|
48
|
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
|
|
49
|
8 # suggestions for improvement and bug fixes welcome at
|
|
|
9 # https://github.com/fubar2/toolfactory
|
|
48
|
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 import argparse
|
|
|
26 import logging
|
|
|
27 import os
|
|
|
28 import re
|
|
|
29 import shutil
|
|
|
30 import subprocess
|
|
|
31 import sys
|
|
|
32 import tarfile
|
|
|
33 import tempfile
|
|
|
34 import time
|
|
|
35
|
|
|
36 import galaxyxml.tool as gxt
|
|
|
37 import galaxyxml.tool.parameters as gxtp
|
|
|
38
|
|
|
39 import lxml
|
|
|
40
|
|
|
41 import yaml
|
|
|
42
|
|
|
43 myversion = "V2.1 July 2020"
|
|
|
44 verbose = True
|
|
|
45 debug = True
|
|
|
46 toolFactoryURL = "https://github.com/fubar2/toolfactory"
|
|
|
47 ourdelim = "~~~"
|
|
50
|
48 ALOT = 10000000 # srsly. command or test overrides use read() so just in case
|
|
49
|
49 STDIOXML = """<stdio>
|
|
|
50 <exit_code range="100:" level="debug" description="shite happens" />
|
|
|
51 </stdio>"""
|
|
48
|
52
|
|
|
53 # --input_files="$input_files~~~$CL~~~$input_formats~~~$input_label
|
|
|
54 # ~~~$input_help"
|
|
|
55 IPATHPOS = 0
|
|
|
56 ICLPOS = 1
|
|
|
57 IFMTPOS = 2
|
|
|
58 ILABPOS = 3
|
|
|
59 IHELPOS = 4
|
|
|
60 IOCLPOS = 5
|
|
|
61
|
|
49
|
62 # --output_files "$otab.history_name~~~$otab.history_format~~~$otab.CL~~~otab.history_test
|
|
48
|
63 ONAMEPOS = 0
|
|
|
64 OFMTPOS = 1
|
|
|
65 OCLPOS = 2
|
|
49
|
66 OTESTPOS = 3
|
|
|
67 OOCLPOS = 4
|
|
|
68
|
|
48
|
69
|
|
|
70 # --additional_parameters="$i.param_name~~~$i.param_value~~~
|
|
|
71 # $i.param_label~~~$i.param_help~~~$i.param_type~~~$i.CL~~~i$.param_CLoverride"
|
|
|
72 ANAMEPOS = 0
|
|
|
73 AVALPOS = 1
|
|
|
74 ALABPOS = 2
|
|
|
75 AHELPPOS = 3
|
|
|
76 ATYPEPOS = 4
|
|
|
77 ACLPOS = 5
|
|
|
78 AOVERPOS = 6
|
|
|
79 AOCLPOS = 7
|
|
|
80
|
|
|
81
|
|
|
82 foo = len(lxml.__version__)
|
|
|
83 # fug you, flake8. Say my name!
|
|
49
|
84 FAKEEXE = "~~~REMOVE~~~ME~~~"
|
|
|
85 # need this until a PR/version bump to fix galaxyxml prepending the exe even
|
|
|
86 # with override.
|
|
|
87
|
|
48
|
88
|
|
|
89 def timenow():
|
|
|
90 """return current time as a string
|
|
|
91 """
|
|
|
92 return time.strftime("%d/%m/%Y %H:%M:%S", time.localtime(time.time()))
|
|
|
93
|
|
|
94
|
|
|
95 def quote_non_numeric(s):
|
|
|
96 """return a prequoted string for non-numerics
|
|
|
97 useful for perl and Rscript parameter passing?
|
|
|
98 """
|
|
|
99 try:
|
|
|
100 _ = float(s)
|
|
|
101 return s
|
|
|
102 except ValueError:
|
|
|
103 return '"%s"' % s
|
|
|
104
|
|
|
105
|
|
|
106 html_escape_table = {"&": "&", ">": ">", "<": "<", "$": r"\$"}
|
|
|
107
|
|
|
108
|
|
|
109 def html_escape(text):
|
|
|
110 """Produce entities within text."""
|
|
|
111 return "".join(html_escape_table.get(c, c) for c in text)
|
|
|
112
|
|
|
113
|
|
|
114 def html_unescape(text):
|
|
|
115 """Revert entities within text. Multiple character targets so use replace"""
|
|
|
116 t = text.replace("&", "&")
|
|
|
117 t = t.replace(">", ">")
|
|
|
118 t = t.replace("<", "<")
|
|
|
119 t = t.replace("\\$", "$")
|
|
|
120 return t
|
|
|
121
|
|
|
122
|
|
|
123 def parse_citations(citations_text):
|
|
|
124 """
|
|
|
125 """
|
|
|
126 citations = [c for c in citations_text.split("**ENTRY**") if c.strip()]
|
|
|
127 citation_tuples = []
|
|
|
128 for citation in citations:
|
|
|
129 if citation.startswith("doi"):
|
|
|
130 citation_tuples.append(("doi", citation[len("doi") :].strip()))
|
|
|
131 else:
|
|
49
|
132 citation_tuples.append(("bibtex", citation[len("bibtex") :].strip()))
|
|
48
|
133 return citation_tuples
|
|
|
134
|
|
|
135
|
|
|
136 class ScriptRunner:
|
|
|
137 """Wrapper for an arbitrary script
|
|
|
138 uses galaxyxml
|
|
|
139
|
|
|
140 """
|
|
|
141
|
|
|
142 def __init__(self, args=None):
|
|
|
143 """
|
|
|
144 prepare command line cl for running the tool here
|
|
|
145 and prepare elements needed for galaxyxml tool generation
|
|
|
146 """
|
|
|
147 self.infiles = [x.split(ourdelim) for x in args.input_files]
|
|
|
148 self.outfiles = [x.split(ourdelim) for x in args.output_files]
|
|
|
149 self.addpar = [x.split(ourdelim) for x in args.additional_parameters]
|
|
|
150 self.args = args
|
|
|
151 self.cleanuppar()
|
|
|
152 self.lastclredirect = None
|
|
|
153 self.lastxclredirect = None
|
|
|
154 self.cl = []
|
|
|
155 self.xmlcl = []
|
|
|
156 self.is_positional = self.args.parampass == "positional"
|
|
49
|
157 if self.args.packages:
|
|
|
158 self.executeme = self.args.packages.split(",")[0].split(":")[0]
|
|
|
159 else:
|
|
|
160 self.executeme = self.args.sysexe
|
|
|
161 assert (
|
|
|
162 self.executeme is not None
|
|
|
163 ), "No system or managed executable passed in. Cannot build"
|
|
48
|
164 aCL = self.cl.append
|
|
49
|
165 aXCL = self.xmlcl.append
|
|
48
|
166 assert args.parampass in [
|
|
|
167 "0",
|
|
|
168 "argparse",
|
|
|
169 "positional",
|
|
49
|
170 ], 'args.parampass must be "0","positional" or "argparse"'
|
|
48
|
171 self.tool_name = re.sub("[^a-zA-Z0-9_]+", "", args.tool_name)
|
|
|
172 self.tool_id = self.tool_name
|
|
50
|
173 self.newtool = gxt.Tool(
|
|
48
|
174 self.args.tool_name,
|
|
|
175 self.tool_id,
|
|
|
176 self.args.tool_version,
|
|
|
177 self.args.tool_desc,
|
|
50
|
178 FAKEEXE,
|
|
48
|
179 )
|
|
|
180 self.tooloutdir = "tfout"
|
|
|
181 self.repdir = "TF_run_report_tempdir"
|
|
|
182 self.testdir = os.path.join(self.tooloutdir, "test-data")
|
|
|
183 if not os.path.exists(self.tooloutdir):
|
|
|
184 os.mkdir(self.tooloutdir)
|
|
|
185 if not os.path.exists(self.testdir):
|
|
|
186 os.mkdir(self.testdir) # make tests directory
|
|
|
187 if not os.path.exists(self.repdir):
|
|
|
188 os.mkdir(self.repdir)
|
|
|
189 self.tinputs = gxtp.Inputs()
|
|
|
190 self.toutputs = gxtp.Outputs()
|
|
|
191 self.testparam = []
|
|
49
|
192 if self.args.script_path:
|
|
|
193 self.prepScript()
|
|
|
194 if self.args.command_override:
|
|
|
195 scos = open(self.args.command_override, "r").readlines()
|
|
|
196 self.command_override = [x.rstrip() for x in scos]
|
|
|
197 else:
|
|
|
198 self.command_override = None
|
|
|
199 if self.args.test_override:
|
|
|
200 stos = open(self.args.test_override, "r").readlines()
|
|
|
201 self.test_override = [x.rstrip() for x in stos]
|
|
|
202 else:
|
|
|
203 self.test_override = None
|
|
50
|
204 if self.args.cl_prefix: # DIY CL start
|
|
49
|
205 clp = self.args.cl_prefix.split(" ")
|
|
|
206 for c in clp:
|
|
|
207 aCL(c)
|
|
|
208 aXCL(c)
|
|
|
209 else:
|
|
56
|
210 if self.args.script_path:
|
|
|
211 aCL(self.executeme)
|
|
|
212 aCL(self.sfile)
|
|
|
213 aXCL(self.executeme)
|
|
|
214 aXCL("$runme")
|
|
48
|
215 else:
|
|
56
|
216 aCL(self.executeme) # this little CL will just run
|
|
|
217 aXCL(self.executeme)
|
|
50
|
218 self.elog = os.path.join(self.repdir, "%s_error_log.txt" % self.tool_name)
|
|
|
219 self.tlog = os.path.join(self.repdir, "%s_runner_log.txt" % self.tool_name)
|
|
48
|
220
|
|
|
221 if self.args.parampass == "0":
|
|
|
222 self.clsimple()
|
|
|
223 else:
|
|
|
224 clsuffix = []
|
|
|
225 xclsuffix = []
|
|
|
226 for i, p in enumerate(self.infiles):
|
|
|
227 if p[IOCLPOS] == "STDIN":
|
|
|
228 appendme = [
|
|
|
229 p[IOCLPOS],
|
|
|
230 p[ICLPOS],
|
|
|
231 p[IPATHPOS],
|
|
|
232 "< %s" % p[IPATHPOS],
|
|
|
233 ]
|
|
|
234 xappendme = [
|
|
|
235 p[IOCLPOS],
|
|
|
236 p[ICLPOS],
|
|
|
237 p[IPATHPOS],
|
|
|
238 "< $%s" % p[ICLPOS],
|
|
|
239 ]
|
|
|
240 else:
|
|
|
241 appendme = [p[IOCLPOS], p[ICLPOS], p[IPATHPOS], ""]
|
|
|
242 xappendme = [p[IOCLPOS], p[ICLPOS], "$%s" % p[ICLPOS], ""]
|
|
|
243 clsuffix.append(appendme)
|
|
|
244 xclsuffix.append(xappendme)
|
|
|
245 for i, p in enumerate(self.outfiles):
|
|
|
246 if p[OOCLPOS] == "STDOUT":
|
|
|
247 self.lastclredirect = [">", p[ONAMEPOS]]
|
|
|
248 self.lastxclredirect = [">", "$%s" % p[OCLPOS]]
|
|
|
249 else:
|
|
|
250 clsuffix.append([p[OOCLPOS], p[OCLPOS], p[ONAMEPOS], ""])
|
|
49
|
251 xclsuffix.append([p[OOCLPOS], p[OCLPOS], "$%s" % p[ONAMEPOS], ""])
|
|
48
|
252 for p in self.addpar:
|
|
49
|
253 clsuffix.append([p[AOCLPOS], p[ACLPOS], p[AVALPOS], p[AOVERPOS]])
|
|
48
|
254 xclsuffix.append(
|
|
|
255 [p[AOCLPOS], p[ACLPOS], '"$%s"' % p[ANAMEPOS], p[AOVERPOS]]
|
|
|
256 )
|
|
|
257 clsuffix.sort()
|
|
|
258 xclsuffix.sort()
|
|
|
259 self.xclsuffix = xclsuffix
|
|
|
260 self.clsuffix = clsuffix
|
|
|
261 if self.args.parampass == "positional":
|
|
|
262 self.clpositional()
|
|
|
263 else:
|
|
|
264 self.clargparse()
|
|
|
265
|
|
|
266 def prepScript(self):
|
|
|
267 rx = open(self.args.script_path, "r").readlines()
|
|
|
268 rx = [x.rstrip() for x in rx]
|
|
|
269 rxcheck = [x.strip() for x in rx if x.strip() > ""]
|
|
|
270 assert len(rxcheck) > 0, "Supplied script is empty. Cannot run"
|
|
|
271 self.script = "\n".join(rx)
|
|
|
272 fhandle, self.sfile = tempfile.mkstemp(
|
|
49
|
273 prefix=self.tool_name, suffix="_%s" % (self.executeme)
|
|
48
|
274 )
|
|
|
275 tscript = open(self.sfile, "w")
|
|
|
276 tscript.write(self.script)
|
|
|
277 tscript.close()
|
|
49
|
278 self.indentedScript = " %s" % "\n".join([" %s" % html_escape(x) for x in rx])
|
|
|
279 self.escapedScript = "%s" % "\n".join([" %s" % html_escape(x) for x in rx])
|
|
|
280 art = "%s.%s" % (self.tool_name, self.executeme)
|
|
48
|
281 artifact = open(art, "wb")
|
|
|
282 artifact.write(bytes(self.script, "utf8"))
|
|
|
283 artifact.close()
|
|
|
284
|
|
|
285 def cleanuppar(self):
|
|
|
286 """ positional parameters are complicated by their numeric ordinal"""
|
|
|
287 for i, p in enumerate(self.infiles):
|
|
|
288 if self.args.parampass == "positional":
|
|
|
289 assert p[ICLPOS].isdigit(), (
|
|
|
290 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
291 % (p[ICLPOS], p[ILABPOS])
|
|
|
292 )
|
|
|
293 p.append(p[ICLPOS])
|
|
|
294 if p[ICLPOS].isdigit() or self.args.parampass == "0":
|
|
|
295 scl = "input%d" % (i + 1)
|
|
|
296 p[ICLPOS] = scl
|
|
|
297 self.infiles[i] = p
|
|
|
298 for i, p in enumerate(
|
|
|
299 self.outfiles
|
|
|
300 ): # trying to automagically gather using extensions
|
|
|
301 if self.args.parampass == "positional" and p[OCLPOS] != "STDOUT":
|
|
|
302 assert p[OCLPOS].isdigit(), (
|
|
|
303 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
304 % (p[OCLPOS], p[ONAMEPOS])
|
|
|
305 )
|
|
|
306 p.append(p[OCLPOS])
|
|
|
307 if p[OCLPOS].isdigit() or p[OCLPOS] == "STDOUT":
|
|
|
308 scl = p[ONAMEPOS]
|
|
|
309 p[OCLPOS] = scl
|
|
|
310 self.outfiles[i] = p
|
|
|
311 for i, p in enumerate(self.addpar):
|
|
|
312 if self.args.parampass == "positional":
|
|
|
313 assert p[ACLPOS].isdigit(), (
|
|
|
314 "Positional parameters must be ordinal integers - got %s for %s"
|
|
|
315 % (p[ACLPOS], p[ANAMEPOS])
|
|
|
316 )
|
|
|
317 p.append(p[ACLPOS])
|
|
|
318 if p[ACLPOS].isdigit():
|
|
|
319 scl = "input%s" % p[ACLPOS]
|
|
|
320 p[ACLPOS] = scl
|
|
|
321 self.addpar[i] = p
|
|
|
322
|
|
|
323 def clsimple(self):
|
|
|
324 """ no parameters - uses < and > for i/o
|
|
|
325 """
|
|
|
326 aCL = self.cl.append
|
|
|
327 aXCL = self.xmlcl.append
|
|
62
|
328
|
|
|
329 if len(self.infiles) > 0:
|
|
|
330 aCL("<")
|
|
|
331 aCL(self.infiles[0][IPATHPOS])
|
|
|
332 aXCL("<")
|
|
|
333 aXCL("$%s" % self.infiles[0][ICLPOS])
|
|
|
334 if len(self.outfiles) > 0:
|
|
|
335 aCL(">")
|
|
|
336 aCL(self.outfiles[0][OCLPOS])
|
|
|
337 aXCL(">")
|
|
|
338 aXCL("$%s" % self.outfiles[0][ONAMEPOS])
|
|
48
|
339
|
|
|
340 def clpositional(self):
|
|
|
341 # inputs in order then params
|
|
|
342 aCL = self.cl.append
|
|
|
343 for (o_v, k, v, koverride) in self.clsuffix:
|
|
|
344 if " " in v:
|
|
|
345 aCL("%s" % v)
|
|
|
346 else:
|
|
|
347 aCL(v)
|
|
|
348 aXCL = self.xmlcl.append
|
|
|
349 for (o_v, k, v, koverride) in self.xclsuffix:
|
|
|
350 aXCL(v)
|
|
|
351 if self.lastxclredirect:
|
|
|
352 aXCL(self.lastxclredirect[0])
|
|
|
353 aXCL(self.lastxclredirect[1])
|
|
|
354
|
|
|
355 def clargparse(self):
|
|
|
356 """ argparse style
|
|
|
357 """
|
|
|
358 aCL = self.cl.append
|
|
|
359 aXCL = self.xmlcl.append
|
|
|
360 # inputs then params in argparse named form
|
|
|
361 for (o_v, k, v, koverride) in self.xclsuffix:
|
|
|
362 if koverride > "":
|
|
|
363 k = koverride
|
|
|
364 elif len(k.strip()) == 1:
|
|
|
365 k = "-%s" % k
|
|
|
366 else:
|
|
|
367 k = "--%s" % k
|
|
|
368 aXCL(k)
|
|
|
369 aXCL(v)
|
|
|
370 for (o_v, k, v, koverride) in self.clsuffix:
|
|
|
371 if koverride > "":
|
|
|
372 k = koverride
|
|
|
373 elif len(k.strip()) == 1:
|
|
|
374 k = "-%s" % k
|
|
|
375 else:
|
|
|
376 k = "--%s" % k
|
|
|
377 aCL(k)
|
|
|
378 aCL(v)
|
|
|
379
|
|
|
380 def getNdash(self, newname):
|
|
|
381 if self.is_positional:
|
|
|
382 ndash = 0
|
|
|
383 else:
|
|
|
384 ndash = 2
|
|
|
385 if len(newname) < 2:
|
|
|
386 ndash = 1
|
|
|
387 return ndash
|
|
|
388
|
|
|
389 def doXMLparam(self):
|
|
|
390 """flake8 made me do this..."""
|
|
|
391 for p in self.outfiles:
|
|
49
|
392 newname, newfmt, newcl, test, oldcl = p
|
|
48
|
393 ndash = self.getNdash(newcl)
|
|
|
394 aparm = gxtp.OutputData(newcl, format=newfmt, num_dashes=ndash)
|
|
|
395 aparm.positional = self.is_positional
|
|
|
396 if self.is_positional:
|
|
|
397 if oldcl == "STDOUT":
|
|
|
398 aparm.positional = 9999999
|
|
|
399 aparm.command_line_override = "> $%s" % newcl
|
|
|
400 else:
|
|
|
401 aparm.positional = int(oldcl)
|
|
|
402 aparm.command_line_override = "$%s" % newcl
|
|
|
403 self.toutputs.append(aparm)
|
|
49
|
404 usetest = None
|
|
|
405 ld = None
|
|
50
|
406 if test > "":
|
|
|
407 if test.startswith("diff"):
|
|
|
408 usetest = "diff"
|
|
|
409 if test.split(":")[1].isdigit:
|
|
|
410 ld = int(test.split(":")[1])
|
|
49
|
411 else:
|
|
|
412 usetest = test
|
|
50
|
413 tp = gxtp.TestOutput(
|
|
|
414 name=newcl,
|
|
|
415 value="%s_sample" % newcl,
|
|
|
416 format=newfmt,
|
|
|
417 compare=usetest,
|
|
|
418 lines_diff=ld,
|
|
|
419 delta=None,
|
|
|
420 )
|
|
48
|
421 self.testparam.append(tp)
|
|
|
422 for p in self.infiles:
|
|
|
423 newname = p[ICLPOS]
|
|
|
424 newfmt = p[IFMTPOS]
|
|
|
425 ndash = self.getNdash(newname)
|
|
|
426 if not len(p[ILABPOS]) > 0:
|
|
|
427 alab = p[ICLPOS]
|
|
|
428 else:
|
|
|
429 alab = p[ILABPOS]
|
|
|
430 aninput = gxtp.DataParam(
|
|
|
431 newname,
|
|
|
432 optional=False,
|
|
|
433 label=alab,
|
|
|
434 help=p[IHELPOS],
|
|
|
435 format=newfmt,
|
|
|
436 multiple=False,
|
|
|
437 num_dashes=ndash,
|
|
|
438 )
|
|
|
439 aninput.positional = self.is_positional
|
|
|
440 self.tinputs.append(aninput)
|
|
|
441 tparm = gxtp.TestParam(name=newname, value="%s_sample" % newname)
|
|
|
442 self.testparam.append(tparm)
|
|
|
443 for p in self.addpar:
|
|
|
444 newname, newval, newlabel, newhelp, newtype, newcl, override, oldcl = p
|
|
|
445 if not len(newlabel) > 0:
|
|
|
446 newlabel = newname
|
|
|
447 ndash = self.getNdash(newname)
|
|
|
448 if newtype == "text":
|
|
|
449 aparm = gxtp.TextParam(
|
|
|
450 newname,
|
|
|
451 label=newlabel,
|
|
|
452 help=newhelp,
|
|
|
453 value=newval,
|
|
|
454 num_dashes=ndash,
|
|
|
455 )
|
|
|
456 elif newtype == "integer":
|
|
|
457 aparm = gxtp.IntegerParam(
|
|
|
458 newname,
|
|
|
459 label=newname,
|
|
|
460 help=newhelp,
|
|
|
461 value=newval,
|
|
|
462 num_dashes=ndash,
|
|
|
463 )
|
|
|
464 elif newtype == "float":
|
|
|
465 aparm = gxtp.FloatParam(
|
|
|
466 newname,
|
|
|
467 label=newname,
|
|
|
468 help=newhelp,
|
|
|
469 value=newval,
|
|
|
470 num_dashes=ndash,
|
|
|
471 )
|
|
|
472 else:
|
|
|
473 raise ValueError(
|
|
|
474 'Unrecognised parameter type "%s" for\
|
|
|
475 additional parameter %s in makeXML'
|
|
|
476 % (newtype, newname)
|
|
|
477 )
|
|
|
478 aparm.positional = self.is_positional
|
|
|
479 if self.is_positional:
|
|
|
480 aninput.positional = int(oldcl)
|
|
|
481 self.tinputs.append(aparm)
|
|
|
482 self.tparm = gxtp.TestParam(newname, value=newval)
|
|
|
483 self.testparam.append(tparm)
|
|
|
484
|
|
|
485 def doNoXMLparam(self):
|
|
49
|
486 """filter style package - stdin to stdout"""
|
|
62
|
487 if len(self.infiles) > 0:
|
|
|
488 alab = self.infiles[0][ILABPOS]
|
|
|
489 if len(alab) == 0:
|
|
|
490 alab = self.infiles[0][ICLPOS]
|
|
|
491 max1s = (
|
|
|
492 "Maximum one input if parampass is 0 but multiple input files supplied - %s"
|
|
|
493 % str(self.infiles)
|
|
|
494 )
|
|
|
495 assert len(self.infiles) == 1, max1s
|
|
|
496 newname = self.infiles[0][ICLPOS]
|
|
|
497 aninput = gxtp.DataParam(
|
|
|
498 newname,
|
|
|
499 optional=False,
|
|
|
500 label=alab,
|
|
|
501 help=self.infiles[0][IHELPOS],
|
|
|
502 format=self.infiles[0][IFMTPOS],
|
|
|
503 multiple=False,
|
|
|
504 num_dashes=0,
|
|
|
505 )
|
|
|
506 aninput.command_line_override = "< $%s" % newname
|
|
|
507 aninput.positional = self.is_positional
|
|
|
508 self.tinputs.append(aninput)
|
|
|
509 tp = gxtp.TestParam(name=newname, value="%s_sample" % newname)
|
|
|
510 self.testparam.append(tp)
|
|
|
511 if len(self.outfiles > 0):
|
|
|
512 newname = self.outfiles[0][OCLPOS]
|
|
|
513 newfmt = self.outfiles[0][OFMTPOS]
|
|
|
514 anout = gxtp.OutputData(newname, format=newfmt, num_dashes=0)
|
|
|
515 anout.command_line_override = "> $%s" % newname
|
|
|
516 anout.positional = self.is_positional
|
|
|
517 self.toutputs.append(anout)
|
|
|
518 tp = gxtp.TestOutput(name=newname, value="%s_sample" % newname, format=newfmt)
|
|
|
519 self.testparam.append(tp)
|
|
48
|
520
|
|
|
521 def makeXML(self):
|
|
|
522 """
|
|
|
523 Create a Galaxy xml tool wrapper for the new script
|
|
|
524 Uses galaxyhtml
|
|
|
525 Hmmm. How to get the command line into correct order...
|
|
|
526 """
|
|
49
|
527 if self.command_override:
|
|
56
|
528 self.newtool.command_override = self.command_override # config file
|
|
48
|
529 else:
|
|
56
|
530 self.newtool.command_override = self.xmlcl
|
|
48
|
531 if self.args.help_text:
|
|
|
532 helptext = open(self.args.help_text, "r").readlines()
|
|
50
|
533 safertext = [html_escape(x) for x in helptext]
|
|
|
534 if self.args.script_path:
|
|
|
535 scrpt = self.script.split('\n')
|
|
62
|
536 scrpt.insert(0,"```\n")
|
|
50
|
537 if len(scrpt) > 300:
|
|
|
538 safertext = safertext + scrpt[:100] + ['>500 lines - stuff deleted','......'] + scrpt[-100:]
|
|
|
539 else:
|
|
|
540 safertext = safertext + scrpt
|
|
|
541 safertext.append("\n```")
|
|
62
|
542 self.newtool.help = "\n".join([x for x in safertext])
|
|
48
|
543 else:
|
|
50
|
544 self.newtool.help = (
|
|
48
|
545 "Please ask the tool author (%s) for help \
|
|
|
546 as none was supplied at tool generation\n"
|
|
|
547 % (self.args.user_email)
|
|
|
548 )
|
|
50
|
549 self.newtool.version_command = None # do not want
|
|
48
|
550 requirements = gxtp.Requirements()
|
|
49
|
551 if self.args.packages:
|
|
|
552 for d in self.args.packages.split(","):
|
|
|
553 if ":" in d:
|
|
|
554 packg, ver = d.split(":")
|
|
|
555 else:
|
|
|
556 packg = d
|
|
|
557 ver = ""
|
|
50
|
558 requirements.append(
|
|
|
559 gxtp.Requirement("package", packg.strip(), ver.strip())
|
|
|
560 )
|
|
|
561 self.newtool.requirements = requirements
|
|
48
|
562 if self.args.parampass == "0":
|
|
|
563 self.doNoXMLparam()
|
|
|
564 else:
|
|
|
565 self.doXMLparam()
|
|
50
|
566 self.newtool.outputs = self.toutputs
|
|
|
567 self.newtool.inputs = self.tinputs
|
|
|
568 if self.args.script_path:
|
|
48
|
569 configfiles = gxtp.Configfiles()
|
|
49
|
570 configfiles.append(gxtp.Configfile(name="runme", text=self.script))
|
|
50
|
571 self.newtool.configfiles = configfiles
|
|
48
|
572 tests = gxtp.Tests()
|
|
|
573 test_a = gxtp.Test()
|
|
|
574 for tp in self.testparam:
|
|
|
575 test_a.append(tp)
|
|
|
576 tests.append(test_a)
|
|
50
|
577 self.newtool.tests = tests
|
|
|
578 self.newtool.add_comment(
|
|
48
|
579 "Created by %s at %s using the Galaxy Tool Factory."
|
|
|
580 % (self.args.user_email, timenow())
|
|
|
581 )
|
|
50
|
582 self.newtool.add_comment("Source in git at: %s" % (toolFactoryURL))
|
|
|
583 self.newtool.add_comment(
|
|
48
|
584 "Cite: Creating re-usable tools from scripts doi: \
|
|
|
585 10.1093/bioinformatics/bts573"
|
|
|
586 )
|
|
50
|
587 exml0 = self.newtool.export()
|
|
49
|
588 exml = exml0.replace(FAKEEXE, "") # temporary work around until PR accepted
|
|
50
|
589 if (
|
|
|
590 self.test_override
|
|
|
591 ): # cannot do this inside galaxyxml as it expects lxml objects for tests
|
|
|
592 part1 = exml.split("<tests>")[0]
|
|
|
593 part2 = exml.split("</tests>")[1]
|
|
|
594 fixed = "%s\n%s\n%s" % (part1, self.test_override, part2)
|
|
49
|
595 exml = fixed
|
|
|
596 xf = open("%s.xml" % self.tool_name, "w")
|
|
48
|
597 xf.write(exml)
|
|
|
598 xf.write("\n")
|
|
|
599 xf.close()
|
|
|
600 # ready for the tarball
|
|
|
601
|
|
|
602 def run(self):
|
|
|
603 """
|
|
50
|
604 generate test outputs by running a command line
|
|
56
|
605 won't work if command or test override in play - planemo is the
|
|
50
|
606 easiest way to generate test outputs for that case so is
|
|
|
607 automagically selected
|
|
48
|
608 """
|
|
|
609 s = "run cl=%s" % str(self.cl)
|
|
|
610 logging.debug(s)
|
|
|
611 scl = " ".join(self.cl)
|
|
|
612 err = None
|
|
|
613 if self.args.parampass != "0":
|
|
56
|
614 if os.path.exists(self.elog):
|
|
|
615 ste = open(self.elog, "a")
|
|
|
616 else:
|
|
|
617 ste = open(self.elog, "w")
|
|
48
|
618 if self.lastclredirect:
|
|
49
|
619 sto = open(self.lastclredirect[1], "wb") # is name of an output file
|
|
48
|
620 else:
|
|
56
|
621 if os.path.exists(self.tlog):
|
|
|
622 sto = open(self.tlog, "a")
|
|
|
623 else:
|
|
|
624 sto = open(self.tlog, "w")
|
|
48
|
625 sto.write(
|
|
56
|
626 "## Executing Toolfactory generated command line = %s\n" % scl
|
|
48
|
627 )
|
|
|
628 sto.flush()
|
|
|
629 p = subprocess.run(self.cl, shell=False, stdout=sto, stderr=ste)
|
|
|
630 sto.close()
|
|
|
631 ste.close()
|
|
|
632 retval = p.returncode
|
|
49
|
633 else: # work around special case - stdin and write to stdout
|
|
62
|
634 if len(self.infiles) > 0:
|
|
|
635 sti = open(self.infiles[0][IPATHPOS], "rb")
|
|
|
636 else:
|
|
|
637 sti = sys.stdin
|
|
|
638 if len(self.outfiles) > 0:
|
|
|
639 sto = open(self.outfiles[0][ONAMEPOS], "wb")
|
|
|
640 else:
|
|
|
641 sto = sys.stdout
|
|
48
|
642 p = subprocess.run(self.cl, shell=False, stdout=sto, stdin=sti)
|
|
|
643 retval = p.returncode
|
|
|
644 sto.close()
|
|
|
645 sti.close()
|
|
|
646 if os.path.isfile(self.tlog) and os.stat(self.tlog).st_size == 0:
|
|
|
647 os.unlink(self.tlog)
|
|
|
648 if os.path.isfile(self.elog) and os.stat(self.elog).st_size == 0:
|
|
|
649 os.unlink(self.elog)
|
|
|
650 if retval != 0 and err: # problem
|
|
|
651 sys.stderr.write(err)
|
|
|
652 logging.debug("run done")
|
|
|
653 return retval
|
|
|
654
|
|
|
655 def planemo_shedload(self):
|
|
|
656 """
|
|
|
657 planemo shed_create --shed_target testtoolshed
|
|
|
658 planemo shed_update --check_diff --shed_target testtoolshed
|
|
|
659 """
|
|
49
|
660 if os.path.exists(self.tlog):
|
|
50
|
661 tout = open(self.tlog, "a")
|
|
49
|
662 else:
|
|
50
|
663 tout = open(self.tlog, "w")
|
|
49
|
664 cll = ["planemo", "shed_create", "--shed_target", "local"]
|
|
|
665 try:
|
|
50
|
666 p = subprocess.run(
|
|
|
667 cll, shell=True, cwd=self.tooloutdir, stdout=tout, stderr=tout
|
|
|
668 )
|
|
49
|
669 except:
|
|
|
670 pass
|
|
48
|
671 if p.returncode != 0:
|
|
49
|
672 print("Repository %s exists" % self.args.tool_name)
|
|
48
|
673 else:
|
|
49
|
674 print("initiated %s" % self.args.tool_name)
|
|
|
675 cll = [
|
|
|
676 "planemo",
|
|
|
677 "shed_upload",
|
|
|
678 "--shed_target",
|
|
|
679 "local",
|
|
|
680 "--owner",
|
|
|
681 "fubar",
|
|
|
682 "--name",
|
|
|
683 self.args.tool_name,
|
|
|
684 "--shed_key",
|
|
|
685 self.args.toolshed_api_key,
|
|
|
686 "--tar",
|
|
|
687 self.newtarpath,
|
|
|
688 ]
|
|
56
|
689 p = subprocess.run(cll, shell=False)
|
|
49
|
690 print("Ran", " ".join(cll), "got", p.returncode)
|
|
|
691 tout.close()
|
|
48
|
692 return p.returncode
|
|
|
693
|
|
49
|
694 def planemo_test(self, genoutputs=True):
|
|
50
|
695 """planemo is a requirement so is available for testing
|
|
|
696 and for generating test outputs if command or test overrides are supplied
|
|
|
697 test outputs are sent to repdir for display
|
|
48
|
698 """
|
|
49
|
699 xreal = "%s.xml" % self.tool_name
|
|
|
700 if os.path.exists(self.tlog):
|
|
50
|
701 tout = open(self.tlog, "a")
|
|
49
|
702 else:
|
|
50
|
703 tout = open(self.tlog, "w")
|
|
49
|
704 if genoutputs:
|
|
|
705 cll = [
|
|
|
706 "planemo",
|
|
|
707 "test",
|
|
|
708 "--galaxy_root",
|
|
|
709 self.args.galaxy_root,
|
|
|
710 "--update_test_data",
|
|
|
711 xreal,
|
|
|
712 ]
|
|
|
713 else:
|
|
56
|
714 cll = ["planemo", "test", "--galaxy_root",
|
|
62
|
715 self.args.galaxy_root,
|
|
50
|
716 xreal,]
|
|
|
717 p = subprocess.run(
|
|
56
|
718 cll, shell=False, cwd=self.tooloutdir, stderr=tout, stdout=tout
|
|
50
|
719 )
|
|
|
720 if genoutputs:
|
|
|
721 with os.scandir(self.testdir) as outs:
|
|
|
722 for entry in outs:
|
|
|
723 if entry.is_file():
|
|
|
724 dest = os.path.join(self.repdir, entry.name)
|
|
|
725 src = os.path.join(self.testdir, entry.name)
|
|
|
726 shutil.copyfile(src, dest)
|
|
|
727 tout.write(
|
|
|
728 "Copied output %s to %s after planemo test\n" % (src, dest)
|
|
|
729 )
|
|
49
|
730 tout.close()
|
|
48
|
731 return p.returncode
|
|
|
732
|
|
|
733 def eph_galaxy_load(self):
|
|
50
|
734 """load the new tool from the local toolshed after planemo uploads it
|
|
48
|
735 """
|
|
49
|
736 if os.path.exists(self.tlog):
|
|
50
|
737 tout = open(self.tlog, "a")
|
|
49
|
738 else:
|
|
50
|
739 tout = open(self.tlog, "w")
|
|
49
|
740 cll = [
|
|
|
741 "shed-tools",
|
|
|
742 "install",
|
|
|
743 "-g",
|
|
|
744 self.args.galaxy_url,
|
|
|
745 "--latest",
|
|
|
746 "-a",
|
|
|
747 self.args.galaxy_api_key,
|
|
|
748 "--name",
|
|
|
749 self.args.tool_name,
|
|
|
750 "--owner",
|
|
|
751 "fubar",
|
|
|
752 "--toolshed",
|
|
|
753 self.args.toolshed_url,
|
|
|
754 "--section_label",
|
|
|
755 "Generated Tools",
|
|
51
|
756
|
|
49
|
757 ]
|
|
|
758 print("running\n", " ".join(cll))
|
|
|
759 p = subprocess.run(cll, shell=False, stderr=tout, stdout=tout)
|
|
48
|
760 if p.returncode != 0:
|
|
49
|
761 print(
|
|
|
762 "Repository %s installation returned %d"
|
|
|
763 % (self.args.tool_name, p.returncode)
|
|
|
764 )
|
|
48
|
765 else:
|
|
49
|
766 print("installed %s" % self.args.tool_name)
|
|
|
767 tout.close()
|
|
48
|
768 return p.returncode
|
|
|
769
|
|
|
770 def writeShedyml(self):
|
|
50
|
771 """for planemo
|
|
|
772 """
|
|
49
|
773 yuser = self.args.user_email.split("@")[0]
|
|
|
774 yfname = os.path.join(self.tooloutdir, ".shed.yml")
|
|
|
775 yamlf = open(yfname, "w")
|
|
|
776 odict = {
|
|
|
777 "name": self.tool_name,
|
|
|
778 "owner": yuser,
|
|
|
779 "type": "unrestricted",
|
|
|
780 "description": self.args.tool_desc,
|
|
50
|
781 "synopsis": self.args.tool_desc,
|
|
|
782 "category": "TF Generated Tools",
|
|
49
|
783 }
|
|
48
|
784 yaml.dump(odict, yamlf, allow_unicode=True)
|
|
|
785 yamlf.close()
|
|
|
786
|
|
50
|
787 def makeTool(self):
|
|
|
788 """write xmls and input samples into place
|
|
|
789 """
|
|
|
790 self.makeXML()
|
|
|
791 if self.args.script_path:
|
|
|
792 stname = os.path.join(self.tooloutdir, "%s" % (self.sfile))
|
|
|
793 if not os.path.exists(stname):
|
|
|
794 shutil.copyfile(self.sfile, stname)
|
|
|
795 xreal = "%s.xml" % self.tool_name
|
|
|
796 xout = os.path.join(self.tooloutdir, xreal)
|
|
|
797 shutil.copyfile(xreal, xout)
|
|
|
798 for p in self.infiles:
|
|
|
799 pth = p[IPATHPOS]
|
|
|
800 dest = os.path.join(self.testdir, "%s_sample" % p[ICLPOS])
|
|
|
801 shutil.copyfile(pth, dest)
|
|
|
802 dest = os.path.join(self.repdir, "%s.%s" % (p[ICLPOS], p[IFMTPOS]))
|
|
|
803 shutil.copyfile(pth, dest)
|
|
49
|
804
|
|
50
|
805 def makeToolTar(self):
|
|
56
|
806 """ move outputs into test-data and prepare the tarball
|
|
50
|
807 """
|
|
|
808 for p in self.outfiles:
|
|
|
809 src = p[ONAMEPOS]
|
|
|
810 if os.path.isfile(src):
|
|
|
811 dest = os.path.join(self.testdir, "%s_sample" % src)
|
|
|
812 shutil.copyfile(src, dest)
|
|
|
813 dest = os.path.join(self.repdir, "%s.%s" % (src, p[OFMTPOS]))
|
|
|
814 shutil.copyfile(src, dest)
|
|
|
815 else:
|
|
|
816 print(
|
|
|
817 "### problem - output file %s not found in tooloutdir %s"
|
|
|
818 % (src, self.tooloutdir)
|
|
|
819 )
|
|
|
820 self.newtarpath = "toolfactory_%s.tgz" % self.tool_name
|
|
|
821 tf = tarfile.open(self.newtarpath, "w:gz")
|
|
|
822 tf.add(name=self.tooloutdir, arcname=self.tool_name)
|
|
|
823 tf.close()
|
|
|
824 shutil.copyfile(self.newtarpath, self.args.new_tool)
|
|
|
825
|
|
|
826 def moveRunOutputs(self):
|
|
|
827 """need to move planemo or run outputs into toolfactory collection
|
|
|
828 """
|
|
|
829 with os.scandir(self.tooloutdir) as outs:
|
|
|
830 for entry in outs:
|
|
|
831 if not entry.is_file() or entry.name.startswith("."):
|
|
|
832 continue
|
|
|
833 if "." in entry.name:
|
|
|
834 nayme, ext = os.path.splitext(entry.name)
|
|
|
835 else:
|
|
|
836 ext = ".txt"
|
|
|
837 ofn = "%s%s" % (entry.name.replace(".", "_"), ext)
|
|
|
838 dest = os.path.join(self.repdir, ofn)
|
|
|
839 src = os.path.join(self.tooloutdir, entry.name)
|
|
|
840 shutil.copyfile(src, dest)
|
|
|
841
|
|
49
|
842
|
|
48
|
843 def main():
|
|
|
844 """
|
|
|
845 This is a Galaxy wrapper. It expects to be called by a special purpose tool.xml as:
|
|
49
|
846 <command interpreter="python">rgBaseScriptWrapper.py --script_path "$scriptPath"
|
|
|
847 --tool_name "foo" --interpreter "Rscript"
|
|
48
|
848 </command>
|
|
|
849 """
|
|
|
850 parser = argparse.ArgumentParser()
|
|
|
851 a = parser.add_argument
|
|
49
|
852 a("--script_path", default=None)
|
|
|
853 a("--history_test", default=None)
|
|
|
854 a("--cl_prefix", default=None)
|
|
|
855 a("--sysexe", default=None)
|
|
|
856 a("--packages", default=None)
|
|
48
|
857 a("--tool_name", default=None)
|
|
|
858 a("--input_files", default=[], action="append")
|
|
|
859 a("--output_files", default=[], action="append")
|
|
|
860 a("--user_email", default="Unknown")
|
|
|
861 a("--bad_user", default=None)
|
|
49
|
862 a("--make_Tool", default="runonly")
|
|
48
|
863 a("--help_text", default=None)
|
|
|
864 a("--tool_desc", default=None)
|
|
|
865 a("--tool_version", default=None)
|
|
|
866 a("--citations", default=None)
|
|
49
|
867 a("--command_override", default=None)
|
|
|
868 a("--test_override", default=None)
|
|
48
|
869 a("--additional_parameters", action="append", default=[])
|
|
|
870 a("--edit_additional_parameters", action="store_true", default=False)
|
|
|
871 a("--parampass", default="positional")
|
|
|
872 a("--tfout", default="./tfout")
|
|
|
873 a("--new_tool", default="new_tool")
|
|
49
|
874 a("--galaxy_url", default="http://localhost:8080")
|
|
50
|
875 a("--galaxy_api_key", default="fakekey")
|
|
48
|
876 a("--toolshed_url", default="http://localhost:9009")
|
|
50
|
877 a("--toolshed_api_key", default="fakekey")
|
|
|
878 a("--galaxy_root", default="/galaxy-central")
|
|
48
|
879
|
|
|
880 args = parser.parse_args()
|
|
|
881 assert not args.bad_user, (
|
|
|
882 'UNAUTHORISED: %s is NOT authorized to use this tool until Galaxy admin adds %s to "admin_users" in the Galaxy configuration file'
|
|
|
883 % (args.bad_user, args.bad_user)
|
|
|
884 )
|
|
49
|
885 assert args.tool_name, "## Tool Factory expects a tool name - eg --tool_name=DESeq"
|
|
48
|
886 assert (
|
|
49
|
887 args.sysexe or args.packages
|
|
48
|
888 ), "## Tool Factory wrapper expects an interpreter or an executable package"
|
|
49
|
889 args.input_files = [x.replace('"', "").replace("'", "") for x in args.input_files]
|
|
48
|
890 # remove quotes we need to deal with spaces in CL params
|
|
|
891 for i, x in enumerate(args.additional_parameters):
|
|
49
|
892 args.additional_parameters[i] = args.additional_parameters[i].replace('"', "")
|
|
48
|
893 r = ScriptRunner(args)
|
|
49
|
894 r.writeShedyml()
|
|
|
895 r.makeTool()
|
|
50
|
896 if args.command_override or args.test_override:
|
|
|
897 retcode = r.planemo_test(genoutputs=True) # this fails :( - see PR
|
|
|
898 r.moveRunOutputs()
|
|
|
899 r.makeToolTar()
|
|
|
900 retcode = r.planemo_test(genoutputs=False)
|
|
|
901 r.moveRunOutputs()
|
|
|
902 if args.make_Tool == "gentestinstall":
|
|
|
903 r.planemo_shedload()
|
|
|
904 r.eph_galaxy_load()
|
|
|
905 else:
|
|
49
|
906 retcode = r.run()
|
|
|
907 if retcode:
|
|
|
908 sys.stderr.write(
|
|
|
909 "## Run failed with return code %d. Cannot build yet. Please fix and retry"
|
|
|
910 % retcode
|
|
|
911 )
|
|
|
912 sys.exit(1)
|
|
|
913 else:
|
|
|
914 r.moveRunOutputs()
|
|
50
|
915 r.makeToolTar()
|
|
|
916 if args.make_Tool in ["gentestinstall", "gentest"]:
|
|
|
917 r.planemo_test(genoutputs=False)
|
|
|
918 r.moveRunOutputs()
|
|
|
919 if args.make_Tool == "gentestinstall":
|
|
|
920 r.planemo_shedload()
|
|
|
921 r.eph_galaxy_load()
|
|
48
|
922
|
|
|
923
|
|
|
924 if __name__ == "__main__":
|
|
|
925 main()
|
