
|
|
58
|
|
|
59 def cheetah_escape(text):
|
|
|
60 """Produce entities within text."""
|
|
|
61 return "".join([cheetah_escape_table.get(c, c) for c in text])
|
|
|
62
|
|
|
63 def parse_citations(citations_text):
|
|
|
64 """"""
|
|
|
65 citations = [c for c in citations_text.split("**ENTRY**") if c.strip()]
|
|
|
66 citation_tuples = []
|
|
|
67 for citation in citations:
|
|
|
68 if citation.startswith("doi"):
|
|
|
69 citation_tuples.append(("doi", citation[len("doi") :].strip()))
|
|
|
70 else:
|
|
|
71 citation_tuples.append(("bibtex", citation[len("bibtex") :].strip()))
|
|
|
72 return citation_tuples
|
|
|
73
|
|
6
|
74
|
|
|
75 class Tool_Conf_Updater():
|
|
4
|
76 # update config/tool_conf.xml with a new tool unpacked in /tools
|
|
|
77 # requires highly insecure docker settings - like write to tool_conf.xml and to tools !
|
|
|
78 # if in a container possibly not so courageous.
|
|
|
79 # Fine on your own laptop but security red flag for most production instances
|
|
|
80
|
|
6
|
81 def __init__(self, args, tool_conf_path, new_tool_archive_path, new_tool_name, tool_dir):
|
|
4
|
82 self.args = args
|
|
5
|
83 self.tool_conf_path = os.path.join(args.galaxy_root,tool_conf_path)
|
|
|
84 self.tool_dir = os.path.join(args.galaxy_root, tool_dir)
|
|
4
|
85 self.our_name = 'ToolFactory'
|
|
|
86 tff = tarfile.open(new_tool_archive_path, "r:*")
|
|
|
87 flist = tff.getnames()
|
|
|
88 ourdir = os.path.commonpath(flist) # eg pyrevpos
|
|
|
89 self.tool_id = ourdir # they are the same for TF tools
|
|
|
90 ourxml = [x for x in flist if x.lower().endswith('.xml')]
|
|
5
|
91 res = tff.extractall()
|
|
4
|
92 tff.close()
|
|
5
|
93 self.run_rsync(ourdir, self.tool_dir)
|
|
4
|
94 self.update_toolconf(ourdir,ourxml)
|
|
|
95
|
|
5
|
96 def run_rsync(self, srcf, dstf):
|
|
|
97 src = os.path.abspath(srcf)
|
|
|
98 dst = os.path.abspath(dstf)
|
|
|
99 if os.path.isdir(src):
|
|
|
100 cll = ['rsync', '-vr', src, dst]
|
|
|
101 else:
|
|
|
102 cll = ['rsync', '-v', src, dst]
|
|
|
103 p = subprocess.run(
|
|
|
104 cll,
|
|
|
105 capture_output=False,
|
|
|
106 encoding='utf8',
|
|
|
107 shell=False,
|
|
|
108 )
|
|
|
109
|
|
4
|
110 def install_deps(self):
|
|
|
111 gi = galaxy.GalaxyInstance(url=self.args.galaxy_url, key=self.args.galaxy_api_key)
|
|
|
112 x = gi.tools.install_dependencies(self.tool_id)
|
|
|
113 print(f"Called install_dependencies on {self.tool_id} - got {x}")
|
|
|
114
|
|
|
115 def update_toolconf(self,ourdir,ourxml): # path is relative to tools
|
|
|
116 updated = False
|
|
5
|
117 localconf = './local_tool_conf.xml'
|
|
|
118 self.run_rsync(self.tool_conf_path,localconf)
|
|
|
119 tree = ET.parse(localconf)
|
|
4
|
120 root = tree.getroot()
|
|
|
121 hasTF = False
|
|
|
122 TFsection = None
|
|
|
123 for e in root.findall('section'):
|
|
|
124 if e.attrib['name'] == self.our_name:
|
|
|
125 hasTF = True
|
|
|
126 TFsection = e
|
|
|
127 if not hasTF:
|
|
|
128 TFsection = ET.Element('section')
|
|
|
129 root.insert(0,TFsection) # at the top!
|
|
|
130 our_tools = TFsection.findall('tool')
|
|
|
131 conf_tools = [x.attrib['file'] for x in our_tools]
|
|
|
132 for xml in ourxml: # may be > 1
|
|
|
133 if not xml in conf_tools: # new
|
|
|
134 updated = True
|
|
|
135 ET.SubElement(TFsection, 'tool', {'file':xml})
|
|
|
136 ET.indent(tree)
|
|
5
|
137 newconf = f"{self.tool_id}_conf"
|
|
|
138 tree.write(newconf, pretty_print=True)
|
|
|
139 self.run_rsync(newconf,self.tool_conf_path)
|
|
4
|
140 if False and self.args.packages and self.args.packages > '':
|
|
|
141 self.install_deps()
|
|
|
142
|
|
6
|
143 class Tool_Factory:
|
|
4
|
144 """Wrapper for an arbitrary script
|
|
|
145 uses galaxyxml
|
|
|
146
|
|
|
147 """
|
|
|
148
|
|
|
149 def __init__(self, args=None): # noqa
|
|
|
150 """
|
|
|
151 prepare command line cl for running the tool here
|
|
|
152 and prepare elements needed for galaxyxml tool generation
|
|
|
153 """
|
|
|
154 self.ourcwd = os.getcwd()
|
|
|
155 self.collections = []
|
|
|
156 if len(args.collection) > 0:
|
|
|
157 try:
|
|
|
158 self.collections = [
|
|
|
159 json.loads(x) for x in args.collection if len(x.strip()) > 1
|
|
|
160 ]
|
|
|
161 except Exception:
|
|
|
162 print(
|
|
|
163 f"--collections parameter {str(args.collection)} is malformed - should be a dictionary"
|
|
|
164 )
|
|
|
165 try:
|
|
|
166 self.infiles = [
|
|
|
167 json.loads(x) for x in args.input_files if len(x.strip()) > 1
|
|
|
168 ]
|
|
|
169 except Exception:
|
|
|
170 print(
|
|
|
171 f"--input_files parameter {str(args.input_files)} is malformed - should be a dictionary"
|
|
|
172 )
|
|
|
173 try:
|
|
|
174 self.outfiles = [
|
|
|
175 json.loads(x) for x in args.output_files if len(x.strip()) > 1
|
|
|
176 ]
|
|
|
177 except Exception:
|
|
|
178 print(
|
|
|
179 f"--output_files parameter {args.output_files} is malformed - should be a dictionary"
|
|
|
180 )
|
|
|
181 try:
|
|
|
182 self.addpar = [
|
|
|
183 json.loads(x) for x in args.additional_parameters if len(x.strip()) > 1
|
|
|
184 ]
|
|
|
185 except Exception:
|
|
|
186 print(
|
|
|
187 f"--additional_parameters {args.additional_parameters} is malformed - should be a dictionary"
|
|
|
188 )
|
|
|
189 try:
|
|
|
190 self.selpar = [
|
|
|
191 json.loads(x) for x in args.selecttext_parameters if len(x.strip()) > 1
|
|
|
192 ]
|
|
|
193 except Exception:
|
|
|
194 print(
|
|
|
195 f"--selecttext_parameters {args.selecttext_parameters} is malformed - should be a dictionary"
|
|
|
196 )
|
|
|
197 self.args = args
|
|
|
198 self.cleanuppar()
|
|
|
199 self.lastxclredirect = None
|
|
|
200 self.xmlcl = []
|
|
|
201 self.is_positional = self.args.parampass == "positional"
|
|
|
202 if self.args.sysexe:
|
|
|
203 if ' ' in self.args.sysexe:
|
|
|
204 self.executeme = self.args.sysexe.split(' ')
|
|
|
205 else:
|
|
|
206 self.executeme = [self.args.sysexe, ]
|
|
|
207 else:
|
|
|
208 if self.args.packages:
|
|
|
209 self.executeme = [self.args.packages.split(",")[0].split(":")[0].strip(), ]
|
|
|
210 else:
|
|
|
211 self.executeme = None
|
|
|
212 aXCL = self.xmlcl.append
|
|
|
213 assert args.parampass in [
|
|
|
214 "0",
|
|
|
215 "argparse",
|
|
|
216 "positional",
|
|
|
217 ], 'args.parampass must be "0","positional" or "argparse"'
|
|
|
218 self.tool_name = re.sub("[^a-zA-Z0-9_]+", "", args.tool_name)
|
|
|
219 self.tool_id = self.tool_name
|
|
|
220 self.newtool = gxt.Tool(
|
|
|
221 self.tool_name,
|
|
|
222 self.tool_id,
|
|
|
223 self.args.tool_version,
|
|
|
224 self.args.tool_desc,
|
|
|
225 FAKEEXE,
|
|
|
226 )
|
|
|
227 self.newtarpath = "%s_toolshed.gz" % self.tool_name
|
|
|
228 self.tooloutdir = "./tfout"
|
|
|
229 self.repdir = "./TF_run_report"
|
|
|
230 self.testdir = os.path.join(self.tooloutdir, "test-data")
|
|
|
231 if not os.path.exists(self.tooloutdir):
|
|
|
232 os.mkdir(self.tooloutdir)
|
|
|
233 if not os.path.exists(self.testdir):
|
|
|
234 os.mkdir(self.testdir)
|
|
|
235 if not os.path.exists(self.repdir):
|
|
|
236 os.mkdir(self.repdir)
|
|
|
237 self.tinputs = gxtp.Inputs()
|
|
|
238 self.toutputs = gxtp.Outputs()
|
|
|
239 self.testparam = []
|
|
|
240 if self.args.script_path:
|
|
|
241 self.prepScript()
|
|
|
242 if self.args.command_override:
|
|
|
243 scos = open(self.args.command_override, "r").readlines()
|
|
|
244 self.command_override = [x.rstrip() for x in scos]
|
|
|
245 else:
|
|
|
246 self.command_override = None
|
|
|
247 if self.args.test_override:
|
|
|
248 stos = open(self.args.test_override, "r").readlines()
|
|
|
249 self.test_override = [x.rstrip() for x in stos]
|
|
|
250 else:
|
|
|
251 self.test_override = None
|
|
|
252 if self.args.script_path:
|
|
|
253 for ex in self.executeme:
|
|
|
254 aXCL(ex)
|
|
|
255 aXCL("$runme")
|
|
|
256 else:
|
|
|
257 for ex in self.executeme:
|
|
|
258 aXCL(ex)
|
|
|
259
|
|
|
260 if self.args.parampass == "0":
|
|
|
261 self.clsimple()
|
|
|
262 else:
|
|
|
263 if self.args.parampass == "positional":
|
|
|
264 self.prepclpos()
|
|
|
265 self.clpositional()
|
|
|
266 else:
|
|
|
267 self.prepargp()
|
|
|
268 self.clargparse()
|
|
|
269
|
|
|
270 def clsimple(self):
|
|
|
271 """no parameters or repeats - uses < and > for i/o"""
|
|
|
272 aXCL = self.xmlcl.append
|
|
|
273 if len(self.infiles) > 0:
|
|
|
274 aXCL("<")
|
|
|
275 aXCL("$%s" % self.infiles[0]["infilename"])
|
|
|
276 if len(self.outfiles) > 0:
|
|
|
277 aXCL(">")
|
|
|
278 aXCL("$%s" % self.outfiles[0]["name"])
|
|
|
279 if self.args.cl_user_suffix: # DIY CL end
|
|
|
280 clp = shlex.split(self.args.cl_user_suffix)
|
|
|
281 for c in clp:
|
|
|
282 aXCL(c)
|
|
|
283
|
|
|
284 def prepargp(self):
|
|
|
285 xclsuffix = []
|
|
|
286 for i, p in enumerate(self.infiles):
|
|
|
287 nam = p["infilename"]
|
|
|
288 if p["origCL"].strip().upper() == "STDIN":
|
|
|
289 xappendme = [
|
|
|
290 nam,
|
|
|
291 nam,
|
|
|
292 "< $%s" % nam,
|
|
|
293 ]
|
|
|
294 else:
|
|
|
295 rep = p["repeat"] == "1"
|
|
|
296 over = ""
|
|
|
297 if rep:
|
|
|
298 over = f'#for $rep in $R_{nam}:\n--{nam} "$rep.{nam}"\n#end for'
|
|
|
299 xappendme = [p["CL"], "$%s" % p["CL"], over]
|
|
|
300 xclsuffix.append(xappendme)
|
|
|
301 for i, p in enumerate(self.outfiles):
|
|
|
302 if p["origCL"].strip().upper() == "STDOUT":
|
|
|
303 self.lastxclredirect = [">", "$%s" % p["name"]]
|
|
|
304 else:
|
|
|
305 xclsuffix.append([p["name"], "$%s" % p["name"], ""])
|
|
|
306 for p in self.addpar:
|
|
|
307 nam = p["name"]
|
|
|
308 rep = p["repeat"] == "1"
|
|
|
309 if rep:
|
|
|
310 over = f'#for $rep in $R_{nam}:\n--{nam} "$rep.{nam}"\n#end for'
|
|
|
311 else:
|
|
|
312 over = p["override"]
|
|
|
313 xclsuffix.append([p["CL"], '"$%s"' % nam, over])
|
|
|
314 for p in self.selpar:
|
|
|
315 xclsuffix.append([p["CL"], '"$%s"' % p["name"], p["override"]])
|
|
|
316 self.xclsuffix = xclsuffix
|
|
|
317
|
|
|
318 def prepclpos(self):
|
|
|
319 xclsuffix = []
|
|
|
320 for i, p in enumerate(self.infiles):
|
|
|
321 if p["origCL"].strip().upper() == "STDIN":
|
|
|
322 xappendme = [
|
|
|
323 "999",
|
|
|
324 p["infilename"],
|
|
|
325 "< $%s" % p["infilename"],
|
|
|
326 ]
|
|
|
327 else:
|
|
|
328 xappendme = [p["CL"], "$%s" % p["infilename"], ""]
|
|
|
329 xclsuffix.append(xappendme)
|
|
|
330 for i, p in enumerate(self.outfiles):
|
|
|
331 if p["origCL"].strip().upper() == "STDOUT":
|
|
|
332 self.lastxclredirect = [">", "$%s" % p["name"]]
|
|
|
333 else:
|
|
|
334 xclsuffix.append([p["CL"], "$%s" % p["name"], ""])
|
|
|
335 for p in self.addpar:
|
|
|
336 nam = p["name"]
|
|
|
337 rep = p["repeat"] == "1" # repeats make NO sense
|
|
|
338 if rep:
|
|
|
339 print(f'### warning. Repeats for {nam} ignored - not permitted in positional parameter command lines!')
|
|
|
340 over = p["override"]
|
|
|
341 xclsuffix.append([p["CL"], '"$%s"' % nam, over])
|
|
|
342 for p in self.selpar:
|
|
|
343 xclsuffix.append([p["CL"], '"$%s"' % p["name"], p["override"]])
|
|
|
344 xclsuffix.sort()
|
|
|
345 self.xclsuffix = xclsuffix
|
|
|
346
|
|
|
347 def prepScript(self):
|
|
|
348 rx = open(self.args.script_path, "r").readlines()
|
|
|
349 rx = [x.rstrip() for x in rx]
|
|
|
350 rxcheck = [x.strip() for x in rx if x.strip() > ""]
|
|
|
351 assert len(rxcheck) > 0, "Supplied script is empty. Cannot run"
|
|
|
352 self.script = "\n".join(rx)
|
|
|
353 fhandle, self.sfile = tempfile.mkstemp(
|
|
|
354 prefix=self.tool_name, suffix="_%s" % (self.executeme[0])
|
|
|
355 )
|
|
|
356 tscript = open(self.sfile, "w")
|
|
|
357 tscript.write(self.script)
|
|
|
358 tscript.close()
|
|
|
359 self.spacedScript = [f" {x}" for x in rx if x.strip() > ""]
|
|
|
360 rx.insert(0,'#raw')
|
|
|
361 rx.append('#end raw')
|
|
|
362 self.escapedScript = rx
|
|
|
363 art = "%s.%s" % (self.tool_name, self.executeme[0])
|
|
|
364 artifact = open(art, "wb")
|
|
|
365 artifact.write(bytes(self.script, "utf8"))
|
|
|
366 artifact.close()
|
|
|
367
|
|
|
368 def cleanuppar(self):
|
|
|
369 """ positional parameters are complicated by their numeric ordinal"""
|
|
|
370 if self.args.parampass == "positional":
|
|
|
371 for i, p in enumerate(self.infiles):
|
|
|
372 assert (
|
|
|
373 p["CL"].isdigit() or p["CL"].strip().upper() == "STDIN"
|
|
|
374 ), "Positional parameters must be ordinal integers - got %s for %s" % (
|
|
|
375 p["CL"],
|
|
|
376 p["label"],
|
|
|
377 )
|
|
|
378 for i, p in enumerate(self.outfiles):
|
|
|
379 assert (
|
|
|
380 p["CL"].isdigit() or p["CL"].strip().upper() == "STDOUT"
|
|
|
381 ), "Positional parameters must be ordinal integers - got %s for %s" % (
|
|
|
382 p["CL"],
|
|
|
383 p["name"],
|
|
|
384 )
|
|
|
385 for i, p in enumerate(self.addpar):
|
|
|
386 assert p[
|
|
|
387 "CL"
|
|
|
388 ].isdigit(), "Positional parameters must be ordinal integers - got %s for %s" % (
|
|
|
389 p["CL"],
|
|
|
390 p["name"],
|
|
|
391 )
|
|
|
392 for i, p in enumerate(self.infiles):
|
|
|
393 infp = copy.copy(p)
|
|
|
394 infp["origCL"] = infp["CL"]
|
|
|
395 if self.args.parampass in ["positional", "0"]:
|
|
|
396 infp["infilename"] = infp["label"].replace(" ", "_")
|
|
|
397 else:
|
|
|
398 infp["infilename"] = infp["CL"]
|
|
|
399 self.infiles[i] = infp
|
|
|
400 for i, p in enumerate(self.outfiles):
|
|
|
401 p["origCL"] = p["CL"] # keep copy
|
|
|
402 self.outfiles[i] = p
|
|
|
403 for i, p in enumerate(self.addpar):
|
|
|
404 p["origCL"] = p["CL"]
|
|
|
405 self.addpar[i] = p
|
|
|
406
|
|
|
407 def clpositional(self):
|
|
|
408 # inputs in order then params
|
|
|
409 aXCL = self.xmlcl.append
|
|
|
410 for (k, v, koverride) in self.xclsuffix:
|
|
|
411 aXCL(v)
|
|
|
412 if self.lastxclredirect:
|
|
|
413 aXCL(self.lastxclredirect[0])
|
|
|
414 aXCL(self.lastxclredirect[1])
|
|
|
415 if self.args.cl_user_suffix: # DIY CL end
|
|
|
416 clp = shlex.split(self.args.cl_user_suffix)
|
|
|
417 for c in clp:
|
|
|
418 aXCL(c)
|
|
|
419
|
|
|
420
|
|
|
421 def clargparse(self):
|
|
|
422 """argparse style"""
|
|
|
423 aXCL = self.xmlcl.append
|
|
|
424 # inputs then params in argparse named form
|
|
|
425
|
|
|
426 for (k, v, koverride) in self.xclsuffix:
|
|
|
427 if koverride > "":
|
|
|
428 k = koverride
|
|
|
429 aXCL(k)
|
|
|
430 else:
|
|
|
431 if len(k.strip()) == 1:
|
|
|
432 k = "-%s" % k
|
|
|
433 else:
|
|
|
434 k = "--%s" % k
|
|
|
435 aXCL(k)
|
|
|
436 aXCL(v)
|
|
|
437 if self.lastxclredirect:
|
|
|
438 aXCL(self.lastxclredirect[0])
|
|
|
439 aXCL(self.lastxclredirect[1])
|
|
|
440 if self.args.cl_user_suffix: # DIY CL end
|
|
|
441 clp = shlex.split(self.args.cl_user_suffix)
|
|
|
442 for c in clp:
|
|
|
443 aXCL(c)
|
|
|
444
|
|
|
445 def getNdash(self, newname):
|
|
|
446 if self.is_positional:
|
|
|
447 ndash = 0
|
|
|
448 else:
|
|
|
449 ndash = 2
|
|
|
450 if len(newname) < 2:
|
|
|
451 ndash = 1
|
|
|
452 return ndash
|
|
|
453
|
|
|
454 def doXMLparam(self): # noqa
|
|
|
455 """Add all needed elements to tool"""
|
|
|
456 for p in self.outfiles:
|
|
|
457 newname = p["name"]
|
|
|
458 newfmt = p["format"]
|
|
|
459 newcl = p["CL"]
|
|
|
460 test = p["test"]
|
|
|
461 oldcl = p["origCL"]
|
|
|
462 test = test.strip()
|
|
|
463 ndash = self.getNdash(newcl)
|
|
|
464 aparm = gxtp.OutputData(
|
|
|
465 name=newname, format=newfmt, num_dashes=ndash, label=newname
|
|
|
466 )
|
|
|
467 aparm.positional = self.is_positional
|
|
|
468 if self.is_positional:
|
|
|
469 if oldcl.upper() == "STDOUT":
|
|
|
470 aparm.positional = 9999999
|
|
|
471 aparm.command_line_override = "> $%s" % newname
|
|
|
472 else:
|
|
|
473 aparm.positional = int(oldcl)
|
|
|
474 aparm.command_line_override = "$%s" % newname
|
|
|
475 self.toutputs.append(aparm)
|
|
|
476 ld = None
|
|
|
477 if test.strip() > "":
|
|
|
478 if test.startswith("diff"):
|
|
|
479 c = "diff"
|
|
|
480 ld = 0
|
|
|
481 if test.split(":")[1].isdigit:
|
|
|
482 ld = int(test.split(":")[1])
|
|
|
483 tp = gxtp.TestOutput(
|
|
|
484 name=newname,
|
|
|
485 value="%s_sample" % newname,
|
|
|
486 compare=c,
|
|
|
487 lines_diff=ld,
|
|
|
488 )
|
|
|
489 elif test.startswith("sim_size"):
|
|
|
490 c = "sim_size"
|
|
|
491 tn = test.split(":")[1].strip()
|
|
|
492 if tn > "":
|
|
|
493 if "." in tn:
|
|
|
494 delta = None
|
|
|
495 delta_frac = min(1.0, float(tn))
|
|
|
496 else:
|
|
|
497 delta = int(tn)
|
|
|
498 delta_frac = None
|
|
|
499 tp = gxtp.TestOutput(
|
|
|
500 name=newname,
|
|
|
501 value="%s_sample" % newname,
|
|
|
502 compare=c,
|
|
|
503 delta=delta,
|
|
|
504 delta_frac=delta_frac,
|
|
|
505 )
|
|
|
506 else:
|
|
|
507 c = test
|
|
|
508 tp = gxtp.TestOutput(
|
|
|
509 name=newname,
|
|
|
510 value="%s_sample" % newname,
|
|
|
511 compare=c,
|
|
|
512 )
|
|
|
513 self.testparam.append(tp)
|
|
|
514 for p in self.infiles:
|
|
|
515 newname = p["infilename"]
|
|
|
516 newfmt = p["format"]
|
|
|
517 ndash = self.getNdash(newname)
|
|
|
518 reps = p.get("repeat", "0") == "1"
|
|
|
519 if not len(p["label"]) > 0:
|
|
|
520 alab = p["CL"]
|
|
|
521 else:
|
|
|
522 alab = p["label"]
|
|
|
523 aninput = gxtp.DataParam(
|
|
|
524 newname,
|
|
|
525 optional=False,
|
|
|
526 label=alab,
|
|
|
527 help=p["help"],
|
|
|
528 format=newfmt,
|
|
|
529 multiple=False,
|
|
|
530 num_dashes=ndash,
|
|
|
531 )
|
|
|
532 aninput.positional = self.is_positional
|
|
|
533 if self.is_positional:
|
|
|
534 if p["origCL"].upper() == "STDIN":
|
|
|
535 aninput.positional = 9999998
|
|
|
536 aninput.command_line_override = "> $%s" % newname
|
|
|
537 else:
|
|
|
538 aninput.positional = int(p["origCL"])
|
|
|
539 aninput.command_line_override = "$%s" % newname
|
|
|
540 if reps:
|
|
|
541 repe = gxtp.Repeat(name=f"R_{newname}", title=f"Add as many {alab} as needed")
|
|
|
542 repe.append(aninput)
|
|
|
543 self.tinputs.append(repe)
|
|
|
544 tparm = gxtp.TestRepeat(name=f"R_{newname}")
|
|
|
545 tparm2 = gxtp.TestParam(newname, value="%s_sample" % newname)
|
|
|
546 tparm.append(tparm2)
|
|
|
547 self.testparam.append(tparm)
|
|
|
548 else:
|
|
|
549 self.tinputs.append(aninput)
|
|
|
550 tparm = gxtp.TestParam(newname, value="%s_sample" % newname)
|
|
|
551 self.testparam.append(tparm)
|
|
|
552 for p in self.addpar:
|
|
|
553 newname = p["name"]
|
|
|
554 newval = p["value"]
|
|
|
555 newlabel = p["label"]
|
|
|
556 newhelp = p["help"]
|
|
|
557 newtype = p["type"]
|
|
|
558 newcl = p["CL"]
|
|
|
559 oldcl = p["origCL"]
|
|
|
560 reps = p["repeat"] == "1"
|
|
|
561 if not len(newlabel) > 0:
|
|
|
562 newlabel = newname
|
|
|
563 ndash = self.getNdash(newname)
|
|
|
564 if newtype == "text":
|
|
|
565 aparm = gxtp.TextParam(
|
|
|
566 newname,
|
|
|
567 label=newlabel,
|
|
|
568 help=newhelp,
|
|
|
569 value=newval,
|
|
|
570 num_dashes=ndash,
|
|
|
571 )
|
|
|
572 elif newtype == "integer":
|
|
|
573 aparm = gxtp.IntegerParam(
|
|
|
574 newname,
|
|
|
575 label=newlabel,
|
|
|
576 help=newhelp,
|
|
|
577 value=newval,
|
|
|
578 num_dashes=ndash,
|
|
|
579 )
|
|
|
580 elif newtype == "float":
|
|
|
581 aparm = gxtp.FloatParam(
|
|
|
582 newname,
|
|
|
583 label=newlabel,
|
|
|
584 help=newhelp,
|
|
|
585 value=newval,
|
|
|
586 num_dashes=ndash,
|
|
|
587 )
|
|
|
588 elif newtype == "boolean":
|
|
|
589 aparm = gxtp.BooleanParam(
|
|
|
590 newname,
|
|
|
591 label=newlabel,
|
|
|
592 help=newhelp,
|
|
|
593 value=newval,
|
|
|
594 num_dashes=ndash,
|
|
|
595 )
|
|
|
596 else:
|
|
|
597 raise ValueError(
|
|
|
598 'Unrecognised parameter type "%s" for\
|
|
|
599 additional parameter %s in makeXML'
|
|
|
600 % (newtype, newname)
|
|
|
601 )
|
|
|
602 aparm.positional = self.is_positional
|
|
|
603 if self.is_positional:
|
|
|
604 aparm.positional = int(oldcl)
|
|
|
605 if reps:
|
|
|
606 repe = gxtp.Repeat(name=f"R_{newname}", title=f"Add as many {newlabel} as needed")
|
|
|
607 repe.append(aparm)
|
|
|
608 self.tinputs.append(repe)
|
|
|
609 tparm = gxtp.TestRepeat(name=f"R_{newname}")
|
|
|
610 tparm2 = gxtp.TestParam(newname, value=newval)
|
|
|
611 tparm.append(tparm2)
|
|
|
612 self.testparam.append(tparm)
|
|
|
613 else:
|
|
|
614 self.tinputs.append(aparm)
|
|
|
615 tparm = gxtp.TestParam(newname, value=newval)
|
|
|
616 self.testparam.append(tparm)
|
|
|
617 for p in self.selpar:
|
|
|
618 newname = p["name"]
|
|
|
619 newval = p["value"]
|
|
|
620 newlabel = p["label"]
|
|
|
621 newhelp = p["help"]
|
|
|
622 newtype = p["type"]
|
|
|
623 newcl = p["CL"]
|
|
|
624 if not len(newlabel) > 0:
|
|
|
625 newlabel = newname
|
|
|
626 ndash = self.getNdash(newname)
|
|
|
627 if newtype == "selecttext":
|
|
|
628 newtext = p["texts"]
|
|
|
629 aparm = gxtp.SelectParam(
|
|
|
630 newname,
|
|
|
631 label=newlabel,
|
|
|
632 help=newhelp,
|
|
|
633 num_dashes=ndash,
|
|
|
634 )
|
|
|
635 for i in range(len(newval)):
|
|
|
636 anopt = gxtp.SelectOption(
|
|
|
637 value=newval[i],
|
|
|
638 text=newtext[i],
|
|
|
639 )
|
|
|
640 aparm.append(anopt)
|
|
|
641 aparm.positional = self.is_positional
|
|
|
642 if self.is_positional:
|
|
|
643 aparm.positional = int(newcl)
|
|
|
644 self.tinputs.append(aparm)
|
|
|
645 tparm = gxtp.TestParam(newname, value=newval)
|
|
|
646 self.testparam.append(tparm)
|
|
|
647 else:
|
|
|
648 raise ValueError(
|
|
|
649 'Unrecognised parameter type "%s" for\
|
|
|
650 selecttext parameter %s in makeXML'
|
|
|
651 % (newtype, newname)
|
|
|
652 )
|
|
|
653 for p in self.collections:
|
|
|
654 newkind = p["kind"]
|
|
|
655 newname = p["name"]
|
|
|
656 newlabel = p["label"]
|
|
|
657 newdisc = p["discover"]
|
|
|
658 collect = gxtp.OutputCollection(newname, label=newlabel, type=newkind)
|
|
|
659 disc = gxtp.DiscoverDatasets(
|
|
|
660 pattern=newdisc, directory=f"{newname}", visible="false"
|
|
|
661 )
|
|
|
662 collect.append(disc)
|
|
|
663 self.toutputs.append(collect)
|
|
|
664 try:
|
|
|
665 tparm = gxtp.TestOutputCollection(newname) # broken until PR merged.
|
|
|
666 self.testparam.append(tparm)
|
|
|
667 except Exception:
|
|
|
668 print("#### WARNING: Galaxyxml version does not have the PR merged yet - tests for collections must be over-ridden until then!")
|
|
|
669
|
|
|
670 def doNoXMLparam(self):
|
|
|
671 """filter style package - stdin to stdout"""
|
|
|
672 if len(self.infiles) > 0:
|
|
|
673 alab = self.infiles[0]["label"]
|
|
|
674 if len(alab) == 0:
|
|
|
675 alab = self.infiles[0]["infilename"]
|
|
|
676 max1s = (
|
|
|
677 "Maximum one input if parampass is 0 but multiple input files supplied - %s"
|
|
|
678 % str(self.infiles)
|
|
|
679 )
|
|
|
680 assert len(self.infiles) == 1, max1s
|
|
|
681 newname = self.infiles[0]["infilename"]
|
|
|
682 aninput = gxtp.DataParam(
|
|
|
683 newname,
|
|
|
684 optional=False,
|
|
|
685 label=alab,
|
|
|
686 help=self.infiles[0]["help"],
|
|
|
687 format=self.infiles[0]["format"],
|
|
|
688 multiple=False,
|
|
|
689 num_dashes=0,
|
|
|
690 )
|
|
|
691 aninput.command_line_override = "< $%s" % newname
|
|
|
692 aninput.positional = True
|
|
|
693 self.tinputs.append(aninput)
|
|
|
694 tp = gxtp.TestParam(name=newname, value="%s_sample" % newname)
|
|
|
695 self.testparam.append(tp)
|
|
|
696 if len(self.outfiles) > 0:
|
|
|
697 newname = self.outfiles[0]["name"]
|
|
|
698 newfmt = self.outfiles[0]["format"]
|
|
|
699 anout = gxtp.OutputData(newname, format=newfmt, num_dashes=0)
|
|
|
700 anout.command_line_override = "> $%s" % newname
|
|
|
701 anout.positional = self.is_positional
|
|
|
702 self.toutputs.append(anout)
|
|
|
703 tp = gxtp.TestOutput(name=newname, value="%s_sample" % newname)
|
|
|
704 self.testparam.append(tp)
|
|
|
705
|
|
|
706 def makeXML(self): # noqa
|
|
|
707 """
|
|
|
708 Create a Galaxy xml tool wrapper for the new script
|
|
|
709 Uses galaxyhtml
|
|
|
710 Hmmm. How to get the command line into correct order...
|
|
|
711 """
|
|
|
712 if self.command_override:
|
|
|
713 self.newtool.command_override = self.command_override # config file
|
|
|
714 else:
|
|
|
715 self.newtool.command_override = self.xmlcl
|
|
|
716 cite = gxtp.Citations()
|
|
|
717 acite = gxtp.Citation(type="doi", value="10.1093/bioinformatics/bts573")
|
|
|
718 cite.append(acite)
|
|
|
719 self.newtool.citations = cite
|
|
|
720 safertext = ""
|
|
|
721 if self.args.help_text:
|
|
|
722 helptext = open(self.args.help_text, "r").readlines()
|
|
|
723 safertext = "\n".join([cheetah_escape(x) for x in helptext])
|
|
|
724 if len(safertext.strip()) == 0:
|
|
|
725 safertext = (
|
|
|
726 "Ask the tool author (%s) to rebuild with help text please\n"
|
|
|
727 % (self.args.user_email)
|
|
|
728 )
|
|
|
729 if self.args.script_path:
|
|
|
730 if len(safertext) > 0:
|
|
|
731 safertext = safertext + "\n\n------\n" # transition allowed!
|
|
|
732 scr = [x for x in self.spacedScript if x.strip() > ""]
|
|
|
733 scr.insert(0, "\n\nScript::\n")
|
|
|
734 if len(scr) > 300:
|
|
|
735 scr = (
|
|
|
736 scr[:100]
|
|
|
737 + [" >300 lines - stuff deleted", " ......"]
|
|
|
738 + scr[-100:]
|
|
|
739 )
|
|
|
740 scr.append("\n")
|
|
|
741 safertext = safertext + "\n".join(scr)
|
|
|
742 self.newtool.help = safertext
|
|
|
743 self.newtool.version_command = f'echo "{self.args.tool_version}"'
|
|
|
744 std = gxtp.Stdios()
|
|
|
745 std1 = gxtp.Stdio()
|
|
|
746 std.append(std1)
|
|
|
747 self.newtool.stdios = std
|
|
|
748 requirements = gxtp.Requirements()
|
|
|
749 if self.args.packages:
|
|
|
750 try:
|
|
|
751 for d in self.args.packages.split(","):
|
|
|
752 ver = ""
|
|
|
753 d = d.replace("==", ":")
|
|
|
754 d = d.replace("=", ":")
|
|
|
755 if ":" in d:
|
|
|
756 packg, ver = d.split(":")
|
|
|
757 else:
|
|
|
758 packg = d
|
|
|
759 requirements.append(
|
|
|
760 gxtp.Requirement("package", packg.strip(), ver.strip())
|
|
|
761 )
|
|
|
762 except Exception:
|
|
|
763 print('### malformed packages string supplied - cannot parse =',self.args.packages)
|
|
|
764 sys.exit(2)
|
|
|
765 self.newtool.requirements = requirements
|
|
|
766 if self.args.parampass == "0":
|
|
|
767 self.doNoXMLparam()
|
|
|
768 else:
|
|
|
769 self.doXMLparam()
|
|
|
770 self.newtool.outputs = self.toutputs
|
|
|
771 self.newtool.inputs = self.tinputs
|
|
|
772 if self.args.script_path:
|
|
|
773 configfiles = gxtp.Configfiles()
|
|
|
774 configfiles.append(
|
|
|
775 gxtp.Configfile(name="runme", text="\n".join(self.escapedScript))
|
|
|
776 )
|
|
|
777 self.newtool.configfiles = configfiles
|
|
|
778 tests = gxtp.Tests()
|
|
|
779 test_a = gxtp.Test()
|
|
|
780 for tp in self.testparam:
|
|
|
781 test_a.append(tp)
|
|
|
782 tests.append(test_a)
|
|
|
783 self.newtool.tests = tests
|
|
|
784 self.newtool.add_comment(
|
|
|
785 "Created by %s at %s using the Galaxy Tool Factory."
|
|
|
786 % (self.args.user_email, timenow())
|
|
|
787 )
|
|
|
788 self.newtool.add_comment("Source in git at: %s" % (toolFactoryURL))
|
|
|
789 exml0 = self.newtool.export()
|
|
|
790 exml = exml0.replace(FAKEEXE, "") # temporary work around until PR accepted
|
|
|
791 if (
|
|
|
792 self.test_override
|
|
|
793 ): # cannot do this inside galaxyxml as it expects lxml objects for tests
|
|
|
794 part1 = exml.split("<tests>")[0]
|
|
|
795 part2 = exml.split("</tests>")[1]
|
|
|
796 fixed = "%s\n%s\n%s" % (part1, "\n".join(self.test_override), part2)
|
|
|
797 exml = fixed
|
|
|
798 # exml = exml.replace('range="1:"', 'range="1000:"')
|
|
|
799 xf = open("%s.xml" % self.tool_name, "w")
|
|
|
800 xf.write(exml)
|
|
|
801 xf.write("\n")
|
|
|
802 xf.close()
|
|
|
803 # ready for the tarball
|
|
|
804
|
|
|
805 def writeShedyml(self):
|
|
|
806 """for planemo"""
|
|
|
807 yuser = self.args.user_email.split("@")[0]
|
|
|
808 yfname = os.path.join(self.tooloutdir, ".shed.yml")
|
|
|
809 yamlf = open(yfname, "w")
|
|
|
810 odict = {
|
|
|
811 "name": self.tool_name,
|
|
|
812 "owner": yuser,
|
|
|
813 "type": "unrestricted",
|
|
|
814 "description": self.args.tool_desc,
|
|
|
815 "synopsis": self.args.tool_desc,
|
|
|
816 "category": "TF Generated Tools",
|
|
|
817 }
|
|
|
818 yaml.dump(odict, yamlf, allow_unicode=True)
|
|
|
819 yamlf.close()
|
|
|
820
|
|
|
821 def makeTool(self):
|
|
|
822 """write xmls and input samples into place"""
|
|
|
823 if self.args.parampass == 0:
|
|
|
824 self.doNoXMLparam()
|
|
|
825 else:
|
|
|
826 self.makeXML()
|
|
|
827 if self.args.script_path:
|
|
|
828 stname = os.path.join(self.tooloutdir, self.sfile)
|
|
|
829 if not os.path.exists(stname):
|
|
|
830 shutil.copyfile(self.sfile, stname)
|
|
|
831 xreal = "%s.xml" % self.tool_name
|
|
|
832 xout = os.path.join(self.tooloutdir, xreal)
|
|
|
833 shutil.copyfile(xreal, xout)
|
|
|
834 for p in self.infiles:
|
|
|
835 pth = p["name"]
|
|
|
836 dest = os.path.join(self.testdir, "%s_sample" % p["infilename"])
|
|
|
837 shutil.copyfile(pth, dest)
|
|
|
838 dest = os.path.join(self.repdir, "%s_sample.%s" % (p["infilename"],p["format"]))
|
|
|
839 shutil.copyfile(pth, dest)
|
|
|
840
|
|
|
841 def makeToolTar(self, report_fail=False):
|
|
|
842 """move outputs into test-data and prepare the tarball"""
|
|
|
843 excludeme = "_planemo_test_report.html"
|
|
|
844
|
|
|
845 def exclude_function(tarinfo):
|
|
|
846 filename = tarinfo.name
|
|
|
847 return None if filename.endswith(excludeme) else tarinfo
|
|
|
848
|
|
|
849 for p in self.outfiles:
|
|
|
850 oname = p["name"]
|
|
|
851 tdest = os.path.join(self.testdir, "%s_sample" % oname)
|
|
|
852 src = os.path.join(self.testdir, oname)
|
|
|
853 if not os.path.isfile(tdest):
|
|
|
854 if os.path.isfile(src):
|
|
|
855 shutil.copyfile(src, tdest)
|
|
|
856 dest = os.path.join(self.repdir, "%s.sample" % (oname))
|
|
|
857 shutil.copyfile(src, dest)
|
|
|
858 else:
|
|
|
859 if report_fail:
|
|
|
860 print(
|
|
|
861 "###Tool may have failed - output file %s not found in testdir after planemo run %s."
|
|
|
862 % (tdest, self.testdir)
|
|
|
863 )
|
|
|
864 tf = tarfile.open(self.newtarpath, "w:gz")
|
|
|
865 tf.add(
|
|
|
866 name=self.tooloutdir,
|
|
|
867 arcname=self.tool_name,
|
|
|
868 filter=exclude_function,
|
|
|
869 )
|
|
|
870 tf.close()
|
|
|
871 shutil.copyfile(self.newtarpath, self.args.new_tool)
|
|
|
872
|
|
|
873 def moveRunOutputs(self):
|
|
|
874 """need to move planemo or run outputs into toolfactory collection"""
|
|
|
875 with os.scandir(self.tooloutdir) as outs:
|
|
|
876 for entry in outs:
|
|
|
877 if not entry.is_file():
|
|
|
878 continue
|
|
|
879 if not entry.name.endswith('.html'):
|
|
|
880 _, ext = os.path.splitext(entry.name)
|
|
|
881 newname = f"{entry.name.replace('.','_')}.txt"
|
|
|
882 dest = os.path.join(self.repdir, newname)
|
|
|
883 src = os.path.join(self.tooloutdir, entry.name)
|
|
|
884 shutil.copyfile(src, dest)
|
|
|
885 if self.args.include_tests:
|
|
|
886 with os.scandir(self.testdir) as outs:
|
|
|
887 for entry in outs:
|
|
|
888 if (not entry.is_file()) or entry.name.endswith(
|
|
|
889 "_planemo_test_report.html"
|
|
|
890 ):
|
|
|
891 continue
|
|
|
892 if "." in entry.name:
|
|
|
893 _, ext = os.path.splitext(entry.name)
|
|
|
894 if ext in [".tgz", ".json"]:
|
|
|
895 continue
|
|
|
896 if ext in [".yml", ".xml", ".yaml"]:
|
|
|
897 newname = f"{entry.name.replace('.','_')}.txt"
|
|
|
898 else:
|
|
|
899 newname = entry.name
|
|
|
900 else:
|
|
|
901 newname = f"{entry.name}.txt"
|
|
|
902 dest = os.path.join(self.repdir, newname)
|
|
|
903 src = os.path.join(self.testdir, entry.name)
|
|
|
904 shutil.copyfile(src, dest)
|
|
|
905
|
|
|
906
|
|
|
907 def main():
|
|
|
908 """
|
|
|
909 This is a Galaxy wrapper.
|
|
|
910 It expects to be called by a special purpose tool.xml
|
|
|
911
|
|
|
912 """
|
|
|
913 parser = argparse.ArgumentParser()
|
|
|
914 a = parser.add_argument
|
|
|
915 a("--script_path", default=None)
|
|
|
916 a("--history_test", default=None)
|
|
|
917 a("--cl_user_suffix", default=None)
|
|
|
918 a("--sysexe", default=None)
|
|
|
919 a("--packages", default=None)
|
|
|
920 a("--tool_name", default="newtool")
|
|
|
921 a("--tool_dir", default=None)
|
|
|
922 a("--input_files", default=[], action="append")
|
|
|
923 a("--output_files", default=[], action="append")
|
|
|
924 a("--user_email", default="Unknown")
|
|
|
925 a("--bad_user", default=None)
|
|
|
926 a("--help_text", default=None)
|
|
|
927 a("--tool_desc", default=None)
|
|
|
928 a("--tool_version", default=None)
|
|
|
929 a("--citations", default=None)
|
|
|
930 a("--command_override", default=None)
|
|
|
931 a("--test_override", default=None)
|
|
|
932 a("--additional_parameters", action="append", default=[])
|
|
|
933 a("--selecttext_parameters", action="append", default=[])
|
|
|
934 a("--edit_additional_parameters", action="store_true", default=False)
|
|
|
935 a("--parampass", default="positional")
|
|
|
936 a("--tfout", default="./tfout")
|
|
|
937 a("--new_tool", default="new_tool")
|
|
|
938 a("--galaxy_root", default="/galaxy-central")
|
|
|
939 a("--galaxy_venv", default="/galaxy_venv")
|
|
|
940 a("--collection", action="append", default=[])
|
|
|
941 a("--include_tests", default=False, action="store_true")
|
|
6
|
942 a("--admin_only", default=False, action="store_true")
|
|
4
|
943 a("--install", default=False, action="store_true")
|
|
|
944 a("--run_test", default=False, action="store_true")
|
|
5
|
945 a("--local_tools", default='tools') # relative to $__root_dir__
|
|
|
946 a("--tool_conf_path", default='config/tool_conf.xml') # relative to $__root_dir__
|
|
4
|
947 a("--galaxy_url", default="http://localhost:8080")
|
|
|
948 a("--toolshed_url", default="http://localhost:9009")
|
|
|
949 # make sure this is identical to tool_sheds_conf.xml
|
|
|
950 # localhost != 127.0.0.1 so validation fails
|
|
|
951 a("--toolshed_api_key", default="fakekey")
|
|
|
952 a("--galaxy_api_key", default="8993d65865e6d6d1773c2c34a1cc207d")
|
|
|
953 args = parser.parse_args()
|
|
6
|
954 if args.admin_only:
|
|
|
955 assert not args.bad_user, (
|
|
|
956 'UNAUTHORISED: %s is NOT authorized to use this tool until Galaxy \
|
|
4
|
957 admin adds %s to "admin_users" in the galaxy.yml Galaxy configuration file'
|
|
6
|
958 % (args.bad_user, args.bad_user)
|
|
4
|
959 )
|
|
|
960 assert args.tool_name, "## Tool Factory expects a tool name - eg --tool_name=DESeq"
|
|
6
|
961 r = Tool_Factory(args)
|
|
4
|
962 r.writeShedyml()
|
|
|
963 r.makeTool()
|
|
|
964 r.makeToolTar()
|
|
|
965 if args.install:
|
|
|
966 #try:
|
|
6
|
967 tcu = Tool_Conf_Updater(args=args, tool_dir=args.local_tools,
|
|
5
|
968 new_tool_archive_path=r.newtarpath, tool_conf_path=args.tool_conf_path,
|
|
4
|
969 new_tool_name=r.tool_name)
|
|
|
970 #except Exception:
|
|
|
971 # print("### Unable to install the new tool. Are you sure you have all the required special settings?")
|
|
|
972
|
|
|
973 if __name__ == "__main__":
|
|
|
974 main()
|
|
|
975
|
