|
48
|
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 import yaml
|
|
|
43
|
|
|
44 myversion = "V2.1 July 2020"
|
|
|
45 verbose = True
|
|
|
46 debug = True
|
|
|
47 toolFactoryURL = "https://github.com/fubar2/toolfactory"
|
|
|
48 ourdelim = "~~~"
|
|
|
49
|
|
|
50 # --input_files="$input_files~~~$CL~~~$input_formats~~~$input_label
|
|
|
51 # ~~~$input_help"
|
|
|
52 IPATHPOS = 0
|
|
|
53 ICLPOS = 1
|
|
|
54 IFMTPOS = 2
|
|
|
55 ILABPOS = 3
|
|
|
56 IHELPOS = 4
|
|
|
57 IOCLPOS = 5
|
|
|
58
|
|
|
59 # --output_files "$otab.history_name~~~$otab.history_format~~~$otab.CL
|
|
|
60 ONAMEPOS = 0
|
|
|
61 OFMTPOS = 1
|
|
|
62 OCLPOS = 2
|
|
|
63 OOCLPOS = 3
|
|
|
64
|
|
|
65 # --additional_parameters="$i.param_name~~~$i.param_value~~~
|
|
|
66 # $i.param_label~~~$i.param_help~~~$i.param_type~~~$i.CL~~~i$.param_CLoverride"
|
|
|
67 ANAMEPOS = 0
|
|
|
68 AVALPOS = 1
|
|
|
69 ALABPOS = 2
|
|
|
70 AHELPPOS = 3
|
|
|
71 ATYPEPOS = 4
|
|
|
72 ACLPOS = 5
|
|
|
73 AOVERPOS = 6
|
|
|
74 AOCLPOS = 7
|
|
|
75
|
|
|
76
|
|
|
77 foo = len(lxml.__version__)
|
|
|
78 # fug you, flake8. Say my name!
|
|
|
79
|
|
|
80 def timenow():
|
|
|
81 """return current time as a string
|
|
|
82 """
|
|
|
83 return time.strftime("%d/%m/%Y %H:%M:%S", time.localtime(time.time()))
|
|
|
84
|
|
|
85
|
|
|
86 def quote_non_numeric(s):
|
|
|
87 """return a prequoted string for non-numerics
|
|
|
88 useful for perl and Rscript parameter passing?
|
|
|
89 """
|
|
|
90 try:
|
|
|
91 _ = float(s)
|
|
|
92 return s
|
|
|
93 except ValueError:
|
|
|
94 return '"%s"' % s
|
|
|
95
|
|
|
96
|
|
|
97 html_escape_table = {"&": "&", ">": ">", "<": "<", "$": r"\$"}
|
|
|
98
|
|
|
99
|
|
|
100 def html_escape(text):
|
|
|
101 """Produce entities within text."""
|
|
|
102 return "".join(html_escape_table.get(c, c) for c in text)
|
|
|
103
|
|
|
104
|
|
|
105 def html_unescape(text):
|
|
|
106 """Revert entities within text. Multiple character targets so use replace"""
|
|
|
107 t = text.replace("&", "&")
|
|
|
108 t = t.replace(">", ">")
|
|
|
109 t = t.replace("<", "<")
|
|
|
110 t = t.replace("\\$", "$")
|
|
|
111 return t
|
|
|
112
|
|
|
113
|
|
|
114 def parse_citations(citations_text):
|
|
|
115 """
|
|
|
116 """
|
|
|
117 citations = [c for c in citations_text.split("**ENTRY**") if c.strip()]
|
|
|
118 citation_tuples = []
|
|
|
119 for citation in citations:
|
|
|
120 if citation.startswith("doi"):
|
|
|
121 citation_tuples.append(("doi", citation[len("doi") :].strip()))
|
|
|
122 else:
|
|
|
123 citation_tuples.append(
|
|
|
124 ("bibtex", citation[len("bibtex") :].strip())
|
|
|
125 )
|
|
|
126 return citation_tuples
|
|
|
127
|
|
|
128
|
|
|
129 class ScriptRunner:
|
|
|
130 """Wrapper for an arbitrary script
|
|
|
131 uses galaxyxml
|
|
|
132
|
|
|
133 """
|
|
|
134
|
|
|
135 def __init__(self, args=None):
|
|
|
136 """
|
|
|
137 prepare command line cl for running the tool here
|
|
|
138 and prepare elements needed for galaxyxml tool generation
|
|
|
139 """
|
|
|
140
|
|
|
141 self.infiles = [x.split(ourdelim) for x in args.input_files]
|
|
|
142 self.outfiles = [x.split(ourdelim) for x in args.output_files]
|
|
|
143 self.addpar = [x.split(ourdelim) for x in args.additional_parameters]
|
|
|
144 self.args = args
|
|
|
145 self.cleanuppar()
|
|
|
146 self.lastclredirect = None
|
|
|
147 self.lastxclredirect = None
|
|
|
148 self.cl = []
|
|
|
149 self.xmlcl = []
|
|
|
150 self.is_positional = self.args.parampass == "positional"
|
|
|
151 aCL = self.cl.append
|
|
|
152 assert args.parampass in [
|
|
|
153 "0",
|
|
|
154 "argparse",
|
|
|
155 "positional",
|
|
|
156 ], 'Parameter passing in args.parampass must be "0","positional" or "argparse"'
|
|
|
157 self.tool_name = re.sub("[^a-zA-Z0-9_]+", "", args.tool_name)
|
|
|
158 self.tool_id = self.tool_name
|
|
|
159 if self.args.interpreter_name:
|
|
|
160 exe = "$runMe"
|
|
|
161 else:
|
|
|
162 exe = self.args.exe_package
|
|
|
163 assert (
|
|
|
164 exe is not None
|
|
|
165 ), "No interpeter or executable passed in - nothing to run so cannot build"
|
|
|
166 self.tool = gxt.Tool(
|
|
|
167 self.args.tool_name,
|
|
|
168 self.tool_id,
|
|
|
169 self.args.tool_version,
|
|
|
170 self.args.tool_desc,
|
|
|
171 exe,
|
|
|
172 )
|
|
|
173 self.tooloutdir = "tfout"
|
|
|
174 self.repdir = "TF_run_report_tempdir"
|
|
|
175 self.testdir = os.path.join(self.tooloutdir, "test-data")
|
|
|
176 if not os.path.exists(self.tooloutdir):
|
|
|
177 os.mkdir(self.tooloutdir)
|
|
|
178 if not os.path.exists(self.testdir):
|
|
|
179 os.mkdir(self.testdir) # make tests directory
|
|
|
180 if not os.path.exists(self.repdir):
|
|
|
181 os.mkdir(self.repdir)
|
|
|
182 self.tinputs = gxtp.Inputs()
|
|
|
183 self.toutputs = gxtp.Outputs()
|
|
|
184 self.testparam = []
|
|
|
185 if (
|
|
|
186 self.args.runmode == "Executable" or self.args.runmode == "system"
|
|
|
187 ):
|
|
|
188 if len(self.args.cl_override) > 0:
|
|
|
189 for x in self.args.cl_override.split(' '):
|
|
|
190 aCL(x)
|
|
|
191 else:
|
|
|
192 aCL(self.args.exe_package) # this little CL will just run
|
|
|
193 else:
|
|
|
194 self.prepScript()
|
|
|
195 aCL(self.args.interpreter_name)
|
|
|
196 aCL(self.sfile)
|
|
|
197
|
|
|
198 self.elog = "%s_error_log.txt" % self.tool_name
|
|
|
199 self.tlog = "%s_runner_log.txt" % self.tool_name
|
|
|
200
|
|
|
201 if self.args.parampass == "0":
|
|
|
202 self.clsimple()
|
|
|
203 else:
|
|
|
204 clsuffix = []
|
|
|
205 xclsuffix = []
|
|
|
206 for i, p in enumerate(self.infiles):
|
|
|
207 if p[IOCLPOS] == "STDIN":
|
|
|
208 appendme = [
|
|
|
209 p[IOCLPOS],
|
|
|
210 p[ICLPOS],
|
|
|
211 p[IPATHPOS],
|
|
|
212 "< %s" % p[IPATHPOS],
|
|
|
213 ]
|
|
|
214 xappendme = [
|
|
|
215 p[IOCLPOS],
|
|
|
216 p[ICLPOS],
|
|
|
217 p[IPATHPOS],
|
|
|
218 "< $%s" % p[ICLPOS],
|
|
|
219 ]
|
|
|
220 else:
|
|
|
221 appendme = [p[IOCLPOS], p[ICLPOS], p[IPATHPOS], ""]
|
|
|
222 xappendme = [p[IOCLPOS], p[ICLPOS], "$%s" % p[ICLPOS], ""]
|
|
|
223 clsuffix.append(appendme)
|
|
|
224 xclsuffix.append(xappendme)
|
|
|
225 # print('##infile i=%d, appendme=%s' % (i,appendme))
|
|
|
226 for i, p in enumerate(self.outfiles):
|
|
|
227 if p[OOCLPOS] == "STDOUT":
|
|
|
228 self.lastclredirect = [">", p[ONAMEPOS]]
|
|
|
229 self.lastxclredirect = [">", "$%s" % p[OCLPOS]]
|
|
|
230 else:
|
|
|
231 clsuffix.append([p[OOCLPOS], p[OCLPOS], p[ONAMEPOS], ""])
|
|
|
232 xclsuffix.append(
|
|
|
233 [p[OOCLPOS], p[OCLPOS], "$%s" % p[ONAMEPOS], ""]
|
|
|
234 )
|
|
|
235 for p in self.addpar:
|
|
|
236 clsuffix.append(
|
|
|
237 [p[AOCLPOS], p[ACLPOS], p[AVALPOS], p[AOVERPOS]]
|
|
|
238 )
|
|
|
239 xclsuffix.append(
|
|
|
240 [p[AOCLPOS], p[ACLPOS], '"$%s"' % p[ANAMEPOS], p[AOVERPOS]]
|
|
|
241 )
|
|
|
242 clsuffix.sort()
|
|
|
243 xclsuffix.sort()
|
|
|
244 self.xclsuffix = xclsuffix
|
|
|
245 self.clsuffix = clsuffix
|
|
|
246 if self.args.parampass == "positional":
|
|
|
247 self.clpositional()
|
|
|
248 else:
|
|
|
249 self.clargparse()
|
|
|
250
|
|
|
251 def prepScript(self):
|
|
|
252 rx = open(self.args.script_path, "r").readlines()
|
|
|
253 rx = [x.rstrip() for x in rx]
|
|
|
254 rxcheck = [x.strip() for x in rx if x.strip() > ""]
|
|
|
255 assert len(rxcheck) > 0, "Supplied script is empty. Cannot run"
|
|
|
256 self.script = "\n".join(rx)
|
|
|
257 fhandle, self.sfile = tempfile.mkstemp(
|
|
|
258 prefix=self.tool_name, suffix="_%s" % (self.args.interpreter_name)
|
|
|
259 )
|
|
|
260 tscript = open(self.sfile, "w")
|
|
|
261 tscript.write(self.script)
|
|
|
262 tscript.close()
|
|
|
263 self.indentedScript = " %s" % "\n".join(
|
|
|
264 [" %s" % html_escape(x) for x in rx]
|
|
|
265 )
|
|
|
266 self.escapedScript = "%s" % "\n".join(
|
|
|
267 [" %s" % html_escape(x) for x in rx]
|
|
|
268 )
|
|
|
269 art = "%s.%s" % (self.tool_name, self.args.interpreter_name)
|
|
|
270 artifact = open(art, "wb")
|
|
|
271 if self.args.interpreter_name == "python":
|
|
|
272 artifact.write(bytes("#!/usr/bin/env python\n", "utf8"))
|
|
|
273 artifact.write(bytes(self.script, "utf8"))
|
|
|
274 artifact.close()
|
|
|
275
|
|
|
276 def cleanuppar(self):
|
|
|
277 """ positional parameters are complicated by their numeric ordinal"""
|
|
|
278 for i, p in enumerate(self.infiles):
|
|
|
279 if self.args.parampass == "positional":
|
|
|
280 assert p[ICLPOS].isdigit(), (
|
|
|
281 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
282 % (p[ICLPOS], p[ILABPOS])
|
|
|
283 )
|
|
|
284 p.append(p[ICLPOS])
|
|
|
285 if p[ICLPOS].isdigit() or self.args.parampass == "0":
|
|
|
286 scl = "input%d" % (i + 1)
|
|
|
287 p[ICLPOS] = scl
|
|
|
288 self.infiles[i] = p
|
|
|
289 for i, p in enumerate(
|
|
|
290 self.outfiles
|
|
|
291 ): # trying to automagically gather using extensions
|
|
|
292 if self.args.parampass == "positional" and p[OCLPOS] != "STDOUT":
|
|
|
293 assert p[OCLPOS].isdigit(), (
|
|
|
294 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
295 % (p[OCLPOS], p[ONAMEPOS])
|
|
|
296 )
|
|
|
297 p.append(p[OCLPOS])
|
|
|
298 if p[OCLPOS].isdigit() or p[OCLPOS] == "STDOUT":
|
|
|
299 scl = p[ONAMEPOS]
|
|
|
300 p[OCLPOS] = scl
|
|
|
301 self.outfiles[i] = p
|
|
|
302 for i, p in enumerate(self.addpar):
|
|
|
303 if self.args.parampass == "positional":
|
|
|
304 assert p[ACLPOS].isdigit(), (
|
|
|
305 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
306 % (p[ACLPOS], p[ANAMEPOS])
|
|
|
307 )
|
|
|
308 p.append(p[ACLPOS])
|
|
|
309 if p[ACLPOS].isdigit():
|
|
|
310 scl = "input%s" % p[ACLPOS]
|
|
|
311 p[ACLPOS] = scl
|
|
|
312 self.addpar[i] = p
|
|
|
313
|
|
|
314 def clsimple(self):
|
|
|
315 """ no parameters - uses < and > for i/o
|
|
|
316 """
|
|
|
317 aCL = self.cl.append
|
|
|
318 aCL("<")
|
|
|
319 aCL(self.infiles[0][IPATHPOS])
|
|
|
320 aCL(">")
|
|
|
321 aCL(self.outfiles[0][OCLPOS])
|
|
|
322 aXCL = self.xmlcl.append
|
|
|
323 aXCL("<")
|
|
|
324 aXCL("$%s" % self.infiles[0][ICLPOS])
|
|
|
325 aXCL(">")
|
|
|
326 aXCL("$%s" % self.outfiles[0][ONAMEPOS])
|
|
|
327
|
|
|
328 def clpositional(self):
|
|
|
329 # inputs in order then params
|
|
|
330 aCL = self.cl.append
|
|
|
331 for (o_v, k, v, koverride) in self.clsuffix:
|
|
|
332 if " " in v:
|
|
|
333 aCL("%s" % v)
|
|
|
334 else:
|
|
|
335 aCL(v)
|
|
|
336 aXCL = self.xmlcl.append
|
|
|
337 for (o_v, k, v, koverride) in self.xclsuffix:
|
|
|
338 aXCL(v)
|
|
|
339 if self.lastxclredirect:
|
|
|
340 aXCL(self.lastxclredirect[0])
|
|
|
341 aXCL(self.lastxclredirect[1])
|
|
|
342
|
|
|
343 def clargparse(self):
|
|
|
344 """ argparse style
|
|
|
345 """
|
|
|
346 aCL = self.cl.append
|
|
|
347 aXCL = self.xmlcl.append
|
|
|
348 # inputs then params in argparse named form
|
|
|
349 for (o_v, k, v, koverride) in self.xclsuffix:
|
|
|
350 if koverride > "":
|
|
|
351 k = koverride
|
|
|
352 elif len(k.strip()) == 1:
|
|
|
353 k = "-%s" % k
|
|
|
354 else:
|
|
|
355 k = "--%s" % k
|
|
|
356 aXCL(k)
|
|
|
357 aXCL(v)
|
|
|
358 for (o_v, k, v, koverride) in self.clsuffix:
|
|
|
359 if koverride > "":
|
|
|
360 k = koverride
|
|
|
361 elif len(k.strip()) == 1:
|
|
|
362 k = "-%s" % k
|
|
|
363 else:
|
|
|
364 k = "--%s" % k
|
|
|
365 aCL(k)
|
|
|
366 aCL(v)
|
|
|
367
|
|
|
368 def getNdash(self, newname):
|
|
|
369 if self.is_positional:
|
|
|
370 ndash = 0
|
|
|
371 else:
|
|
|
372 ndash = 2
|
|
|
373 if len(newname) < 2:
|
|
|
374 ndash = 1
|
|
|
375 return ndash
|
|
|
376
|
|
|
377 def doXMLparam(self):
|
|
|
378 """flake8 made me do this..."""
|
|
|
379 for p in self.outfiles:
|
|
|
380 newname, newfmt, newcl, oldcl = p
|
|
|
381 ndash = self.getNdash(newcl)
|
|
|
382 aparm = gxtp.OutputData(newcl, format=newfmt, num_dashes=ndash)
|
|
|
383 aparm.positional = self.is_positional
|
|
|
384 if self.is_positional:
|
|
|
385 if oldcl == "STDOUT":
|
|
|
386 aparm.positional = 9999999
|
|
|
387 aparm.command_line_override = "> $%s" % newcl
|
|
|
388 else:
|
|
|
389 aparm.positional = int(oldcl)
|
|
|
390 aparm.command_line_override = "$%s" % newcl
|
|
|
391 self.toutputs.append(aparm)
|
|
|
392 tp = gxtp.TestOutput(
|
|
|
393 name=newcl, value="%s_sample" % newcl, format=newfmt
|
|
|
394 )
|
|
|
395 self.testparam.append(tp)
|
|
|
396 for p in self.infiles:
|
|
|
397 newname = p[ICLPOS]
|
|
|
398 newfmt = p[IFMTPOS]
|
|
|
399 ndash = self.getNdash(newname)
|
|
|
400 if not len(p[ILABPOS]) > 0:
|
|
|
401 alab = p[ICLPOS]
|
|
|
402 else:
|
|
|
403 alab = p[ILABPOS]
|
|
|
404 aninput = gxtp.DataParam(
|
|
|
405 newname,
|
|
|
406 optional=False,
|
|
|
407 label=alab,
|
|
|
408 help=p[IHELPOS],
|
|
|
409 format=newfmt,
|
|
|
410 multiple=False,
|
|
|
411 num_dashes=ndash,
|
|
|
412 )
|
|
|
413 aninput.positional = self.is_positional
|
|
|
414 self.tinputs.append(aninput)
|
|
|
415 tparm = gxtp.TestParam(name=newname, value="%s_sample" % newname)
|
|
|
416 self.testparam.append(tparm)
|
|
|
417 for p in self.addpar:
|
|
|
418 newname, newval, newlabel, newhelp, newtype, newcl, override, oldcl = p
|
|
|
419 if not len(newlabel) > 0:
|
|
|
420 newlabel = newname
|
|
|
421 ndash = self.getNdash(newname)
|
|
|
422 if newtype == "text":
|
|
|
423 aparm = gxtp.TextParam(
|
|
|
424 newname,
|
|
|
425 label=newlabel,
|
|
|
426 help=newhelp,
|
|
|
427 value=newval,
|
|
|
428 num_dashes=ndash,
|
|
|
429 )
|
|
|
430 elif newtype == "integer":
|
|
|
431 aparm = gxtp.IntegerParam(
|
|
|
432 newname,
|
|
|
433 label=newname,
|
|
|
434 help=newhelp,
|
|
|
435 value=newval,
|
|
|
436 num_dashes=ndash,
|
|
|
437 )
|
|
|
438 elif newtype == "float":
|
|
|
439 aparm = gxtp.FloatParam(
|
|
|
440 newname,
|
|
|
441 label=newname,
|
|
|
442 help=newhelp,
|
|
|
443 value=newval,
|
|
|
444 num_dashes=ndash,
|
|
|
445 )
|
|
|
446 else:
|
|
|
447 raise ValueError(
|
|
|
448 'Unrecognised parameter type "%s" for\
|
|
|
449 additional parameter %s in makeXML'
|
|
|
450 % (newtype, newname)
|
|
|
451 )
|
|
|
452 aparm.positional = self.is_positional
|
|
|
453 if self.is_positional:
|
|
|
454 aninput.positional = int(oldcl)
|
|
|
455 self.tinputs.append(aparm)
|
|
|
456 self.tparm = gxtp.TestParam(newname, value=newval)
|
|
|
457 self.testparam.append(tparm)
|
|
|
458
|
|
|
459 def doNoXMLparam(self):
|
|
|
460 alab = self.infiles[0][ILABPOS]
|
|
|
461 if len(alab) == 0:
|
|
|
462 alab = self.infiles[0][ICLPOS]
|
|
|
463 max1s = (
|
|
|
464 "Maximum one input if parampass is 0 - more than one input files supplied - %s"
|
|
|
465 % str(self.infiles)
|
|
|
466 )
|
|
|
467 assert len(self.infiles) == 1, max1s
|
|
|
468 newname = self.infiles[0][ICLPOS]
|
|
|
469 aninput = gxtp.DataParam(
|
|
|
470 newname,
|
|
|
471 optional=False,
|
|
|
472 label=alab,
|
|
|
473 help=self.infiles[0][IHELPOS],
|
|
|
474 format=self.infiles[0][IFMTPOS],
|
|
|
475 multiple=False,
|
|
|
476 num_dashes=0,
|
|
|
477 )
|
|
|
478 aninput.command_line_override = "< $%s" % newname
|
|
|
479 aninput.positional = self.is_positional
|
|
|
480 self.tinputs.append(aninput)
|
|
|
481 tp = gxtp.TestParam(name=newname, value="%s_sample" % newname)
|
|
|
482 self.testparam.append(tp)
|
|
|
483 newname = self.outfiles[0][OCLPOS]
|
|
|
484 newfmt = self.outfiles[0][OFMTPOS]
|
|
|
485 anout = gxtp.OutputData(newname, format=newfmt, num_dashes=0)
|
|
|
486 anout.command_line_override = "> $%s" % newname
|
|
|
487 anout.positional = self.is_positional
|
|
|
488 self.toutputs.append(anout)
|
|
|
489 tp = gxtp.TestOutput(
|
|
|
490 name=newname, value="%s_sample" % newname, format=newfmt
|
|
|
491 )
|
|
|
492 self.testparam.append(tp)
|
|
|
493
|
|
|
494 def makeXML(self):
|
|
|
495 """
|
|
|
496 Create a Galaxy xml tool wrapper for the new script
|
|
|
497 Uses galaxyhtml
|
|
|
498 Hmmm. How to get the command line into correct order...
|
|
|
499 """
|
|
|
500 if self.args.cl_override:
|
|
|
501 self.tool.command_line_override = self.args.cl_override.split(' ') + self.xmlcl
|
|
|
502 else:
|
|
|
503 self.tool.command_line_override = self.xmlcl
|
|
|
504 if self.args.interpreter_name:
|
|
|
505 self.tool.interpreter = self.args.interpreter_name
|
|
|
506 if self.args.help_text:
|
|
|
507 helptext = open(self.args.help_text, "r").readlines()
|
|
|
508 helptext = [html_escape(x) for x in helptext]
|
|
|
509 self.tool.help = "".join([x for x in helptext])
|
|
|
510 else:
|
|
|
511 self.tool.help = (
|
|
|
512 "Please ask the tool author (%s) for help \
|
|
|
513 as none was supplied at tool generation\n"
|
|
|
514 % (self.args.user_email)
|
|
|
515 )
|
|
|
516 self.tool.version_command = None # do not want
|
|
|
517 requirements = gxtp.Requirements()
|
|
|
518
|
|
|
519 if self.args.interpreter_name:
|
|
|
520 if self.args.dependencies:
|
|
|
521 for d in self.args.dependencies.split(','):
|
|
|
522 requirements.append(
|
|
|
523 gxtp.Requirement(
|
|
|
524 "package", d, ""
|
|
|
525 )
|
|
|
526 )
|
|
|
527 if self.args.interpreter_name == "python":
|
|
|
528 requirements.append(
|
|
|
529 gxtp.Requirement(
|
|
|
530 "package", "python", self.args.interpreter_version
|
|
|
531 )
|
|
|
532 )
|
|
|
533 elif self.args.interpreter_name not in ["bash", "sh"]:
|
|
|
534 requirements.append(
|
|
|
535 gxtp.Requirement(
|
|
|
536 "package",
|
|
|
537 self.args.interpreter_name,
|
|
|
538 self.args.interpreter_version,
|
|
|
539 )
|
|
|
540 )
|
|
|
541 else:
|
|
|
542 if self.args.exe_package and self.args.parampass != "system":
|
|
|
543 requirements.append(
|
|
|
544 gxtp.Requirement(
|
|
|
545 "package",
|
|
|
546 self.args.exe_package,
|
|
|
547 self.args.exe_package_version,
|
|
|
548 )
|
|
|
549 )
|
|
|
550 self.tool.requirements = requirements
|
|
|
551 if self.args.parampass == "0":
|
|
|
552 self.doNoXMLparam()
|
|
|
553 else:
|
|
|
554 self.doXMLparam()
|
|
|
555 self.tool.outputs = self.toutputs
|
|
|
556 self.tool.inputs = self.tinputs
|
|
|
557 if self.args.runmode not in ["Executable", "system"]:
|
|
|
558 configfiles = gxtp.Configfiles()
|
|
|
559 configfiles.append(gxtp.Configfile(name="runMe", text=self.script))
|
|
|
560 self.tool.configfiles = configfiles
|
|
|
561 tests = gxtp.Tests()
|
|
|
562 test_a = gxtp.Test()
|
|
|
563 for tp in self.testparam:
|
|
|
564 test_a.append(tp)
|
|
|
565 tests.append(test_a)
|
|
|
566 self.tool.tests = tests
|
|
|
567 self.tool.add_comment(
|
|
|
568 "Created by %s at %s using the Galaxy Tool Factory."
|
|
|
569 % (self.args.user_email, timenow())
|
|
|
570 )
|
|
|
571 self.tool.add_comment("Source in git at: %s" % (toolFactoryURL))
|
|
|
572 self.tool.add_comment(
|
|
|
573 "Cite: Creating re-usable tools from scripts doi: \
|
|
|
574 10.1093/bioinformatics/bts573"
|
|
|
575 )
|
|
|
576 exml = self.tool.export()
|
|
|
577 xf = open('%s.xml' % self.tool_name, "w")
|
|
|
578 xf.write(exml)
|
|
|
579 xf.write("\n")
|
|
|
580 xf.close()
|
|
|
581 # ready for the tarball
|
|
|
582
|
|
|
583 def makeTooltar(self):
|
|
|
584 """
|
|
|
585 a tool is a gz tarball with eg
|
|
|
586 /toolname/tool.xml /toolname/tool.py /toolname/test-data/test1_in.foo ...
|
|
|
587 NOTE names for test inputs and outputs are munged here so must
|
|
|
588 correspond to actual input and output names used on the generated cl
|
|
|
589 """
|
|
|
590 retval = self.run()
|
|
|
591 if retval:
|
|
|
592 sys.stderr.write(
|
|
|
593 "## Run failed. Cannot build yet. Please fix and retry"
|
|
|
594 )
|
|
|
595 sys.exit(1)
|
|
|
596 self.makeXML()
|
|
|
597 for p in self.infiles:
|
|
|
598 pth = p[IPATHPOS]
|
|
|
599 dest = os.path.join(self.testdir, "%s_sample" % p[ICLPOS])
|
|
|
600 shutil.copyfile(pth, dest)
|
|
|
601 for p in self.outfiles:
|
|
|
602 pth = p[OCLPOS]
|
|
|
603 if p[OOCLPOS] == "STDOUT" or self.args.parampass == "0":
|
|
|
604 pth = p[ONAMEPOS]
|
|
|
605 dest = os.path.join(self.testdir, "%s_sample" % p[ONAMEPOS])
|
|
|
606 shutil.copyfile(pth, dest)
|
|
|
607 dest = os.path.join(self.tooloutdir, p[ONAMEPOS])
|
|
|
608 shutil.copyfile(pth, dest)
|
|
|
609 else:
|
|
|
610 pth = p[OCLPOS]
|
|
|
611 dest = os.path.join(self.testdir, "%s_sample" % p[OCLPOS])
|
|
|
612 shutil.copyfile(pth, dest)
|
|
|
613 dest = os.path.join(self.tooloutdir, p[OCLPOS])
|
|
|
614 shutil.copyfile(pth, dest)
|
|
|
615 if os.path.exists(self.tlog) and os.stat(self.tlog).st_size > 0:
|
|
|
616 shutil.copyfile(self.tlog, os.path.join(self.tooloutdir, "test1_log_outfiletxt"))
|
|
|
617 if self.args.runmode not in ["Executable", "system"]:
|
|
|
618 stname = os.path.join(self.tooloutdir, "%s" % (self.sfile))
|
|
|
619 if not os.path.exists(stname):
|
|
|
620 shutil.copyfile(self.sfile, stname)
|
|
|
621 xreal = '%s.xml' % self.tool_name
|
|
|
622 xout = os.path.join(self.tooloutdir,xreal)
|
|
|
623 shutil.copyfile(xreal, xout)
|
|
|
624 self.newtarpath = "toolfactory_%s.tgz" % self.tool_name
|
|
|
625 tf = tarfile.open(self.newtarpath, "w:gz")
|
|
|
626 tf.add(name=self.tooloutdir, arcname=self.tool_name)
|
|
|
627 tf.close()
|
|
|
628 shutil.copyfile(self.newtarpath, self.args.new_tool)
|
|
|
629 shutil.copyfile(xreal,"tool_xml.txt")
|
|
|
630 repoutnames = [x[OCLPOS] for x in self.outfiles]
|
|
|
631 with os.scandir('.') as outs:
|
|
|
632 for entry in outs:
|
|
|
633 if entry.name.endswith('.tgz') or not entry.is_file():
|
|
|
634 continue
|
|
|
635 if entry.name in repoutnames:
|
|
|
636 if '.' in entry.name:
|
|
|
637 ofne = os.path.splitext(entry.name)[1]
|
|
|
638 else:
|
|
|
639 ofne = '.txt'
|
|
|
640 ofn = '%s%s' % (entry.name.replace('.','_'),ofne)
|
|
|
641 shutil.copyfile(entry.name,os.path.join(self.repdir,ofn))
|
|
|
642 elif entry.name == "%s.xml" % self.tool_name:
|
|
|
643 shutil.copyfile(entry.name,os.path.join(self.repdir,"new_tool_xml.xml"))
|
|
|
644 return retval
|
|
|
645
|
|
|
646 def run(self):
|
|
|
647 """
|
|
|
648
|
|
|
649 """
|
|
|
650 s = "run cl=%s" % str(self.cl)
|
|
|
651
|
|
|
652 logging.debug(s)
|
|
|
653 scl = " ".join(self.cl)
|
|
|
654 err = None
|
|
|
655 if self.args.parampass != "0":
|
|
|
656 ste = open(self.elog, "wb")
|
|
|
657 if self.lastclredirect:
|
|
|
658 sto = open(
|
|
|
659 self.lastclredirect[1], "wb"
|
|
|
660 ) # is name of an output file
|
|
|
661 else:
|
|
|
662 sto = open(self.tlog, "wb")
|
|
|
663 sto.write(
|
|
|
664 bytes(
|
|
|
665 "## Executing Toolfactory generated command line = %s\n"
|
|
|
666 % scl,
|
|
|
667 "utf8",
|
|
|
668 )
|
|
|
669 )
|
|
|
670 sto.flush()
|
|
|
671 p = subprocess.run(self.cl, shell=False, stdout=sto, stderr=ste)
|
|
|
672 sto.close()
|
|
|
673 ste.close()
|
|
|
674 tmp_stderr = open(self.elog, "rb")
|
|
|
675 err = ""
|
|
|
676 buffsize = 1048576
|
|
|
677 try:
|
|
|
678 while True:
|
|
|
679 err += str(tmp_stderr.read(buffsize))
|
|
|
680 if not err or len(err) % buffsize != 0:
|
|
|
681 break
|
|
|
682 except OverflowError:
|
|
|
683 pass
|
|
|
684 tmp_stderr.close()
|
|
|
685 retval = p.returncode
|
|
|
686 else: # work around special case of simple scripts that take stdin and write to stdout
|
|
|
687 sti = open(self.infiles[0][IPATHPOS], "rb")
|
|
|
688 sto = open(self.outfiles[0][ONAMEPOS], "wb")
|
|
|
689 # must use shell to redirect
|
|
|
690 p = subprocess.run(self.cl, shell=False, stdout=sto, stdin=sti)
|
|
|
691 retval = p.returncode
|
|
|
692 sto.close()
|
|
|
693 sti.close()
|
|
|
694 if os.path.isfile(self.tlog) and os.stat(self.tlog).st_size == 0:
|
|
|
695 os.unlink(self.tlog)
|
|
|
696 if os.path.isfile(self.elog) and os.stat(self.elog).st_size == 0:
|
|
|
697 os.unlink(self.elog)
|
|
|
698 if retval != 0 and err: # problem
|
|
|
699 sys.stderr.write(err)
|
|
|
700 logging.debug("run done")
|
|
|
701 return retval
|
|
|
702
|
|
|
703 def install_load(self):
|
|
|
704 testres = self.planemo_test()
|
|
|
705 if testres == 0:
|
|
|
706 self.planemo_shedload()
|
|
|
707 self,eph_galaxy_load()
|
|
|
708 else:
|
|
|
709 stderr.write('Planemo test failed - tool %s was not installed' % self.args.tool_name)
|
|
|
710
|
|
|
711
|
|
|
712 def planemo_shedload(self):
|
|
|
713 """
|
|
|
714 planemo shed_create --shed_target testtoolshed
|
|
|
715 planemo shed_update --check_diff --shed_target testtoolshed
|
|
|
716 """
|
|
|
717 cll = ['planemo', 'shed_create', '--shed_target','local']
|
|
|
718 p = subprocess.run(cll, shell=False,cwd=self.tooloutdir )
|
|
|
719 if p.returncode != 0:
|
|
|
720 print('Repository %s exists' % self.args.tool_name)
|
|
|
721 else:
|
|
|
722 print('initiated %s' % self.args.tool_name)
|
|
|
723 cll = ['planemo', 'shed_upload', '--shed_target','local','-r','--owner','fubar',
|
|
|
724 '--name',self.args.tool_name,'--shed_key',self.args.toolshed_api_key,'--tar',self.newtarpath]
|
|
|
725 p = subprocess.run(cll, shell=False,cwd=self.tooloutdir)
|
|
|
726 print('Ran',' '.join(cll),'got',p.returncode)
|
|
|
727 return p.returncode
|
|
|
728
|
|
|
729
|
|
|
730 def planemo_test(self):
|
|
|
731 """planemo is a requirement so is available
|
|
|
732 """
|
|
|
733 cll = ['planemo', 'test', '--galaxy_root', self.args.galaxy_root,
|
|
|
734 self.args.tool_name]
|
|
|
735 p = subprocess.run(cll, shell=False)
|
|
|
736 ols = os.listdir('.')
|
|
|
737 for fn in ols:
|
|
|
738 if '.' in fn:
|
|
|
739 ofne = os.path.splitext(fn)[1]
|
|
|
740 else:
|
|
|
741 ofne = '.txt'
|
|
|
742 ofn = '%s%s' % (fn.replace('.','_'),ofne)
|
|
|
743 shutil.copyfile(fn,os.path.join(self.repdir,ofn))
|
|
|
744 return p.returncode
|
|
|
745
|
|
|
746
|
|
|
747 def eph_galaxy_load(self):
|
|
|
748 """
|
|
|
749 """
|
|
|
750 cll = ['shed-tools', 'install', '-g', self.args.galaxy_url, '--latest',
|
|
|
751 '-a', self.args.galaxy_api_key , '--name', self.args.tool_name, '--owner','fubar',
|
|
|
752 '--toolshed', self.args.toolshed_url,
|
|
|
753 '--section_label','Generated Tools','--install_tool_dependencies',]
|
|
|
754 print('running\n',' '.join(cll))
|
|
|
755 p = subprocess.run(cll, shell=False)
|
|
|
756 if p.returncode != 0:
|
|
|
757 print('Repository %s installation returned %d' % (self.args.tool_name,p.retcode))
|
|
|
758 else:
|
|
|
759 print('installed %s' % self.args.tool_name)
|
|
|
760 return p.returncode
|
|
|
761
|
|
|
762 def writeShedyml(self):
|
|
|
763 yuser = self.args.user_email.split('@')[0]
|
|
|
764 yfname = os.path.join(self.tooloutdir,'.shed.yml')
|
|
|
765 yamlf = open(yfname, 'w')
|
|
|
766 odict = {'name':self.tool_name,'owner':yuser,'type':'unrestricted','description':self.args.tool_desc}
|
|
|
767 yaml.dump(odict, yamlf, allow_unicode=True)
|
|
|
768 yamlf.close()
|
|
|
769
|
|
|
770 def main():
|
|
|
771 """
|
|
|
772 This is a Galaxy wrapper. It expects to be called by a special purpose tool.xml as:
|
|
|
773 <command interpreter="python">rgBaseScriptWrapper.py --script_path "$scriptPath" --tool_name "foo" --interpreter "Rscript"
|
|
|
774 </command>
|
|
|
775 """
|
|
|
776 parser = argparse.ArgumentParser()
|
|
|
777 a = parser.add_argument
|
|
|
778 a("--script_path", default="")
|
|
|
779 a("--dependencies", default="")
|
|
|
780 a("--cl_override", default="")
|
|
|
781 a("--tool_name", default=None)
|
|
|
782 a("--interpreter_name", default=None)
|
|
|
783 a("--interpreter_version", default=None)
|
|
|
784 a("--exe_package", default=None)
|
|
|
785 a("--exe_package_version", default=None)
|
|
|
786 a("--input_files", default=[], action="append")
|
|
|
787 a("--output_files", default=[], action="append")
|
|
|
788 a("--user_email", default="Unknown")
|
|
|
789 a("--bad_user", default=None)
|
|
|
790 a("--make_Tool", default=None)
|
|
|
791 a("--help_text", default=None)
|
|
|
792 a("--tool_desc", default=None)
|
|
|
793 a("--tool_version", default=None)
|
|
|
794 a("--citations", default=None)
|
|
|
795 a("--additional_parameters", action="append", default=[])
|
|
|
796 a("--edit_additional_parameters", action="store_true", default=False)
|
|
|
797 a("--parampass", default="positional")
|
|
|
798 a("--tfout", default="./tfout")
|
|
|
799 a("--new_tool", default="new_tool")
|
|
|
800 a("--runmode", default=None)
|
|
|
801 a("--galaxy_url", default="http://localhost")
|
|
|
802 a("--galaxy_api_key", default='fakekey')
|
|
|
803 a("--toolshed_url", default="http://localhost:9009")
|
|
|
804 a("--toolshed_api_key", default=None)
|
|
|
805 a("--planemo_test", default="yes")
|
|
|
806
|
|
|
807 args = parser.parse_args()
|
|
|
808 assert not args.bad_user, (
|
|
|
809 'UNAUTHORISED: %s is NOT authorized to use this tool until Galaxy admin adds %s to "admin_users" in the Galaxy configuration file'
|
|
|
810 % (args.bad_user, args.bad_user)
|
|
|
811 )
|
|
|
812 assert (
|
|
|
813 args.tool_name
|
|
|
814 ), "## Tool Factory expects a tool name - eg --tool_name=DESeq"
|
|
|
815 assert (
|
|
|
816 args.interpreter_name or args.exe_package
|
|
|
817 ), "## Tool Factory wrapper expects an interpreter or an executable package"
|
|
|
818 assert args.exe_package or (
|
|
|
819 len(args.script_path) > 0 and os.path.isfile(args.script_path)
|
|
|
820 ), "## Tool Factory wrapper expects a script path - eg --script_path=foo.R if no executable"
|
|
|
821 args.input_files = [
|
|
|
822 x.replace('"', "").replace("'", "") for x in args.input_files
|
|
|
823 ]
|
|
|
824 # remove quotes we need to deal with spaces in CL params
|
|
|
825 for i, x in enumerate(args.additional_parameters):
|
|
|
826 args.additional_parameters[i] = args.additional_parameters[i].replace(
|
|
|
827 '"', ""
|
|
|
828 )
|
|
|
829 r = ScriptRunner(args)
|
|
|
830 if args.make_Tool:
|
|
|
831 r.writeShedyml()
|
|
|
832 retcode = r.makeTooltar()
|
|
|
833 if retcode == 0:
|
|
|
834 if args.planemo_test == "yes":
|
|
|
835 r.install_load()
|
|
|
836 else:
|
|
|
837 retcode = r.run()
|
|
|
838 # if retcode:
|
|
|
839 # sys.exit(retcode) # indicate failure to job runner
|
|
|
840
|
|
|
841
|
|
|
842 if __name__ == "__main__":
|
|
|
843 main()
|
