|
43
|
1 #!/usr/bin/env python
|
|
|
2 # rgToolFactory.py
|
|
|
3 # see https://github.com/fubar2/toolfactory
|
|
|
4 #
|
|
|
5 # copyright ross lazarus (ross stop lazarus at gmail stop com) May 2012
|
|
|
6 #
|
|
|
7 # all rights reserved
|
|
|
8 # Licensed under the LGPL
|
|
|
9 # suggestions for improvement and bug fixes welcome at https://github.com/fubar2/toolfactory
|
|
|
10 #
|
|
|
11 # July 2020: BCC was fun and I feel like rip van winkle after 5 years.
|
|
|
12 # Decided to
|
|
|
13 # 1. Fix the toolfactory so it works - done for simplest case
|
|
|
14 # 2. Fix planemo so the toolfactory function works
|
|
|
15 # 3. Rewrite bits using galaxyxml functions where that makes sense - done
|
|
|
16 #
|
|
|
17 # removed all the old complications including making the new tool use this same script
|
|
|
18 # galaxyxml now generates the tool xml https://github.com/hexylena/galaxyxml
|
|
|
19 # No support for automatic HTML file creation from arbitrary outputs
|
|
|
20 # essential problem is to create two command lines - one for the tool xml and a different
|
|
|
21 # one to run the executable with the supplied test data and settings
|
|
|
22 # Be simpler to write the tool, then run it with planemo and soak up the test outputs.
|
|
|
23
|
|
|
24
|
|
|
25
|
|
|
26 import argparse
|
|
|
27 import logging
|
|
|
28 import os
|
|
|
29 import re
|
|
|
30 import shutil
|
|
|
31 import subprocess
|
|
|
32 import sys
|
|
|
33 import tarfile
|
|
|
34 import tempfile
|
|
|
35 import time
|
|
|
36
|
|
|
37 import galaxyxml.tool as gxt
|
|
|
38 import galaxyxml.tool.parameters as gxtp
|
|
|
39
|
|
|
40 import lxml
|
|
|
41
|
|
|
42 myversion = "V2.1 July 2020"
|
|
|
43 verbose = True
|
|
|
44 debug = True
|
|
|
45 toolFactoryURL = "https://github.com/fubar2/toolfactory"
|
|
|
46 ourdelim = "~~~"
|
|
|
47
|
|
|
48 # --input_files="$input_files~~~$CL~~~$input_formats~~~$input_label
|
|
|
49 # ~~~$input_help"
|
|
|
50 IPATHPOS = 0
|
|
|
51 ICLPOS = 1
|
|
|
52 IFMTPOS = 2
|
|
|
53 ILABPOS = 3
|
|
|
54 IHELPOS = 4
|
|
|
55 IOCLPOS = 5
|
|
|
56
|
|
|
57 # --output_files "$otab.history_name~~~$otab.history_format~~~$otab.CL
|
|
|
58 ONAMEPOS = 0
|
|
|
59 OFMTPOS = 1
|
|
|
60 OCLPOS = 2
|
|
|
61 OOCLPOS = 3
|
|
|
62
|
|
|
63 # --additional_parameters="$i.param_name~~~$i.param_value~~~
|
|
|
64 # $i.param_label~~~$i.param_help~~~$i.param_type~~~$i.CL~~~i$.param_CLoverride"
|
|
|
65 ANAMEPOS = 0
|
|
|
66 AVALPOS = 1
|
|
|
67 ALABPOS = 2
|
|
|
68 AHELPPOS = 3
|
|
|
69 ATYPEPOS = 4
|
|
|
70 ACLPOS = 5
|
|
|
71 AOVERPOS = 6
|
|
|
72 AOCLPOS = 7
|
|
|
73
|
|
|
74
|
|
|
75 foo = len(lxml.__version__)
|
|
|
76 # fug you, flake8. Say my name!
|
|
|
77
|
|
|
78 def timenow():
|
|
|
79 """return current time as a string
|
|
|
80 """
|
|
|
81 return time.strftime("%d/%m/%Y %H:%M:%S", time.localtime(time.time()))
|
|
|
82
|
|
|
83
|
|
|
84 def quote_non_numeric(s):
|
|
|
85 """return a prequoted string for non-numerics
|
|
|
86 useful for perl and Rscript parameter passing?
|
|
|
87 """
|
|
|
88 try:
|
|
|
89 _ = float(s)
|
|
|
90 return s
|
|
|
91 except ValueError:
|
|
|
92 return '"%s"' % s
|
|
|
93
|
|
|
94
|
|
|
95 html_escape_table = {"&": "&", ">": ">", "<": "<", "$": r"\$"}
|
|
|
96
|
|
|
97
|
|
|
98 def html_escape(text):
|
|
|
99 """Produce entities within text."""
|
|
|
100 return "".join(html_escape_table.get(c, c) for c in text)
|
|
|
101
|
|
|
102
|
|
|
103 def html_unescape(text):
|
|
|
104 """Revert entities within text. Multiple character targets so use replace"""
|
|
|
105 t = text.replace("&", "&")
|
|
|
106 t = t.replace(">", ">")
|
|
|
107 t = t.replace("<", "<")
|
|
|
108 t = t.replace("\\$", "$")
|
|
|
109 return t
|
|
|
110
|
|
|
111
|
|
|
112 def parse_citations(citations_text):
|
|
|
113 """
|
|
|
114 """
|
|
|
115 citations = [c for c in citations_text.split("**ENTRY**") if c.strip()]
|
|
|
116 citation_tuples = []
|
|
|
117 for citation in citations:
|
|
|
118 if citation.startswith("doi"):
|
|
|
119 citation_tuples.append(("doi", citation[len("doi") :].strip()))
|
|
|
120 else:
|
|
|
121 citation_tuples.append(
|
|
|
122 ("bibtex", citation[len("bibtex") :].strip())
|
|
|
123 )
|
|
|
124 return citation_tuples
|
|
|
125
|
|
|
126
|
|
|
127 class ScriptRunner:
|
|
|
128 """Wrapper for an arbitrary script
|
|
|
129 uses galaxyxml
|
|
|
130
|
|
|
131 """
|
|
|
132
|
|
|
133 def __init__(self, args=None):
|
|
|
134 """
|
|
|
135 prepare command line cl for running the tool here
|
|
|
136 and prepare elements needed for galaxyxml tool generation
|
|
|
137 """
|
|
|
138
|
|
|
139 self.infiles = [x.split(ourdelim) for x in args.input_files]
|
|
|
140 self.outfiles = [x.split(ourdelim) for x in args.output_files]
|
|
|
141 self.addpar = [x.split(ourdelim) for x in args.additional_parameters]
|
|
|
142 self.args = args
|
|
|
143 self.cleanuppar()
|
|
|
144 self.lastclredirect = None
|
|
|
145 self.lastxclredirect = None
|
|
|
146 self.cl = []
|
|
|
147 self.xmlcl = []
|
|
|
148 self.is_positional = self.args.parampass == "positional"
|
|
|
149 aCL = self.cl.append
|
|
|
150 assert args.parampass in [
|
|
|
151 "0",
|
|
|
152 "argparse",
|
|
|
153 "positional",
|
|
|
154 ], 'Parameter passing in args.parampass must be "0","positional" or "argparse"'
|
|
|
155 self.tool_name = re.sub("[^a-zA-Z0-9_]+", "", args.tool_name)
|
|
|
156 self.tool_id = self.tool_name
|
|
|
157 if self.args.interpreter_name:
|
|
|
158 exe = "$runMe"
|
|
|
159 else:
|
|
|
160 exe = self.args.exe_package
|
|
|
161 assert (
|
|
|
162 exe is not None
|
|
|
163 ), "No interpeter or executable passed in - nothing to run so cannot build"
|
|
|
164 self.tool = gxt.Tool(
|
|
|
165 self.args.tool_name,
|
|
|
166 self.tool_id,
|
|
|
167 self.args.tool_version,
|
|
|
168 self.args.tool_desc,
|
|
|
169 exe,
|
|
|
170 )
|
|
|
171 self.tinputs = gxtp.Inputs()
|
|
|
172 self.toutputs = gxtp.Outputs()
|
|
|
173 self.testparam = []
|
|
|
174 if (
|
|
|
175 self.args.runmode == "Executable" or self.args.runmode == "system"
|
|
|
176 ): # binary - no need
|
|
|
177 aCL(self.args.exe_package) # this little CL will just run
|
|
|
178 else:
|
|
|
179 self.prepScript()
|
|
|
180 self.elog = "%s_error_log.txt" % self.tool_name
|
|
|
181 self.tlog = "%s_runner_log.txt" % self.tool_name
|
|
|
182
|
|
|
183 if self.args.parampass == "0":
|
|
|
184 self.clsimple()
|
|
|
185 else:
|
|
|
186 clsuffix = []
|
|
|
187 xclsuffix = []
|
|
|
188 for i, p in enumerate(self.infiles):
|
|
|
189 if p[IOCLPOS] == "STDIN":
|
|
|
190 appendme = [
|
|
|
191 p[IOCLPOS],
|
|
|
192 p[ICLPOS],
|
|
|
193 p[IPATHPOS],
|
|
|
194 "< %s" % p[IPATHPOS],
|
|
|
195 ]
|
|
|
196 xappendme = [
|
|
|
197 p[IOCLPOS],
|
|
|
198 p[ICLPOS],
|
|
|
199 p[IPATHPOS],
|
|
|
200 "< $%s" % p[ICLPOS],
|
|
|
201 ]
|
|
|
202 else:
|
|
|
203 appendme = [p[IOCLPOS], p[ICLPOS], p[IPATHPOS], ""]
|
|
|
204 xappendme = [p[IOCLPOS], p[ICLPOS], "$%s" % p[ICLPOS], ""]
|
|
|
205 clsuffix.append(appendme)
|
|
|
206 xclsuffix.append(xappendme)
|
|
|
207 # print('##infile i=%d, appendme=%s' % (i,appendme))
|
|
|
208 for i, p in enumerate(self.outfiles):
|
|
|
209 if p[OOCLPOS] == "STDOUT":
|
|
|
210 self.lastclredirect = [">", p[ONAMEPOS]]
|
|
|
211 self.lastxclredirect = [">", "$%s" % p[OCLPOS]]
|
|
|
212 else:
|
|
|
213 clsuffix.append([p[OOCLPOS], p[OCLPOS], p[ONAMEPOS], ""])
|
|
|
214 xclsuffix.append(
|
|
|
215 [p[OOCLPOS], p[OCLPOS], "$%s" % p[ONAMEPOS], ""]
|
|
|
216 )
|
|
|
217 for p in self.addpar:
|
|
|
218 clsuffix.append(
|
|
|
219 [p[AOCLPOS], p[ACLPOS], p[AVALPOS], p[AOVERPOS]]
|
|
|
220 )
|
|
|
221 xclsuffix.append(
|
|
|
222 [p[AOCLPOS], p[ACLPOS], '"$%s"' % p[ANAMEPOS], p[AOVERPOS]]
|
|
|
223 )
|
|
|
224 clsuffix.sort()
|
|
|
225 xclsuffix.sort()
|
|
|
226 self.xclsuffix = xclsuffix
|
|
|
227 self.clsuffix = clsuffix
|
|
|
228 if self.args.parampass == "positional":
|
|
|
229 self.clpositional()
|
|
|
230 else:
|
|
|
231 self.clargparse()
|
|
|
232
|
|
|
233 def prepScript(self):
|
|
|
234 aCL = self.cl.append
|
|
|
235 rx = open(self.args.script_path, "r").readlines()
|
|
|
236 rx = [x.rstrip() for x in rx]
|
|
|
237 rxcheck = [x.strip() for x in rx if x.strip() > ""]
|
|
|
238 assert len(rxcheck) > 0, "Supplied script is empty. Cannot run"
|
|
|
239 self.script = "\n".join(rx)
|
|
|
240 fhandle, self.sfile = tempfile.mkstemp(
|
|
|
241 prefix=self.tool_name, suffix="_%s" % (self.args.interpreter_name)
|
|
|
242 )
|
|
|
243 tscript = open(self.sfile, "w")
|
|
|
244 tscript.write(self.script)
|
|
|
245 tscript.close()
|
|
|
246 self.indentedScript = " %s" % "\n".join(
|
|
|
247 [" %s" % html_escape(x) for x in rx]
|
|
|
248 )
|
|
|
249 self.escapedScript = "%s" % "\n".join(
|
|
|
250 [" %s" % html_escape(x) for x in rx]
|
|
|
251 )
|
|
|
252 art = "%s.%s" % (self.tool_name, self.args.interpreter_name)
|
|
|
253 artifact = open(art, "wb")
|
|
|
254 if self.args.interpreter_name == "python":
|
|
|
255 artifact.write(bytes("#!/usr/bin/env python\n", "utf8"))
|
|
|
256 artifact.write(bytes(self.script, "utf8"))
|
|
|
257 artifact.close()
|
|
|
258 aCL(self.args.interpreter_name)
|
|
|
259 aCL(self.sfile)
|
|
|
260
|
|
|
261 def cleanuppar(self):
|
|
|
262 """ positional parameters are complicated by their numeric ordinal"""
|
|
|
263 for i, p in enumerate(self.infiles):
|
|
|
264 if self.args.parampass == "positional":
|
|
|
265 assert p[ICLPOS].isdigit(), (
|
|
|
266 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
267 % (p[ICLPOS], p[ILABPOS])
|
|
|
268 )
|
|
|
269 p.append(p[ICLPOS])
|
|
|
270 if p[ICLPOS].isdigit() or self.args.parampass == "0":
|
|
|
271 scl = "input%d" % (i + 1)
|
|
|
272 p[ICLPOS] = scl
|
|
|
273 self.infiles[i] = p
|
|
|
274 for i, p in enumerate(
|
|
|
275 self.outfiles
|
|
|
276 ): # trying to automagically gather using extensions
|
|
|
277 if self.args.parampass == "positional" and p[OCLPOS] != "STDOUT":
|
|
|
278 assert p[OCLPOS].isdigit(), (
|
|
|
279 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
280 % (p[OCLPOS], p[ONAMEPOS])
|
|
|
281 )
|
|
|
282 p.append(p[OCLPOS])
|
|
|
283 if p[OCLPOS].isdigit() or p[OCLPOS] == "STDOUT":
|
|
|
284 scl = p[ONAMEPOS]
|
|
|
285 p[OCLPOS] = scl
|
|
|
286 self.outfiles[i] = p
|
|
|
287 for i, p in enumerate(self.addpar):
|
|
|
288 if self.args.parampass == "positional":
|
|
|
289 assert p[ACLPOS].isdigit(), (
|
|
|
290 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
291 % (p[ACLPOS], p[ANAMEPOS])
|
|
|
292 )
|
|
|
293 p.append(p[ACLPOS])
|
|
|
294 if p[ACLPOS].isdigit():
|
|
|
295 scl = "input%s" % p[ACLPOS]
|
|
|
296 p[ACLPOS] = scl
|
|
|
297 self.addpar[i] = p
|
|
|
298
|
|
|
299 def clsimple(self):
|
|
|
300 """ no parameters - uses < and > for i/o
|
|
|
301 """
|
|
|
302 aCL = self.cl.append
|
|
|
303 aCL("<")
|
|
|
304 aCL(self.infiles[0][IPATHPOS])
|
|
|
305 aCL(">")
|
|
|
306 aCL(self.outfiles[0][OCLPOS])
|
|
|
307 aXCL = self.xmlcl.append
|
|
|
308 aXCL("<")
|
|
|
309 aXCL("$%s" % self.infiles[0][ICLPOS])
|
|
|
310 aXCL(">")
|
|
|
311 aXCL("$%s" % self.outfiles[0][ONAMEPOS])
|
|
|
312
|
|
|
313 def clpositional(self):
|
|
|
314 # inputs in order then params
|
|
|
315 aCL = self.cl.append
|
|
|
316 for (o_v, k, v, koverride) in self.clsuffix:
|
|
|
317 if " " in v:
|
|
|
318 aCL("%s" % v)
|
|
|
319 else:
|
|
|
320 aCL(v)
|
|
|
321 aXCL = self.xmlcl.append
|
|
|
322 for (o_v, k, v, koverride) in self.xclsuffix:
|
|
|
323 aXCL(v)
|
|
|
324 if self.lastxclredirect:
|
|
|
325 aXCL(self.lastxclredirect[0])
|
|
|
326 aXCL(self.lastxclredirect[1])
|
|
|
327
|
|
|
328 def clargparse(self):
|
|
|
329 """ argparse style
|
|
|
330 """
|
|
|
331 aCL = self.cl.append
|
|
|
332 aXCL = self.xmlcl.append
|
|
|
333 # inputs then params in argparse named form
|
|
|
334 for (o_v, k, v, koverride) in self.xclsuffix:
|
|
|
335 if koverride > "":
|
|
|
336 k = koverride
|
|
|
337 elif len(k.strip()) == 1:
|
|
|
338 k = "-%s" % k
|
|
|
339 else:
|
|
|
340 k = "--%s" % k
|
|
|
341 aXCL(k)
|
|
|
342 aXCL(v)
|
|
|
343 for (o_v, k, v, koverride) in self.clsuffix:
|
|
|
344 if koverride > "":
|
|
|
345 k = koverride
|
|
|
346 elif len(k.strip()) == 1:
|
|
|
347 k = "-%s" % k
|
|
|
348 else:
|
|
|
349 k = "--%s" % k
|
|
|
350 aCL(k)
|
|
|
351 aCL(v)
|
|
|
352
|
|
|
353 def getNdash(self, newname):
|
|
|
354 if self.is_positional:
|
|
|
355 ndash = 0
|
|
|
356 else:
|
|
|
357 ndash = 2
|
|
|
358 if len(newname) < 2:
|
|
|
359 ndash = 1
|
|
|
360 return ndash
|
|
|
361
|
|
|
362 def doXMLparam(self):
|
|
|
363 """flake8 made me do this..."""
|
|
|
364 for p in self.outfiles:
|
|
|
365 newname, newfmt, newcl, oldcl = p
|
|
|
366 ndash = self.getNdash(newcl)
|
|
|
367 aparm = gxtp.OutputData(newcl, format=newfmt, num_dashes=ndash)
|
|
|
368 aparm.positional = self.is_positional
|
|
|
369 if self.is_positional:
|
|
|
370 if oldcl == "STDOUT":
|
|
|
371 aparm.positional = 9999999
|
|
|
372 aparm.command_line_override = "> $%s" % newcl
|
|
|
373 else:
|
|
|
374 aparm.positional = int(oldcl)
|
|
|
375 aparm.command_line_override = "$%s" % newcl
|
|
|
376 self.toutputs.append(aparm)
|
|
|
377 tp = gxtp.TestOutput(
|
|
|
378 name=newcl, value="%s_sample" % newcl, format=newfmt
|
|
|
379 )
|
|
|
380 self.testparam.append(tp)
|
|
|
381 for p in self.infiles:
|
|
|
382 newname = p[ICLPOS]
|
|
|
383 newfmt = p[IFMTPOS]
|
|
|
384 ndash = self.getNdash(newname)
|
|
|
385 if not len(p[ILABPOS]) > 0:
|
|
|
386 alab = p[ICLPOS]
|
|
|
387 else:
|
|
|
388 alab = p[ILABPOS]
|
|
|
389 aninput = gxtp.DataParam(
|
|
|
390 newname,
|
|
|
391 optional=False,
|
|
|
392 label=alab,
|
|
|
393 help=p[IHELPOS],
|
|
|
394 format=newfmt,
|
|
|
395 multiple=False,
|
|
|
396 num_dashes=ndash,
|
|
|
397 )
|
|
|
398 aninput.positional = self.is_positional
|
|
|
399 self.tinputs.append(aninput)
|
|
|
400 tparm = gxtp.TestParam(name=newname, value="%s_sample" % newname)
|
|
|
401 self.testparam.append(tparm)
|
|
|
402 for p in self.addpar:
|
|
|
403 newname, newval, newlabel, newhelp, newtype, newcl, override, oldcl = p
|
|
|
404 if not len(newlabel) > 0:
|
|
|
405 newlabel = newname
|
|
|
406 ndash = self.getNdash(newname)
|
|
|
407 if newtype == "text":
|
|
|
408 aparm = gxtp.TextParam(
|
|
|
409 newname,
|
|
|
410 label=newlabel,
|
|
|
411 help=newhelp,
|
|
|
412 value=newval,
|
|
|
413 num_dashes=ndash,
|
|
|
414 )
|
|
|
415 elif newtype == "integer":
|
|
|
416 aparm = gxtp.IntegerParam(
|
|
|
417 newname,
|
|
|
418 label=newname,
|
|
|
419 help=newhelp,
|
|
|
420 value=newval,
|
|
|
421 num_dashes=ndash,
|
|
|
422 )
|
|
|
423 elif newtype == "float":
|
|
|
424 aparm = gxtp.FloatParam(
|
|
|
425 newname,
|
|
|
426 label=newname,
|
|
|
427 help=newhelp,
|
|
|
428 value=newval,
|
|
|
429 num_dashes=ndash,
|
|
|
430 )
|
|
|
431 else:
|
|
|
432 raise ValueError(
|
|
|
433 'Unrecognised parameter type "%s" for\
|
|
|
434 additional parameter %s in makeXML'
|
|
|
435 % (newtype, newname)
|
|
|
436 )
|
|
|
437 aparm.positional = self.is_positional
|
|
|
438 if self.is_positional:
|
|
|
439 aninput.positional = int(oldcl)
|
|
|
440 self.tinputs.append(aparm)
|
|
|
441 self.tparm = gxtp.TestParam(newname, value=newval)
|
|
|
442 self.testparam.append(tparm)
|
|
|
443
|
|
|
444 def doNoXMLparam(self):
|
|
|
445 alab = self.infiles[0][ILABPOS]
|
|
|
446 if len(alab) == 0:
|
|
|
447 alab = self.infiles[0][ICLPOS]
|
|
|
448 max1s = (
|
|
|
449 "Maximum one input if parampass is 0 - more than one input files supplied - %s"
|
|
|
450 % str(self.infiles)
|
|
|
451 )
|
|
|
452 assert len(self.infiles) == 1, max1s
|
|
|
453 newname = self.infiles[0][ICLPOS]
|
|
|
454 aninput = gxtp.DataParam(
|
|
|
455 newname,
|
|
|
456 optional=False,
|
|
|
457 label=alab,
|
|
|
458 help=self.infiles[0][IHELPOS],
|
|
|
459 format=self.infiles[0][IFMTPOS],
|
|
|
460 multiple=False,
|
|
|
461 num_dashes=0,
|
|
|
462 )
|
|
|
463 aninput.command_line_override = "< $%s" % newname
|
|
|
464 aninput.positional = self.is_positional
|
|
|
465 self.tinputs.append(aninput)
|
|
|
466 tp = gxtp.TestParam(name=newname, value="%s_sample" % newname)
|
|
|
467 self.testparam.append(tp)
|
|
|
468 newname = self.outfiles[0][OCLPOS]
|
|
|
469 newfmt = self.outfiles[0][OFMTPOS]
|
|
|
470 anout = gxtp.OutputData(newname, format=newfmt, num_dashes=0)
|
|
|
471 anout.command_line_override = "> $%s" % newname
|
|
|
472 anout.positional = self.is_positional
|
|
|
473 self.toutputs.append(anout)
|
|
|
474 tp = gxtp.TestOutput(
|
|
|
475 name=newname, value="%s_sample" % newname, format=newfmt
|
|
|
476 )
|
|
|
477 self.testparam.append(tp)
|
|
|
478
|
|
|
479 def makeXML(self):
|
|
|
480 """
|
|
|
481 Create a Galaxy xml tool wrapper for the new script
|
|
|
482 Uses galaxyhtml
|
|
|
483 Hmmm. How to get the command line into correct order...
|
|
|
484 """
|
|
|
485 self.tool.command_line_override = self.xmlcl
|
|
|
486 if self.args.interpreter_name:
|
|
|
487 self.tool.interpreter = self.args.interpreter_name
|
|
|
488 if self.args.help_text:
|
|
|
489 helptext = open(self.args.help_text, "r").readlines()
|
|
|
490 helptext = [html_escape(x) for x in helptext]
|
|
|
491 self.tool.help = "".join([x for x in helptext])
|
|
|
492 else:
|
|
|
493 self.tool.help = (
|
|
|
494 "Please ask the tool author (%s) for help \
|
|
|
495 as none was supplied at tool generation\n"
|
|
|
496 % (self.args.user_email)
|
|
|
497 )
|
|
|
498 self.tool.version_command = None # do not want
|
|
|
499 requirements = gxtp.Requirements()
|
|
|
500
|
|
|
501 if self.args.interpreter_name:
|
|
|
502 if self.args.interpreter_name == "python":
|
|
|
503 requirements.append(
|
|
|
504 gxtp.Requirement(
|
|
|
505 "package", "python", self.args.interpreter_version
|
|
|
506 )
|
|
|
507 )
|
|
|
508 elif self.args.interpreter_name not in ["bash", "sh"]:
|
|
|
509 requirements.append(
|
|
|
510 gxtp.Requirement(
|
|
|
511 "package",
|
|
|
512 self.args.interpreter_name,
|
|
|
513 self.args.interpreter_version,
|
|
|
514 )
|
|
|
515 )
|
|
|
516 else:
|
|
|
517 if self.args.exe_package and self.args.parampass != "system":
|
|
|
518 requirements.append(
|
|
|
519 gxtp.Requirement(
|
|
|
520 "package",
|
|
|
521 self.args.exe_package,
|
|
|
522 self.args.exe_package_version,
|
|
|
523 )
|
|
|
524 )
|
|
|
525 self.tool.requirements = requirements
|
|
|
526 if self.args.parampass == "0":
|
|
|
527 self.doNoXMLparam()
|
|
|
528 else:
|
|
|
529 self.doXMLparam()
|
|
|
530 self.tool.outputs = self.toutputs
|
|
|
531 self.tool.inputs = self.tinputs
|
|
|
532 if self.args.runmode not in ["Executable", "system"]:
|
|
|
533 configfiles = gxtp.Configfiles()
|
|
|
534 configfiles.append(gxtp.Configfile(name="runMe", text=self.script))
|
|
|
535 self.tool.configfiles = configfiles
|
|
|
536 tests = gxtp.Tests()
|
|
|
537 test_a = gxtp.Test()
|
|
|
538 for tp in self.testparam:
|
|
|
539 test_a.append(tp)
|
|
|
540 tests.append(test_a)
|
|
|
541 self.tool.tests = tests
|
|
|
542 self.tool.add_comment(
|
|
|
543 "Created by %s at %s using the Galaxy Tool Factory."
|
|
|
544 % (self.args.user_email, timenow())
|
|
|
545 )
|
|
|
546 self.tool.add_comment("Source in git at: %s" % (toolFactoryURL))
|
|
|
547 self.tool.add_comment(
|
|
|
548 "Cite: Creating re-usable tools from scripts doi: \
|
|
|
549 10.1093/bioinformatics/bts573"
|
|
|
550 )
|
|
|
551 exml = self.tool.export()
|
|
|
552 xf = open('%s.xml' % self.tool_name, "w")
|
|
|
553 xf.write(exml)
|
|
|
554 xf.write("\n")
|
|
|
555 xf.close()
|
|
|
556 # ready for the tarball
|
|
|
557
|
|
|
558 def makeTooltar(self):
|
|
|
559 """
|
|
|
560 a tool is a gz tarball with eg
|
|
|
561 /toolname/tool.xml /toolname/tool.py /toolname/test-data/test1_in.foo ...
|
|
|
562 NOTE names for test inputs and outputs are munged here so must
|
|
|
563 correspond to actual input and output names used on the generated cl
|
|
|
564 """
|
|
|
565 retval = self.run()
|
|
|
566 if retval:
|
|
|
567 sys.stderr.write(
|
|
|
568 "## Run failed. Cannot build yet. Please fix and retry"
|
|
|
569 )
|
|
|
570 sys.exit(1)
|
|
|
571 tdir = "tfout"
|
|
|
572 if not os.path.exists(tdir):
|
|
|
573 os.mkdir(tdir)
|
|
|
574 self.makeXML()
|
|
|
575 testdir = os.path.join(tdir, "test-data")
|
|
|
576 if not os.path.exists(testdir):
|
|
|
577 os.mkdir(testdir) # make tests directory
|
|
|
578 for p in self.infiles:
|
|
|
579 pth = p[IPATHPOS]
|
|
|
580 dest = os.path.join(testdir, "%s_sample" % p[ICLPOS])
|
|
|
581 shutil.copyfile(pth, dest)
|
|
|
582 for p in self.outfiles:
|
|
|
583 pth = p[OCLPOS]
|
|
|
584 if p[OOCLPOS] == "STDOUT" or self.args.parampass == "0":
|
|
|
585 pth = p[ONAMEPOS]
|
|
|
586 dest = os.path.join(testdir, "%s_sample" % p[ONAMEPOS])
|
|
|
587 shutil.copyfile(pth, dest)
|
|
|
588 dest = os.path.join(tdir, p[ONAMEPOS])
|
|
|
589 shutil.copyfile(pth, dest)
|
|
|
590 else:
|
|
|
591 pth = p[OCLPOS]
|
|
|
592 dest = os.path.join(testdir, "%s_sample" % p[OCLPOS])
|
|
|
593 shutil.copyfile(pth, dest)
|
|
|
594 dest = os.path.join(tdir, p[OCLPOS])
|
|
|
595 shutil.copyfile(pth, dest)
|
|
|
596
|
|
|
597 if os.path.exists(self.tlog) and os.stat(self.tlog).st_size > 0:
|
|
|
598 shutil.copyfile(self.tlog, os.path.join(testdir, "test1_log_outfiletxt"))
|
|
|
599 if self.args.runmode not in ["Executable", "system"]:
|
|
|
600 stname = os.path.join(tdir, "%s" % (self.sfile))
|
|
|
601 if not os.path.exists(stname):
|
|
|
602 shutil.copyfile(self.sfile, stname)
|
|
|
603 xreal = '%s.xml' % self.tool_name
|
|
|
604 xout = os.path.join(tdir,xreal)
|
|
|
605 shutil.copyfile(xreal, xout)
|
|
|
606 tarpath = "toolfactory_%s.tgz" % self.tool_name
|
|
|
607 tf = tarfile.open(tarpath, "w:gz")
|
|
|
608 tf.add(name=tdir, arcname=self.tool_name)
|
|
|
609 tf.close()
|
|
|
610 shutil.copyfile(tarpath, self.args.new_tool)
|
|
|
611 shutil.copyfile(xreal,"tool_xml.txt")
|
|
|
612 repdir = "TF_run_report_tempdir"
|
|
|
613 if not os.path.exists(repdir):
|
|
|
614 os.mkdir(repdir)
|
|
|
615 repoutnames = [x[OCLPOS] for x in self.outfiles]
|
|
|
616 with os.scandir('.') as outs:
|
|
|
617 for entry in outs:
|
|
|
618 if entry.name.endswith('.tgz') or not entry.is_file():
|
|
|
619 continue
|
|
|
620 if entry.name in repoutnames:
|
|
|
621 shutil.copyfile(entry.name,os.path.join(repdir,entry.name))
|
|
|
622 elif entry.name == "%s.xml" % self.tool_name:
|
|
|
623 shutil.copyfile(entry.name,os.path.join(repdir,"new_tool_xml"))
|
|
|
624 return retval
|
|
|
625
|
|
|
626 def run(self):
|
|
|
627 """
|
|
|
628 Some devteam tools have this defensive stderr read so I'm keeping with the faith
|
|
|
629 Feel free to update.
|
|
|
630 """
|
|
|
631 s = "run cl=%s" % str(self.cl)
|
|
|
632
|
|
|
633 logging.debug(s)
|
|
|
634 scl = " ".join(self.cl)
|
|
|
635 err = None
|
|
|
636 if self.args.parampass != "0":
|
|
|
637 ste = open(self.elog, "wb")
|
|
|
638 if self.lastclredirect:
|
|
|
639 sto = open(
|
|
|
640 self.lastclredirect[1], "wb"
|
|
|
641 ) # is name of an output file
|
|
|
642 else:
|
|
|
643 sto = open(self.tlog, "wb")
|
|
|
644 sto.write(
|
|
|
645 bytes(
|
|
|
646 "## Executing Toolfactory generated command line = %s\n"
|
|
|
647 % scl,
|
|
|
648 "utf8",
|
|
|
649 )
|
|
|
650 )
|
|
|
651 sto.flush()
|
|
|
652 p = subprocess.run(self.cl, shell=False, stdout=sto, stderr=ste)
|
|
|
653 sto.close()
|
|
|
654 ste.close()
|
|
|
655 tmp_stderr = open(self.elog, "rb")
|
|
|
656 err = ""
|
|
|
657 buffsize = 1048576
|
|
|
658 try:
|
|
|
659 while True:
|
|
|
660 err += str(tmp_stderr.read(buffsize))
|
|
|
661 if not err or len(err) % buffsize != 0:
|
|
|
662 break
|
|
|
663 except OverflowError:
|
|
|
664 pass
|
|
|
665 tmp_stderr.close()
|
|
|
666 retval = p.returncode
|
|
|
667 else: # work around special case of simple scripts that take stdin and write to stdout
|
|
|
668 sti = open(self.infiles[0][IPATHPOS], "rb")
|
|
|
669 sto = open(self.outfiles[0][ONAMEPOS], "wb")
|
|
|
670 # must use shell to redirect
|
|
|
671 p = subprocess.run(self.cl, shell=False, stdout=sto, stdin=sti)
|
|
|
672 retval = p.returncode
|
|
|
673 sto.close()
|
|
|
674 sti.close()
|
|
|
675 if os.path.isfile(self.tlog) and os.stat(self.tlog).st_size == 0:
|
|
|
676 os.unlink(self.tlog)
|
|
|
677 if os.path.isfile(self.elog) and os.stat(self.elog).st_size == 0:
|
|
|
678 os.unlink(self.elog)
|
|
|
679 if p.returncode != 0 and err: # problem
|
|
|
680 sys.stderr.write(err)
|
|
|
681 logging.debug("run done")
|
|
|
682 return retval
|
|
|
683
|
|
|
684
|
|
|
685 def main():
|
|
|
686 """
|
|
|
687 This is a Galaxy wrapper. It expects to be called by a special purpose tool.xml as:
|
|
|
688 <command interpreter="python">rgBaseScriptWrapper.py --script_path "$scriptPath" --tool_name "foo" --interpreter "Rscript"
|
|
|
689 </command>
|
|
|
690 """
|
|
|
691 parser = argparse.ArgumentParser()
|
|
|
692 a = parser.add_argument
|
|
|
693 a("--script_path", default="")
|
|
|
694 a("--tool_name", default=None)
|
|
|
695 a("--interpreter_name", default=None)
|
|
|
696 a("--interpreter_version", default=None)
|
|
|
697 a("--exe_package", default=None)
|
|
|
698 a("--exe_package_version", default=None)
|
|
|
699 a("--input_files", default=[], action="append")
|
|
|
700 a("--output_files", default=[], action="append")
|
|
|
701 a("--user_email", default="Unknown")
|
|
|
702 a("--bad_user", default=None)
|
|
|
703 a("--make_Tool", default=None)
|
|
|
704 a("--help_text", default=None)
|
|
|
705 a("--tool_desc", default=None)
|
|
|
706 a("--tool_version", default=None)
|
|
|
707 a("--citations", default=None)
|
|
|
708 a("--additional_parameters", action="append", default=[])
|
|
|
709 a("--edit_additional_parameters", action="store_true", default=False)
|
|
|
710 a("--parampass", default="positional")
|
|
|
711 a("--tfout", default="./tfout")
|
|
|
712 a("--new_tool", default="new_tool")
|
|
|
713 a("--runmode", default=None)
|
|
|
714 args = parser.parse_args()
|
|
|
715 assert not args.bad_user, (
|
|
|
716 'UNAUTHORISED: %s is NOT authorized to use this tool until Galaxy admin adds %s to "admin_users" in the Galaxy configuration file'
|
|
|
717 % (args.bad_user, args.bad_user)
|
|
|
718 )
|
|
|
719 assert (
|
|
|
720 args.tool_name
|
|
|
721 ), "## Tool Factory expects a tool name - eg --tool_name=DESeq"
|
|
|
722 assert (
|
|
|
723 args.interpreter_name or args.exe_package
|
|
|
724 ), "## Tool Factory wrapper expects an interpreter or an executable package"
|
|
|
725 assert args.exe_package or (
|
|
|
726 len(args.script_path) > 0 and os.path.isfile(args.script_path)
|
|
|
727 ), "## Tool Factory wrapper expects a script path - eg --script_path=foo.R if no executable"
|
|
|
728 args.input_files = [
|
|
|
729 x.replace('"', "").replace("'", "") for x in args.input_files
|
|
|
730 ]
|
|
|
731 # remove quotes we need to deal with spaces in CL params
|
|
|
732 for i, x in enumerate(args.additional_parameters):
|
|
|
733 args.additional_parameters[i] = args.additional_parameters[i].replace(
|
|
|
734 '"', ""
|
|
|
735 )
|
|
|
736 r = ScriptRunner(args)
|
|
|
737 if args.make_Tool:
|
|
|
738 retcode = r.makeTooltar()
|
|
|
739 else:
|
|
|
740 retcode = r.run()
|
|
|
741 if retcode:
|
|
|
742 sys.exit(retcode) # indicate failure to job runner
|
|
|
743
|
|
|
744
|
|
|
745 if __name__ == "__main__":
|
|
|
746 main()
|
