Expand the glyph name list and remove duplicates

Fixes #88

src/glyphlist-export.py

@@ -1,18 +1,41 @@
-f = open("glyphlist-extended.txt")
-lines = f.readlines()
-import re
 glyphlist = []
-for l in lines:
-    if l[0] == '#' or l[0] == '\n':
-        continue
-    name, code = re.split('[; ]+', l)[0:2]
-    glyphlist.append((name,int(code,16)))
+glyphs_seen = {}
+def read_glyphs(name):
+    f = open(name)
+    lines = f.readlines()
+    import re
+    for l in lines:
+        if l[0] == '#' or l[0] == '\n':
+            continue
+        split = re.split('[; ,]+', l)
+        name = split[0]
+        val = int(split[1], 16)
+        if val > 0xffff:
+            val = int(split[-1], 16)
+        if val == 0xf766 and name != "Fsmall":
+            continue
+        if name in glyphs_seen:
+            continue
+        glyphs_seen[name] = True
+        glyphlist.append((name,val))
+read_glyphs("glyphlist-extended.txt")
+read_glyphs("texglyphlist.txt")
+read_glyphs("additional.txt")
+# there are some conflicts between these files
+# e.g. tildewide=0x02dc, vs tildewide=0x0303
+# for now we just ignore the subsequent ones
+glyphlist.append(('mapsto', 0x21A6))
+glyphlist = list(set(glyphlist))
 glyphlist.sort()
-print "/* Autogenerated from https://github.com/michal-h21/htfgen/commits/master/glyphlist-extended.txt */"
+print "/* Autogenerated from:"
+print "     https://github.com/michal-h21/htfgen/commits/master/glyphlist-extended.txt"
+print "     https://github.com/2ion/lcdf-typetools/blob/master/texglyphlist.txt"
+print "     https://github.com/apache/pdfbox/blob/trunk/pdfbox/src/main/resources/org/apache/pdfbox/resources/glyphlist/additional.txt"
+print " */"
 print "pub fn name_to_unicode(name: &str) -> Option<u16> {"
 print "    let names = ["
 print ",\n".join('(\"%s\", 0x%04x)' % (g[0], g[1]) for g in glyphlist)
 print "    ];"
-print "    let result = names.binary_search_by_key(&name, |&(name,code)| &name);"
+print "    let result = names.binary_search_by_key(&name, |&(name,_code)| &name);"
 print "    result.ok().map(|indx| names[indx].1)"
 print "}"

src/glyphnames.rs

@@ -1,5 +1,8 @@
-/* Autogenerated from https://github.com/michal-h21/htfgen/commits/master/glyphlist-extended.txt
-   and https://github.com/2ion/lcdf-typetools/blob/master/texglyphlist.txt */
+/* Autogenerated from:
+     https://github.com/michal-h21/htfgen/commits/master/glyphlist-extended.txt
+     https://github.com/2ion/lcdf-typetools/blob/master/texglyphlist.txt
+     https://github.com/apache/pdfbox/blob/trunk/pdfbox/src/main/resources/org/apache/pdfbox/resources/glyphlist/additional.txt
+ */
 pub fn name_to_unicode(name: &str) -> Option<u16> {
     let names = [
 ("A", 0x0041),
@@ -228,10 +231,8 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("Gdot", 0x0120),
 ("Gdotaccent", 0x0120),
 ("Gecyrillic", 0x0413),
-("Germandbls", 0x0053),
 ("Germandbls", 0x1e9e),
 ("Germandblssmall", 0xd803),
-("Germandblssmall", 0xf773),
 ("Ghadarmenian", 0x0542),
 ("Ghemiddlehookcyrillic", 0x0494),
 ("Ghestrokecyrillic", 0x0492),
@@ -561,7 +562,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("SF540000", 0x256a),
 ("SS", 0x0053),
 ("SSsmall", 0xd803),
-("SSsmall", 0xf773),
 ("Sacute", 0x015a),
 ("Sacutedotaccent", 0x1e64),
 ("Sampigreek", 0x03e0),
@@ -1108,7 +1108,15 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("anbopomofo", 0x3122),
 ("angbopomofo", 0x3124),
 ("angbracketleft", 0x27e8),
+("angbracketleftBig", 0x2329),
+("angbracketleftBigg", 0x2329),
+("angbracketleftbig", 0x2329),
+("angbracketleftbigg", 0x2329),
 ("angbracketright", 0x27e9),
+("angbracketrightBig", 0x232a),
+("angbracketrightBigg", 0x232a),
+("angbracketrightbig", 0x232a),
+("angbracketrightbigg", 0x232a),
 ("angkhankhuthai", 0x0e5a),
 ("angle", 0x2220),
 ("anglebracketleft", 0x3008),
@@ -1175,6 +1183,8 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("arrowheadleftmod", 0x02c2),
 ("arrowheadrightmod", 0x02c3),
 ("arrowheadupmod", 0x02c4),
+("arrowhookleft", 0x21aa),
+("arrowhookright", 0x21a9),
 ("arrowhorizex", 0xf8e7),
 ("arrowleft", 0x2190),
 ("arrowleftbothalf", 0x21bd),
@@ -1211,7 +1221,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("arrowupleftofdown", 0x21c5),
 ("arrowupright", 0x2197),
 ("arrowupwhite", 0x21e7),
-("arrowvertex", 0x2195),
 ("arrowvertex", 0xf8e6),
 ("ascendercompwordmark", 0xd80a),
 ("asciicircum", 0x005e),
@@ -1339,15 +1348,17 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("bokatakana", 0x30dc),
 ("bparen", 0x249d),
 ("bqsquare", 0x33c3),
-("braceex", 0x007c),
 ("braceex", 0xf8f4),
+("bracehtipdownleft", 0xfe37),
+("bracehtipdownright", 0xfe37),
+("bracehtipupleft", 0xfe38),
+("bracehtipupright", 0xfe38),
 ("braceleft", 0x007b),
 ("braceleftBig", 0x007b),
 ("braceleftBigg", 0x007b),
 ("braceleftbig", 0x007b),
 ("braceleftbigg", 0x007b),
 ("braceleftbt", 0xf8f3),
-("braceleftmid", 0x007c),
 ("braceleftmid", 0xf8f2),
 ("braceleftmonospace", 0xff5b),
 ("braceleftsmall", 0xfe5b),
@@ -1359,7 +1370,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("bracerightbig", 0x007d),
 ("bracerightbigg", 0x007d),
 ("bracerightbt", 0xf8fe),
-("bracerightmid", 0x2016),
 ("bracerightmid", 0xf8fd),
 ("bracerightmonospace", 0xff5d),
 ("bracerightsmall", 0xfe5c),
@@ -1372,6 +1382,7 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("bracketleftbigg", 0x005b),
 ("bracketleftbt", 0xf8f0),
 ("bracketleftex", 0xf8ef),
+("bracketleftmath", 0x005b),
 ("bracketleftmonospace", 0xff3b),
 ("bracketlefttp", 0xf8ee),
 ("bracketright", 0x005d),
@@ -1381,6 +1392,7 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("bracketrightbigg", 0x005d),
 ("bracketrightbt", 0xf8fb),
 ("bracketrightex", 0xf8fa),
+("bracketrightmath", 0x005d),
 ("bracketrightmonospace", 0xff3d),
 ("bracketrighttp", 0xf8f9),
 ("breve", 0x02d8),
@@ -1571,6 +1583,7 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("controlHT", 0x0009),
 ("controlLF", 0x000a),
 ("controlNAK", 0x0015),
+("controlNULL", 0x0000),
 ("controlRS", 0x001e),
 ("controlSI", 0x000f),
 ("controlSO", 0x000e),
@@ -1727,7 +1740,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("dhook", 0x0257),
 ("dialytikatonos", 0x0385),
 ("dialytikatonoscmb", 0x0344),
-("diamond", 0x2662),
 ("diamond", 0x2666),
 ("diamondmath", 0x22c4),
 ("diamondsolid", 0x2666),
@@ -1771,7 +1783,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("dotbelowcomb", 0x0323),
 ("dotkatakana", 0x30fb),
 ("dotlessi", 0x0131),
-("dotlessj", 0x0237),
 ("dotlessj", 0xf6be),
 ("dotlessjstrokehook", 0x0284),
 ("dotmath", 0x22c5),
@@ -1908,6 +1919,7 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("equal", 0x003d),
 ("equaldotleftright", 0x2252),
 ("equaldotrightleft", 0x2253),
+("equalmath", 0x003d),
 ("equalmonospace", 0xff1d),
 ("equalorfollows", 0x22df),
 ("equalorgreater", 0x2a96),
@@ -2059,7 +2071,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("followornoteqvlnt", 0x22e9),
 ("follows", 0x227b),
 ("followsequal", 0x227d),
-("followsequal", 0x2ab0),
 ("followsorcurly", 0x227d),
 ("followsorequal", 0x227f),
 ("fongmanthai", 0x0e4f),
@@ -2279,6 +2290,7 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("hatwide", 0x0302),
 ("hatwider", 0x0302),
 ("hatwiderr", 0x0302),
+("hatwidest", 0x0302),
 ("hbar", 0x0127),
 ("hbopomofo", 0x310f),
 ("hbrevebelow", 0x1e2b),
@@ -2289,7 +2301,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("hdotaccent", 0x1e23),
 ("hdotbelow", 0x1e25),
 ("he", 0x05d4),
-("heart", 0x2661),
 ("heart", 0x2665),
 ("heartsuitblack", 0x2665),
 ("heartsuitwhite", 0x2661),
@@ -2496,7 +2507,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("integraltp", 0x2320),
 ("intercal", 0x22ba),
 ("interrobang", 0x203d),
-("interrobangdown", 0x2e18),
 ("interrobangdown", 0xd80b),
 ("intersection", 0x2229),
 ("intersectiondbl", 0x22d2),
@@ -2839,6 +2849,7 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("male", 0x2642),
 ("maltesecross", 0x2720),
 ("mansyonsquare", 0x3347),
+("mapsto", 0x21a6),
 ("maqafhebrew", 0x05be),
 ("mars", 0x2642),
 ("masoracirclehebrew", 0x05af),
@@ -3281,9 +3292,9 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("parenleftbig", 0x0028),
 ("parenleftbigg", 0x0028),
 ("parenleftbt", 0xf8ed),
-("parenleftex", 0x007c),
 ("parenleftex", 0xf8ec),
 ("parenleftinferior", 0x208d),
+("parenleftmath", 0x0028),
 ("parenleftmonospace", 0xff08),
 ("parenleftsmall", 0xfe59),
 ("parenleftsuperior", 0x207d),
@@ -3296,9 +3307,9 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("parenrightbig", 0x0029),
 ("parenrightbigg", 0x0029),
 ("parenrightbt", 0xf8f8),
-("parenrightex", 0x007c),
 ("parenrightex", 0xf8f7),
 ("parenrightinferior", 0x208e),
+("parenrightmath", 0x0029),
 ("parenrightmonospace", 0xff09),
 ("parenrightsmall", 0xfe5a),
 ("parenrightsuperior", 0x207e),
@@ -3360,8 +3371,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("phagujarati", 0x0aab),
 ("phagurmukhi", 0x0a2b),
 ("phi", 0x03c6),
-("phi", 0x03d5),
-("phi1", 0x03c6),
 ("phi1", 0x03d5),
 ("phieuphacirclekorean", 0x327a),
 ("phieuphaparenkorean", 0x321a),
@@ -3398,6 +3407,7 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("plus", 0x002b),
 ("plusbelowcmb", 0x031f),
 ("pluscircle", 0x2295),
+("plusmath", 0x002b),
 ("plusminus", 0x00b1),
 ("plusmod", 0x02d6),
 ("plusmonospace", 0xff0b),
@@ -3420,7 +3430,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("precedeornoteqvlnt", 0x22e8),
 ("precedes", 0x227a),
 ("precedesequal", 0x227c),
-("precedesequal", 0x2aaf),
 ("precedesorcurly", 0x227c),
 ("precedesorequal", 0x227e),
 ("prescription", 0x211e),
@@ -3544,6 +3553,8 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("radicalbigg", 0x221a),
 ("radicalbt", 0x221a),
 ("radicalex", 0xf8e5),
+("radicaltp", 0x221a),
+("radicalvertex", 0x221a),
 ("radoverssquare", 0x33ae),
 ("radoverssquaredsquare", 0x33af),
 ("radsquare", 0x33ad),
@@ -4170,6 +4181,7 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("tildewide", 0x0303),
 ("tildewider", 0x0303),
 ("tildewiderr", 0x0303),
+("tildewidest", 0x02dc),
 ("timescircle", 0x2297),
 ("tipehahebrew", 0x0596),
 ("tipehalefthebrew", 0x0596),
@@ -4214,10 +4226,8 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("triangledownsld", 0x25bc),
 ("triangleinv", 0x25bd),
 ("triangleleft", 0x25b9),
-("triangleleft", 0x25c1),
 ("triangleleftequal", 0x22b4),
 ("triangleleftsld", 0x25c0),
-("triangleright", 0x25b7),
 ("triangleright", 0x25c3),
 ("trianglerightequal", 0x22b5),
 ("trianglerightsld", 0x25b6),
@@ -4263,7 +4273,6 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("twelveparen", 0x247f),
 ("twelveperiod", 0x2493),
 ("twelveroman", 0x217b),
-("twelveudash", 0xd80c),
 ("twelveudash", 0xf6de),
 ("twentycircle", 0x2473),
 ("twentyhangzhou", 0x5344),
@@ -4422,6 +4431,8 @@ pub fn name_to_unicode(name: &str) -> Option<u16> {
 ("verticallinelowmod", 0x02cc),
 ("verticallinemod", 0x02c8),
 ("vewarmenian", 0x057e),
+("vextenddouble", 0x2225),
+("vextendsingle", 0x2223),
 ("vhook", 0x028b),
 ("vikatakana", 0x30f8),
 ("viramabengali", 0x09cd),