#! /usr/bin/gawk -f
# Last edited on 2022-11-17 13:05:18 by stolfi

# Reads from {stdin} one or more files produced by {}. }
# Splits each entry into a separate file "{odir}/{key}.bib" where 
# the {key} is built from the author names and year.


# Must be executed with "-f ~/lib/read_table.gawk".
# Expects a file called "key-map.tbl" that maps synthetic
# bib keys to the keys used in {stolfi,others}.bib.
# See comments in {read_table} for the file format.

BEGIN {
  abort = -1;
  if (odir == "") { arg_error("must define {odir}"); }
  ifile = ""; # Previous input file name.
  alf = "abcdefghijklmnopqrstuvxyz"; # Used to convert numbers to "aa", "ab", etc.
  split("", kc); # The number of times bare key {k} occurred is {kc[k]}.
  
  # Read the key mapping table {key_map}: 
  tbl_fname = "key-map.tbl"
  printf "reading %s ...\n", tbl_fname > "/dev/stderr"
  split("", key_map);
  read_table(tbl_fname, 0, key_map, 1);
  n_mapped = 0; # Counts keys that were actually mapped.
  
  # Initialize the entry data:
  clear_entry();
}

(abort >= 0) { exit(abort); }

(FILENAME != ifile) { 
  # Report input file name change:
  printf "== %s ==\n", FILENAME > "/dev/stderr"
  ifile = FILENAME;
}

# General cleanup:
//{ 
  # Remove boldface markup:
  gsub(/<[\/]?b>/, "", $0);
  # Repace HTML #-codes:
  gsub(/[&]lt;/, "<", $0);
  gsub(/[&]gt;/, ">", $0);
  gsub(/[&]qt;/, "'", $0);
  gsub(/[&][#]39;/, "'", $0);
  gsub(/[&]amp;/, "&", $0);
  # Remove final </br>, </p>, </li>:
  gsub(/<[\/]p> *$/, "", $0);
  gsub(/<br[\/]> *$/, "", $0);
}

/^ *<li>/ { 
  if (ti != "") { data_error("missing </li> at end of entry"); }
  if (! match($0, /^ *<li> *<p> *<a *href="(.*)" *> *(.*)<[\/]a> *$/, fld)) { 
    data_error("bad <li> line format");
  }
  ur = cleanup_url(fld[1]);
  ti = cleanup_title(fld[2]);
  next;
}

/^ *<[\/]li>/ { 
  if (ti == "") { data_error("spurious </li> line"); }
  dump_entry();
  clear_entry();
  next;
}

(ti != "") {
  if (au == "") {
    # Assume it is the author line:
    if (match($0, /^ *<i> *(.*)[ ][-][ ] *(.*)[,]? *([12][0129][0-9][0-9]) *[ ][-][ ] *(.*) *<[\/]i> *$/, fld)) {
      # Normal Google Scholar hit:
      au = cleanup_authors(fld[1]);
      jn = cleanup_journal(fld[2]);
      yr = cleanup_year(fld[3]);
      sr = cleanup_source(fld[4]);
    } else if (match($0, /^ *<i> *(.*)[ ][-][ ] *(.*) *<[\/]i> *$/, fld)) {
      # Incomplete hit (no journal or year):
      au = cleanup_authors(fld[1]);
      jn = "";
      yr = "2022"; # Just to have a valid year.
      sr = cleanup_source(fld[2]);
    } else {
      data_error("bad author line format");
    }
  } else {
    # Assume quote line:
    qt = append_quote(qt, $0);
  }
  if (match($0, /<[\/]li>/)) { data_error("mark </li> not on separate line"); }
}

END {
  if (abort >= 0) { exit(abort); }
  if (ti != "") { data_error("missing </li> at end of entry"); }
  printf "%d keys were remapped\n", n_mapped > "/dev/stderr"
}

function clear_entry() {
  ti = ""; # Title.
  au = ""; # Author list with "and" as in Bibtex.
  jn = ""; # Journal name, book title, etc.
  yr = ""; # Year.
  sr = ""; # Source of Google Scholar data.
  ur = ""; # URL as per Google Scholar.
  qt = ""; # Quotes from paper.
}
         
function cleanup_url(x) {
  # gsub(/[?]casa_token=[^&]*[&]/, "?", x);
  # gsub(/[?]casa_token=[^&]*$/, "", x);
  gsub(/[&]amp;/, "&", x);
  return x;
} 

function cleanup_title(x) {
  gsub(/^[ ]+/, "", x);
  gsub(/[ ]+$/, "", x);
  
  # https://www.morganclaypool.com/doi/pdf/10.2200/S00983ED1V01Y202001MAS032
  gsub(/2[.]5.*INTERVAL RECIPROCAL/, "Affine Arithmetic Based Solution of Uncertain Static and Dynamic Problems", x);
  
  # http://repositorio.ufjf.br/jspui/bitstream/ufjf/2339/1/biancamariacostaaraujo.pdf
  gsub(/MESTRADO EM ENGENHARIA EL.*TRICA/, "Aritméticas Intervalares Aplicadas à Solução do Problema de Fluxo de Potência via Equações de Injeção de Corrente", x);
  
  return x;
} 

function cleanup_authors(x) {
  gsub(/^[ ]+/, "", x);
  gsub(/[ ]+$/, "", x);

  # https://www.morganclaypool.com/doi/pdf/10.2200/S00983ED1V01Y202001MAS032
  gsub(/^D.* .*x.* Affine.*$/, "S Chakraverty, S Rout", x);
  
  # http://repositorio.ufjf.br/jspui/bitstream/ufjf/2339/1/biancamariacostaaraujo.pdf
  gsub(/^AIA.*CORRENTE.*$/,  "Bianca Maria Costa Araújo", x);
  return x;
} 

function cleanup_journal(x) {
  gsub(/^[ ]+/, "", x);
  gsub(/[ ]+$/, "", x);
  return x;
} 

function cleanup_year(x) {
  if (x !~ /^(1[89][0-9][0-9]|20([01][0-9]|2[012]))$/) { data_error("invalid year \"" x "\""); }
  return x;
} 

function cleanup_source(x) {
  gsub(/^[ ]+/, "", x);
  gsub(/[ ]+$/, "", x);
  return x;
} 
    
function append_quote(x,y,  z) {    
  gsub(/^[ ]+/, "", y)
  gsub(/[ ]+$/, "", y);
  gsub(/[ ][ ]+/, " ", y);
  if (y == "") { return x; }
  if (x != "") { x = (x "  "); }
  z = (x y);
  return z;
}

function dump_entry(   ydir,fname,txau,key) {
  txau = texify_authors(au);
  key = make_key(txau, yr);
  ydir = (odir "/" yr);
  system(("mkdir -p " ydir));
  fname = (ydir "/" key ".bib");
  # printf "  %s\n", key > "/dev/stderr";
  printf "@entry{%s,\n", key > fname;
  printf "  author = {%s},\n", txau >> fname;
  printf "  title = {%s},\n", texify_text(ti) >> fname;
  printf "  journal = {%s},\n", texify_text(jn) >> fname;
  printf "  volume = {},\n" >> fname;
  printf "  number = {},\n" >> fname;
  printf "  pages = {},\n" >> fname;
  printf "  year = %s,\n", yr >> fname;
  printf "  month = ,\n" >> fname;
  printf "  doi = {},\n" >> fname;
  printf "  comment = {},\n" >> fname;
  printf "  abstract = {},\n" >> fname;
  printf "  url = {{\\url{%s}}},\n", ur >> fname;
  printf "  quotes = {%s}\n", texify_text(qt) >> fname;
  printf "}\n" >> fname;
  close(fname);
}

function texify_text(x) {
  
  x = texify_unicode(x);

  gsub(/à/, "{\\`a}", x);
  gsub(/á/, "{\\'a}", x);
  gsub(/ä/, "{\\\"a}", x);
  gsub(/ã/, "{\\~a}", x);
  gsub(/â/, "{\\^a}", x);

  gsub(/é/, "{\\'e}", x);
  gsub(/è/, "{\\`e}", x);
  gsub(/ë/, "{\\\"e}", x);
  gsub(/ê/, "{\\^e}", x);
 
  gsub(/í/, "{\\'\\i}", x);
  gsub(/ì/, "{\\`\\i}", x);
  gsub(/î/, "{\\^\\i}", x);
  gsub(/ï/, "{\\\"\\i}", x);

  gsub(/ó/, "{\\'o}", x);
  gsub(/ò/, "{\\`o}", x);
  gsub(/ö/, "{\\\"o}", x);
  gsub(/õ/, "{\\~o}", x);
  gsub(/ô/, "{\\^o}", x);
  gsub(/ø/, "{\\o}", x);

  gsub(/ú/, "{\\'u}", x);
  gsub(/ù/, "{\\`u}", x);
  gsub(/ü/, "{\\\"u}", x);
  
  gsub(/ý/, "{\\'y}", x);
  gsub(/ÿ/, "{\\\"y}", x);

  gsub(/ç/, "{\\c{c}}", x);
  gsub(/ñ/, "{\\~n}", x);
  gsub(/æ/, "{\\ae}", x);
  gsub(/ß/, "{\\ss}", x);
  
  gsub(/Á/, "{\\'A}", x);
  gsub(/À/, "{\\`A}", x);
  gsub(/Ä/, "{\\\"A}", x);
  gsub(/Ã/, "{\\~A}", x);
  gsub(/Â/, "{\\^A}", x);

  gsub(/É/, "{\\'E}", x);
  gsub(/È/, "{\\`E}", x);
  gsub(/Ë/, "{\\\"E}", x);
  gsub(/Ê/, "{\\^E}", x);
 
  gsub(/Í/, "{\\'\\I}", x);
  gsub(/Ì/, "{\\`\\I}", x);
  gsub(/Î/, "{\\^\\I}", x);
  gsub(/Ï/, "{\\\"\\I}", x);

  gsub(/Ó/, "{\\'O}", x);
  gsub(/Ò/, "{\\`O}", x);
  gsub(/Ö/, "{\\\"O}", x);
  gsub(/Õ/, "{\\~O}", x);
  gsub(/Ô/, "{\\^O}", x);
  gsub(/Ø/, "{\\O}", x);

  gsub(/Ú/, "{\\'U}", x);
  gsub(/Ù/, "{\\`U}", x);
  gsub(/Ü/, "{\\\"U}", x);
  gsub(/Û/, "{\\^U}", x);
  
  gsub(/Ý/, "{\\'Y}", x);

  gsub(/Ç/, "{\\C{C}}", x);
  gsub(/Ñ/, "{\\~N}", x);
  gsub(/Æ/, "{\\AE}", x);
  
  gsub(/·/, "{\\cdot}", x);
  gsub(/×/, "{\\times}", x);
  gsub(/±/, "{\\pm}", x);
  
  gsub(/[\\]uc\{0163\}/, "{\\k{t}}", x);
  gsub(/[\\]uc\{0160\}/, "{\\v{S}}", x);
 
  gsub(/[\240]/, " ", x);
  
  x = gensub(/\([^\\]\)[&]/, "\\1{\\\\&}", "g", x); # Assumes it is in text mode.
  gsub(/[%]/, "{\\%}", x);     # Protects "%" from being TeX comment.

  if (x !~ /^[\040-\176]*$/) { data_error("non-ascii character in TEX string \"" x "\""); }
  
  return x;
}

function texify_unicode(x,  n,k,i,ck,ci,m,h,u,y) {
  # Map CJK unicode chars to "\\cjk{HHHH}" where {HHHH}" is a 40digit hex codepoint.
  n = length(x); k = 1; y = ""
  while (k <= n) {
    ck = substr(x,k,1);
    # Determine the number of bytes in the next utf-8 encoded char:
    if ((ck >= "\300") && (ck <= "\337")) {
      m = 2;
    } else if ((ck >= "\340") && (ck <= "\357")) {
      m = 3;
    } else if ((ck >= "\360") && (ck <= "\367")) {
      m = 4;
    } else if ((ck >= "\370") && (ck <= "\373")) {
      m = 5;
    } else if ((ck >= "\374") && (ck <= "\375")) {
      m = 6;
    } else {
      m = 1;
    }
    if (length(x) < m) { data_error("truncated utf-8 character"); }
    if (m == 1) {
      # One-byte ascii character:
      y = (y ck);
    } else {
      # Next {m-1} bytes must start with bits "01" that is octal "\200" to "\277"
      for (i = 1; i < m; i++) {
        ci = substr(x,k+i,1);
        if ((ci < "\200") || (ci  > "\277")) { data_error("invalid utf-8 continuation char"); }
      }
      ci = substr(x,k+1,1);
      if ((m == 2) && (ck <= "\303")) {
        # If it is in the range +0080 to +00ff, assume latin-1 and just copy:
        y = (y ck ci);
      } else { 
        # Beyond unicode +00ff.
        # Convert the utf-8 encoding to hex unicode point:
        # printf "  x = [[%s]] k = %d x[k:m] = %s\n", x, k, substr(x,k,m) > "/dev/stderr"
        h = hexcode(substr(x,k,m));
        y = (y "\\uc{" h "}");
      }
    }
    # Skip the utf-8 encoded character:
    k = k + m;
  }
  return y;
}

function hexcode(c,  m,v,h,cmd) {
  # Returns the hex unicode point (4 or more dhex digits) of the multi-byte string {c}
  # that is supposed to be the unicode of a CJK character in utf-8 encoding.
  # This is a big crock.
  m = length(c)
  # printf "  c = %s\n", c > "/dev/stderr"
  # Write to disk:
  system("rm -f .tmpchar")
  printf "%s", c > ".tmpchar"
  close(".tmpchar")
  # getline v < ".tmpchar";  close(".tmpchar"); printf "  v = %s\n", v > "/dev/stderr"
  # Convert to 4-byte unicode chars and and then display the hexadecimal:
  cmd = "cat .tmpchar | recode u8..UCS-4 | od --endian=big -t x4 > .tmphex";
  # printf "  cmd = [[%s]]\n", cmd > "/dev/stderr";
  system(cmd);
  getline h < ".tmphex"
  close(".tmphex")
  # printf "  h = [[%s]]\n", h > "/dev/stderr"
  # Remove file position offset:
  gsub(/^ *[0-9][0-9][0-9][0-9]+ +/, "", h);
  # Remove encoding indicator, if any:
  gsub(/^fe ff +/, "", h);
  # Remove leading "00" or "0000" (but not "000000"):
  gsub(/^00/, "", h);
  gsub(/^00/, "", h);
  # Return the hex code:
  # printf "  c = %s\n", c > "/dev/stderr"
  # Write to disk:
  system("rm -f .tmpchar")
  printf "%s", c > ".tmpchar"
  close(".tmpchar")
  # getline v < ".tmpchar";  close(".tmpchar"); printf "  v = %s\n", v > "/dev/stderr"
  # Convert to 4-byte unicode chars and and then display the hexadecimal:
  cmd = "cat .tmpchar | recode u8..UCS-4 | od --endian=big -t x4 > .tmphex";
  # printf "  cmd = [[%s]]\n", cmd > "/dev/stderr";
  system(cmd);
  getline h < ".tmphex"
  close(".tmphex")
  # printf "  h = [[%s]]\n", h > "/dev/stderr"
  # Remove file position offset:
  gsub(/^ *[0-9][0-9][0-9][0-9]+ +/, "", h);
  # Remove encoding indicator, if any:
  gsub(/^fe ff +/, "", h);
  # Remove leading "00" or "0000" (but not "000000"):
  gsub(/^00/, "", h);
  gsub(/^00/, "", h);
  # Return the hex code:
  # printf "  h = [[%s]]\n", h > "/dev/stderr"
  return h;
}
 
function texify_authors(au) {
  #printf "in au = [[%s]]\n", au > "/dev/stderr"
  au = texify_text(au);
  # printf "tx au = [[%s]]\n", au > "/dev/stderr"
  # CJK comma:
  gsub(/[\\]uc\{ff0c\}/, ", ", au);
  # Special authors:
  # https://cir.nii.ac.jp/crid/1570009752446900992
  gsub(/[\\]uc\{5965\}[\\]uc\{7530\}[\\]uc\{5c1a\}[\\]uc\{4f38\}/, "H Okuda", au);
  gsub(/[\\]uc\{5b89\}[\\]uc\{7559\}[\\]uc\{8aa0\}[\\]uc\{543e\}/, "S Yasutome", au);
  gsub(/[\\]uc\{90fd\}[\\]uc\{5009\}[\\]uc\{4fe1\}[\\]uc\{6a39\}/, "N Tokura", au); 
  
  # https://cir.nii.ac.jp/crid/1571417127447012352 = https://cir.nii.ac.jp/crid/1571417127447012352
  # https://ci.nii.ac.jp/naid/110003292159/ = https://cir.nii.ac.jp/crid/1571698602293259008
  gsub(/[\\]uc\{5d0e\}[\\]uc\{5c71\}[\\]uc\{8cb4\}[\\]uc\{884c\}/, "T Sakiyama", au);
  gsub(/[\\]uc\{67cf\}[\\]uc\{6728\}[\\]uc\{96c5\}[\\]uc\{82f1\}/, "M Kashiwagi", au);
  
  # https://ci.nii.ac.jp/naid/110003292221/ = https://cir.nii.ac.jp/crid/1520009410272491776
  gsub(/[\\]uc\{795e\}[\\]uc\{6ca2\}[\\]uc\{96c4\}[\\]uc\{667a\}/, "Y Kanazawa", au);
  gsub(/[\\]uc\{5927\}[\\]uc\{77f3\}[\\]uc\{9032\}[\\]uc\{4e00\}/, "S Oishi", au);
  
  # https://www.ieice.org/ken/paper/20080131raBF/
  gsub(/[\\]uc\{5185\}[\\]uc\{6751\}[\\]uc\{5275\}/, "H Uchimura", au);
  gsub(/[\\]uc\{67cf\}[\\]uc\{6728\}[\\]uc\{5553\}[\\]uc\{4e00\}[\\]uc\{90ce\}/, "K Kashiwagi", au);
  
  # https://ipsj.ixsq.nii.ac.jp/ej/?item_id=152845
  gsub(/[\\]uc\{5bae\}[\\]uc\{5cf6\}[\\]uc\{4fe1\}[\\]uc\{4e5f\}/, "S Miyajima", au);
  gsub(/[\\]uc\{5bae\}[\\]uc\{7530\}[\\]uc\{5b5d\}[\\]uc\{5bcc\}/, "T Miyata", au);
  
  # https://ci.nii.ac.jp/naid/110003291360/ = https://cir.nii.ac.jp/crid/1520853832584743680

  # https://www.ieice.org/ken/paper/200909242aoU/
  gsub(/[\\]uc\{65b0\}[\\]uc\{5bae\}[\\]uc\{5f18\}[\\]uc\{654f\}/, "H Shingu", au);
  gsub(/[\\]uc\{795e\}[\\]uc\{6fa4\}[\\]uc\{96c4\}[\\]uc\{667a\}/, "Y Kanazawa", au);
  
  # https://ir.nctu.edu.tw/handle/11536/38189
  gsub(/[\\]uc\{9673\}[\\]uc\{5f65\}[\\]uc\{5b87\}/, "YY Chen", au);
  gsub(/[\\]uc\{5468\}[\\]uc\{666f\}[\\]uc\{63da\}/, "JY Jou", au);
  
  # https://chuo-u.repo.nii.ac.jp/?item_id=4004
  gsub(/[\\]uc\{4e09\}[\\]uc\{5cf6\}[\\]uc\{548c\}[\\]uc\{535a\}/, "K Mishima", au);
  
  # https://library.naist.jp/library/Gakunai/IEICE/general/2012/Settings/pdf/a_02_013.pdf
  gsub(/[\\]uc\{4e95\}[\\]uc\{539f\}[\\]uc\{6d69\}[\\]uc\{4ecb\}/, "K Ihara", au);

  # https://www.airitilibrary.com/Publication/alDetailedMesh?docid=U0017-2907201910202500
  gsub(/[\\]uc\{66fe\}[\\]uc\{9865\}/, "Z Hao", au);
  
  # http://jssst.or.jp/files/user/taikai/2016/GENERAL/general5-2.pdf
  gsub(/[\\]uc\{677e\}[\\]uc\{672c\}[\\]uc\{7fd4\}[\\]uc\{592a\}/, "S Matsumoto", au);
  gsub(/[\\]uc\{4e0a\}[\\]uc\{7530\}[\\]uc\{548c\}[\\]uc\{7d00\}/, "K Ueda", au);
  
  # https://oatd.org/oatd/record?record=handle%5C%3A2065%5C%2F637
  gsub(/[\\]uc\{6c34\}[\\]uc\{53e3\}[\\]uc\{7fa9\}[\\]uc\{96c4\}/, "Y Minakuchi", au);
  
  # https://www.ieice.org/ken/paper/20140306MBla/
  
  # https://journal.bupt.edu.cn/CN/Y2015/V38/I2/69
  gsub(/[\\]uc\{8c22\}[\\]uc\{6c38\}[\\]uc\{5f3a\}/, "YQ Xie", au);
  gsub(/[\\]uc\{9648\}[\\]uc\{5efa\}[\\]uc\{519b\}/, "JJ Chen", au);
  gsub(/[\\]uc\{66f9\}[\\]uc\{9e3f\}[\\]uc\{94a7\}/, "HJ Cao", au);
  
  # https://elib.spbstu.ru/dl/3/2020/vr/vr20-3728.pdf/info
  gsub(/[\\]uc\{0414\}[\\]uc\{0413\} [\\]uc\{0414\}[\\]uc\{0435\}[\\]uc\{043c\}[\\]uc\{0447\}[\\]uc\{0435\}[\\]uc\{043d\}[\\]uc\{043a\}[\\]uc\{043e\}/, "DD Georgievich", au);
  
  # http://xddl.ncepujournal.com/cn/article/doi/10.19725/j.cnki.1007-2322.2019.1072
  gsub(/[\\]uc\{5173\}[\\]uc\{4f73\}[\\]uc\{6b23\}/, "JX Guan", au);
  gsub(/[\\]uc\{8fb9\}[\\]uc\{7ade\}/, "J Bian", au);
  gsub(/[\\]uc\{674e\}[\\]uc\{56fd\}[\\]uc\{5e86\}/, "GQ Li", au);
  gsub(/[\\]uc\{738b\}[\\]uc\{9e64\}/, "H Wang", au);
  
  # http://dgjsxb.ces-transaction.com/CN/article/downloadArticleFile.do?attachType=PDF&id=6382
  gsub(/[\\]uc\{675c\}[\\]uc\{840d\}[\\]uc\{9759\}/, "PJ Du", au);
  gsub(/[\\]uc\{6768\}[\\]uc\{660e\}/, "M Yang", au);
  gsub(/[\\]uc\{66f9\}[\\]uc\{826f\}[\\]uc\{6676\}/, "LJ Cao", au);
  gsub(/[\\]uc\{7fdf\}[\\]uc\{9e64\}[\\]uc\{5cf0\}/, "HF Zhai", au);
  gsub(/[\\]uc\{6768\}[\\]uc\{4f73\}[\\]uc\{5cfb\}/, "JJ Yang", au);
  
  # http://verifiedby.me/kv/affine/affine.pdf
  
  # https://ieeexplore.ieee.org/abstract/document/6673287/
  gsub(/[\\]uc\{5bae\}[\\]uc\{524d\}[\\]uc\{77e5\}[\\]uc\{9686\}/, "T Miyamae", au);
  
  gsub(/, *$/, "", au);
  gsub(/, */, " and ", au);

  # printf "cn au = [[%s]]\n", au > "/dev/stderr"
  return au;
}

function make_key(txau,yr,  xa,na,af,i,ky,gn,sn,zi,nd,xd) {
  # Makes a Bibtex key from texified author names {txau} and year {yr}.
  # Appends seq tag "aa", "ab", "ac" ... for disamb.
  # Then maps the key through the global {key_map} table, if
  # it is defined there.
  
  # Also increments the global counter {n_mapped}.
  
  # Convert iso-latin-1 special chars to TeX:
  xa = txau;
  # printf "in xa = [[%s]]\n", xa > "/dev/stderr"
  # Replace special letter macros by ascii letters:
  gsub(/[\\]o\b/, "oe", xa);
  gsub(/[\\]ae\b/, "ae", xa);
  gsub(/[\\]i\b/, "i", xa);
  gsub(/[\\]["]u/, "ue", xa);
  gsub(/[\\]["]a/, "ae", xa);
  gsub(/[\\]["]o/, "oe", xa);

  gsub(/[\\]O\b/, "OE", xa);
  gsub(/[\\]AE\b/, "AE", xa);
  gsub(/[\\]I\b/, "I", xa);
  gsub(/[\\]["]U/, "UE", xa);
  gsub(/[\\]["]A/, "AE", xa);
  gsub(/[\\]["]O/, "OE", xa);
  # Remove all accent macros:
  gsub(/[\\][a-zA-Z]+/, "", xa);
  gsub(/[\\]./, "", xa);
  gsub(/[{}]/, "", xa);
  # printf "rm xa = [[%s]]\n", xa > "/dev/stderr"
  # Remove hyphens, periods, apostrophes:
  gsub(/[-.']/, "", xa);
  # Split authors:
  na = split(xa, af, / and /)
  # Assemble bare key {ky}:
  ky = ""
  for (i = 1; i <= na; i++) {
    gn = af[i]; gsub(/[ ].*$/, "", gn);     # Initials.
    sn = af[i]; gsub(/^[-A-Z]*[ ]/, "", sn); # Last name.
    gsub(/^ *[dD][ieaos]+ +/, "", sn); # Remove preposition from surname.
    gsub(/^ *[Vv][oa][n] +/, "", sn);  # Remove german/dutch preposition from surname.
    gsub(/^ *[Ll][ea]+ +/, "", sn);    # Remove detached article from surname.
    gsub(/[ ]/, "", sn); # Join multiword surnames.
    # Append initials to name and pad with "x" to ensure 3 letters:
    zi = tolower(sn gn "xxx");
    # Take first 3 letters:
    zi = substr(zi, 1, 3);
    if (length(zi) != 3) { prog_error("could not get 3 letters \"" af[i] "\" --> \"" zi "\""); }
    ky = (ky zi "-")
  }
  ky = (ky substr(yr, 3,2)); # Append year 00-99.
  # Check for duplicates and add disamb suffix "aa", "ab", etc:
  if (ky in kc) { nd = kc[ky]; } else { nd = 0; }
  kc[ky] = nd+1;
  xd = (substr(alf, 1+int(nd/26), 1) substr(alf, 1+(nd%26), 1))
  ky = (ky "-" xd);
  printf "  %s", ky > "/dev/stderr";
  if (ky in key_map) {
    # Map through the key table:
    ky = key_map[ky]
    printf " --> %s\n", ky > "/dev/stderr";
    n_mapped++;
  } else {
    printf " seems new\n" > "/dev/stderr";
  }
  return ky
}
 
function data_error(msg) { 
  printf "%s:%s: ** %s\n", FILENAME, FNR, msg > "/dev/stderr"; 
  printf "  «%s»\n", $0 > "/dev/stderr"; 
  abort = 1;
  exit(abort);
} 
          
function arg_error(msg) { 
  printf "** %s\n", msg > "/dev/stderr"; 
  abort = 1;
  exit(abort);
} 
  
          
function prog_error(msg) { 
  printf "** PROG ERROR: %s\n", msg > "/dev/stderr"; 
  abort = 1;
  exit(abort);
}