# HG changeset patch
# User Martin C. Frith
# Date 1510116368 -32400
# Node ID 21fbf377641f0ca1f055bea6c9bd4889e7c4dfca
# Parent  8d52cef22afab61d9b1168e72cd0e79720717e57
Refactor last-dotplot

diff -r 8d52cef22afa -r 21fbf377641f scripts/last-dotplot
--- a/scripts/last-dotplot	Wed Nov 08 12:17:50 2017 +0900
+++ b/scripts/last-dotplot	Wed Nov 08 13:46:08 2017 +0900
@@ -11,6 +11,7 @@
 
 import gzip
 from fnmatch import fnmatchcase
+from operator import itemgetter
 import subprocess
 import itertools, optparse, os, re, sys
 
@@ -213,22 +214,22 @@
     '''Return miscellaneous information about the sequences.'''
     if sortOpt == 1:
         seqNames.sort(key=natural_sort_key)
-    seqSizes = [seqLimits[i][1] - seqLimits[i][0] for i in seqNames]
+    rangeSizes = [seqLimits[i][1] - seqLimits[i][0] for i in seqNames]
     for i in seqNames:
         r = seqLimits[i]
         out = i, str(r[0]), str(r[1])
         warn("\t".join(out))
     warn("")
     if sortOpt == 2:
-        seqRecords = sorted(zip(seqSizes, seqNames), reverse=True)
-        seqSizes = [i[0] for i in seqRecords]
+        seqRecords = sorted(zip(rangeSizes, seqNames), reverse=True)
+        rangeSizes = [i[0] for i in seqRecords]
         seqNames = [i[1] for i in seqRecords]
     seqRanges = [(i, seqLimits[i][0], seqLimits[i][1]) for i in seqNames]
     labelData = list(rangeLabels(seqRanges, labelOpt,
                                  font, fontsize, image_mode, textRot))
     margin = max(i[2] for i in labelData)
     # xxx the margin may be too big, because some labels may get omitted
-    return seqNames, seqSizes, labelData, margin
+    return seqRanges, rangeSizes, labelData, margin
 
 def div_ceil(x, y):
     '''Return x / y rounded up.'''
@@ -488,12 +489,16 @@
         draw.text(position, i[3], font=font, fill=opts.text_color)
     return im
 
-def rangesPerSeq(seqNames, seqLimits, rangePixBegs, bpPerPix):
-    for i, j in zip(seqNames, rangePixBegs):
-        beg, end = seqLimits[i]
-        origin = bpPerPix * j - beg
-        r = beg, end, origin
-        yield i, [r]
+def rangesWithOrigins(sortedRanges, rangePixBegs, bpPerPix):
+    for i, j in zip(sortedRanges, rangePixBegs):
+        seqName, rangeBeg, rangeEnd = i
+        origin = bpPerPix * j - rangeBeg
+        yield seqName, (rangeBeg, rangeEnd, origin)
+
+def rangesPerSeq(sortedRanges, rangePixBegs, bpPerPix):
+    a = rangesWithOrigins(sortedRanges, rangePixBegs, bpPerPix)
+    for k, v in itertools.groupby(a, itemgetter(0)):
+        yield k, [i[1] for i in v]
 
 def getFont(opts):
     if opts.fontfile:
@@ -533,12 +538,12 @@
     textRot1 = "vertical".startswith(opts.rot1)
     i1 = getSeqInfo(opts.sort1, seqNames1, seqLimits1,
                     font, opts.fontsize, image_mode, opts.labels1, textRot1)
-    seqNames1, rangeSizes1, labelData1, tMargin = i1
+    sortedRanges1, rangeSizes1, labelData1, tMargin = i1
 
     textRot2 = "horizontal".startswith(opts.rot2)
     i2 = getSeqInfo(opts.sort2, seqNames2, seqLimits2,
                     font, opts.fontsize, image_mode, opts.labels2, textRot2)
-    seqNames2, rangeSizes2, labelData2, lMargin = i2
+    sortedRanges2, rangeSizes2, labelData2, lMargin = i2
 
     maxPixels1 = opts.width  - lMargin
     maxPixels2 = opts.height - tMargin
@@ -549,13 +554,11 @@
 
     p1 = pixelData(rangeSizes1, bpPerPix, opts.border_pixels, lMargin)
     rangePixBegs1, rangePixLens1, width = p1
-    rangeDict1 = dict(rangesPerSeq(seqNames1, seqLimits1, rangePixBegs1,
-                                   bpPerPix))
+    rangeDict1 = dict(rangesPerSeq(sortedRanges1, rangePixBegs1, bpPerPix))
 
     p2 = pixelData(rangeSizes2, bpPerPix, opts.border_pixels, tMargin)
     rangePixBegs2, rangePixLens2, height = p2
-    rangeDict2 = dict(rangesPerSeq(seqNames2, seqLimits2, rangePixBegs2,
-                                   bpPerPix))
+    rangeDict2 = dict(rangesPerSeq(sortedRanges2, rangePixBegs2, bpPerPix))
 
     warn("width:  " + str(width))
     warn("height: " + str(height))