analyze: Customizeable triadfreq output

Add commandline args to triadfreq.

4 files changed, 50 insertions, 46 deletions

diff --git a/lulua/plot.py b/lulua/plot.py
index 13077e0..453c6eb 100644
--- a/lulua/plot.py
+++ b/lulua/plot.py
@@ -18,7 +18,7 @@
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 # THE SOFTWARE.
 
-import sys, argparse, json, unicodedata, pickle, logging
+import sys, argparse, json, unicodedata, pickle, logging, math
 from operator import itemgetter
 from bokeh.plotting import figure
 from bokeh.models import ColumnDataSource, LinearAxis, Range1d
@@ -26,7 +26,7 @@ from bokeh.embed import json_item
 
 from .layout import *
 from .keyboard import defaultKeyboards
-from .util import limit
+from .util import limit, displayText
 from .writer import Writer
 from .carpalx import Carpalx, model01
 
@@ -104,6 +104,17 @@ def letterfreq (args):
     return 0
 
 def triadfreq (args):
+    """ Dump triad frequency stats to stdout """
+    sorter = dict (
+        weight=lambda x: x['weight'],
+        effort=lambda x: x['effort'].effort,
+        # increase impact of extremely “bad” triads using math.pow
+        combined=lambda x: (x['weight']/weightSum)*math.pow (x['effort'].effort, 2)
+        )
+    def noLimit (l, n):
+        yield from l
+    limiter = limit if args.limit > 0 else noLimit
+
     stats = pickle.load (sys.stdin.buffer)
 
     # XXX: add layout to stats?
@@ -126,32 +137,19 @@ def triadfreq (args):
 
     # triads that contribute to x% of the weight
     topTriads = list ()
-    topTriadsCutoff = 0.50
     topTriadsWeight = 0
     for data in sorted (binned.values (), key=lambda x: x['weight'], reverse=True):
-        if topTriadsWeight < weightSum*topTriadsCutoff:
+        if topTriadsWeight < weightSum*args.cutoff:
             topTriads.append (data)
             topTriadsWeight += data['weight']
 
-    # get top triads (by weight)
-    print ('by weight')
-    for data in limit (sorted (binned.values (), key=lambda x: x['weight'], reverse=True), 20):
-        print (data['textTriad'], data['weight'], data['effort'].effort)
-
-    logging.info (f'{len (topTriads)}/{len (stats["triads"].triads)} triads contribute to {topTriadsCutoff*100}% of the typing')
-
-    print ('by effort')
-    # only base layer
-    includeBaseLayer = iter (topTriads)
-    sortByEffort = sorted (includeBaseLayer, key=lambda x: x['effort'].effort, reverse=True)
-    for data in limit (sortByEffort, 20):
-        print (data['textTriad'], data['weight'], data['effort'].effort)
-
-    print ('by effort and weight')
-    includeBaseLayer = iter (topTriads)
-    sortByEffortWeight = sorted (includeBaseLayer, key=lambda x: (x['weight']/weightSum)*x['effort'].effort, reverse=True)
-    for data in limit (sortByEffortWeight, 20):
-        print (data['textTriad'], data['weight'], data['effort'].effort)
+    logging.info (f'{len (topTriads)}/{len (stats["triads"].triads)} triads '
+            'contribute to {args.cutoff*100}% of the typing')
+
+    # final output
+    sortByEffort = sorted (iter (topTriads), key=sorter[args.sort], reverse=args.reverse)
+    for data in limiter (sortByEffort, args.limit):
+        print (''.join (map (displayText, data['textTriad'])), data['weight'], data['effort'].effort)
 
     return 0
 
diff --git a/lulua/render.py b/lulua/render.py
index cbe553b..16dc543 100644
--- a/lulua/render.py
+++ b/lulua/render.py
@@ -18,7 +18,7 @@
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 # THE SOFTWARE.
 
-import argparse, sys, unicodedata, logging
+import argparse, sys, logging
 from collections import namedtuple, defaultdict
 from operator import attrgetter
 from datetime import datetime
@@ -30,7 +30,7 @@ import yaml
 from .layout import LITTLE, RING, MIDDLE, INDEX, THUMB, GenericLayout, defaultLayouts
 from .writer import Writer
 from .keyboard import defaultKeyboards
-from .util import first
+from .util import first, displayText
 
 RendererSettings = namedtuple ('RendererSetting', ['buttonMargin', 'middleGap', 'buttonWidth', 'rounded', 'shadowOffset'])
 
@@ -99,25 +99,6 @@ class Renderer:
         return btn.width * self.settings.buttonWidth
 
     def _addButton (self, btn):
-        def toDisplayText (text):
-            if text is None:
-                return text
-            if len (text) == 1 and unicodedata.combining (text) != 0:
-                # add circle if combining
-                return '\u25cc' + text
-            invMap = {
-                '\t': '⭾',
-                '\n': '↳',
-                ' ': '\u2423',
-                '\u200e': '[LRM]', # left to right mark
-                '\u061c': '[ALM]', # arabic letter mark
-                '\u202c': '[PDF]', # pop directional formatting
-                "\u2066": '[LRI]', # left-to-right isolate (lri)
-                "\u2067": '[RLI]', # right-to-left isolate (rli)
-                "\u2069": '[PDI]', # pop directional isolate (pdi)
-                }
-            return invMap.get (text, text)
-
         xoff, yoff = self.cursor
         settings = self.settings
         width = self.buttonWidth (btn)
@@ -137,7 +118,7 @@ class Renderer:
             buttonText = [layerToArrow[i]]
             gclass.append ('modifier')
         else:
-            buttonText = list (map (toDisplayText, self.layout.getButtonText (btn)))
+            buttonText = list (map (displayText, self.layout.getButtonText (btn)))
 
         # background rect
         if any (buttonText):
diff --git a/lulua/stats.py b/lulua/stats.py
index fc2b202..6689616 100644
--- a/lulua/stats.py
+++ b/lulua/stats.py
@@ -234,6 +234,10 @@ def main ():
     sp = subparsers.add_parser('letterfreq')
     sp.set_defaults (func=letterfreq)
     sp = subparsers.add_parser('triadfreq')
+    sp.add_argument('-c', '--cutoff', type=float, default=0.5, help='Only include the top x% of all triads')
+    sp.add_argument('-r', '--reverse', action='store_true', help='Reverse sorting order')
+    sp.add_argument('-s', '--sort', choices={'weight', 'effort', 'combined'}, default='weight', help='Sorter')
+    sp.add_argument('-n', '--limit', type=int, default=0, help='Sorter')
     sp.set_defaults (func=triadfreq)
 
     logging.basicConfig (level=logging.INFO)
diff --git a/lulua/util.py b/lulua/util.py
index dd35c23..1e2156d 100644
--- a/lulua/util.py
+++ b/lulua/util.py
@@ -22,11 +22,12 @@
 Misc utilities
 """
 
-import os, yaml, pkg_resources
+import os, yaml, pkg_resources, unicodedata
 
 first = lambda x: next (iter (x))
 
 def limit (l, n):
+    """ Limit the number of items drawn from iterable l to n. """
     it = iter (l)
     for i in range (n):
         yield next (it)
@@ -65,3 +66,23 @@ class YamlLoader:
         for res in pkg_resources.resource_listdir (__package__, self.defaultDir):
             yield self.__getitem__ (res, onlyRes=True)
 
+def displayText (text):
+    """ Convert text into a string that is always renderable without combining,
+    control or invisible characters """
+    if text is None:
+        return text
+    if len (text) == 1 and unicodedata.combining (text) != 0:
+        # add circle if combining
+        return '\u25cc' + text
+    invMap = {
+        '\t': '⭾',
+        '\n': '↳',
+        ' ': '\u2423',
+        '\u200e': '[LRM]', # left to right mark
+        '\u061c': '[ALM]', # arabic letter mark
+        '\u202c': '[PDF]', # pop directional formatting
+        "\u2066": '[LRI]', # left-to-right isolate (lri)
+        "\u2067": '[RLI]', # right-to-left isolate (rli)
+        "\u2069": '[PDI]', # pop directional isolate (pdi)
+        }
+    return invMap.get (text, text)