Improve markup of links to HARIVAMSA Calcutta edition.

Ref: #49

pwg_ls2/hariv/change_abnormal.py

@@ -0,0 +1,211 @@
+#-*- coding:utf-8 -*-
+"""change_abnormal.py
+   generate change transactions for lines in entries which
+   are considered abnormal.  The
+"""
+from __future__ import print_function
+import sys, re,codecs
+from parseheadline import parseheadline
+
+class Entry(object):
+ Ldict = {}
+ def __init__(self,lines,linenum1,linenum2):
+  # linenum1,2 are int
+  self.metaline = lines[0]
+  self.lend = lines[-1]  # the <LEND> line
+  self.datalines = lines[1:-1]  # the non-meta lines
+  # parse the meta line into a dictionary
+  #self.meta = Hwmeta(self.metaline)
+  self.metad = parseheadline(self.metaline)
+  self.linenum1 = linenum1
+  self.linenum2 = linenum2
+  #L = self.meta.L
+  L = self.metad['L']
+  if L in self.Ldict:
+   print("Entry init error: duplicate L",L,linenum1)
+   exit(1)
+  self.Ldict[L] = self
+  self.lsarr = []
+
+def init_entries(filein):
+ # slurp lines
+ with codecs.open(filein,encoding='utf-8',mode='r') as f:
+  lines = [line.rstrip('\r\n') for line in f]
+ recs=[]  # list of Entry objects
+ inentry = False  
+ idx1 = None
+ idx2 = None
+ for idx,line in enumerate(lines):
+  if inentry:
+   if line.startswith('<LEND>'):
+    idx2 = idx
+    entrylines = lines[idx1:idx2+1]
+    linenum1 = idx1 + 1
+    linenum2 = idx2 + 1
+    entry = Entry(entrylines,linenum1,linenum2)
+    recs.append(entry)
+    # prepare for next entry
+    idx1 = None
+    idx2 = None
+    inentry = False
+   elif line.startswith('<L>'):  # error
+    print('init_entries Error 1. Not expecting <L>')
+    print("line # ",idx+1)
+    print(line.encode('utf-8'))
+    exit(1)
+   else: 
+    # keep looking for <LEND>
+    continue
+  else:
+   # inentry = False. Looking for '<L>'
+   if line.startswith('<L>'):
+    idx1 = idx
+    inentry = True
+   elif line.startswith('<LEND>'): # error
+    print('init_entries Error 2. Not expecting <LEND>')
+    print("line # ",idx+1)
+    print(line.encode('utf-8'))
+    exit(1)
+   else: 
+    # keep looking for <L>
+    continue
+ # when all lines are read, we should have inentry = False
+ if inentry:
+  print('init_entries Error 3. Last entry not closed')
+  print('Open entry starts at line',idx1+1)
+  exit(1)
+
+ print(len(lines),"lines read from",filein)
+ print(len(recs),"entries found")
+ return recs
+
+
+class LSchange(object):
+ def __init__(self,entry,iline,ls):
+  self.entry = entry
+  self.iline = iline
+  self.ls = ls
+
+def find_abnormals_hariv(lspfx,entries):
+ replacements = (('.','[.]'), ('(','\('),  (')','\)'))
+ tmp = lspfx
+ for old,new in replacements:
+  tmp = tmp.replace(old,new)
+ #regexnorm = re.compile(r'^<ls>%s ([0-9]+), ( fg+[.])?</ls>$'%tmp)
+ """
+ regexdata = [
+  [r'<ls>%s.*?</ls>'%tmp,
+   [ (r'<ls>%s ([0-9]+), ([0-9]+[.]?)</ls>'%tmp , '1a'),
+   (r'<ls>%s ([0-9]+), ([0-9]+[.]?) fgg?[.]</ls>' % tmp, '1b'),
+   (r'<ls>%s ([0-9]+), ([0-9]+[.,]?) v[.] l[.]</ls>' % tmp, '1c'),
+   (r'<ls>%s</ls>'%tmp , '1d'),
+   (r'<ls>%s[^<]*ed[.] Bomb[.].*?</ls>'%tmp , '1e'),
+   ]
+  ],
+  [r'<ls n="%s[^"]*">.*?</ls>'%tmp,
+   [(r'<ls n="%s">([0-9]+), ([0-9]+[.]?)</ls>'%tmp , '2a'),
+   (r'<ls n="%s">([0-9]+), ([0-9]+[.]?) fgg?[.]</ls>' % tmp, '2b'),
+   (r'<ls n="%s">([0-9]+), ([0-9]+[.,]?) v[.] l[.]</ls>' % tmp, '2c'),
+   (r'<ls n="%s ([0-9]+),">([0-9]+[.]?)</ls>'%tmp , '2d'),
+   (r'<ls n="%s ([0-9]+),">([0-9]+[.]?) fgg?[.]</ls>' % tmp, '2e'),
+   (r'<ls n="%s ([0-9]+),">([0-9]+[.,]?) v[.] l[.]</ls>' % tmp, '2f'),
+ 
+   ]
+  ]
+ ]
+ """
+ regexdata = [
+  [r'<ls>%s.*?</ls>'%tmp,
+   [ (r'<ls>%s ([0-9]+[.]?)</ls>'%tmp , '1a'),
+   (r'<ls>%s ([0-9]+[.]?) fgg?[.]</ls>' % tmp, '1b'),
+   (r'<ls>%s ([0-9]+[.,]?) v[.] l[.]</ls>' % tmp, '1c'),
+   (r'<ls>%s</ls>'%tmp , '1d'),
+   (r'<ls>%s[^<]*ed[.] Bomb[.].*?</ls>'%tmp , '1e'),
+   ]
+  ],
+  [r'<ls n="%s[^"]*">.*?</ls>'%tmp,
+   [(r'<ls n="%s">([0-9]+[.]?)</ls>'%tmp , '2a'),
+   (r'<ls n="%s">([0-9]+[.]?) fgg?[.]</ls>' % tmp, '2b'),
+   (r'<ls n="%s">([0-9]+[.,]?) v[.] l[.]</ls>' % tmp, '2c'),
+   #(r'<ls n="%s ([0-9]+),">([0-9]+[.]?)</ls>'%tmp , '2d'),
+   #(r'<ls n="%s ([0-9]+),">([0-9]+[.]?) fgg?[.]</ls>' % tmp, '2e'),
+   #(r'<ls n="%s ([0-9]+),">([0-9]+[.,]?) v[.] l[.]</ls>' % tmp, '2f'),
+
+   ]
+  ]
+ ]
+ abnormals = []
+ normals = []
+ for entry in entries:
+  #text = '\n'.join(entry.datalines)
+  for iline,line in enumerate(entry.datalines):
+   abnormal = False
+   for regex1,regexnorms in regexdata:
+    lsarr = re.findall(regex1,line)
+    #abnormal = 
+    for ls in lsarr:
+     normal = False
+     for regex1a,regextype in regexnorms:
+      if re.search(regex1a,ls):
+       #normal = LSinstance(entry,ls,regextype)
+       #normals.append(normal)
+       normal = True
+       break
+     if normal == False:
+      abnormal = True
+      break # for ls in lsarr
+    if abnormal:
+     break # for regex1a
+   #
+   if abnormal:
+    # generate change transaction for this line
+    instance = LSchange(entry,iline,ls)
+    abnormals.append(instance)
+ print(len(abnormals),'abnormal lines found')
+ return abnormals
+
+def normals_summary(normals):
+ d = {} 
+ for lsinstance in normals:
+  t = lsinstance.type
+  if t not in d:
+   d[t] = 0
+  d[t] = d[t] + 1
+ types = sorted(d.keys())
+ tot = 0
+ for t in types:
+  print(d[t],"ls instances of type",t)
+  tot = tot + d[t]
+ print('totals=',tot)
+
+def write_abnormals(fileout,abnormals):
+ with codecs.open(fileout,"w","utf-8") as f:
+  for x in abnormals:
+   entry = x.entry
+   iline = x.iline
+   ls = x.ls # the abnormal ls
+   lnum = entry.linenum1+iline+1
+   metaline = re.sub(r'<k2>.*$','',entry.metaline)
+   line = entry.datalines[iline]
+   outarr = []
+   outarr.append('; --------------------------------')
+   outarr.append('; %s' % metaline)
+   outarr.append('; Abnormal ls: %s' %ls)
+   outarr.append('%s old %s' %(lnum,line))
+   outarr.append(';')
+   outarr.append('%s new %s' %(lnum,line))
+   for out in outarr:
+    f.write(out+'\n')
+ print(len(abnormals),'change transactions',fileout)
+
+if __name__=="__main__":
+ lspfx = sys.argv[1]
+ filein = sys.argv[2] #  xxx.txt (path to digitization of xxx)
+ #filebib = sys.argv[2]  # pwbib_input.txt
+ fileout = sys.argv[3] #
+ entries = init_entries(filein)
+ if lspfx == 'HARIV.':
+  abnormals = find_abnormals_hariv(lspfx,entries)
+  write_abnormals(fileout,abnormals)
+ else:
+  print('Not implemented for lspfx = %s' %lspfx)

pwg_ls2/hariv/change_abnormal.txt

@@ -0,0 +1,156 @@
+; --------------------------------
+; <L>371<pc>1-0030<k1>agnija
+; Abnormal ls: <ls>HARIV. S. 927, Z. 5</ls>
+3232 old <ls>HARIV. S. 927, Z. 5</ls>, v. u. 
+;
+3232 new <ls>HARIV. S. 927, Z. 5</ls>, v. u. 
+; --------------------------------
+; <L>521<pc>1-0039<k1>agra
+; Abnormal ls: <ls>HARIV. S. 927</ls>
+4342 old <ls>HARIV. S. 927</ls>, ult. erscheint {#agra#} im Voc. als ein Beiname <is>Viṣṇu's</is>.
+;
+4342 new <ls>HARIV. S. 927</ls>, ult. erscheint {#agra#} im Voc. als ein Beiname <is>Viṣṇu's</is>.
+; --------------------------------
+; <L>526<pc>1-0040<k1>agraja
+; Abnormal ls: <ls>HARIV. S. 927</ls>
+4386 old <ls>HARIV. S. 927</ls>, ult. 
+;
+4386 new <ls>HARIV. S. 927</ls>, ult. 
+; --------------------------------
+; <L>873<pc>1-0066<k1>aja
+; Abnormal ls: <ls>HARIV. S. 927, Z. 4</ls>
+7646 old <ls>HARIV. S. 927, Z. 4</ls>, v. u. 
+;
+7646 new <ls>HARIV. S. 927, Z. 4</ls>, v. u. 
+; --------------------------------
+; <L>924<pc>1-0069<k1>ajaya
+; Abnormal ls: <ls>HARIV. S. 927, Z. 4</ls>
+8021 old <ls>HARIV. S. 927, Z. 4</ls>, v. u. 
+;
+8021 new <ls>HARIV. S. 927, Z. 4</ls>, v. u. 
+; --------------------------------
+; <L>1016<pc>1-0074<k1>ajEkapAd
+; Abnormal ls: <ls>HARIV. S. 928, Z. 1.</ls>
+8755 old <ls>HARIV. S. 928, Z. 1.</ls> 
+;
+8755 new <ls>HARIV. S. 928, Z. 1.</ls> 
+; --------------------------------
+; <L>5286<pc>1-0384<k1>ambarIza
+; Abnormal ls: <ls n="HARIV.">83, N. 6.</ls>
+44108 old <ls n="HARIV.">83, N. 6.</ls> <is>Māṃdhātar's</is> 
+;
+44108 new <ls n="HARIV.">83, N. 6.</ls> <is>Māṃdhātar's</is> 
+; --------------------------------
+; <L>11578<pc>1-0937<k1>unnetar
+; Abnormal ls: <ls>HARIV. LANGL. II, 297</ls>
+106622 old <ls>13.</ls> <ls>HARIV. LANGL. II, 297</ls>; der gedr. Text <ls>HARIV. 11363</ls> : {#sunetar#} .
+;
+106622 new <ls>13.</ls> <ls>HARIV. LANGL. II, 297</ls>; der gedr. Text <ls>HARIV. 11363</ls> : {#sunetar#} .
+; --------------------------------
+; <L>14182<pc>2-0009<k1>kakudmin
+; Abnormal ls: <ls>HARIV. S. 927, Z. 4.</ls>
+130173 old <ls>HARIV. S. 927, Z. 4.</ls> v. u. 
+;
+130173 new <ls>HARIV. S. 927, Z. 4.</ls> v. u. 
+; --------------------------------
+; <L>14857<pc>2-0063<k1>kapAlin
+; Abnormal ls: <ls>HARIV. LANGL. I, 513.</ls>
+136842 old <ls>VYĀḌI</ls> zu <ls>210.</ls> <ls>HARIV. LANGL. I, 513.</ls> 
+;
+136842 new <ls>VYĀḌI</ls> zu <ls>210.</ls> <ls>HARIV. LANGL. I, 513.</ls> 
+; --------------------------------
+; <L>14891<pc>2-0065<k1>kapila
+; Abnormal ls: <ls n="HARIV.">S. 927, Z. 5 v. u.</ls>
+137119 old <ls>HARIV. 788.</ls> <ls n="HARIV.">2219.</ls> <ls n="HARIV.">7595.</ls> <ls n="HARIV.">11495.</ls> <ls n="HARIV.">12439.</ls> <ls n="HARIV.">S. 927, Z. 5 v. u.</ls> <ls>R. 1, 41, 25.</ls> <ls>RAGH. 3, 50.</ls> <ls>BHĀG. P. 1, 3, 10. 3, 24, 19.</ls> <ls>VP. 378. fg.</ls> ein Sohn <is>Vitatha's</is> 
+;
+137119 new <ls>HARIV. 788.</ls> <ls n="HARIV.">2219.</ls> <ls n="HARIV.">7595.</ls> <ls n="HARIV.">11495.</ls> <ls n="HARIV.">12439.</ls> <ls n="HARIV.">S. 927, Z. 5 v. u.</ls> <ls>R. 1, 41, 25.</ls> <ls>RAGH. 3, 50.</ls> <ls>BHĀG. P. 1, 3, 10. 3, 24, 19.</ls> <ls>VP. 378. fg.</ls> ein Sohn <is>Vitatha's</is> 
+; --------------------------------
+; <L>19351<pc>2-0448<k1>kola
+; Abnormal ls: <ls>HARIV. LANGL. 1, 68</ls>
+181940 old <ls>HARIV. LANGL. 1, 68</ls> und <ls>ŚKDR.</ls> nach <ls>HARIV.</ls>; die gedr. Ausg.: {#kolisarpAH#} st. {#kolAH sarpAH#}, wie 
+;
+181940 new <ls>HARIV. LANGL. 1, 68</ls> und <ls>ŚKDR.</ls> nach <ls>HARIV.</ls>; die gedr. Ausg.: {#kolisarpAH#} st. {#kolAH sarpAH#}, wie 
+; --------------------------------
+; <L>28074<pc>3-0172<k1>jvAlAjihva
+; Abnormal ls: <ls>HARIV. LANGL. I, 513</ls>
+276910 old <ls>VYĀḌI</ls> zu <ls>H. 210</ls> (vgl. <ls>HARIV. LANGL. I, 513</ls>). N. pr. eines <is>Dānava</is> 
+;
+276910 new <ls>VYĀḌI</ls> zu <ls>H. 210</ls> (vgl. <ls>HARIV. LANGL. I, 513</ls>). N. pr. eines <is>Dānava</is> 
+; --------------------------------
+; <L>30856<pc>3-0424<k1>trikakud
+; Abnormal ls: <ls>HARIV. S. 927, Z. 4</ls>
+305856 old <ls n="MBH. 12,">13252.</ls> <ls n="MBH.">13, 6956.</ls> <ls>HARIV. S. 927, Z. 4</ls> v. u. 
+;
+305856 new <ls n="MBH. 12,">13252.</ls> <ls n="MBH.">13, 6956.</ls> <ls>HARIV. S. 927, Z. 4</ls> v. u. 
+; --------------------------------
+; <L>43424<pc>4-0572<k1>parRaka
+; Abnormal ls: <ls>HARIV. LANGL. II, 376</ls>
+432891 old <ls>HARIV. 14165</ls>; vgl. <ls>HARIV. LANGL. II, 376</ls>, wo die Calc. Ausg. {#parRinI#} hat.
+;
+432891 new <ls>HARIV. 14165</ls>; vgl. <ls>HARIV. LANGL. II, 376</ls>, wo die Calc. Ausg. {#parRinI#} hat.
+; --------------------------------
+; <L>45424<pc>4-0731<k1>piSitASin
+; Abnormal ls: <ls>HARIV. LANGL. I, 513.</ls>
+451966 old <ls>VYĀḌI</ls> zu <ls>H. 210</ls>; vgl. <ls>HARIV. LANGL. I, 513.</ls>
+;
+451966 new <ls>VYĀḌI</ls> zu <ls>H. 210</ls>; vgl. <ls>HARIV. LANGL. I, 513.</ls>
+; --------------------------------
+; <L>46726<pc>4-0839<k1>pUrRAyus
+; Abnormal ls: <ls>HARIV. LANGL. II, 481</ls>
+464771 old <ls>MBH. 1, 2554.</ls> <ls>HARIV. LANGL. II, 481</ls> (die Calc. Ausg. <ls n="HARIV.">14156</ls>
+;
+464771 new <ls>MBH. 1, 2554.</ls> <ls>HARIV. LANGL. II, 481</ls> (die Calc. Ausg. <ls n="HARIV.">14156</ls>
+; --------------------------------
+; <L>49502<pc>4-1047<k1>pramardana
+; Abnormal ls: <ls>HARIV. LANGL. I, 513.</ls>
+489797 old <ls>VYĀḌI</ls> zu <ls>H. 210</ls>; vgl. <ls>HARIV. LANGL. I, 513.</ls> eines <is>Vidyādhara</is> 
+;
+489797 new <ls>VYĀḌI</ls> zu <ls>H. 210</ls>; vgl. <ls>HARIV. LANGL. I, 513.</ls> eines <is>Vidyādhara</is> 
+; --------------------------------
+; <L>54819<pc>5-0277<k1>BAsvant
+; Abnormal ls: <ls n="HARIV."> 5185.</ls>
+541191 old <ls n="HARIV.">1331</ls> <ls n="HARIV."> 5185.</ls> <ls n="HARIV.">10995.</ls> <ls>R. 1, 44, 30. 2, 83, 6.</ls> <ls>R. GORR. 2, 100, 16.</ls> {#BUzaRAni#} 
+;
+541191 new <ls n="HARIV.">1331</ls> <ls n="HARIV."> 5185.</ls> <ls n="HARIV.">10995.</ls> <ls>R. 1, 44, 30. 2, 83, 6.</ls> <ls>R. GORR. 2, 100, 16.</ls> {#BUzaRAni#} 
+; --------------------------------
+; <L>54948<pc>5-0294<k1>BImaka
+; Abnormal ls: <ls>HARIV. LANGL. I, 513.</ls>
+543102 old <ls>HARIV. 9561.</ls> <ls>HARIV. LANGL. I, 513.</ls> {#BIzaka#} 
+;
+543102 new <ls>HARIV. 9561.</ls> <ls>HARIV. LANGL. I, 513.</ls> {#BIzaka#} 
+; --------------------------------
+; <L>58325<pc>5-0619<k1>mahAkAla
+; Abnormal ls: <ls>HARIV. LANGL. I,512.</ls>
+579488 old <ls n="HARIV.">12502.</ls> <ls>HARIV. LANGL. I,512.</ls> <ls>KATHĀS. 50,147.</ls> <ls>PAÑCAR. 1,15,7.</ls> <ls>Verz. d. Oxf. H. 45,a,7.</ls> {#°gaRotpatti#} 
+;
+579488 new <ls n="HARIV.">12502.</ls> <ls>HARIV. LANGL. I,512.</ls> <ls>KATHĀS. 50,147.</ls> <ls>PAÑCAR. 1,15,7.</ls> <ls>Verz. d. Oxf. H. 45,a,7.</ls> {#°gaRotpatti#} 
+; --------------------------------
+; <L>100618<pc>7-0281<k1>SUra
+; Abnormal ls: <ls>HARIV. lith. Ausg. 33, 56.</ls>
+947904 old <ls>HARIV. lith. Ausg. 33, 56.</ls> {#SUrABIrAH#} ({#SUdrA°?#}) 
+;
+947904 new <ls>HARIV. lith. Ausg. 33, 56.</ls> {#SUrABIrAH#} ({#SUdrA°?#}) 
+; --------------------------------
+; <L>100642<pc>7-0282<k1>SUravIra
+; Abnormal ls: <ls>HARIV. lith. Ausg. 33, 56.</ls>
+948079 old <ls>HARIV. lith. Ausg. 33, 56.</ls>
+;
+948079 new <ls>HARIV. lith. Ausg. 33, 56.</ls>
+; --------------------------------
+; <L>108891<pc>7-0963<k1>sAvarRa
+; Abnormal ls: <ls>HARIV. lith Ausg. 7, 43.</ls>
+1030782 old <ls>HARIV. lith Ausg. 7, 43.</ls> 
+;
+1030782 new <ls>HARIV. lith Ausg. 7, 43.</ls> 
+; --------------------------------
+; <L>111836<pc>7-1143<k1>susaMyukta
+; Abnormal ls: <ls>HARIV. lith. Ausg. 21, 37.</ls>
+1055950 old <ls>HARIV. lith. Ausg. 21, 37.</ls> so v. a. {%in richtigem Zahlenverhältniss zu einander stehend%} 
+;
+1055950 new <ls>HARIV. lith. Ausg. 21, 37.</ls> so v. a. {%in richtigem Zahlenverhältniss zu einander stehend%} 
+; --------------------------------
+; <L>116085<pc>7-1519<k1>har
+; Abnormal ls: <ls>HARIV. 2, 3.</ls>
+1102535 old <ls>NĪLAK.</ls> zu <ls>MBH. 3, 10247</ls> und zu <ls>HARIV. 2, 3.</ls> 
+;
+1102535 new <ls>NĪLAK.</ls> zu <ls>MBH. 3, 10247</ls> und zu <ls>HARIV. 2, 3.</ls>