|
30
|
1 # rgToolFactory.py
|
|
|
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 https://github.com/fubar2/toolfactory
|
|
|
9 #
|
|
|
10 # July 2020: BCC was fun and I feel like rip van winkle after 5 years.
|
|
|
11 # Decided to
|
|
|
12 # 1. Fix the toolfactory so it works - done for simplest case
|
|
|
13 # 2. Fix planemo so the toolfactory function works
|
|
|
14 # 3. Rewrite bits using galaxyxml functions where that makes sense - done
|
|
|
15 #
|
|
|
16 # removed all the old complications including making the new tool use this same script
|
|
|
17 # galaxyxml now generates the tool xml https://github.com/hexylena/galaxyxml
|
|
|
18 # No support for automatic HTML file creation from arbitrary outputs
|
|
|
19 # TODO: add option to run that code as a post execution hook
|
|
|
20 # TODO: add additional history input parameters - currently only one
|
|
|
21
|
|
36
|
22
|
|
|
23 import argparse
|
|
|
24 import logging
|
|
|
25 import os
|
|
|
26 import re
|
|
|
27 import shutil
|
|
30
|
28 import subprocess
|
|
36
|
29 import sys
|
|
|
30 import tarfile
|
|
|
31 import tempfile
|
|
30
|
32 import time
|
|
36
|
33
|
|
30
|
34 import galaxyxml.tool as gxt
|
|
|
35 import galaxyxml.tool.parameters as gxtp
|
|
36
|
36 import lxml
|
|
30
|
37
|
|
36
|
38 foo = lxml.__name__ # fug you, flake8. Say my name! Please accept the PR, Helena!
|
|
32
|
39
|
|
30
|
40 progname = os.path.split(sys.argv[0])[1]
|
|
36
|
41 myversion = "V2.1 July 2020"
|
|
30
|
42 verbose = True
|
|
|
43 debug = True
|
|
36
|
44 toolFactoryURL = "https://github.com/fubar2/toolfactory"
|
|
|
45 ourdelim = "~~~"
|
|
30
|
46
|
|
|
47 # --input_files="$input_files~~~$CL~~~$input_formats~~~$input_label~~~$input_help"
|
|
|
48 IPATHPOS = 0
|
|
|
49 ICLPOS = 1
|
|
|
50 IFMTPOS = 2
|
|
|
51 ILABPOS = 3
|
|
|
52 IHELPOS = 4
|
|
|
53 IOCLPOS = 5
|
|
|
54 # --output_files "$otab.history_name~~~$otab.history_format~~~$otab.CL
|
|
|
55 ONAMEPOS = 0
|
|
|
56 OFMTPOS = 1
|
|
|
57 OCLPOS = 2
|
|
|
58 OOCLPOS = 3
|
|
|
59
|
|
36
|
60 # --additional_parameters="$i.param_name~~~$i.param_value~~~$i.param_label~~~$i.param_help~~~$i.param_type~~~$i.CL"
|
|
30
|
61 ANAMEPOS = 0
|
|
|
62 AVALPOS = 1
|
|
|
63 ALABPOS = 2
|
|
|
64 AHELPPOS = 3
|
|
|
65 ATYPEPOS = 4
|
|
|
66 ACLPOS = 5
|
|
|
67 AOCLPOS = 6
|
|
|
68
|
|
36
|
69
|
|
30
|
70 def timenow():
|
|
|
71 """return current time as a string
|
|
|
72 """
|
|
36
|
73 return time.strftime("%d/%m/%Y %H:%M:%S", time.localtime(time.time()))
|
|
30
|
74
|
|
|
75
|
|
|
76 def quote_non_numeric(s):
|
|
|
77 """return a prequoted string for non-numerics
|
|
|
78 useful for perl and Rscript parameter passing?
|
|
|
79 """
|
|
|
80 try:
|
|
|
81 _ = float(s)
|
|
|
82 return s
|
|
|
83 except ValueError:
|
|
|
84 return '"%s"' % s
|
|
|
85
|
|
|
86
|
|
36
|
87 html_escape_table = {"&": "&", ">": ">", "<": "<", "$": r"\$"}
|
|
30
|
88
|
|
|
89
|
|
|
90 def html_escape(text):
|
|
|
91 """Produce entities within text."""
|
|
|
92 return "".join(html_escape_table.get(c, c) for c in text)
|
|
|
93
|
|
|
94
|
|
|
95 def html_unescape(text):
|
|
|
96 """Revert entities within text. Multiple character targets so use replace"""
|
|
36
|
97 t = text.replace("&", "&")
|
|
|
98 t = t.replace(">", ">")
|
|
|
99 t = t.replace("<", "<")
|
|
|
100 t = t.replace("\\$", "$")
|
|
30
|
101 return t
|
|
|
102
|
|
|
103
|
|
|
104 def parse_citations(citations_text):
|
|
|
105 """
|
|
|
106 """
|
|
|
107 citations = [c for c in citations_text.split("**ENTRY**") if c.strip()]
|
|
|
108 citation_tuples = []
|
|
|
109 for citation in citations:
|
|
|
110 if citation.startswith("doi"):
|
|
|
111 citation_tuples.append(("doi", citation[len("doi"):].strip()))
|
|
|
112 else:
|
|
36
|
113 citation_tuples.append(("bibtex", citation[len("bibtex"):].strip()))
|
|
30
|
114 return citation_tuples
|
|
|
115
|
|
|
116
|
|
|
117 class ScriptRunner:
|
|
|
118 """Wrapper for an arbitrary script
|
|
|
119 uses galaxyxml
|
|
|
120
|
|
|
121 """
|
|
36
|
122
|
|
30
|
123 def __init__(self, args=None):
|
|
|
124 """
|
|
|
125 prepare command line cl for running the tool here
|
|
|
126 and prepare elements needed for galaxyxml tool generation
|
|
|
127 """
|
|
|
128
|
|
|
129 self.infiles = [x.split(ourdelim) for x in args.input_files]
|
|
|
130 self.outfiles = [x.split(ourdelim) for x in args.output_files]
|
|
|
131 self.addpar = [x.split(ourdelim) for x in args.additional_parameters]
|
|
|
132 self.args = args
|
|
|
133 self.cleanuppar()
|
|
|
134 self.lastclredirect = None
|
|
|
135 self.lastxclredirect = None
|
|
|
136 self.cl = []
|
|
|
137 self.xmlcl = []
|
|
36
|
138 self.is_positional = self.args.parampass == "positional"
|
|
30
|
139 aCL = self.cl.append
|
|
36
|
140 assert args.parampass in [
|
|
|
141 "0",
|
|
|
142 "argparse",
|
|
|
143 "positional",
|
|
|
144 ], 'Parameter passing in args.parampass must be "0","positional" or "argparse"'
|
|
|
145 self.tool_name = re.sub("[^a-zA-Z0-9_]+", "", args.tool_name)
|
|
30
|
146 self.tool_id = self.tool_name
|
|
36
|
147 self.xmlfile = "%s.xml" % self.tool_name
|
|
|
148 if self.args.interpreter_name:
|
|
|
149 exe = "$runMe"
|
|
|
150 else:
|
|
|
151 exe = self.args.exe_package
|
|
|
152 assert (
|
|
|
153 exe is not None
|
|
|
154 ), "No interpeter or executable passed in - nothing to run so cannot build"
|
|
|
155 self.tool = gxt.Tool(
|
|
|
156 self.args.tool_name,
|
|
|
157 self.tool_id,
|
|
|
158 self.args.tool_version,
|
|
|
159 self.args.tool_desc,
|
|
|
160 exe,
|
|
|
161 )
|
|
|
162 self.tinputs = gxtp.Inputs()
|
|
|
163 self.toutputs = gxtp.Outputs()
|
|
|
164 self.testparam = []
|
|
|
165 if (
|
|
|
166 self.args.runmode == "Executable" or self.args.runmode == "system"
|
|
|
167 ): # binary - no need
|
|
30
|
168 aCL(self.args.exe_package) # this little CL will just run
|
|
36
|
169 else:
|
|
|
170 self.prepScript()
|
|
30
|
171 self.elog = "%s_error_log.txt" % self.tool_name
|
|
|
172 self.tlog = "%s_runner_log.txt" % self.tool_name
|
|
|
173
|
|
36
|
174 if self.args.parampass == "0":
|
|
30
|
175 self.clsimple()
|
|
|
176 else:
|
|
|
177 clsuffix = []
|
|
|
178 xclsuffix = []
|
|
|
179 for i, p in enumerate(self.infiles):
|
|
|
180 appendme = [p[IOCLPOS], p[ICLPOS], p[IPATHPOS]]
|
|
|
181 clsuffix.append(appendme)
|
|
36
|
182 xclsuffix.append([p[IOCLPOS], p[ICLPOS], "$%s" % p[ICLPOS]])
|
|
|
183 # print('##infile i=%d, appendme=%s' % (i,appendme))
|
|
30
|
184 for i, p in enumerate(self.outfiles):
|
|
|
185 if p[OOCLPOS] == "STDOUT":
|
|
36
|
186 self.lastclredirect = [">", p[ONAMEPOS]]
|
|
|
187 self.lastxclredirect = [">", "$%s" % p[OCLPOS]]
|
|
|
188 # print('##outfiles i=%d lastclredirect = %s' % (i,self.lastclredirect))
|
|
30
|
189 else:
|
|
36
|
190 appendme = [p[OOCLPOS], p[OCLPOS], p[ONAMEPOS]]
|
|
|
191 clsuffix.append(appendme)
|
|
|
192 xclsuffix.append([p[OOCLPOS], p[OCLPOS], "$%s" % p[ONAMEPOS]])
|
|
|
193 # print('##outfiles i=%d' % i,'appendme',appendme)
|
|
|
194 for p in self.addpar:
|
|
30
|
195 appendme = [p[AOCLPOS], p[ACLPOS], p[AVALPOS]]
|
|
|
196 clsuffix.append(appendme)
|
|
|
197 xclsuffix.append([p[AOCLPOS], p[ACLPOS], '"$%s"' % p[ANAMEPOS]])
|
|
36
|
198 # print('##adpar %d' % i,'appendme=',appendme)
|
|
30
|
199 clsuffix.sort()
|
|
|
200 xclsuffix.sort()
|
|
|
201 self.xclsuffix = xclsuffix
|
|
|
202 self.clsuffix = clsuffix
|
|
36
|
203 if self.args.parampass == "positional":
|
|
30
|
204 self.clpositional()
|
|
|
205 else:
|
|
|
206 self.clargparse()
|
|
36
|
207
|
|
|
208 def prepScript(self):
|
|
|
209 aCL = self.cl.append
|
|
|
210 rx = open(self.args.script_path, "r").readlines()
|
|
|
211 rx = [x.rstrip() for x in rx]
|
|
|
212 rxcheck = [x.strip() for x in rx if x.strip() > ""]
|
|
|
213 assert len(rxcheck) > 0, "Supplied script is empty. Cannot run"
|
|
|
214 self.script = "\n".join(rx)
|
|
|
215 fhandle, self.sfile = tempfile.mkstemp(
|
|
|
216 prefix=self.tool_name, suffix=".%s" % (self.args.interpreter_name)
|
|
|
217 )
|
|
|
218 tscript = open(self.sfile, "w")
|
|
|
219 tscript.write(self.script)
|
|
|
220 tscript.close()
|
|
|
221 self.indentedScript = " %s" % "\n".join([" %s" % html_escape(x) for x in rx])
|
|
|
222 self.escapedScript = "%s" % "\n".join([" %s" % html_escape(x) for x in rx])
|
|
|
223 art = "%s.%s" % (self.tool_name, self.args.interpreter_name)
|
|
|
224 artifact = open(art, "wb")
|
|
|
225 artifact.write(bytes(self.script, "utf8"))
|
|
|
226 artifact.close()
|
|
|
227 aCL(self.args.interpreter_name)
|
|
|
228 aCL(self.sfile)
|
|
|
229
|
|
30
|
230 def cleanuppar(self):
|
|
|
231 """ positional parameters are complicated by their numeric ordinal"""
|
|
36
|
232 for i, p in enumerate(self.infiles):
|
|
|
233 if self.args.parampass == "positional":
|
|
|
234 assert p[ICLPOS].isdigit(), (
|
|
|
235 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
236 % (p[ICLPOS], p[ILABPOS])
|
|
|
237 )
|
|
30
|
238 p.append(p[ICLPOS])
|
|
|
239 if p[ICLPOS].isdigit() or self.args.parampass == "0":
|
|
36
|
240 scl = "input%d" % (i + 1)
|
|
30
|
241 p[ICLPOS] = scl
|
|
|
242 self.infiles[i] = p
|
|
36
|
243 for i, p in enumerate(
|
|
|
244 self.outfiles
|
|
|
245 ): # trying to automagically gather using extensions
|
|
|
246 if self.args.parampass == "positional" and p[OCLPOS] != "STDOUT":
|
|
|
247 assert p[OCLPOS].isdigit(), (
|
|
|
248 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
249 % (p[OCLPOS], p[ONAMEPOS])
|
|
|
250 )
|
|
30
|
251 p.append(p[OCLPOS])
|
|
|
252 if p[OCLPOS].isdigit() or p[OCLPOS] == "STDOUT":
|
|
|
253 scl = p[ONAMEPOS]
|
|
|
254 p[OCLPOS] = scl
|
|
|
255 self.outfiles[i] = p
|
|
36
|
256 for i, p in enumerate(self.addpar):
|
|
|
257 if self.args.parampass == "positional":
|
|
|
258 assert p[ACLPOS].isdigit(), (
|
|
|
259 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
260 % (p[ACLPOS], p[ANAMEPOS])
|
|
|
261 )
|
|
30
|
262 p.append(p[ACLPOS])
|
|
|
263 if p[ACLPOS].isdigit():
|
|
36
|
264 scl = "input%s" % p[ACLPOS]
|
|
30
|
265 p[ACLPOS] = scl
|
|
|
266 self.addpar[i] = p
|
|
36
|
267
|
|
30
|
268 def clsimple(self):
|
|
|
269 """ no parameters - uses < and > for i/o
|
|
|
270 """
|
|
|
271 aCL = self.cl.append
|
|
36
|
272 aCL("<")
|
|
30
|
273 aCL(self.infiles[0][IPATHPOS])
|
|
36
|
274 aCL(">")
|
|
30
|
275 aCL(self.outfiles[0][OCLPOS])
|
|
|
276 aXCL = self.xmlcl.append
|
|
36
|
277 aXCL("<")
|
|
|
278 aXCL("$%s" % self.infiles[0][ICLPOS])
|
|
|
279 aXCL(">")
|
|
|
280 aXCL("$%s" % self.outfiles[0][ONAMEPOS])
|
|
30
|
281
|
|
|
282 def clpositional(self):
|
|
|
283 # inputs in order then params
|
|
|
284 aCL = self.cl.append
|
|
36
|
285 for (o_v, k, v) in self.clsuffix:
|
|
30
|
286 if " " in v:
|
|
|
287 aCL("%s" % v)
|
|
|
288 else:
|
|
|
289 aCL(v)
|
|
|
290 aXCL = self.xmlcl.append
|
|
36
|
291 for (o_v, k, v) in self.xclsuffix:
|
|
30
|
292 aXCL(v)
|
|
|
293 if self.lastxclredirect:
|
|
|
294 aXCL(self.lastxclredirect[0])
|
|
|
295 aXCL(self.lastxclredirect[1])
|
|
|
296
|
|
|
297 def clargparse(self):
|
|
|
298 """ argparse style
|
|
|
299 """
|
|
|
300 aCL = self.cl.append
|
|
|
301 aXCL = self.xmlcl.append
|
|
|
302 # inputs then params in argparse named form
|
|
36
|
303 for (o_v, k, v) in self.xclsuffix:
|
|
34
|
304 if len(k.strip()) == 1:
|
|
36
|
305 k = "-%s" % k
|
|
34
|
306 else:
|
|
36
|
307 k = "--%s" % k
|
|
30
|
308 aXCL(k)
|
|
|
309 aXCL(v)
|
|
36
|
310 for (o_v, k, v) in self.clsuffix:
|
|
30
|
311 if len(k.strip()) == 1:
|
|
36
|
312 k = "-%s" % k
|
|
30
|
313 else:
|
|
36
|
314 k = "--%s" % k
|
|
30
|
315 aCL(k)
|
|
|
316 aCL(v)
|
|
|
317
|
|
36
|
318 def getNdash(self, newname):
|
|
|
319 if self.is_positional:
|
|
|
320 ndash = 0
|
|
|
321 else:
|
|
|
322 ndash = 2
|
|
|
323 if len(newname) < 2:
|
|
|
324 ndash = 1
|
|
|
325 return ndash
|
|
30
|
326
|
|
36
|
327 def doXMLparam(self):
|
|
|
328 """flake8 made me do this..."""
|
|
|
329 for p in self.outfiles:
|
|
|
330 newname, newfmt, newcl, oldcl = p
|
|
|
331 ndash = self.getNdash(newcl)
|
|
|
332 aparm = gxtp.OutputData(newcl, format=newfmt, num_dashes=ndash)
|
|
|
333 aparm.positional = self.is_positional
|
|
|
334 if self.is_positional:
|
|
|
335 if oldcl == "STDOUT":
|
|
|
336 aparm.positional = 9999999
|
|
|
337 aparm.command_line_override = "> $%s" % newcl
|
|
|
338 else:
|
|
|
339 aparm.positional = int(oldcl)
|
|
|
340 aparm.command_line_override = "$%s" % newcl
|
|
|
341 self.toutputs.append(aparm)
|
|
|
342 tp = gxtp.TestOutput(name=newcl, value="%s_sample" % newcl, format=newfmt)
|
|
|
343 self.testparam.append(tp)
|
|
|
344 for p in self.infiles:
|
|
|
345 newname = p[ICLPOS]
|
|
|
346 newfmt = p[IFMTPOS]
|
|
|
347 ndash = self.getNdash(newname)
|
|
|
348 if not len(p[ILABPOS]) > 0:
|
|
|
349 alab = p[ICLPOS]
|
|
|
350 else:
|
|
|
351 alab = p[ILABPOS]
|
|
|
352 aninput = gxtp.DataParam(
|
|
|
353 newname,
|
|
|
354 optional=False,
|
|
|
355 label=alab,
|
|
|
356 help=p[IHELPOS],
|
|
|
357 format=newfmt,
|
|
|
358 multiple=False,
|
|
|
359 num_dashes=ndash,
|
|
|
360 )
|
|
|
361 aninput.positional = self.is_positional
|
|
|
362 self.tinputs.append(aninput)
|
|
|
363 tparm = gxtp.TestParam(name=newname, value="%s_sample" % newname)
|
|
|
364 self.testparam.append(tparm)
|
|
|
365 for p in self.addpar:
|
|
|
366 newname, newval, newlabel, newhelp, newtype, newcl, oldcl = p
|
|
|
367 if not len(newlabel) > 0:
|
|
|
368 newlabel = newname
|
|
|
369 ndash = self.getNdash(newname)
|
|
|
370 if newtype == "text":
|
|
|
371 aparm = gxtp.TextParam(
|
|
|
372 newname,
|
|
|
373 label=newlabel,
|
|
|
374 help=newhelp,
|
|
|
375 value=newval,
|
|
|
376 num_dashes=ndash,
|
|
|
377 )
|
|
|
378 elif newtype == "integer":
|
|
|
379 aparm = gxtp.IntegerParam(
|
|
|
380 newname,
|
|
|
381 label=newname,
|
|
|
382 help=newhelp,
|
|
|
383 value=newval,
|
|
|
384 num_dashes=ndash,
|
|
|
385 )
|
|
|
386 elif newtype == "float":
|
|
|
387 aparm = gxtp.FloatParam(
|
|
|
388 newname,
|
|
|
389 label=newname,
|
|
|
390 help=newhelp,
|
|
|
391 value=newval,
|
|
|
392 num_dashes=ndash,
|
|
|
393 )
|
|
|
394 else:
|
|
|
395 raise ValueError(
|
|
|
396 'Unrecognised parameter type "%s" for\
|
|
|
397 additional parameter %s in makeXML'
|
|
|
398 % (newtype, newname)
|
|
|
399 )
|
|
|
400 aparm.positional = self.is_positional
|
|
|
401 if self.is_positional:
|
|
|
402 aninput.positional = int(oldcl)
|
|
|
403 self.tinputs.append(aparm)
|
|
|
404 self.tparm = gxtp.TestParam(newname, value=newval)
|
|
|
405 self.testparam.append(tparm)
|
|
|
406
|
|
|
407 def doNoXMLparam(self):
|
|
|
408 alab = self.infiles[0][ILABPOS]
|
|
|
409 if len(alab) == 0:
|
|
|
410 alab = self.infiles[0][ICLPOS]
|
|
|
411 max1s = (
|
|
|
412 "Maximum one input if parampass is 0 - more than one input files supplied - %s"
|
|
|
413 % str(self.infiles)
|
|
|
414 )
|
|
|
415 assert len(self.infiles) == 1, max1s
|
|
|
416 newname = self.infiles[0][ICLPOS]
|
|
|
417 aninput = gxtp.DataParam(
|
|
|
418 newname,
|
|
|
419 optional=False,
|
|
|
420 label=alab,
|
|
|
421 help=self.infiles[0][IHELPOS],
|
|
|
422 format=self.infiles[0][IFMTPOS],
|
|
|
423 multiple=False,
|
|
|
424 num_dashes=0,
|
|
|
425 )
|
|
|
426 aninput.command_line_override = "< $%s" % newname
|
|
|
427 aninput.positional = self.is_positional
|
|
|
428 self.tinputs.append(aninput)
|
|
|
429 tp = gxtp.TestParam(name=newname, value="%s_sample" % newname)
|
|
|
430 self.testparam.append(tp)
|
|
|
431 newname = self.outfiles[0][OCLPOS]
|
|
|
432 newfmt = self.outfiles[0][OFMTPOS]
|
|
|
433 anout = gxtp.OutputData(newname, format=newfmt, num_dashes=0)
|
|
|
434 anout.command_line_override = "> $%s" % newname
|
|
|
435 anout.positional = self.is_positional
|
|
|
436 self.toutputs.append(anout)
|
|
|
437 tp = gxtp.TestOutput(name=newname, value="%s_sample" % newname, format=newfmt)
|
|
|
438 self.testparam.append(tp)
|
|
30
|
439
|
|
|
440 def makeXML(self):
|
|
|
441 """
|
|
|
442 Create a Galaxy xml tool wrapper for the new script
|
|
|
443 Uses galaxyhtml
|
|
|
444 Hmmm. How to get the command line into correct order...
|
|
|
445 """
|
|
36
|
446 self.tool.command_line_override = self.xmlcl
|
|
30
|
447 if self.args.interpreter_name:
|
|
36
|
448 self.tool.interpreter = self.interp
|
|
30
|
449 if self.args.help_text:
|
|
36
|
450 helptext = open(self.args.help_text, "r").readlines()
|
|
30
|
451 helptext = [html_escape(x) for x in helptext]
|
|
36
|
452 self.tool.help = "".join([x for x in helptext])
|
|
30
|
453 else:
|
|
36
|
454 self.tool.help = (
|
|
|
455 "Please ask the tool author (%s) for help \
|
|
|
456 as none was supplied at tool generation\n"
|
|
|
457 % (self.args.user_email)
|
|
|
458 )
|
|
|
459 self.tool.version_command = None # do not want
|
|
30
|
460 requirements = gxtp.Requirements()
|
|
36
|
461
|
|
30
|
462 if self.args.interpreter_name:
|
|
36
|
463 if self.args.interpreter_name == "python":
|
|
|
464 requirements.append(
|
|
|
465 gxtp.Requirement("package", "python", self.args.interpreter_version)
|
|
|
466 )
|
|
|
467 elif self.args.interpreter_name not in ["bash", "sh"]:
|
|
|
468 requirements.append(
|
|
|
469 gxtp.Requirement(
|
|
|
470 "package",
|
|
|
471 self.args.interpreter_name,
|
|
|
472 self.args.interpreter_version,
|
|
|
473 )
|
|
|
474 )
|
|
30
|
475 else:
|
|
|
476 if self.args.exe_package and self.args.parampass != "system":
|
|
36
|
477 requirements.append(
|
|
|
478 gxtp.Requirement(
|
|
|
479 "package", self.args.exe_package, self.args.exe_package_version
|
|
|
480 )
|
|
|
481 )
|
|
|
482 self.tool.requirements = requirements
|
|
|
483 if self.args.parampass == "0":
|
|
|
484 self.doXMLNoparam()
|
|
30
|
485 else:
|
|
36
|
486 self.doXMLParam()
|
|
|
487 self.tool.outputs = self.toutputs
|
|
|
488 self.tool.inputs = self.tinputs
|
|
|
489 if self.args.runmode not in ["Executable", "system"]:
|
|
30
|
490 configfiles = gxtp.Configfiles()
|
|
|
491 configfiles.append(gxtp.Configfile(name="runMe", text=self.script))
|
|
36
|
492 self.tool.configfiles = configfiles
|
|
30
|
493 tests = gxtp.Tests()
|
|
|
494 test_a = gxtp.Test()
|
|
36
|
495 for tp in self.testparam:
|
|
30
|
496 test_a.append(tp)
|
|
|
497 tests.append(test_a)
|
|
36
|
498 self.tool.tests = tests
|
|
|
499 self.tool.add_comment(
|
|
|
500 "Created by %s at %s using the Galaxy Tool Factory."
|
|
|
501 % (self.args.user_email, timenow())
|
|
|
502 )
|
|
|
503 self.tool.add_comment("Source in git at: %s" % (toolFactoryURL))
|
|
|
504 self.tool.add_comment(
|
|
|
505 "Cite: Creating re-usable tools from scripts doi: 10.1093/bioinformatics/bts573"
|
|
|
506 )
|
|
|
507 exml = self.tool.export()
|
|
|
508 xf = open(self.xmlfile, "w")
|
|
30
|
509 xf.write(exml)
|
|
36
|
510 xf.write("\n")
|
|
30
|
511 xf.close()
|
|
|
512 # ready for the tarball
|
|
|
513
|
|
|
514 def makeTooltar(self):
|
|
|
515 """
|
|
|
516 a tool is a gz tarball with eg
|
|
|
517 /toolname/tool.xml /toolname/tool.py /toolname/test-data/test1_in.foo ...
|
|
|
518 NOTE names for test inputs and outputs are munged here so must
|
|
|
519 correspond to actual input and output names used on the generated cl
|
|
|
520 """
|
|
|
521 retval = self.run()
|
|
|
522 if retval:
|
|
36
|
523 sys.stderr.write("## Run failed. Cannot build yet. Please fix and retry")
|
|
30
|
524 sys.exit(1)
|
|
36
|
525 tdir = "tfout"
|
|
30
|
526 if not os.path.exists(tdir):
|
|
|
527 os.mkdir(tdir)
|
|
|
528 self.makeXML()
|
|
36
|
529 testdir = os.path.join(tdir, "test-data")
|
|
30
|
530 if not os.path.exists(testdir):
|
|
|
531 os.mkdir(testdir) # make tests directory
|
|
|
532 for p in self.infiles:
|
|
|
533 pth = p[IPATHPOS]
|
|
36
|
534 dest = os.path.join(testdir, "%s_sample" % p[ICLPOS])
|
|
30
|
535 shutil.copyfile(pth, dest)
|
|
|
536 for p in self.outfiles:
|
|
|
537 pth = p[OCLPOS]
|
|
36
|
538 if p[OOCLPOS] == "STDOUT" or self.args.parampass == "0":
|
|
30
|
539 pth = p[ONAMEPOS]
|
|
36
|
540 dest = os.path.join(testdir, "%s_sample" % p[ONAMEPOS])
|
|
30
|
541 shutil.copyfile(pth, dest)
|
|
|
542 dest = os.path.join(tdir, p[ONAMEPOS])
|
|
36
|
543 shutil.copyfile(pth, dest)
|
|
30
|
544 else:
|
|
|
545 pth = p[OCLPOS]
|
|
36
|
546 dest = os.path.join(testdir, "%s_sample" % p[OCLPOS])
|
|
30
|
547 shutil.copyfile(pth, dest)
|
|
|
548 dest = os.path.join(tdir, p[OCLPOS])
|
|
36
|
549 shutil.copyfile(pth, dest)
|
|
30
|
550
|
|
|
551 if os.path.exists(self.tlog) and os.stat(self.tlog).st_size > 0:
|
|
36
|
552 shutil.copyfile(self.tlog, os.path.join(testdir, "test1_log.txt"))
|
|
|
553 if self.args.runmode not in ["Executable", "system"]:
|
|
|
554 stname = os.path.join(tdir, "%s" % (self.sfile))
|
|
30
|
555 if not os.path.exists(stname):
|
|
|
556 shutil.copyfile(self.sfile, stname)
|
|
36
|
557 xtname = os.path.join(tdir, self.xmlfile)
|
|
30
|
558 if not os.path.exists(xtname):
|
|
|
559 shutil.copyfile(self.xmlfile, xtname)
|
|
36
|
560 tarpath = "toolfactory_%s.tgz" % self.tool_name
|
|
|
561 tf = tarfile.open(tarpath, "w:gz")
|
|
|
562 tf.add(name=tdir, arcname=self.tool_name)
|
|
30
|
563 tf.close()
|
|
|
564 shutil.copyfile(tarpath, self.args.new_tool)
|
|
|
565 return retval
|
|
|
566
|
|
|
567 def run(self):
|
|
|
568 """
|
|
|
569 Some devteam tools have this defensive stderr read so I'm keeping with the faith
|
|
|
570 Feel free to update.
|
|
|
571 """
|
|
36
|
572 s = "run cl=%s" % str(self.cl)
|
|
|
573
|
|
30
|
574 logging.debug(s)
|
|
36
|
575 scl = " ".join(self.cl)
|
|
30
|
576 err = None
|
|
36
|
577 if self.args.parampass != "0":
|
|
|
578 ste = open(self.elog, "wb")
|
|
30
|
579 if self.lastclredirect:
|
|
36
|
580 sto = open(self.lastclredirect[1], "wb") # is name of an output file
|
|
30
|
581 else:
|
|
36
|
582 sto = open(self.tlog, "wb")
|
|
30
|
583 sto.write(
|
|
36
|
584 bytes(
|
|
|
585 "## Executing Toolfactory generated command line = %s\n" % scl,
|
|
|
586 "utf8",
|
|
|
587 )
|
|
|
588 )
|
|
30
|
589 sto.flush()
|
|
36
|
590 p = subprocess.run(self.cl, shell=False, stdout=sto, stderr=ste)
|
|
30
|
591 sto.close()
|
|
|
592 ste.close()
|
|
36
|
593 tmp_stderr = open(self.elog, "rb")
|
|
|
594 err = ""
|
|
30
|
595 buffsize = 1048576
|
|
|
596 try:
|
|
|
597 while True:
|
|
|
598 err += str(tmp_stderr.read(buffsize))
|
|
|
599 if not err or len(err) % buffsize != 0:
|
|
|
600 break
|
|
|
601 except OverflowError:
|
|
|
602 pass
|
|
|
603 tmp_stderr.close()
|
|
|
604 retval = p.returncode
|
|
|
605 else: # work around special case of simple scripts that take stdin and write to stdout
|
|
36
|
606 sti = open(self.infiles[0][IPATHPOS], "rb")
|
|
|
607 sto = open(self.outfiles[0][ONAMEPOS], "wb")
|
|
30
|
608 # must use shell to redirect
|
|
|
609 p = subprocess.run(self.cl, shell=False, stdout=sto, stdin=sti)
|
|
|
610 retval = p.returncode
|
|
|
611 sto.close()
|
|
|
612 sti.close()
|
|
|
613 if os.path.isfile(self.tlog) and os.stat(self.tlog).st_size == 0:
|
|
|
614 os.unlink(self.tlog)
|
|
|
615 if os.path.isfile(self.elog) and os.stat(self.elog).st_size == 0:
|
|
|
616 os.unlink(self.elog)
|
|
|
617 if p.returncode != 0 and err: # problem
|
|
|
618 sys.stderr.write(err)
|
|
36
|
619 logging.debug("run done")
|
|
30
|
620 return retval
|
|
|
621
|
|
|
622
|
|
|
623 def main():
|
|
|
624 """
|
|
|
625 This is a Galaxy wrapper. It expects to be called by a special purpose tool.xml as:
|
|
|
626 <command interpreter="python">rgBaseScriptWrapper.py --script_path "$scriptPath" --tool_name "foo" --interpreter "Rscript"
|
|
|
627 </command>
|
|
|
628 """
|
|
|
629 parser = argparse.ArgumentParser()
|
|
|
630 a = parser.add_argument
|
|
36
|
631 a("--script_path", default="")
|
|
|
632 a("--tool_name", default=None)
|
|
|
633 a("--interpreter_name", default=None)
|
|
|
634 a("--interpreter_version", default=None)
|
|
|
635 a("--exe_package", default=None)
|
|
|
636 a("--exe_package_version", default=None)
|
|
|
637 a("--input_files", default=[], action="append")
|
|
|
638 a("--output_files", default=[], action="append")
|
|
|
639 a("--user_email", default="Unknown")
|
|
|
640 a("--bad_user", default=None)
|
|
|
641 a("--make_Tool", default=None)
|
|
|
642 a("--help_text", default=None)
|
|
|
643 a("--tool_desc", default=None)
|
|
|
644 a("--tool_version", default=None)
|
|
|
645 a("--citations", default=None)
|
|
|
646 a("--additional_parameters", action="append", default=[])
|
|
|
647 a("--edit_additional_parameters", action="store_true", default=False)
|
|
|
648 a("--parampass", default="positional")
|
|
|
649 a("--tfout", default="./tfout")
|
|
|
650 a("--new_tool", default="new_tool")
|
|
|
651 a("--runmode", default=None)
|
|
30
|
652 args = parser.parse_args()
|
|
36
|
653 assert not args.bad_user, (
|
|
|
654 'UNAUTHORISED: %s is NOT authorized to use this tool until Galaxy admin adds %s to "admin_users" in the Galaxy configuration file'
|
|
|
655 % (args.bad_user, args.bad_user)
|
|
|
656 )
|
|
|
657 assert args.tool_name, "## Tool Factory expects a tool name - eg --tool_name=DESeq"
|
|
|
658 assert (
|
|
|
659 args.interpreter_name or args.exe_package
|
|
|
660 ), "## Tool Factory wrapper expects an interpreter or an executable package"
|
|
|
661 assert args.exe_package or (
|
|
|
662 len(args.script_path) > 0 and os.path.isfile(args.script_path)
|
|
|
663 ), "## Tool Factory wrapper expects a script path - eg --script_path=foo.R if no executable"
|
|
|
664 args.input_files = [x.replace('"', "").replace("'", "") for x in args.input_files]
|
|
30
|
665 # remove quotes we need to deal with spaces in CL params
|
|
|
666 for i, x in enumerate(args.additional_parameters):
|
|
36
|
667 args.additional_parameters[i] = args.additional_parameters[i].replace('"', "")
|
|
30
|
668 r = ScriptRunner(args)
|
|
|
669 if args.make_Tool:
|
|
|
670 retcode = r.makeTooltar()
|
|
|
671 else:
|
|
|
672 retcode = r.run()
|
|
|
673 if retcode:
|
|
|
674 sys.exit(retcode) # indicate failure to job runner
|
|
|
675
|
|
|
676
|
|
|
677 if __name__ == "__main__":
|
|
|
678 main()
|
