Merge branch 'master' of mit.cs.uchicago.edu:ethanmertz/CS-123-Final

README.txt

@@ -13,11 +13,9 @@ Python subdirectory: Contains the files used for our final Python analysis
 	-topvariables.py: Computes the least unique variable names.
 	-topvariables_intersection.py: Checks which variable files have the largest mean number of common variables with other files.
 
-histogram.py: Used to construct the histograms for the report.
-
 index.py: Creates the index file of tuples for the MapReduce jobs. Methodology detailed in the report.
 
-plot_graph.py: Used to create the distance plot in the report.
+plot_graph.py: Used to create the distance plot and histograms for the report.
 
 runsim.sh: Used to send specific files for subset testing when there are other files in the directory so sending the entire directory is inefficient.
 

Results/similarfuncs/pyfile1026.txt

+reEscape|s|return re.sub(r"([\(\)\[\]\.\*\+\?\{\}\^\$])", r"\\\1", s)

Results/similarfuncs/pyfile1596.txt

+testOutfile2|self|self._TestOutFile(GTEST_OUTPUT_2_TEST, EXPECTED_XML_2)

Results/similarfuncs/pyfile2085.txt

+reEscape|s|return re.sub(r"([\(\)\[\]\.\*\+\?\{\}\^\$])", r"\\\1", s)

Results/similarfuncs/pyfile4352.txt

+_is_builtin_type|self, type|return type in [KW_INT, KW_CHAR, KW_BOOLEAN, KW_VOID]

Results/similarfuncs/pyfile4399.txt

+rot|v|return [-v[3],v[0],v[1],v[2]]

Results/similarfuncs/pyfile4402.txt

+rot|v|return [-v[3],v[0],v[1],v[2]]

Results/similarfuncs/pyfile4405.txt

+rot|v|return [-v[3],v[0],v[1],v[2]]

Results/similarfuncs/pyfile4446.txt

+cslogo||print '''
+ | |||||||||   `--------'     |          O ..
+   `\_,-------, ___@ PlayWeb   ______|
+     / XXXXXX /`|     /
+    / XXXXXX /  `\   /
+   / XXXXXX /\______(
+  / XXXXXX /        
+ / XXXXXX /
+(________(    For example:
+ `------'         cutescan.py  yaseng.me 

Results/similarfuncs/pyfile585.txt

+testShuffleChangesTestOrder|self|self.assert_(SHUFFLED_ALL_TESTS != ALL_TESTS, SHUFFLED_ALL_TESTS)
+self.assert_(SHUFFLED_ACTIVE_TESTS != ACTIVE_TESTS, SHUFFLED_ACTIVE_TESTS)
+self.assert_(SHUFFLED_FILTERED_TESTS != FILTERED_TESTS,
+             SHUFFLED_FILTERED_TESTS)
+self.assert_(SHUFFLED_SHARDED_TESTS != SHARDED_TESTS,
+             SHUFFLED_SHARDED_TESTS)

Results/similarfuncs/pyfile607.txt

+testOutfile2|self|self._TestOutFile(GTEST_OUTPUT_2_TEST, EXPECTED_XML_2)

Results/topfunctionfiles.txt

+"pyfunctions/pyfile4868.txt"	0.45227296555961644
+"pyfunctions/pyfile4867.txt"	0.4522679227959639
+"pyfunctions/pyfile4832.txt"	0.4522656746303916
+"pyfunctions/pyfile4829.txt"	0.45226564475591263
+"pyfunctions/pyfile4824.txt"	0.45226410122512023
+"pyfunctions/pyfile4863.txt"	0.4522573011096878
+"pyfunctions/pyfile4838.txt"	0.4522560748014902
+"pyfunctions/pyfile4834.txt"	0.4522548385799899
+"pyfunctions/pyfile4835.txt"	0.45225150914007517
+"pyfunctions/pyfile4861.txt"	0.45225105302079377
+"pyfunctions/pyfile4860.txt"	0.4522489289211038
+"pyfunctions/pyfile4841.txt"	0.4522444888611511
+"pyfunctions/pyfile4842.txt"	0.45224308549564535
+"pyfunctions/pyfile4843.txt"	0.45224184093524655
+"pyfunctions/pyfile4859.txt"	0.4522410155894634
+"pyfunctions/pyfile4844.txt"	0.4522400971609809
+"pyfunctions/pyfile4858.txt"	0.45223796317899556
+"pyfunctions/pyfile4847.txt"	0.45223450113225316
+"pyfunctions/pyfile4857.txt"	0.45223323404196697
+"pyfunctions/pyfile4856.txt"	0.4522317621550111
+"pyfunctions/pyfile4848.txt"	0.45222745662516817
+"pyfunctions/pyfile4906.txt"	0.4522019827167537
+"pyfunctions/pyfile4908.txt"	0.45218356662435466
+"pyfunctions/pyfile4911.txt"	0.4521704245413058
+"pyfunctions/pyfile4913.txt"	0.4521553039190678
+"pyfunctions/pyfile4914.txt"	0.4521493960651507
+"pyfunctions/pyfile4915.txt"	0.45214908480809884
+"pyfunctions/pyfile4921.txt"	0.4521346495527042
+"pyfunctions/pyfile4920.txt"	0.4521314614444177
+"pyfunctions/pyfile4923.txt"	0.45212348730652485
+"pyfunctions/pyfile4925.txt"	0.4521197695543568
+"pyfunctions/pyfile4926.txt"	0.45211943861521137
+"pyfunctions/pyfile4930.txt"	0.45211726537407737
+"pyfunctions/pyfile4932.txt"	0.4521118250319259
+"pyfunctions/pyfile4934.txt"	0.4520934752908221
+"pyfunctions/pyfile4941.txt"	0.45205912412055277
+"pyfunctions/pyfile4944.txt"	0.45205748240145965
+"pyfunctions/pyfile4948.txt"	0.45204049596370427
+"pyfunctions/pyfile4953.txt"	0.45201664960202226
+"pyfunctions/pyfile4962.txt"	0.45199147325765543
+"pyfunctions/pyfile4963.txt"	0.4519804335074533
+"pyfunctions/pyfile4967.txt"	0.45196019799633486
+"pyfunctions/pyfile4876.txt"	0.45150183281781125
+"pyfunctions/pyfile4864.txt"	0.45149932970866036
+"pyfunctions/pyfile4754.txt"	0.4514613751084486
+"pyfunctions/pyfile4839.txt"	0.4514448992923634
+"pyfunctions/pyfile4761.txt"	0.45144126388551403
+"pyfunctions/pyfile4782.txt"	0.45138747818913655
+"pyfunctions/pyfile4833.txt"	0.4512999358397714
+"pyfunctions/pyfile4929.txt"	0.4512406459859192
+"pyfunctions/pyfile4965.txt"	0.4511234283827964
+"pyfunctions/pyfile2307.txt"	0.45079337386427276
+"pyfunctions/pyfile4412.txt"	0.45067075560742664
+"pyfunctions/pyfile4319.txt"	0.45028930109106696
+"pyfunctions/pyfile870.txt"	0.4502163493624254
+"pyfunctions/pyfile4088.txt"	0.4498077713800801
+"pyfunctions/pyfile4772.txt"	0.4497156135689474
+"pyfunctions/pyfile4918.txt"	0.44942474461139764
+"pyfunctions/pyfile1615.txt"	0.44924669150427277
+"pyfunctions/pyfile1859.txt"	0.44923285970436144
+"pyfunctions/pyfile4549.txt"	0.4491588985824687
+"pyfunctions/pyfile4527.txt"	0.4491249089186674
+"pyfunctions/pyfile626.txt"	0.44844625099652824
+"pyfunctions/pyfile624.txt"	0.4450547594906591
+"pyfunctions/pyfile4415.txt"	0.44456213486322615
+"pyfunctions/pyfile1613.txt"	0.44389569788523214
+"pyfunctions/pyfile1374.txt"	0.4431732735192146
+"pyfunctions/pyfile385.txt"	0.4428557102193677
+"pyfunctions/pyfile533.txt"	0.4394317874062247
+"pyfunctions/pyfile4980.txt"	0.4390631651186325
+"pyfunctions/pyfile1522.txt"	0.43896426638240316
+"pyfunctions/pyfile4154.txt"	0.43851491087482514
+"pyfunctions/pyfile4481.txt"	0.43621991369328056
+"pyfunctions/pyfile4059.txt"	0.4346032631231313
+"pyfunctions/pyfile4058.txt"	0.4344369954172888
+"pyfunctions/pyfile2780.txt"	0.4341901711340027
+"pyfunctions/pyfile3211.txt"	0.4297320963043031
+"pyfunctions/pyfile4346.txt"	0.4268909575959104
+"pyfunctions/pyfile4054.txt"	0.4250602972604692
+"pyfunctions/pyfile4064.txt"	0.4250180057833518
+"pyfunctions/pyfile4165.txt"	0.42498476989635076
+"pyfunctions/pyfile4214.txt"	0.4249106096576094
+"pyfunctions/pyfile604.txt"	0.42434023758717926
+"pyfunctions/pyfile1593.txt"	0.42393750627413884
+"pyfunctions/pyfile565.txt"	0.42150106423370126
+"pyfunctions/pyfile4396.txt"	0.4213796008618865
+"pyfunctions/pyfile1554.txt"	0.4213025729661326
+"pyfunctions/pyfile4408.txt"	0.4161380800000597
+"pyfunctions/pyfile4325.txt"	0.4152889804022673
+"pyfunctions/pyfile1574.txt"	0.40956063055991926
+"pyfunctions/pyfile4446.txt"	0.40942442841417825
+"pyfunctions/pyfile585.txt"	0.4083175165091745
+"pyfunctions/pyfile4352.txt"	0.4047567157733136
+"pyfunctions/pyfile1026.txt"	0.4010799066543849
+"pyfunctions/pyfile2085.txt"	0.3999315016261221
+"pyfunctions/pyfile1596.txt"	0.3851350228321844
+"pyfunctions/pyfile4399.txt"	0.3840959647592031
+"pyfunctions/pyfile4402.txt"	0.38409378514112485
+"pyfunctions/pyfile4405.txt"	0.3840740854745839
+"pyfunctions/pyfile607.txt"	0.3838987731690707

find_top_python_variables.py

-import re
-
-d = {}
-with open("plot_graph.py") as p:
-	for line in p:
-		var = re.findall("[\s]*([\w, ]+)[\s]*=[^=]", line)
-		if var:
-			d[var[0].strip()] = d.get(var[0].strip(), 0) + 1	
-l = []
-for item in d:
-	l.append((d[item], item))
-nl = sorted(l)
-
-for i in range(100):
-	print(nl[i])
-

histograms.py

-import matplotlib.pyplot as plt
-import numpy as np
-import plotly.plotly as py  # tools to communicate with Plotly's server
-import plotly
-plotly.tools.set_credentials_file(username='ravibakhai', api_key='IBBKlxSe6zkg0osw8RFz')
-'''
-Takes a list of integers and outputs a histogram
-'''
-
-numpy_hist = plt.figure()
-
-plt.hist([1, 2, 1], bins=[0, 1, 2, 3])
-
-plot_url = py.plot_mpl(numpy_hist, filename='numpy-bins')
\ No newline at end of file