wip

analyze_batch.py

@@ -7,34 +7,33 @@ from math import ceil
 
 import matplotlib.pyplot as plt
 import numpy as np
-from vaderSentiment.vaderSentiment import SentimentIntensityAnalyzer
 
-from common import calc_intervals, imprt, FigSaver
+from common import calc_intervals, imprt, printnoln, rprint, DAYS_NEW_USER, IMAGE_MAGICK
 from loader import load, dmt, cms
 
-printnoln = lambda text: print(text, end='', flush=True)
-rprint = lambda text: print('\r' + text)
-
-DAYS_NEW_USER = 7
-OLD_USER_YEAR = 3
 OLD_USER_PERCENTILE = 0.95
 
-analyser = SentimentIntensityAnalyzer()
-figsaver = FigSaver()
 colors = ['red', 'green', 'blue', 'orange', 'deeppink']
 
 
-def main(folder):
+def main(folder, intervl):
     users, posts, firstcontrib, sumcontrib = load(folder)
 
-    intervals = calc_intervals(posts)
+    intervals = calc_intervals(posts, intervl)
     cachedsentiments = imprt(folder + "/output/sentiments.py").answers
 
-    outfolder = folder + "/output/batch/"
-    os.system("mkdir -p " + outfolder)
+    outputdir = folder + "/output/batch/"
+    os.system("mkdir -p " + outputdir)
 
     postcounts = range(1, 5 + 1)
+
+    magickpost = {i: IMAGE_MAGICK for i in postcounts}
+    magickold = IMAGE_MAGICK
+    magickglobal = IMAGE_MAGICK
+
     for (option_date_from, option_date_to) in intervals:
+        magickdate = IMAGE_MAGICK
+
         # get questions for option_date_from <= creation date < option_date_to
         newposts = dmt(posts).filter(lambda p: option_date_from <= p['CreationDate'] < option_date_to, "filter posts by dates").getresults()
         if len(newposts) == 0:
@@ -51,8 +50,8 @@ def main(folder):
         gpos = []
         gcom = []
 
-        goutfilenamenewusers = outfolder + "batch_newusers_" + folder.split("/")[-1] + "_" + option_date_from.strftime("%d-%m-%Y") + "_" + option_date_to.strftime("%d-%m-%Y")
-        goutfilenameoldusers = outfolder + "batch_oldusers_" + folder.split("/")[-1] + "_" + option_date_from.strftime("%d-%m-%Y") + "_" + option_date_to.strftime("%d-%m-%Y")
+        goutfilenamenewusers = outputdir + "batch_newusers_" + folder.split("/")[-1] + "_" + option_date_from.strftime("%d-%m-%Y") + "_" + option_date_to.strftime("%d-%m-%Y")
+        goutfilenameoldusers = outputdir + "batch_oldusers_" + folder.split("/")[-1] + "_" + option_date_from.strftime("%d-%m-%Y") + "_" + option_date_to.strftime("%d-%m-%Y")
 
         for option_posts in postcounts:
             # print(option_date_from.strftime("%d-%m-%Y") + " to " + option_date_to.strftime("%d-%m-%Y") + " - #posts: " + str(option_posts))
@@ -122,6 +121,9 @@ def main(folder):
             fig.savefig(outfilename + ".png", bbox_inches='tight')
             plt.close(fig)
             rprint("computing toxic levels: post " + str(len(filteredposts)) + "/" + str(len(filteredposts)) + " ... plotting ... saving ... took " + str(cms() - start) + "ms")
+            magickpost[option_posts] += " " + outfilename + ".png"
+            magickdate += " " + outfilename + ".png"
+        os.system(magickdate + " " + goutfilenamenewusers + ".pdf")
 
         # global
         start = cms()
@@ -146,6 +148,7 @@ def main(folder):
         gfig.savefig(goutfilenamenewusers + ".png", bbox_inches='tight')
         plt.close(gfig)
         rprint("global plot post ... plotting ... saving ... took " + str(cms() - start) + "ms")
+        magickglobal += " " + goutfilenamenewusers + ".png"
 
         # for old users ---------------------------------------------------------------------------------
         start = cms()
@@ -192,18 +195,16 @@ def main(folder):
 
         # plt.show()
         fig.suptitle("Sentiment of answers to posts by most posting users (95%tile)\nPosts created between " + option_date_from.strftime("%d-%m-%Y") + " to " + option_date_to.strftime("%d-%m-%Y"))
-        # figsaver.save(fig, goutfilenameoldusers + ".png", bbox_inches='tight')
         printnoln("\rcomputing toxic levels: post " + str(len(filteredposts)) + "/" + str(len(filteredposts)) + " ... plotting ... saving ...")
         fig.savefig(goutfilenameoldusers + ".png", bbox_inches='tight')
         plt.close(fig)
         rprint("computing toxic levels: post " + str(len(filteredposts)) + "/" + str(len(filteredposts)) + " ... plotting ... saving ... took " + str(cms() - start) + "ms")
+        magickold += " " + goutfilenameoldusers + ".png"
 
-        figsaver.join()
-    figsaver.join()
-
-
-def computeToxLevel(text):
-    return analyser.polarity_scores(text)
+    os.system(magickglobal + " batch_newusers.pdf")
+    os.system(magickold + " batch_oldusers.pdf")
+    for (i, cmd) in magickpost.items():
+        os.system(cmd + " " + "batch_newusers_" + i + ".pdf")
 
 
 def dumptoxlevels(lvls, filename):
@@ -222,5 +223,20 @@ if __name__ == "__main__":
     if not os.path.isdir(folder):
         print(folder + " is not a folder")
         sys.exit(1)
+    interval = 3
+    if len(sys.argv) >= 3:
+        if sys.argv[2].startswith("-i"):
+            interval = sys.argv[2][2:]
+            try:
+                interval = int(interval)
+            except ValueError:
+                print("-i: int required")
+                sys.exit(1)
+            if interval < 1 or interval > 12:
+                print("-i: only 1 - 12")
+                sys.exit(1)
+        else:
+            print("unknown parameter: " + sys.argv[2])
+            sys.exit(1)
 
-    main(folder)
+    main(folder, interval)